Advances in Food Research Vol 5 Compress

ADVANCES IN FOOD RESEARCH
VOLUME V
This Page Intentionally Left Blank
ADVANCES IN
FOOD RESEARCH
VOLUME V
Edited by
E. M. MRAK G. F. STEWART
University of California University of California
Davis, California Davis, California
Editorial Board
E. C. BATE-SMITH S. LEPKOVSHY
W. H. COOK B. E. PROCTOR
W. F. GEDDES EDWARD
SELTZER
M. A. JOSLYN P. F. SHARP
A. J. KLUWER W. M. URBAIN
0. 13. WILLIAMS
1954
ACADEMIC PRESS INC., PUBLISHERS
NEW YORK, N. Y.
Copyright 1854, by
ACADEMIC PRESS INC.
125 EAST 2 3 STREET
~ ~
N E W YORK 10, N. Y.
ALL RIQHTS RESERVED

NO PART OF THIS BOOK MAY B E REPRODUCED I N ANY FORM,
BY PHOTOBTAT, MICROFILM, OR ANY OTHER MEANS,
WITHOUT WRITTEN PERMISSION FROM T H E PUBLISHERS.
Library of Congress Catalog Card Number: 48-7808
PRINTED I N T H E UNITED STATES OF AMERICA

CONTRIBUTORS
TO VOLUME
V
MAYNARD A. AMERINE,Department of Viticulture and Enology, College
of Agriculture, University of California.
E. C. BATE-SMITH, Low Temperature Research Station, University of
Cambridge, and Deparlment of ScientiJic and Industrial Research,
Cambridge, England.
J. B. S . BRAVERMAN, Department of Food Technology, University of
California, Berkeley, California.
C. 0. CHICHESTER,
Department of Food Technology, University of Cali-
fornia, Berkeley, California.
HARRYGEHMAN, Research Department, Chemical Division, Corn Products
Refining Company, Argo, Illinois.
M. A. JOSLYN,Department of Food Technology, University of California,
Berkeley, California.
G. MACKINNEY,Department of Food Technology, University of Cali-
fornia, Berkeley, California.
ELIZABETH M. OSMAN, Department of Foods and Nutrition, Michigan
State College, East Lansing, Michigan.
B. OSTLE, Montana State College, Bozeman, Montana.
ROBERTG. TISCHER,
Iowa State College, Ames, Iowa.
BETTY M. WATTS,Department of Food and Nutrition, Florida State
University, Tallahassee, Florida.
FOREWORD
Publication of Volume V provides a convenient and timely oppor-

tunity for the editors to pause and review our efforts to make Advances
in Food Research a significant factor in promoting scientific progress in
foods and nutrition. First, let us take a look at our original objective-to
assist in ‘(thecoordination and integration of food research t o promote an
orderly and systematic development of scientific knowledge in this impor-
tant field.” This need certainly remains; to us, it seems more urgent than
ever and we feel that a number of the articles published in the first five
volumes have been helpful in promoting a more orderly and systematic
development in a number of fields. The frequency with which some of
these have been quoted in scientific literature attests to that feeling.
We find that we have adhered rather closely to our original ideas about
subject matter coverage. In fact, one or more articles have appeared in all
categories: Food Acceptance, Nutrition, Agriculture, Microbiology and
Public Health, Biochemistry and Histology, Food Technology and Engi-
neering, Entomology and Zoology and commodities. Reflecting on our
original choices brings no better group to mind. Our main concern is to
bring together articles in this wide range of subject matter that are inter-
esting and challenging to the reader audience. Volume V represents our
latest effort to achieve this goal.
Volume V again presents a wide variety of subject matter which we
hope will be both stimulating and interesting to our readers. Functional
subjects range from the biochemistry of the flavonoids to statistical
methods for food research. Commodity coverage ranges from meat to
wine. It is the editors’ opinion that all are worthy reviews and should
make interesting and informative reading and reference sources for all
interested in the advancing frontiers of food research.
The review by Watts on in situ oxidation of animal fats is most timely
and important. The author brings t o bear her own research experiences
along with those of researchers from many countries to clarify and
integrate knowledge of the oxidative changes in meat. Both the theoreti-
cal and practical aspects of the problem are discussed. Meat technologists
should be especially appreciative for the way in which this author has
critically reviewed and summarized the many conflicting statements in
the literature about oxidation problems.
vii
...
Vlll FOREWORD
The wide use of sulfites in the food industry and the lack of critical
information about their effects on foods prompted the editors to seek
contributions in this field. Gehman and Osman and Joslyn and Braver-
man have come forward with two excellent articles relating to both the
fundamental and practical aspects of using sulfites in foods. The former
authors have confined themselves to the all-important subject of sugar-
sulfite reactions in foods while the latter discuss the many uses for, and
effects of sulfites.
The need for better statistical techniques for designing and carrying
out experimental work on foods is now becoming well-recognized. The
editors feel fortunate to be able to present the contribution by Ostle and
Tischer on this subject. These workers are not only conversant with the
literature in the field but have been active in it themselves. Excellent
examples of statistical problems encountered in food research have been
extracted from the literature. Critical examinations of these have been
made and suggestions for improved designs and analyses outlined.
In presenting Bate-Smith’s article on the flavonoids the editors are
pursuing a plan to have reviewed the literature on all of the important
natural food pigments. The present article is a well written account of
the distribution and properties of the flavonoids.
Equally important to understanding more about the chemistry of
pigments is to be able to understand their effects on the appearance of
foods. MacKinney and Chichester have tackled the problem of reviewing
work on color measurement in a masterly way. One is struck by the
enormous difficulty of defining color rigorously as seen by the human eye,
and the great efforts that have been made to do so for certain commodi-
ties. This is an extremely interesting and important field, the results of
which are just beginning to have their impact on food research in general.
Amerine has painstakingly reviewed the enormous literature on the
organic chemical constituents of wine. One is struck by the tremendous
literature in this field. Workers in other commodity fields can but be
envious of the numerous studies that have been devoted to the analyses
of wine.
September, 1954 G. F. STEWART

E. M. MRAK
CONTENTS
Contributors to Volume V . . . . . . . . . . . . . . . . . . . . . v
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Oxidative Rancidity and Discoloration in Meat
BY BETTYM . WATTS.Department of Food and Nutrition.
Florida State University. Tallahassee. Florida
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
I1. Oxidative Rancidity in Meat . . . . . . . . . . . . . . . . . . . . 3
I11. Oxidative Discolorations . . . . . . . . . . . . . . . . . . . . . . 12
IV . The Coupled Oxidation of Hemoglobin and Unsaturated Fats . . . . . . 22
V. Antioxidants . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
VI . Effect of Various Meat Constituents and Additives on Rancidity and
Discoloration . . . . . . . . . . . . . . . . . . . . . . . . . . 30
VII . Physical Factors Affecting Oxidative Changes . . . . . . . . . . . . . 36
VIII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
The Chemistry of the Sugar-Sulfite Reaction and Its Relationship to
Food Problems
BY HARRYGEHMAN AND ELIZABETH .
M OSMAN,Research Department,
Chemical Division, Corn Products Rejining Company, Argo, Illinoi8, and
Department of Foods and Nutrition, Michigan State College, East Lansing, Michigan
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
I1. Nature of the Sugar-Bisulfite Addition Compounds . . . . . . . . . . 55
I11. Analytical Procedures . . . . . . . . . . . . . . . . . . . . . . . 59
IV . Reaction Equilibrium Constant . . . . . . . . . . . . . . . . . . . 63
V. Reaction Velocity Constants . . . . . . . . . . . . . . . . . . . . 71
VI . Application of the Sugar-Sulfite Reaction to Food Problems . . . . . . . 75
VII . Needed Research . . . . . . . . . . . . . . . . . . . . . . . . . 91
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
The Chemistry and Technology of the Pretreatment and Preservation of Fruit and
Vegetable Products with Sulfur Dioxide and Sulfites
BY M . A . JOSLYN .
AND J B . S. RRAVERMAN, Department of Food Technology.
University of California, Berkeley, California
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
I1. Chemistry of Sulfur Dioxide, Sulfites and Their Organic Compounds. . . 99
I11. Determination of Sulfur Dioxide in Fruit and Vegetable Products . . . . 113
IV. Sulfur Dioxide as Preservative and Sanitizing Agent . . . . . . . . . . 123
V . Sulfur Dioxide as a n Inhibitor of Enzymic and Nonenaymic Browning . . 128
.
VI Source and Application of Sulfur Dioxide . . . . . . . . . . . . . . . 130
VII . Sulfur Dioxide in Fruit Juices, Syrups. Concentrates, and Purees . . . . 134
VIII . Sulfur Dioxides in Wine and Vinegar Making . . . . . . . . . . . . . 136
.
I X Sulfur Dioxide in Dehydrated and Dried Fruit and Vegetable Products . 141
.
X Sulfur Dioxide in “Brining” of Cherries and “Barrelling” of Fruit . . . 143
XI . Sulfur Dioxide in Transportation and Storage of Grapes and in Other
Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
ix
X CONTENTS
Statistical Methods in Food Research

B Y B . OSTLEAND ROBERTG . TISCHER. Montana State College.
Bozeman. Montana and Iowa State College. Ames. Iowa
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
I1. Statistics and Research . . . . . . . . . . . . . . . . . . . . . . . 162
I11. Basic Statistical Concepts . . . . . . . . . . . . . . . . . . . . . 164
IV. Analysis of Variance . . . . . . . . . . . . . . . . . . . . . . . . 173
V. Regression Techniques . . . . . . . . . . . . . . . . . . . . . . . 210
V I . Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
VII . Analysis of Covariance . . . . . . . . . . . . . . . . . . . . . . . 246
VIII . Other Statistical Techniques . . . . . . . . . . . . . . . . . . . . 249
I X . Needs for Future Research . . . . . . . . . . . . . . . . . . . . . 253
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Flavonoid Compounds in Foods
BY E. C. BATE-SMITH. Low Temperature Research Station. University of Cambridge. and
Department of Scientijk and Industrial Research. Cambridge. England
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
I1. Properties of Flavonoid Compounds Significant in Foods . . . . . . . . 268
111. The Genetic Situation . . . . . . . . . . . . . . . . . . . . . . . 278
I V . Systematic Distribution . . . . . . . . . . . . . . . . . . . . . . 283
V. Some Special Cases . . . . . . . . . . . . . . . . . . . . . . . . 289
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
The Color Problem in Foods
BY G . MACKINNEY AND C . 0. CHICHESTER. Department of Food Technology.
University of California. Berkeley
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
I1. Color in Selected Commodities . . . . . . . . . . . . . . . . . . . 311
I11. Some Theoretical Considerations . . . . . . . . . . . . . . . . . . 331
IV . General Considerations. . . . . . . . . . . . . . . . . . . . . . . 337
.
V Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . 338
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
..
Composition of Wines I Organic Constituents
BY MAYNARD A. AMERINE.Department of Viticulture and Enology.
College of Agriculture. University of California
I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
I1. General Methods of Analysis . . . . . . . . . . . . . . . . . . . . 356
I11. Alcohols and Related Compounds . . . . . . . . . . . . . . . . . . 358
IV. Aldehydes and Related Compounds . . . . . . . . . . . . . . . . . 382
V. Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
VI . Carbohydrates . . . . . . . . . . . . . . . . . . . . . . . . . . 418
VII . Esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
VIII . Polyhydroxyphenols . . . . . . . . . . . . . . . . . . . . . . . . 435
I X. Nitrogenous Compounds. . . . . . . . . . . . . . . . . . . . . . 448
X . Enzymes. Vitamins. and Aromatic Constituents . . . . . . . . . . . . 454
XI . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . 466
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Oxidative Rancidity and Discoloration in Meat
BY BETTY M. WATTS
Department of Food and Nutrition, Florida State University
Tallahassee, Floridu
CONTENTS
Page
I . Introduction.. ................... ........ 1
11. Oxidative Rancidity in Meat.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1. The Nature of the Oxidative Process .......... 4
2. Species Variations in the Susceptibili o Oxidation 4
3. Influence of Rations on Fatty Acid Content of Animal Body Fats. . 5
4. Deposition of Antioxidants in Animal Tissues.. . . . . . . . . . . . . . . . . . . . 7
5. Distribution of Fat in Meat as a Factor in Rancidification.. . . . . . . . 8
6. Methods for Evaluating Oxidative Changes in Meat Fats. . . . . . . . . . 9
111. Oxidative Discolorations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1. Normal Pigments of Fresh Meat.. . . . . . . . . . . . . . . . . . . . . . .
2. Oxidation Products of Heme Pigments.. . . . . . . . . . . . . . . . . .
3. The Pigments of Cured Meats; Their Oxidation.. . . . . . . . . .
4. Methods for Investigation of Color Changes in Meat. . . . . .
IV. The Coupled Oxidation of Hemoglobin and Unsaturated Fats.. . . . . . . . . . 22
V. Antioxidants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1. Classification and Mode of Action of F a t Antioxidants.. . . . . . . . . . . . 24
2. Application to Meats.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3. Use of Antioxidants for Color Protection
VI. Effect of Various Meat Constituents and Addi
coloration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1. Changesin p H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2. Salts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3. Metals.. . . . . . . . . . . .
4. gmoke . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Oxygen Tension. ...

1. ........ 36
Light . . . . . . . . . . . . . .
2.
Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. 37
Packaging Problems.. . . . . . . . . . . . . . . . . .
4. 38
VIII. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
I. INTRODUCTION
This review is concerned with two types of oxidative changes which
occur in meat, namely, oxidation of the fat, resulting in rancidity, and
1
2 BETTY M. WATTS
oxidation of the heme pigments, resulting in discoloration. The two are

closely related. I n fact, each can accelerate the other, but each can also
proceed independently of the other. Together they account for a large
part of the economic loss due to substandard or spoiled meat.
A relatively large part of the space has been devoted to the more
fundamental work on these reactions and their interrelationships in meat
since it is believed that control can best be achieved when the causes of
deterioration are thoroughly understood. I n general, the oxidative
changes discussed are those of chemical rather than microbiological
origin, although it is recognized that the same changes are often brought
about by bacteria, and it is not always possible to determine whether or
riot bacteria are implicated.
When the author began work on the oxidative deterioration of meat
some years ago i t was with the implicit acceptance of two ideas th a t have
since proved to be highly fallacious. These misconceptions seemingly have
colored the thinking of other workers in the field also. It is, therefore,
worth-while t o hold them up for inspection before plunging into the
literature on the subject.
One such error was the idea that the fresh meat of any species is a
more or less uniform starting point for experimentation. According to this
idea, pork is pork; it should be possible after suitable experimentation to
state that pork treated in a specified manner and held under a given set
of conditions will keep for such and such a time before showing signs of
rancidity or other oxidative change.
A number of painful experiences have served t o impress upon the
author the fallaciousness of this assumption. Time schedules for the
analysis of experimental samples have had to be revised repeatedly when
controls which were expected to be rancid in two months still showed no
oxidative deterioration after as many years. Optimistic claims made for
the stability of meat treated in a specified way have had to be abandoned
when another lot of meat treated in the same way failed t o measure up.
It was not unusual to find that inherent variations in the meat itself were
of far greater importance in determining the keeping time than any of the
experimental treatments under investigation.
Such “biological variations” are not to be classed with acts of provi-
dence-outside the realm of useful experimentation. On the contrary,
investigation into their causes promises t o be a fruitful line of attack.
As the reasons for such variations are established, selection of meat of
known history for specific purposes should become a reality.
A second important misconception was the idea that the various
oxidations which occur in meat, resulting in off-odors, off-flavors, and dis-
colorations, could be lumped together for purposes of control. Although
OXIDATIVE RANCIDITY A N D DISCOLORATION I N MEAT 3
the manifestations of the various oxidative changes might be different
and the time of their appearance and rate of their progress might also
differ, i t was thought, nevertheless, th at there should be a set of optimum
conditions which would retard oxidative changes in general. There should
be a group of “pro-oxidants” which accelerate such changes and so were
to be avoided, and another group of (‘antioxidants,’’presumably belong-
ing t o the class of reducing agents, which would retard oxidative changes
in general.
On almost every page of this review are facts which contradict any
such simple assumption. A lowering of the pH within the range of normal
meat may discolor fresh meat, improve the color of cured meat, accelerate
the oxidation of fat present in raw meat, but have no effect on the fa t in
cooked meat. Copper and iron, traditional “ pro-oxidants” of fat oxidation
can also accelerate reduction of methemoglobin. Ascorbic acid, one of the
strongest of biological reducing agents can accelerate oxidation of both
fat and hemoglobin, although, to be sure, the same compound, under a
slightly different set of conditions is a good antioxidant. Even the partial
removal of oxygen itself accelerates oxidation of hemoglobin.
These apparent contradictions are not raised to create the impression
that the whole subject is hopelessly muddled. Such is not the case. Many
broad generalizations have been drawn and have resulted in useful
applications. However, it seems unlikely that there will be found any
antioxidant or group of antioxidants suitable for all conditions. It is
essential to examine carefully the various oxidative reactions which take
place before i t will be possible to select the appropriate remedy for each
situation. Meats need to be tailored for the type of product for which
they are intended, whether i t is to be sold over the counter as fresh meat,
cured, frozen, dried or canned by conventional procedures, or made into
new products not now on the market. The tailoring will require a n inti-
mate knowledge of the complex reactions involved.
11. OXIDATIVERANCIDITY
IN MEAT
I n view of the fact that Lea (1939) has published a comprehensive and
critical monograph on the broad topic of rancidity in edible fats, no
attempt will be made to review this earlier work in detail except as i t may
be necessary for purposes of orientation. Furthermore, while great strides
have been made in clarifying the chemistry of the oxidation of un-
saturated fatty acids in the past decade, this work as i t applies t o the
extracted fats from animal tissues has been ably reviewed by Bailey
(1951). Therefore only the salient facts without extensive documentation
will be presented here in order to leave space for more comprehensive
coverage of variations in the composition of the fat of meat animals which
4 BETTY M. WATTS
might be expected to affect rancidity, and special problems in the control

of rancidity in meat arising from the fact that the fat is part of a hetero-
geneous system containing accelerators and inhibitors of rancidity.
1. The Nature of the Oxidative Process

Rancidity, a t least as the term is used in meat products, results from
the oxidative decomposition of unsaturated fats. The first step in this
decomposition is the addition of oxygen at a carbon atom adjacent to an
unsaturated carbon to form a hydroperoxide:
-CH*.CH=CH.CHZ- + 0 2 --t -CH(OOH)CH=CH.CH*-
Although this reaction can occur in fatty acids having a single double
bond, such as oleic, the methylene group between two double-bonded
carbons is very much more susceptible to oxidative attack than carbons
adjacent to a single double bond. Thus, linoleic acid, with one active
methylene group, oxidizes ten to, twelve times as rapidly as oleic acid.
Linolenic acid, with two such labile carbons, oxidizes twice as fast as
linoleic (Gunstone and Hilditch, 1945).
The hydroperoxides formed are intermediates in the oxidative process.
They do not themselves contribute to the rancid odor, but they are
unstable and break down to a great variety of decomposition products,
some of which contribute to rancidity. Much recent work has been
devoted to the isolation and identification of such decomposition prod-
ucts, but the literature is too voluminous to review here.
The course of the oxidation of any animal fat follows a typical pattern.
A period of very limited oxygen uptake (the induction period) is followed
by B phase of rapid oxidation (see Fig. 4). The induction period is due
to the presence in animal fats of varying amounts of a natural anti-
oxidant, alpha tocopherol. The activity of this and other antioxidants is
discussed in Section V.
d. Species Variations in the Susceptibility of Animal Fats to Oxidation

As might be expected from the above discussion, the susceptibility of
any natural fat to oxidative rancidity depends upon its degree of un-
saturation (particularly with respect to fatty acids having more than one
double bond) and upon its antioxidant content. It has long been known
that certain fats, especially those of pork and poultry, are much more
easily oxidized than other animal fats. Rancidity is not a pressing problem
with beef or lamb, although Steinberg et al. (1949) showed that the
palatability changes which take place on freezer storage of beef are
correlated with the availability of oxygen, and Hines et al. (1951) found
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 5
deterioration in beef and lamb as well as pork in freezer storage to be due

primarily to oxidation of the fat.
These species differences are probably t o be attributed largely t o the
fatty acid composition. Chang and Watts (1952), in agreement with
earlier studies, found pork and poultry fats t o be much more unsaturated
than beef or lamb fats. The linoleic acid content of the fat from beef and
lamb cuts varied between 1 and 2 % of the total triglycerides, whereas
with pork it ranged from 7 to 10% and with poultry from 18 to 31%.
Although there may also be species differences in the tocopherol content
of the fat, such data as are now available do not indicate any clear-cut
trend (Lange, 1950). The tocopherol content of all animal body fats is
small, approximately 0.0005 to 0.003 %, as compared to vegetable oils,
most of which fall in the range 0.05 to 0.10%.
3 . Influence of Rations o n Fatty Acid Content of A n i m a l Body Fats
Since the early work of Burr and Burr (1929, 1930) it has been recog-
nized that animals are unable to synthesize fatty acids containing more
than one double bond if none is supplied in the diet, although they
apparently have the ability to utilize dietary fatty acids containing two
or three double bonds to synthesize more highly unsaturated ones
(Widmer and Holman, 1950). Thus, while saturated fatty acids and oleic
acid can be synthesized from carbohydrates, fatty acids containing two,
three, four, or more double bonds which appear in the body fat are
derived from unsaturated fat in the diet. Certain species, includingthe
hog, chicken, and turkey, tend to deposit fatty acids from the diet in
their body fat to a much greater extent than other species.
Early work on the composition of body fats of hogs as related t o diet
and other factors is reviewed both by Lea (1939) and by Bailey (1951).
When the oil or fat content, of the feed is low, the pig synthesizes a firm
fat, but when considerable portions of fat are fed, the pork fat reflects the
composition of that fed. The linoleic acid content of hog fa t could be
varied from 2 % (in low fat rations) to 32% (on soybean rations). Hogs
having “firm” fat may be produced by feeding soybeans to young animals
and then switching t o a carbohydrate ration for the fattening period
(Hostetler and Halverson, 1940), but it seems probable th a t this repre-
sents a dilution, not a replacement, of soft fat with hard. This ability t o
store unsaturated fatty acids extends also t o linolenic acid. Beadle et al.
(1948) found large amouiits of trienoic acid in yellow hog fat and in f a t
of rats fed linseed oil.
Poultry fat presents the same general picture of wide variability in
fatty acid composition. The early work of Cruickshank (1934) and
several more recent studies (Kummerow et al., 1948; Hite et al., 1949;
6 BETTY M. WATTS
Klose et al., 1951) demonstrate conclusively the deposition of large

amounts of unsaturated fatty acids from the ration in the body fats of
chickens and turkeys. Linolenic acid, which is not normally present in
animal fats t o any appreciable extent, was deposited in poultry skin fa t
when present in the diet (Hite et al., 1949). Since this fatty acid is nor-
mally present in such feed constituents as linseed meal, soybean meal,
alfalfa, and whole wheat, it may be expected to be a variable constituent
of poultry fats. Chang and Watts (1952) have also indicated the presence
of highly unsaturated fatty acids with five and six double bonds in
poultry fats, although it is not yet possible to say whether they are of
dietary origin or are synthesized by the fowl from fatty acids with fewer
double bonds.
The effects of these dietary variations in the fatty acid composition on
the susceptibility of the fat and especially of the meat to rancidity are
less clear-cut. One complicating factor is the fact th a t dietary sources of
linoleic and linolenic acids are usually also good sources of tocopherol
(Hove and Harris, 1951), and thus deposition of the antioxidant may to
some extent counteract the greater unsaturation. I n addition, when the
fat is present in meat, variations in the aqueous medium surrounding the
fat undoubtedly play an important part.
Shrewsbury et al. (1942) found in studies on the stability of hog fat
that the soft fat (peanut-fed, iodine number 73-80) was consistently less
stable than the hard (corn-fed, iodine number 5 8 4 0 ) both fresh and after
frozen storage. However, there was more variation in the keeping time of
hard fats from different lots of pork than in the keeping time of hard
versus soft in any one lot. Peroxide values for fats extracted from the
meats after freezer storage for 12 to 16 months were low and showed no
significant differences between hard and soft fats. Brady et al. (1946) and
Palmer et al. (1953) showed positive correlation between the softness of
the fat and the susceptibility t o rancidity of bacon and frozen ground
pork, respectively.
Kummerow et al. (1948) found th at the feeding of highly unsaturated
fatty acids was detrimental to fat stability of eviscerated frozen turkeys,
as determined both by peroxide values on the extracted skin fat and also
by organoleptic tests on the cooked carcass. Klose et al. (1951) observed
fishy flavors in roasted turkeys fed linseed oil as well as fish oils. The
fishy flavors in this case were present in the fresh roasted turkeys as well
as in those cooked after being stored in the freezer. Peroxide values of
birds fed linseed oil increased very rapidly in freezer storage.
I n addition t o direct deposition of unsaturated fatty acids from the
diet in the body fat of meat animals, recent work has indicated th a t
various dietary supplements may influence the amount and composition
of body fats by a n indirect effect on metabolic processes. Hite et al. (1949)

fed supplements of ethanolamine and choline to poultry. The fat from the
supplemented groups contained less of the 3 and 4 double-bond f a tty
acids and showed longer induction periods. Kummerow et al. (1949) were
able t o influence both the amount and also the composition of fat from
rats on purified rations by a large number of supplements (gallates,
tocopherol, butylhydroxyanisole, ascorbic acid, and especially pyri-
doxine). Sufficient data are not available to form any clear picture of the
mode of action of these supplements as yet, With the exception of to-
copherol they are not stored in the fat.
4. Deposition of Antioxidants in Animal Tissues
Barnes et al. (1943), Lundberg et al. (1944a), and Hanson et al. (1944)
found that of a large number of antioxidants fed to rats only the tocoph-
erols were stored in the rat adipose tissue, and the stability of the ex-
tracted rat fat (provided the fatty acid composition was not changed)
depended entirely on the tocopherol content of the diet.
This early work on laboratory animals stimulated attempts t o im-
prove the keeping quality of rendered fat and meat by increasing the
tocopherol in the fat and other tissues of meat animals through the feeding
of tocopherols. Watts et al. (1946) fed tocopherol supplements to pigs,
both on natural and purified rations. The total amounts fed (0.007 to
0.02 % by weight of the pigs) gave some slight increase in stability of the
rendered fat, but the magnitude of the effect was too small t o be of great
practical significance, although the larger amount is considerably more
than could be achieved by any manipulation of natural rations.
The feeding of still larger amounts of tocopherol in relation t o the
weight of the experimental meat animals has resulted in more significant
increases in tocopherol in the fat and various other tissues. Major and
Watts (1948) improved the stability of rabbit fat in animals on purified
rations by feeding or injecting tocopherol. Bratzler et al. (1950) showed a
greatly increased tocopherol content of a number of hog tissues as well a s
TABLEI
Rancidity Development and Tocopherol Storage in Turkey Tissues*’t
Tocopherol Acceptability Peroxides, m.eq./kg. Tocopherols, mg./kg.
total fed,$ g. rating Skin fat Abdominal fat Heart Leg
4.2 Excellent 1 4.8 107 24
.4 Poor 1.8 5.9 60 15
0 Poor 2 6.7 52 13
E depleted Poor 7.3 19.4 68 12
* Criddle and Morgsn (1951).
t After 9 months freezer storage.
t: Dose divided over 35 days just prior to slaughter.
8 BETTY M. WATTS
fat from various parts of t,he body by feeding large amounts of tocopherol
to animals on purified rations. Criddle and Morgan (1951) fed various
levels of tocopherol to turkeys on natural rations and demonstrated not
only increased tocopherol storage in various tissues a t the higher levels
but also improvement of stability of the fat on freezer storage. This in
turn correlated with improved acceptability of the cooked meat after
storage. A typical experiment is shown in Table I.
On the whole, the results of tocopherol feeding have been rather dis-
appointing. It does not seem that very much improvement in the stability
of meat fats can be achieved by manipulating ration components to
secure naturally high levels of tocopherol. Although there is no doubt that
additional tocopherol storage can be achieved in a variety of meat
animals by feeding large tocopherol supplements and that such storage
will improve the stability of the fat, this method is limited and economi-
cally wasteful. Only a very small fraction of the large doses of tocopherol
fed are actually stored in the carcass.
5 . Distribution of Fat in Meat as a Factor in Rancidij'ication
In addition to these inherent characteristics of the fat itself, contact
of the fat in meat with an aqueous solution containing surface-active
substances, accelerators and inhibitors of rancidity, creates a very differ-
ent situation from conditions which exist in a container of rendered lard.
The author has noted on a number of occasions that the keeping time of
fat rendered from pork tissues did not correlate with rancidity develop-
ment in the ground meat. Schreiber et al. (1947) reported that the sta-
bility of fat, as measured by accelerated tests on the extracted fat from
fresh birds, was not a good indication of the stability of poultry fat in situ
during freezer storage.
The orientation of unsaturated fatty acids at an interface can have a
profound effect on their oxidative behavior, even when the nonfat phase
contains no known pro- or antioxidants. Table I1 shows the relative rates
of oxidation of several fatty acids in bulk versus thin layers in contact
TABLEI1
Relative Rates of Oxidation of Pure Fatty Acids under Various Conditions*
Days to turn rancid at 24' C.t
Distribution of fatty acids Oleic Iinoleic Linolenic
Exposed in bulk in watch glasses 13 1 <1
Thin layer absorbed on filter paper 5 2.3 1.6
Thin layer in contact with aqueous phase1 5-8 3-4 2-3
* Lehmann and Watts (1952).
t As determined by half bleaching of dissolved carotene.
$ Range of various experiments in whioh distilled water and buffers at pH 5.6 and 7.5 were employed
a9 the aqueous phase.
with various aqueous solutions or absorbed on filter paper (relatively

inert, but containing polar hydroxy groups). Whereas the oxidation of
oleic acid is accelerated by its orientation a t the interface, the more
highly unsaturated fatty acids containing one or two active methylene
groups oxidize less rapidly under these conditions. Apparently the active
methylene groups are partially masked a t such interfaces.
The effect of surface orientation can be demonstrated very clearly
when the aqueous phase contains accelerators of rancidity. Hughes and
Rideal (1933) spread monolayers of oleic acid on the surface of aqueous
solutions of potassium permanganate. When the double bond touched
the water, oxidation proceeded very rapidly. I n more compressed films,
the reactive double bonds were removed from the water surface and
oxidation was greatly retarded. Eleostearic acid, containing three con-
jugated double bonds, lay flat on the surface and oxidized very rapidly.
Haurowitz and Schwerin (1941) called attention t o the importance of
interfacial orientation of linoleic acid a t the oil water interface in the
hemin-catalyzed oxidation of this fatty acid.
Slight increases in tocopherol content which confer only limited im-
provement in stability of the rendered fat may have a much greater effect
on the keeping quality of meat through synergistic activity with con-
stituents of the muscle juice. Watts and Wong (1951), by increasing the
tocopherol level of lard in contact with hemoglobin solution u p to 0.005%,
brought about a slight (50%) increase in keeping time. I n contrast, when
the hemoglobin solution also contained ascorbic acid, which acts syner-
gistically with tocopherol, there was more than a n eightfold increase in
keeping time. A number of normal constituents of muscle juice may func-
tion in this way.
Much more information needs t o be obtained on the effect of varia-
tions in muscle juice constituents on the stability of f a t with which it is in
contact in model systems which can be controlled, before it will be of
much avail to attempt to change the media surrounding the fat in meat
by dietary supplements.
6. Methods for Evaluating Oxidative Changes in Meat Fats
With the exception of organoleptic evaluations, all tests for rancidity
in meat require the separation of the fat from other meat constituents,
In most of the early studies this was accomplished by a preliminary drying
of the meat followed b y extraction of the f a t with a fat solvent, usually in
some form of continuous extraction apparatus. The disadvantages of this
procedure have been recognized by many workers. Oxidation can take
place during the drying process as well as during the subsequent lengthy
extraction. Decomposition of preformed peroxides can also occur in hot
10 BETTY M. WATTS
solvents (Watt et al., 1949). More recent studies have avoided both the
drying and the hot extraction and have reduced the time required for
extraction to a few minutes by cold blending the sample and solvent in a
Waring-type blender, preferably with a drying agent (Rockwood et al.,
1947; Watts and Peng, 1947b).
If the purpose is a rating of the degree to which the fat has undergone
oxidation at the time of its extraction from the meat, as in meat storage
studies, the fat can usually be analyzed for some product of oxidation
without removing the solvent. The peroxide value remains the most
widely used and generally satisfactory of such tests, in spite of the fact
that peroxides are intermediates in the process of oxidative decomposition
and are not themselves responsible for the rancid odor. Lea (1939) has
reviewed the earlier work on this test. It is commonly carried out by
estimating the amount of free iodine liberated by the oxidizing action
of the fat peroxides on potassium iodide. A somewhat more sensitive test
for peroxides was proposed by Lips et al. (1943) and independently by
Sumner (1943) based on the oxidizing action of the fat peroxide on ferrous
iron, followed by colorimetric estimation of the ferric iron as its colored
complex with thiocyanate. Lea (1945) improved the method by excluding
oxygen. Volz and Gortner (1947) have increased the sensitivity of the
original iodometric procedure by carrying out the reaction in a single
phase, using isopropanol as solvent.
The limitations inherent in the peroxide test as an objective method
for rancidity have been discussed by Lea (1939, 1946a) and by Stansby
(1941). The principal objection to its use is the fact that, because per-
oxides are intermediates, they are related to rancidity only as long as they
are formed from the fresh oil a t a rate greater than that of their decom-
position to rancid products. If peroxides are breaking down more rapidly
than they are forming, the peroxide number will decrease with increasing
rancidity. This can occur, for example, when fats which have been stored
at a low temperature are moved to a higher, or when fats in which
peroxides have accumulated in the dark are exposed to light.
Mainly with the idea of overcoming these objections to the peroxide
value as a measure of rancidity, various tests have been developed for
decomposition products of oxidized fats. The widely used Kreis test,
which depends upon the development of a red color when rancid fats are
treated with phloroglucinol, has recently been shown to be given by
malonic dialdehyde and other closely related constituents of oxidizing fats
(Pstton et al., 1951). The test has been greatly improved in recent years
by solution of all reactants in a single phase and colorimetric estimation
of the color produced (Walters el al., 1938; Pool and Prater, 1945; Watts
and Major, 1946).
The reaction of oxidation products of linolenic acid with thiobarbituric

acid t o give an orange red color has been used by Abramson (1949),
Wilbur et al. (1949), and others to follow unsaturated fat oxidation in
various tissues. Patton et al. (1951) present evidence th a t this test, like
the older Kreis test, also measures malonic dialdehyde.
In addition to the several tests for aldehydes of low and intermediate
molecular weight, Lea (1939) and Pool and Klose (1951) have described
a method for the estimation of monocarbonyl compounds in rancid foods
based on their reaction with dinitrophenylhydrazine. Table I11 compares
results of this test with peroxide values on turkey fats of varying degrees
of oxidation.
TABLE I11
Comparison of Monocarbonyl Compounds and Peroxides in Turkey Fat*
Storage Monocarbonyl
Individual Storage time, temperature, compounds, Peroxide values
birds months OF. millimoles/kg. millimoles/kg.
1 12 - 30 0.05 0.0
2 12 -30 0.14 4.4
3 6 0 0.42 10.9
4 6 0 0.54 19.5
5 6 0 7.34 240.0
* Pool and Klose (1951).
Unfortunately, there is as yet no real evidence that any of these tests
show any better correlation with organoleptic rancidity under changing
conditions of storage than does the peroxide value. I n fact, the Kreis test
is even more sensitive than the peroxide value to changes in temperature
and gives very high values with fats containing more highly unsaturated
fatty acids. Comparisons of the peroxide test with one or more of the
aldehyde tests and sometimes with organoleptic rancidity have been
made on meat or poultry fats under a variety of storage conditions
(White, 1941a; Watts and Major, 1946; Vail and Conrad, 1948; Mackey
et al., 1952).
Frequently i t is desired to determine not the degree of oxidation which
an extracted f a t has already undergone but its susceptibility to oxidation
under a standard set of conditions. This is the usual aim, for example,
when i t is desired t o compare fats from animals on different rations or t o
compare the effects of antioxidants. This may be done by storing the fat
under the desired set of conditions and following the course of oxidation
with any of the tests described above. It may also be accomplished by
manometric measurements of oxygen consumed (French el al., 1935;
Nagy et al., 1944; Banks, 1944; Stirton et al., 1945) or by following the
rate of bleaching of fats to which a carotinoid pigment has been added,
12 BETTY M. WATTS
since such pigments are oxidized very rapidly when fats have passed their
induction period (Hove and Hove, 1944a; Lovern, 1946; Watts and Peng,
1947a; Bickoff et al., 1952).
Most of the studies on fat stability have been carried out with the dry
fats. Of more direct application to the control of rancidity in meats are
artificial systems in which meat fat is brought into contact with a n
aqueous phase. Lea (1937), used emulsions of muscle juice in lard and in
some cases glass slides coated with lard and then immersed in thin layers
of the aqueous solution. Scarborough and Watts (1949) and Lehmann and
Watts (1951) have described a method for achieving contact by bringing
together filter papers saturated with the fat and aqueous phases, respec-
tively. Carotene dissolved in the fat serves as a convenient indicator of
the onset of rancidity. The method is crude but does allow the rapid
evaluation of antioxidants, meat constituents, or additives under condi-
tions which more closely resemble those in meat. The Warburg mano-
metric technique has also been used for investigation of aqueous fat
systems (Banks, 1944), but in this case the fat-water interface is con-
stantly changed by the shaking.
111. OXIDATIVEDISCOLORATIONS
1. Normal Pigments of Fresh Meat
The typical reddish colorations of both fresh and cured meats are
contributed by myoglobin and hemoglobin and their derivatives. Al-
though more than 90% of the pigment present in fresh meats is myoglobin
rather than hemoglobin (Shenk et al. 1934), a much greater volume of
work has been done on the latter pigment, since it can be obtained easily
in large quantities by laking washed red blood corpuscles. Myoglobin is
much more difficult to obtain in a sufficient state of purity. Reliable in-
formation concerning its properties awaited its crystallization by Theorell
(1932). Millikan (1939), in a review which summarizes the existing in-
formation up t o that time, compares the muscle and blood pigments in a
number of important properties. I n structure both have the same pros-
thetic group, ie., heme, an iron porphyrin. However, the globin to which
the heme is attached is different in the two pigments and, whereas myo-
globin consists of a single heme compound (molecular weight 17,500) , the
hemoglobin molecule contains four hemes, each associated with globin.
The molecular weight of hemoglobin is thus 68,000.
Both pigments combine reversibly with oxygen, carbon monoxide, and
nitric oxide t o form bright red oxyhemo (or myo) globin, carbon mon-
oxide hem0 (or myo) globin, and nitric oxide hem0 (or myo) globin,
respectively. I n all of these compounds the iron remains in the ferrous
OXIDATIVE RANCIDITY AND DISCOLORATION I N M E A T 13
form and the equilibrium is governed in each case by the partial pressure
of the respective gas. Both hemoglobin and myoglobin may lose an elec-
tron, becoming oxidized to the corresponding brown “met ” pigments.
Both may undergo changes of the porphyrin ring t o give green or grey
decomposition products (see Section 111, 2).
Whereas the two pigments undergo the same qualitative reactions,
there are several important quantitative differences between them which
should be recognized in applying data obtained on the blood pigment t o
meat. Myoglobin has a much greater affinity for oxygen than does hemo-
globin, but a much lower affinity for carbon monoxide (and probably also
for nitric oxide, although data seem to be lacking). Myoglobin oxidizes
to the brown ferric compound fourteen to sixteen times as rapidly as
hemoglobin when exposed t o atmospheric oxygen (Kiese and Kaeske,
1942). The displacement of the absorption maxima of myoglobin toward
longer wavelengths and its greater stability in alkaline solutions are the
bases for the spectrophotometric differentiation of the two pigments in
their mixtures (Shenk et al., 1934; Watson, 1935; Fanelli, 1949).
The color of fresh meat is typically the bright red of the oxygenated
heme pigments a t surfaces exposed to air and the purplish red of the
reduced pigments in the interior. These variations in color, as a function
of oxygen tension, are normal and characteristic of fresh meat; neverthe-
less the purplish red color, especially in ground meats such as hamburger,
is often objectionable to consumers and is not distinguished from the
dulling or browning due to methemoglobin formation.
Coleman and Steffen (1949), in a patent assigned to Armour and
Company, propose the use of niacin to bring about a uniform bright red
color throughout fresh meats, owing to the formation of “ a new pigment
reaction product which is bright red in color.” No information is available
on the nature of this compound. The amount of niacin recommended
(0.3 g. per pound of meat) is much larger than the amounts normally
occurring in meat. Hopkins et al. (1950) from the same laboratory have
patented a process for maintaining uniform red color in ground meats by
introducing oxygen mechanically. This is accomplished by freezing the
ground meat, breaking it into small pieces while frozen, and subjecting
the pieces to pressure sufficient to shape them into patties but not suffi-
cient t o drive out the entrapped air.
2. Oxidation Products of Heme Pigments
Two types of oxidative changes are chiefly responsible for the ab-
normal brown, grey, and green discoloration of meat. One involves the
oxidation of the ferrous iron in the heme compound t o the ferric condition;
the second is a direct attack by oxygen on the porphyrin ring.’
14 BETTY M. WATTS
The most commonly encountered type of discoloration is th a t of the

brown oxidation products, methemoglobin and metmyoglobin, formed
from the normal blood and muscle pigments by oxidation of the iron t o
the ferric state (Brooks, 1929, 1938). Various aspects of this oxidation are
discussed at length in several recent reviews (Granick and Gilder, 1947;
Theorell, 1947; Wyman, 1948; Granick, 1949; Lemberg and Legge, 1949;
Haurowitz, 1950).
The question as to why the heme pigments sometimes combine
reversibly with atmospheric oxygen to form the bright red oxygenated
pigments of normal meat and at other times become oxidized to the
brown ferric compounds is not clear. The ability of hemoglobin and myo-
globin t o combine reversibly with oxygen depends upon their specific
protein linkage with native globin. Other heme proteins of tissues (per-
oxidase, catalase) do not possess this ability even though the iron por-
phyrin portions of the molecule are identical.
Denaturation of the globin destroys the ability of hemoglobin or
myoglobin t o combine reversibly with oxygen and greatly increases the
susceptibility of these pigments to true oxidation. P a rt of the oxygen
liberated by the denaturation of oxyhemoglobin oxidizes the iron; part
probably attacks globin itself (Holden, 1936). The ferrous, denatured
globin hemochromogen formed by denaturation of hemoglobin under
reducing conditions is much more susceptible to oxidation than is hemo-
globin itself. The oxidation potential of the ferrous-ferric hemochromogen
system is -0.098V a t p H 7.06, whereas the corresponding potential of the
hemoglobin-methemoglobin system is between +O. 144' and +O. 152
(Lemberg and Legge, 1949).
It is probable that even partial and reversible denaturation of the
globin, which may not be accompanied by coagulation, can accelerate
the rate of oxidation. The many agents, not themselves direct oxidizing
agents, which are known to accelerate oxidation of the iron of oxyhemo-
globin, may act indirectly by disturbing, a t least temporarily, the bonds
between heme and globin. Granick (1949) has pointed out th a t the drugs
which bring about methemoglobin formation are those which denature
globin. Factors which accelerate oxidation of fresh meat pigments, such
as heat, freezing, acid, salt, ultraviolet light, and certain metals, are
known to denature globin.
Irrespective of the mechanism of methemoglobin formation, it is
known that this pigment is continually formed and reduced in the blood
of living animals (Cox and Wendel, 1942). The reducing mechanisms
normally operative in the living animal have been studied extensively,
usually with the objective either of combating pathological conditions
of methemoglobinemia or of preserving whole blood for intravenous
injection. Much of this work has been reviewed by Granick (1949),

Lemberg and Legge (1949), and Bodansky (1951).
Reduction of methemoglobin in red cells is dependent upon the
glycolytic system of the cells. Reduced diphosphopyridine nucleotide
(DPN-H,) produced during glycolysis can reduce methemoglobin, but
not directly. Electron mediators, normally flavines, are essential. Several
additives have proved effective in accelerating methemoglobin reduction
through this system and have been suggested for use under varying
conditions. Met,hylene blue catalyzes reduction of methemoglobin by
DPN-H, (Gutman et al., 1947). Various substrates, including glucose
and other hexoses, lactate, fumerate, malate, citrate, etc., can, under some
conditions, accelerate methemoglobin reduction, presumably by in-
creasing the rate of reduction of DPN (Spicer et al., 1949; Pennell and
Smith, 1949; Gibson, 1948; Gutman et al., 1947). The addition of nico-
tinamide has been useful in preventing hydrolysis of D P N (Gutman
et al., 1947).
All of the above work has been done on blood or its derivatives.
Similar studies on the formation and reduction of metmyoglobin in
tissues are for the most, part lacking. Jensen (1935) patented the addition
to meat (by arterial pumping just after slaughter) of various organic acids
or their salts, with the idea of increasing reducing conditions within the
meat through action of cellular dehydrogeriases on the added acids.
Methemoglobin may also be reduced directly by various reducing
agents such as sodium sulfite (Jensen and Urbain, 1936a), titanous
citrate (Ramsay, 1944), dithionite (Lemberg and Legge, 1949), glycer-
aldehyde (Kiese, 1943), cysteine (Kiese, 1943), and ascorbic acid (Gibson,
1943; Kiese, 1943). Only the latter has demonstrated usefulness in meats
(see Section V, 3).
I n addition to brown and grey discolorations due to the formation of
methemoglobin, very objectionable greenish pigments may appear in
meats. Lemberg and Legge (1949) have reviewed in detail various trans-
formations of the heme pigments which result in green compounds. All of
these involve an attack on the porphyrin ring, usually a t the a methene
bridge. The essential change seems t o be elimination of the double bond
a t this point, thus interrupting the series of conjugated double bonds
which comprises the porphyrin ring and destroying the resonance struc-
ture. It is not essential that the ring be ruptured a t this point; in fact,
there is good evidence that at least two green compounds, choleglobin
(Lemberg and Legge, 1950) and sulfhemoglobin, retain their closed
porphyrin rings. Further oxidation, involving opening of the ring with
splitting out of the a methene carbon atom, can then occur (verdohemes) .
Figure 1 shows suggested structures of choleglobin and verdoheme. Iron
16 BETTY M. WATTS
is easily removed from the opened ring t o give the bile pigment biliverdin.
Other methene bridges may be attacked in the same way, giving rise to
three, two, and one pyrrole fragments which range in color from yellow
to colorless.
P P
"\ ' 0 /nr

bl OH\:
H
Choleglolin
(green) \
n2ol
/ "erdoherne
(green)
P P
Heme
(red)
FIG.1. The formation of green porphyrins from hemoglobin. The side chains on the
porphyrin ring are abbreviated, i.e., M = methyl, P = propionic acid, V = vinyl. I n
both green porphyrins the conjugated double-bond system of the porphyrin ring is
interrupted a t the a rnethene bridge.
Such attacks on the porphyrin ring may occur in meat under a variety
of conditions. Sulfhemoglobin (formed directly upon addition of sulfide or
thiosulfates in the presence of oxygen) has been attributed to hydrogen-
sulfide bacteria, discussed by Jensen (1945). Any reaction which will
produce hydrogen peroxide under conditions where it is not readily
decomposed by catalase (as in cured meats where catalase is absent)
results in such rapid and intense greening that the reaction has been
proposed as a delicate test for heme pigments by Jensen and Urbain
(193613). Here again, peroxide-forming bacteria have been implicated
(Jensen and Urbain, 1936a; Jensen, 1945; Niven et al., 1949; Niven,
1951). Methemoglobin and metmyoglobin, as well as hemoglobin and
myoglobin, form unstable addition compounds with hydrogen peroxide,
which then decompose with destruction of the heme nucleus (Keilin and
Hartree, 1950).
A similar greening reaction occurs when various hydrogen donors are
brought into contact with oxyhemoglobin or oxymyoglobin a t physio-
logical temperatures. This coupled oxidation has been studied most
extensively with ascorbic acid and hemoglobin (Lemberg et al., 1939,
1941 ; Kiese and Kaeske, 1942; Foulkes and Lemberg, 1949; Takeya,
1949; Kikuchi, 1950; Watts and Lehmann, 1952a). I n this case the green-
ing is caused by the formation of a n unstable hydrogen peroxide deriva-
tive of hemoglobin (Lemberg et al., 1939).
3. The Pigments of Cured Meats; Their Oxidation

The general chemistry of the meat-curing process is well covered by
Urbain (1951) and the part played b y bacteria in this process by Jensen
(1945). Jensen (1949) has also described the curing process and various
types of cured meat products in nontechnical language.
The cured meat pigment is nitric oxide hemochromogen, formed b y the
heat denaturation of nitric oxide hem0 (or myo) globin (Haldane, 1901).
The latter compound is a ferrous heme derivative similar to oxymyoglobin
except that oxygen is replaced by nitric oxide. Nitric oxide hemoglobin
can be prepared directly by passing nitric oxide gas through hemoglobin
solutions under anaerobic conditions (Keilin and Hartree, 1937; Urbain
and Jensen, 1940) or more conveniently by the addition t o hemoglobin
solutions of a strong reducing agent and a nitrite salt (Jensen and Urbain,
193th). The pigment is bright red with a n absorption spectrum very
similar t o t ha t of oxyhemoglobin (Fig. 5).
While there is no doubt th at nitric oxide can combine directly with
reduced myoglobin to form nitric oxide myoglobin, which retains its
redness on heat denaturation, the sequence of events leading t o the forma-
tion of the cured meat pigment in meats is much less clear. The active
curing ingredient, nitrite, reacts %withoxyhemoglobin to form methemo-
globin (Greenberg et al., 1943) according to the reaction:
NOz- + 2Hh02+ NOa- + Hbd.3 + 02
This reaction is very rapid in the acid range of normal meat. To the
extent t o which the meat pigment is in the oxygenated form when brought
into contact with curing salts, this is the first reaction which occurs. Even
in the absence of oxygen, nitrite reacts with hemoglobin to give one mole-
cule of nitric oxide hemoglobin and one of methemoglobin, if substances
18 BETTY M. WATTS
capable of reducing both methemoglobin and nitrite are not present

(Brooks, 1938).
I n comminuted meats such as frankfurters, the entire mass turns grey
upon mixing with the curing salts. Color fixation (normal pink color of
cured meat) takes place during the gradual heating in the smokehouse
and the subsequent cooking. The heat treatment given most cured meat
products is an important factor in developing this color. Winkler and
Hopkins (1940) made objective measurements on a photoelectric com-
parator of the “total brightness” of bacon, i.e., intensity of the reflected
light in samples heated to various temperatures for different lengths of
time (Table IV).
TABLEIV
Total Brightness (Average) of Bacon Samples at Conclusion of Heat Treatment*
Duration of
heat treatment, Temperature, “C.
hours 20 40 50 60 70 80
5 121 120 126 154 163 168
10 115 115 134 155 163 158
20 113 128 139 152 157 143
40 - 143 158 161 169 128
Mean 116 126 139 156 163 149
* Winkler and Hopkina (1940).
Current theories of the chemistry of meat curing assume that reduc-
tion of metmyoglobin is brought about by cellular reducing systems
during early stages of the heating. The reduced myoglobin then combines
with nitric oxide (similarly formed by reduction of nitrite) and the result-
ing nitric oxide myoglobin is converted by further heating t o the corre-
sponding red (pink) denatured globin hemochromogen. The latter is
believed t o be less subject to oxidation than the undenatured pigment
because reactivity is reduced by loss of solubility on coagulation.
It seems doubtful that this is an adequate explanation of the observed
facts. Not only heat but many other factors known to accelerate globin
denaturation, such as freezing, salt, acid, and certain metals, also acceler-
ate the formation of nitric oxide myoglobin or nitric oxide denatured
globin hemochromogen. This acceleration occurs not only in meat but
also in relatively pure methemoglobin preparations in the presence of
nitrite and a reducing agent such as ascorbic acid, and the acceleration
is very evident even a t such early stages in the denaturation of the globin
that no coagulation has occurred (the solution remaining optically clear).
Thus, the same agents which bring about oxidation of oxyhemoglobin to
brown ferric pigments, also bring about formation of red ferrous nitric
oxide derivatives from methemoglobin in the presence of curing salts
(Watts and Lehmann, 1952a, b). With respect to these factors, the condi-
tions necessary to develop and retain cured meat color are exactly the
reverse of those necessary t o protect fresh meat color.
The mechanism by which protein denaturing agents are able to effect
reduction of the ferric iron in the presence of nitrite (or nitric oxide) is
obscure. It is probable that an intermediate is involved. Keilin and
Hartree (1937) demonstrated that nitric oxide combined not only with
028-
,/'
024-
.
'0
,/
020-
3
YI
0
H 016-
-"
0"
0 12 -
0 08 -
I 002% NOI-
0 0.02% Fe (CNk,'
Wavelength, mu
FIG.2. Absorption spectra of compounds formed upon addition of nitrite t o oxy-

hemoglobin (Watts and Faulkner, 1953). Ferricyanide and the lower concentration
of nitrite produce the typical absorption spectrum of methemoglobin. The curve
obtained with the higher concentration of nitrite probably represents a mixture of
methemoglobin and methemoglobin nitrite.
reduced hemoglobin t o give nitric oxide hemoglobin but also with met-
hemoglobin t o give an unstable nitric oxide methemoglobin which under-
goes a slow autoreduction to nitric oxide hemoglobin. Barnard (1937) has
presented evidence for the formation of a methemoglobin nitrite t o
explain changes from a brown to a red color in solutions containing a high
ratio of nitrite to hemoglobin. Further evidence for this compound is
presented in Fig. 2. At a molar nitrite-to-hemoglobin ratio of 5 to 1, the
solution is brown and the absorption spectrum is practically identical with
that of methemoglobin prepared by the addition of ferricyanide. Upon
the addition of a 50-to-1 ratio of nitrite, the solution is much redder and
has a single peak a t 540 mp.
Further investigation of the formation, stability to oxidation, and
other properties of compounds of methemoglobin with oxides of nitrogen
20 BETTY M. WATTS
should be profitable in the control not only of the normal curing process
but also of such abnormal conditions as “nitrite burn.”
The oxidation products of nitric oxide hemoglobin are the same as
those from hemoglobin. The transformation of nitric oxide hemoglobin
FIQ.3. Derivatives of myoglobin of importance in meats. In the outer circle are

represented the insoluble hemochromogens obtained by coagulation. Only in the case
of the cured meat pigment is the denatured compound red. Dotted portions represent
possible intermediates between metmyoglobin and the cured meat pigment.
to methemoglobin in the presence of oxygen is very rapid, even though
nitric oxide hemoglobin is stable indefinitely in the absence of oxygen
and does not lose its combined nitric oxide when subjected to a high
vacuum (Urbain and Jensen, 1940).
OXIDATIVE RANCIDITY AND DISCOLORATION IN MEAT 21
Nitric oxide hemoglobin and the corresponding denatured hemo-

chromogen are highly subject to attack of the porphyrin ring by pre-
formed hydrogen peroxide. The greater susceptibility of the cured meat as
compared to fresh meat pigments to greening by hydrogen peroxide
(Jensen and Urbain, 1936a) is at least partly to be explained by the
catalase activity of blood and fresh meat and the destruction of this
enzyme in formation of the cured pigment. On the other hand, the cured
meat pigment is not greened by hydrogen donors such as ascorbic acid
(Watts and Lehmann, 1952s).
Figure 3 is a schematic representation of the interrelationships of the
pigments of fresh and cured meats and their oxidation products.
4. Methods for Investigation of Color Changes in Meat

Spectrophotometric methods for the analysis of hemoglobin and
various derivatives in solutions have been widely employed in routine
clinical work as well as in research studies. The literature is voluminous
and will not be reviewed here. Similar methods are applicable to studies
of myoglobin in muscle extracts. Bowen (1949) published absorption
spectra and extinction coefficients of myoglobin and a number of its
derivatives in wavelengths ranging from 1000 to 450 mp. These apply, of
course, only to clear solutions of the pigments in question. Husaini et al.
(1950) give a simple procedure for preparing clear muscle extracts for
myoglobin determination.
Winkler (1939a) and Winkler et al. (1940) described an objective
method for following color changes a t meat surfaces, using a photoelectric
color comparator. Reflected light in the red, green, and blue regions of the
spectrum, defined by standard colored filters, was measured photoelec-
trically and expressed as per cent of the amount scattered in the same
spectral regions from a standard white surface under the same light
intensity. Total reflection, obtained by adding values from all three
regions, was used as a measure of the brightness of the meat surface. A
number of commercial instruments, similar in principle to that described
by Winkler, are now available.
Using a reflectance attachment to the Beckman spectrophotometer,
Ramsbottom et al. (1951) followed fading of the surfaces of cured meats
by measuring the ratio of light reflected a t 650 mp to that a t 570 mp.
Reflectance curves are needed over the entire visible region of the spec-
trum, from meat surfaces on which the pigments have been converted by
suitable'means to known forms, such as nitric oxide myoglobin and met-
myoglobin, before any of the above methods can be widely applied in
meat studies.
22 BETTY M. 'WATTS
IV. THE COUPLEDOXIDATIONOF HEMOGLOBIN

FATS
A N D UNSATURATED
I n addition to the independent oxidation of unsaturated fats and of

the heme pigments in meat, there is also a reaction between the two
which brings about their mutual oxidation, accelerating both rancidity
and color loss.
Robinson (1924) first described the catalytic effect of the hemes on
the oxygen uptake of linseed oil. Hemoglobin, methemoglobin, and hemin
all had about the same accelerating effect in equivalent concentrations.
Inorganic iron had a relatively slight effect and the porphyrins, after
removal of iron, none a t all.
The more rapid oxidation of the fat is accompanied by concomitant
oxidation of the hemoglobin. Niell and Hastings (1925) used linseed oil
to accelerate oxidation of hemoglobin to methemoglobin. Haurowitz et
al. (1941) demonstrated the destruction of the porphyrin during its pro-
longed reaction with unsaturated fat. The color faded and inorganic iron
was released, but neither porphyrins nor bile pigments could be identified
as cleavage products. The reaction was limited to fatty acids more un-
saturated than oleic. The following figures were obtained after hemo-
globin and unsaturated fatty acids were shaken in a Warburg for 2$6
hours :
Catalyst Destroyed,
Fatty Acid 02 Absorption, cu. mm. %
linoleic 351 65
oleic 17 8
The mechanism of the reaction has not been extensively studied.

Barron and Lyman (1938) attributed the catalytic effect to initiation of
new reaction chains by the heme compounds. Banks (1944) suggests th a t
the active catalyst is a combination of heme and fat peroxides. The reac-
tion takes place only in heterogenous systems (Haurowitz and Schwerin,
1941 ; Lovern, 1946) ; if heme compounds and fatty acids are dissolved in
the same solvent the rapid oxidation does not occur. It may be that the
much greater unit efficiency of the iron in hemoglobin as compared t o
inorganic iron as an electron transfer medium for the oxidation of un-
saturated fats is due entirely to concentration and orientation of hemo-
globin a t the interface, in contact with the unsaturated fat (Harper,
1953).
Since hemoglobin and myoglobin are brought into intimate contact
with fat in meats, this coupled reaction might be expected to contribute
both t o rancidity and discoloration. Extracts from pork tissue, both
muscle and fat, have been found by a number of workers to accelerate fa t
OXIDATIVE RANCIDITY A N D DISCOLORATION IN MEAT 23
oxidation. Lea (1937), who first observed this, attributed it to a lipoxidase.

Watts and Peng (1947a) accounted for the accelerating effect of extracts
from pork muscle by the myoglobin present. On the other hand, Reiser
(1949) believed the catalytic effect of aqueous extracts of bacon adipose
tissue on fat oxidation to be due to an enzyme. He based his conclusions
on the heat lability of the catalyst and the fact that after removal of heme
pigments catalysis still occurred. Chang and Watts (1949) ascribe this
loss of activity on heating to coagulation of the hemoglobin. Tappel (1952)
came t o the conclusion that the catalyst in both muscle and fat is a heme
pigment, not a lipoxidase of the type known to occur in plants, since the
catalytic effect here is apparent in heterogenous systems only, whereas
lipoxidase is even more active in homogeneous solutions.
The importance of this reaction in contributing to the oxidative
changes which take place in meat is difficult to evaluate. It is certainly a
contributing factor in the deterioration of ground meats preserved by
freezing (Watts and Peng, 1947b), but may be of less importance in large
cuts, frozen whole (Watts et al., 1948). Klose et al. (1950) found that tur-
key dark meat turned rancid in storage much more rapidly than the
white meat.
Lea (1937) and Watts and Peng (1947a) observed that the rancidifying
effect of muscle extracts fell off with increasing pH (within the range of
normal meat). Experiments on frozen ground pork adjusted to different
pH values (Watts and Peng, 1947b) have demonstrated the same close
correlation of rancidity and pH. Fading of the color accompanies ran-
cidification. pH ranks with tocopherol content and fatty acid make-up
of the fat as a major cause of variation in freezer storage life of meat from
different carcasses.
Heating hemoglobin solutions, muscle extracts, or meat enough to
coagulate the hemoglobin or myoglobin eliminates their catalytic effect
on fat oxidation. The heating does not destroy the iron porphyrin which
TABLEV
The Effect of pH on Rancidity Development in Raw and Precooked Pork Sausage*
Peroxide value after
4.5 months atorage,
Lactic acid pH of raw millimolea/kg.
added % meatt Raw Cooked
None 6.5 2.0 3.3
0.031 6.4 1.6 3.7
0.103 6.1 5.9 2.9
0.206 5.6 16.9 3.6
0.617 4.8 25.2 4.7
* W a t t s and Peng (1947b).
t pH of cooked samples was usually 0.1 to 0.2 higher than raw.
24 BETTY M. WATTS
is the active catalyst, but presumably inactivates it by rendering it

insoluble as the globin is coagulated. Changes in p H no longer affect the
rate of rancidification of meat after cooking (Table V).
As would be expected considering the general effectiveness of all iron
porphyrins so far tried as catalysts of fat oxidation, nitric oxide hemo-
globin accelerates rancidity to the same extent as hemoglobin at the same
concentration (Chang and Watts, 1949). Tappel (1952) found that
extracts from cured pork as well as from raw pork accelerated oxidation
of linoleic acid. No information is available on color changes in nitric
oxide hemoglobin during the course of the catalytic process.
V. ANTIOXIDANTS
The development of new chemicals for the protection of fats from
oxidative changes has progressed a t a very rapid rate during the past
decade. Hilditch (1944) has reviewed some of the British work on sta-
bilization of dried foods, including meats, with antioxidants. Lundberg
(1947) made a survey of the antioxidants proposed for use in foods a t
that time and Riemenschneider (1947) reviewed briefly the activity of
antioxidants of interest t o cereal chemists. Unfortunately, there does
not seem to be a recent comprehensive review of the subject. Space
limitations will not permit more than a brief rbum6 here, directed par-
ticularly at the possible usefulness of these compounds in meat.
1. Classification and Mode of Action of Fat Antioxidants

Most compounds which have a direct antioxidant effect on pure un-
saturated fatty acids or their glycerides are phenolic substances. The
antioxidants of this type which have been approved by the Bureau of
Animal Industry for use in lard are: the naturally occurring tocopherols
0.03% (Olcott and Emerson, 1937; Golumbic, 1941, 1943; Hove and
Hove, 1944b); gum guaiac 0.1% (Newton and Grettie, 1933; Doegey,
1943; Black, 1950) ; nordihydroguaiaretic acid (NDGA) 0.01 % (Lund-
berg et al., 1944b), propyl gallate 0.01% (Golumbic, 1942; Boehm and
Williams, 1943) and butylhydroxyanisole (BHA) 0.02 % (Kraybill et al.,
1949; Dugan et al., 1951; Rosenwald and Chenicek, 1951).
Compounds of this type extend the induction period of oxidizing fats,
presumably by absorbing the activating energy of fat peroxides, thus
breaking chain reactions which might otherwise extend to several hun-
dreds or even thousands of fat molecules. The antioxidants are themselves
oxidized during this process (Filer et al., 1944; Lundberg et al., 1947;
Mahon and Chapman, 1953). Figure 4, taken from the data of Mahon and
and Chapman, illustrates the increase in the induction period of a sample
of lard brought about by an antioxidant mixture of BHA and propyl

gallate and the fate of the antioxidants during this period.
In addition to the phenolic antioxidants, there are a large number of
compounds of widely different chemical composition which have no pro-
tective effect when added to pure triglycerides but which enhance the
keeping qualities of animal fats if added along with a phenolic antioxidant
and which are therefore termed "synergistic " antioxidants. Members of
this group which have been approved as additives to lard are citric acid
250
0.04 200
B .0.03
9
+
-
B
.-
mc
L1
;0.02
\
Control"
mroxide
value
xide
due
150
100
$
E
%
-
m
al
3
a \l I e
82
0.01 50
J I I ' I I
0
0 5 10 15 20 25 30 35
Time in weeks
FIG. 4. Peroxide values and antioxidant destruction in lard (Mahon and Chap-
man, 1953). The control contained no added antioxidant. The experimental sample
contained 0.04% butylhydroxyanisole (BHA), 0.012% propyl gallate (PG), and
0.008% citric acid. Stored at 61" C.
0.005% (Lindsey and Maxwell, 1949) ; phosphoric acid 0.005% (Eckey,

1934, 1935; Kraybill and Beadle, 1948) ; thiodipropionic acid 0.01 % and
its esters (O'Leary, 1946) and lecithin in any amount (Evans, 1935;
Olcott and Mattill, 1936). Many other compounds have been reported to
have synergistic activity, including normal constituents of meat such as
amino acids (Clausen et al., 1947), ascorbic acid (Golumbic and Mattill,
1941; Calkins and Matthill, 1944), nicotinic acid (Taub and Simone,
1947) , and para-aminobenzoic acid (Norris, 1949).
The mode of action of this varied group of compounds is not well
understood and is probably not the same for all synergists. Many have
the ability of combining with metals which would otherwise accelerate
oxidation. Some, such as ascorbic acid, may reduce the oxidized forms of
26 BETTY M. WATTS
the primary antioxidant (Golumbic and Mattill, 1941). Others, which are
not themselves reducing agents, i.e. phosphoric acid and various organic
acids, may form fat-soluble complexes with primary phenolic antioxidants.
The complexes may diffuse into the fat and there react with activated fat
molecules, absorbing the excess energy and so breaking the chain reaction
(Calkins, 1947).
2 . Application to Meats
While all of the compounds approved for use in lard have been
thoroughly tested for toxicity and have established their usefulness in
protecting the rendered fat, none has been given official sanction as an
additive to meat. The problem of protecting meats is more complicated
than that of protecting rendered fats. It is essential that the antioxidant
chosen should protect the fat when it is in contact with muscle juice.
Further, the antioxidant must be capable of uniform distribution in the
meat.
A number of phenolic antioxidants retarded the oxidation of unsaturated
fats in contact with hemoglobin solutions (Barron and Lyman, 1938;
Banks, 1944; Chang and Watts, 1949). I n contrast, the water-soluble
synergistic antioxidants, with the exception of ascorbic acid, had no
effect on the hemoglobin-catalyzed oxidation, even when the fat con-
tained added tocopherol, although many were active when the hemo-
globin was coagulated by heat and so might be expected to be effective in
cooked meats. These studies were made on artificial systems where
tocopherol or other phenolic antioxidants could be introduced directly
into the fat and the synergists into the aqueous phase.
The problem of getting the antioxidant into tissue fat has not been
adequately solved. The most effective phenolic inhibitors are practically
insoluble in water. Attempts at utilizing them in meat generally involve
either their solution in a fat which is then applied to the surface of meat
cuts or their dispersion with various carriers and emulsifying agents in
curing brines, cooking waters, and comminuted meats.
For example, Smith et al. (1945) and Brady et al. (1946) lengthened the
induction period of bacon slices by applying vegetable oils and phenolic
antioxidants to the surface. Davis and Bywaters (1951) prolonged the
freezer storage life of eviscerated broilers by dipping or spraying them
with a solution of melted vegetable fat containing NDGA, ascorbic acid,
and a vegetable gel. Fonyo (1950) patented a treatment for indigenous
tissue fats which consists of NDGA emulsified in sorbitan derivatives of
various fatty acids or polyethylene glycols. The solution is then diluted
with water or brine. Komarik and Hall (1951) patented an accelerated
curing process for bacon which includes a preliminary soaking in an
aqueous bath in which various antioxidants are dispersed. A patent

issued t o Cornwell (1951) describes a protective coating for hams in
which antioxidants as well as bactericidal and fungicidal compounds are
dispersed in a 2 to 10% solution of hydroxyethylcellulose. Klose et al.
(1952) protected turkeys in freezer storage by coating them with various
antioxidant mixtures in propylene glycol and gelatin.
Cooked meats preserved by freezing or drying respond well to anti-
oxidant treatment. Lea (1944) extended the life of dehydrated pork by
incorporating ethyl gallate and gum guaiac into it before drying a t con-
centrations of 0.02-0.1 %. The antioxidants were dissolved in alcohol,
which in turn was mixed with hot fat and added to the cooked meat.
Morgan and Watts (1948) made use of the natural antioxidants present
in soybean flour as well as added gum guaiac, tocopherol, and ascorbic
acid t o protect dehydrated pork scrapple. Lineweaver et al. (1952) added
a commercial antioxidant preparation consisting of 20% BHA, 6%
propyl gallate, and 4 % citric acid in 70 % propylene glycol to the water in
which turkeys were cooked before preservation as the frozen creamed
product. A t a level of 0.005% of the weight of meat they were able to
show excellent protection over storage periods up to twelve months.
3. CTse of Antioxidants f o r (lolor Protection
The fact that an antioxidant may be successful in protecting the fat
from rancidity does not mean that it will also retard oxidative discolora-
tion. There is no reason to suppose that any of the phenolic antioxidants
will have a beneficial effect on color. The polyhydroxy phenols do not
reduce methemoglobin; in fact, their oxidation products, the quinones,
catalyze the oxidation of hemoglobin t o methemoglobin (Fishberg, 1948).
Wiesman and Ziemba (1946) found th at the addition of NDGA t o frozen
pork sausage actually made the sausage look worse. The author has fre-
quently observed increased methemoglobin formation in frozen fresh
meats treated with phenolic antioxidants, even though the antioxidants
may have been giving protection against rancidity. Kraft and Wander-
stock (1950), on the other hand, found that NDGA dissolved in coconut
oil and brushed on the meat surface, protected the surface color of round
steaks exposed to light. However, the protection obtained was erratic and
no information was given on the effect of the oil carrier alone.
The only antioxidant which has shown any great promise in the pro-
tection of meat color is ascorbic acid (Bauernfeind, 1953). This compound
reduces methemoglobin and metmyoglobin and has been claimed as a
color-st,abilizing agent in fresh ground meats (Coleman et al., 1951). Its
use in fresh meats seems to be limited to the refrigerated product. A t
higher temperatures or in the freezer it can accelerate oxidation of hemo-
28 BETTY M. WATTS
globin (Watts and Lehmann, 1952a, b). On the other hand, in the presence
of nitrite, ascorbic acid accelerates methemoglobin reduction at all tem-
peratures and probably reduces nitrite t o nitric oxide (Lugg, 1950),
although little published information is available on this step. It can
develop cured meat color under conditions where color fixation would
otherwise be incomplete and also protect cured meat surfaces from fading
when exposed to oxygen.
50
I I I 4 I I I100
660 640 620 600 580 560 540 520
Wavelength, m r
FIG.5. Changes in the absorption spectrum of oxyhemoglobin brought about by

ascorbic acid (Riedesel and Watts, 1952). All solutions except the original oxyhemo-
globin were stored for 3 hours a t 45" C. (113"F.) before reading.
A. Oxyhemoglobin alone.
B. Oxyhemoglobin plus 0.02% nitrite. This is the typical absorption spectrum of
methemoglobin. Solution is brown in color.
C. Oxyhemoglobin plus 0.1 % ascorbic acid. Solution greenish brown. A mixture of
methemoglobin and choleglobin.
D. Oxyhemoglobin plus 0.02% nitrite and 0.1 % ascorbic acid. Typical absorption
spectrum of nitric oxide hemoglobin. Bright red.
Figure 5 shows the absorption spectra obtained when ascorbic acid

and nitrite, individually and together, were added to aliquots of an oxy-
hemoglobin solution a t 45" C. (1 13" F.). Nitrite alone converted the oxy-
hemoglobin immediately to methemoglobin. Ascorbic acid alone gave a
mixture of methemoglobin and choleglobin. Nitrite and ascorbic acid,
added together, gave the typical absorption spectrum of nitric oxide
hemoglobin.
In view of its usefulness in preserving meat color the behavior of
ascorbic acid with tissue fats becomes of particular interest. Unfortu-
nately, ascorbic acid (Abramson, 1949) and its homologs (Table VI) can
accelerate oxidation when brought into contact with animal fats low in
tocopherol. Ascorbic acid itself brings about greater oxidation when
introduced into aqueous fat systems (Scarborough and Watts, 1949),
whereas the fat-soluble ascorbyl palmitate oxidizes fat alone (Nagy et al.,
1945). Dehydroascorbic acid and d-isoascorbic acid behave like ascorbic
acid. The mechanism of accelerated fat oxidation by these compounds is
not clear.
TABLEVI
The Effect of Ascorbic Acid and Related Compounds on the Oxidation of Lard*
Days to turn rancid1 at 45" C.
Lard in contact with
Ascorbyl compound Plain lard aqueous solution, pH 5.8
Control 4.5 5.0
Ascorbic acid 6.5 0.50
d-Isoascorbic acid 5.0 0.50
Ascorbyl palmitate 1.5 2.5
Dehydroascorbic acid 5.0 0.70
* Lehmann and Watt8 (1962).
t Aa indicated by half bleaching of carotene.
Retardation rather than acceleration of rancidity occurs if the level of

tocopherol or other phenolic antioxidant i n the fat is raised sufficiently
high (Krukovsky, 1949; Watts and Wong, 1951) or if various compounds
such as ethylenediaminetetraacetic acid or polyphosphates (Lehmann and
Watts, 1951), which have in common the ability to complex metal ions,
are added with the ascorbic acid. Frequently the antioxidant effect of the
ascorbic acid will be evidenced later after an initial period of acceleration,
suggesting that the inhibitor is an intermediate in the oxidative decom-
position of the ascorbic acid.
When added to meats, ascorbic acid accelerates rancidity in some lots
of meat and inhibits it in others. Even in the same lot of meat it may
show an acceleration over the control after a short storage period but
inhibition after a longer period. It is probable that if i t could be introduced
into the meat along with a suitable phenolic antioxidant, it would always
inhibit oxidation. The difficulties of effecting good distribution of the
phenolic antioxidants have already been discussed. The combination of
ascorbic acid with certain artificial smokes has consistently given good
protection in frozen pork products (Watts and Faulkner, 1954). For
example, the following peroxide values were obtained in frozen pork
sausage after six months storage:
Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.1
0.1 % Ascorbic acid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70.5
0.02% Commercial liquid smoke preparation. . . . . . . . . 2 0 . 0
Ascorbic acid and smoke preparation. . . . . . . . . . . . . . . 5 . 9
30 BETTY M. WATTS
When antioxidants are introduced into meats with the aid of special
solvents or emulsifying agents, it is obviously necessary to test the added
dispersing agent not only for toxicity but also for effect on rancidity and
discoloration. Propylene glycol, for example, has been widely used in
experimental studies as a vehicle for introducing antioxidants into meats,
since it will dissolve both fat-soluble phenolic substances and also water-
soluble synergists. It is the solvent employed in several commercial anti-
oxidant mixtures (Bent2 et al., 1952). Klose et al. (1952) point out that
some commercial samples of propylene glycol accelerate rancidity. It
may also accelerate methemoglobin formation in frozen raw (but not
cured) meats (Watts and Faulkner, 1953). Furthermore, solution of
antioxidants in a small amount of propylene glycol does not insure
their uniform distribution in ground meats or curing brines, since
dilution of the propylene glycol by the curing brine or muscle juice
may cause separation of the less soluble antioxidants unless emulsify-
ing agents are used. Further work on suitable carriers for antioxidants
is badly needed.
VI. EFFECTOF VARIOUS MEAT CONSTITUENTS

AND ADDITIVES
ON RANCIDITYAND DISCOLORATION
1. Changes in p H
The pH of normal meat varies from approximately 5.2 to 6.6, owing
largely to differences in amount of glycogen available for transformation
into lactic acid a t the time of slaughter. Factors influencing the glycogen
content and pH have been reviewed by Bate-Smith (1949). Animals
which have been rested and well fed before slaughter have larger amounts
of glycogen stored in the muscle and consequently produce meat of a
lower pH.
The optimum pH for meat depends on the purpose for which it is
intended. Fresh meat should be a t the upper end of the normal pH range
if it is to be preserved by freezing, since the reaction between myoglobin
and unsaturated fat is inhibited in this range (see table V) and both dis-
coloration and also fat oxidation are retarded. In addition, drip losses are
minimized (Sair and Cook, 1938).
Lower pH values also directly accelerate oxidation of fresh meat pig-
ments in the absence of fat. Brooks (1938), Greenwood et al. (1940),
Watts and Lehmann (1952a), and others have observed the more rapid
formation of methemoglobin a t lower pH values. Winkler (1939a), study-
ing the relation between pH and light reflectance with several types of
meat (pork, beef, and mutton) obtained optimum pink color a t a pH of
approximately 6. In large cuts of meat, muscles which normally have the
highest pH are those which fade least on freezer storage (Watts et al.,
1948).
On the other hand, with pork intended for conventional curing treat-
ment, a low pH seems to be preferred. At elevated pH values conductivity
is low and penetration of curing salts into the muscle is impeded; con-
sequently the meat is more subject to bacterial spoilage (Callow, 1947;
Ingram, 1948; Gibbons and Rose, 1950). Even in homogenous hemo-
globin solutions and in comminuted meats where distribution of the
curing salts is not a serious problem, low pH accelerates color fixation
(Watts and Lehmann, 1952a, b). Attempts t o lower the pH beyond the
range of normal meat by use of acidified brines have not produced suc-
cessful cured products, owing to loss of nitric oxide from solution (Duis-
berg and Miller, 1943; Ingram, 1919b).
There are advantages to the use of higher pH values even with cured
meats provided that distribution of the curing salts can be attained and
the color developed by appropriate heat treatment. Urbain and Jensen
(1940) found that in solutions of pure, preformed nitric oxide hemoglobin,
oxidation to methemoglobin was less rapid a t high pH values, although
the pH had t o be raised above the normal limits for meat t o get much
protection. Brissey (1952) recommends the addition of alkaline phos-
phates t o the curing brines of hams to be sold as “boiled” ham, since
elevation of the pH during cure increased retention of juice during the
following cooking period. Extensive bacteriological tests in the Swift &z
Co. laboratories showed that under these conditions the elevated p H had
no detrimental effect on the keeping quality of the ham (Jensen, 1953).
2. Salts
No common meat additive has a more profound or more puzzling
effect on oxidative changes than sodium chloride. The accelerating effect
of salt on rancidity has been widely noted in a variety of foods, including
both fresh and cured meats. Lea (1939) has reviewed the earlier work.
White (1941b) and Gaddis (1952) found that the salt used in the curing
process accelerated rancidity in bacon (Fig. 6). A number of workers have
observed rapid development of rancidity and accompanying discoloration
in salted fresh pork (Dubois and Tressler, 1943; Wiesman and Ziemba,
1946; Watts and Peng, 1947a; Watts and Lehmann, 1952b).
Wills and Conochie (1946), in attempting t o explain the accelerating
effect of salt in butter, proposed a general theory for the mechanism of such
oxidations based on a reaction between fat-hydroperoxide and hydrogen
and chloride ions, resulting in the formation of chlorine;
lt-0-0-H + 2C1- + 2 H + + Clz + HzO + R-OH
32 BETTY M. WATTS
The free chlorine resulting from the reaction then brings about further
oxidation of the fat. This is the only theory so far proposed for the fut-
peroxidizing effect of salt.
The accelerating effect of salt on rancidity is evidenced only when
conditions are such that the salt is brought into contact with the fat over
a wide surface. Since salt is not adsorbed at the interface between fat and
Inside samples
OOUtSlde sampler
Uncured rgreen")
- 162%average salt content ,...."
- -- 2 87x average $a11 content
........ 4 00%average salt content ...'......', ,
.....", , '
..'...., ,
/
i'
..'..'
...'..'
_---
,-*-.
3
-. -. -. -.-. -*-
19
-
~ ~- L7x L
y
Weeks
FIG.6. Effect of salt on oxidation of bacon fat (Gaddis, 1952). Peroxide formation
in outside 4 mm. and inside fat from bacon containing different amounts of salt
during 3 weeks of curing a t 38" F. (3.3"C.) and subsequent freezer storage at 0" F.
( - 17.8' C.).
water, dilute salt solutions brought into contact with unsaturated fat
may actually retard oxidation (Chang and Watts, 1950). However, when
these same solutions are partially dried so that the salt is thrown out of
solution as a thin surface film, acceleration of rancidity is extremely rapid.
This may be an important factor in the rancidification of frozen and
dehydrated meats containing salt.
In addition to its effect on fat oxidation, salt also accelerates the
oxidation of hemoglobin and myoglobin to the ferric form, whether or not
fat is present. Coleman (1951) has summarized the rather extensive
literature on this point.
In cured meats, salt, like the hydrogen ion, apparently has just the
opposite effect, i.e., improves the cured color. Woodcock and White
(1943) observed this protective effect on Wiltshire bacon. Watts and

Lehmann (1952a, b) found that in the absence of nitrite, salt accelerated
methemoglobin formation, both in hemoglobin solutions and in meat, but
in the presence of nitrite and ascorbic acid, it accelerated the formation
of the bright red cured meat pigment. Salt, like acid, can bring about
a partial denaturation of globin. This would account for its opposite
effect on fresh and cured meat pigments.
Like hydrogen ions, salt has B number of other important functions in
cured meat which are outside the scope of this article. Callow (1947) has
summarized earlier studies on swelling and texture changes of muscle
tissue in salt. Ingram (1949a) and Hankins et al. (1950) discuss the salty
flavor in bacon. Bulman and Ayres (1952) have contributed to the already
extensive literature on the preservative effect of sodium chloride and
other curing salts.
The nitrate and nitrite used in cured meats appear t o have little
effect on rancidity in the small concentrations in which they are used and
at pH values within the normal range of meat (Lea, 1939; Watts and
Lehmann, 1952b).
3 . Metals
The earlier literature on the effect of metals on rancidity in fats has
been well covered by Lea (1939). The accelerating effect of copper and
iron on oxidation of fats and fat-containing foods is well established by a
wealth of experimental evidence. Other metals, particularly vanadium,
cobalt, and manganese, may also be strong pro-oxidants but are less
likely to contaminate foods. These metals can accelerate rancidity either
in the form of the solid metal in contact with the fat or as water- or fat-
soluble salts in heterogenous systems. Aluminum, tin, and zinc have little
catalytic activity under most conditions, although tin can act as a surface
catalyst for dry fats (Lea, 1946b).
Recent work on this subject has been concerned mainly with the use
in fats or fat-containing foods of various chemicals which can form pre-
cipitates or soluble nonionized complexes with metals and so diminish
their catalytic activity. Many of the well-known synergistic antioxidants
are metal complexing agents. Dutton et al. (1948) have studied the
effectiveness of various polycarboxylic acids and polyhydric alcohols on
improving the stability of soybean oil. Citric acid and sorbitol were
found to counteract the effect of pro-oxidant metals. More recently poly-
phosphates (Watts, 1950; Lehmann and Watts, 1951) and ethylene-
diaminetetraacetic acid (Trevor, 1949; Watts and Wong, 1951) have
been suggested as antioxidants for fats in heterogenous systems. Both are
metal sequestering agents. The phytates, which form both soluble and
34 BETTY M . WATTS
insoluble salts with metals (Cohee and Steffen, 1949) did not show anti-
oxidant activity in aqueous fat systems under the same conditions
(Lehmann and Watts, 1952).
It is not necessarily true that formation of nonionized complexes will
reduce the catalytic activity of metals on fat oxidation. The great increase
in fat oxidase activity of iron porphyrins over inorganic iron has already
been discussed. Still more active catalysts are formed by complexing iron
with phenanthroline (Simon et al., 1944).
The part played by metals in oxidative changes of meats is virtually
an unexplored field. Published information on metal contamination of
meats or on the effect of added metals or metal sequestering agents in
meat is very limited. Chang and Watts (1949) found that citric acid and
several polyphosphates which act as synergistic antioxidants in aqueous
fat systems did not retard fat oxidation when catalyzed by hemoglobin,
but were effective after the hemoglobin was coagulated by heat.
In the above experiments, the polyphosphates had an adverse effect
on color, accelerating oxidation of oxyhemoglobin. On the other hand,
Hall (1950) recommended the use of polyphosphatcs for preserving the
color in frankfurters. Weiss et al. (1953) found that, whereas copper and
iron accelerated the oxidation of oxyhemoglobin, these same metals as
well as zinc catalyzed the reduction of methemoglobin and color fixation
by ascorbic acid in the presence of nitrite. The addition of a metal
sequestering agent, ethylenediaminetetraacetic acid, interfered with
formation of nitric oxide hemoglobin.
4. Smoke
It has been known for many years that smoking of flesh foods in-
creases their resistance t o rancidity. Lea (1933), White (1941b, 1944),
Smith et al. (1945), Grant and White (1949), Gaddis (1952), and a number
of others have demonstrated the antioxidant effect of the smoke treat-
ment by appropriate chemical tests on the fat exposed to the smoke.
Jensen (1945) has reviewed the literature on the smoking process, the
chemical composition of wood smoke, and its penetration into meat.
Many classes of compounds have been found, including unidentified
phenolic substances. Presumably the antioxidants belong t o this latter
class, although this is by no means certain. The concentration of smoke
constituents is highest on the outside of a smoked ham; very little pene-
tration of the smoke to the center tissues takes place. Thus, while uncut
hams or sides of bacon are protected from oxidative changes, slices from
the smoked meat may be unprotected over much of their surface area.
However, Gaddis (1952) observed some protection of the inner portion
of sides of bacon as well as the outer four mm. strip (Fig. 6). Johnson
OXIDATIVE R A N C I D I T Y AND DISCOLORATION I N M E A T 35
(1930) recommends the smoking of individual slices of bacon for more

uniform and efficient distribution of the smoke.
Several artificial smoke flavors or “liquid smokes” are available on
the market. These differ in their method of manufacture and in the
quality and intensity of the smoke flavor, although all are derived from
thermal decomposition of hardwood. The effect of several of these prepa-
rations on fat oxidation has been tested (Watts and Faulkner, 1954) and
found t o vary from no protectioii or even a slight pro-oxidant effect a t
one extreme t o a strong antioxidant effect in concentrations of a few
hundredths per cent of the commercial preparation a t the other. The
antioxidant activity of the latter preparation thus approaches that of the
pure phenolic inhibitors. Discolorations with iron salts, such as occur
with gallates and other polyhydroxy phenols, do not take place with these
liquid smokes.
T o date, no imitation smoke flavor has been permitted by the Bureau
of Animal Industry in meat products being shipped across state bounda-
ries. Although these imitation smokes can certainly not perform the many
other functions of the conventional smoking process, such as partial
drying of the meat and surface gloss, there would seem t o be some ad-
vantage in their use as antioxidants as well as flavor constituents in
curing brines, particularly since, unlike many of the best phenolic in-
hibitors, they can be dispersed in the brine and thus more uniformly
distributed throughout the meat.
There is little mention in the literature of specific effects of smoke
ingredients on color of meat. However, the smoking process itself, since
heat treatment is involved, is important in developing cured meat color
(Table IV). The phenolic constituents of smoke, like other phenolic
inhibitors of fat oxidation, cause browning of fresh meat in freezer
storage.
5 . Spices
Many natural spices and other condiments have an antioxidant effect
when added t o fats. Sethi and Aggarwal (19.50) obtained protection of
ground nut oil with chilies, cinnamon, ginger, turmeric, nutmeg, black
pepper, cloves, and mace. Chipault et al. (1952) tested thirty-two ground
spices and found th at most of them had antioxidant properties with lard.
Rosemary and sage were particularly effective. However, when the
natural spices were extracted, fractions containing the odor components
did not have a strong antioxidant effect.
Dubois and Tressler (1943) and Atkinson el al. (1947) extended the
keeping time of frozen pork sausage by the use of spices. Sage and black
pepper were good antioxidants. It should be noted, however, th a t the
36 BETTY M. WATTS
antioxidant effect of the combined spices is not usually as great as the pro-
oxidant effect of the salt, so that untreated pork usually keeps better
than seasoned sausage.
VII. PHYSICAL
FACTORS
AFFECTINGOXIDATIVE
CHANQES
1 . Oxygen Tension
Niell and Hastings (1925) demonstrated more rapid conversion of
hemoglobin t o methemoglobin a t intermediate rather than a t very high
or very low oxygen tensions. This was true not only for the spontaneous
oxidation of laked blood corpuscles but also for the oxidation of hemo-
globin catalyzed by unsaturated fats.
Brooks (1929, 1935, 1936, 1938) has studied extensively the penetra-
tion of oxygen into muscle tissues and the effect of such penetration on
meat color. In early experiments Brooks (1929) devised a simple tech-
nique for following oxygen penetration and pigment oxidation. Slices of
tissue were placed on a glass slide between thin rods of glass and com-
pressed with a cover glass. A Zeiss microspectroscope allowed examination
of tissue pigments at different distances from the air tissue interface. A
steady state was achieved after oxygen had penetrated to a depth of
approximately 2 mm. Further penetration occurred only after long
standing and waa attributed to slow decrease in oxygen consumption by
the tissue. The interior of the slice remained completely reduced. Met-
hemoglobin formed only in the thin region of oxygen penetration and
most rapidly in the inner part of this region. The outer part was largely
oxyhemoglobin. Freezing and thawing the tissue had no effect on the rate
or depth of penetration but did accelerate methemoglobin formation in
the region of oxygen penetration.
Unlike hemoglobin and myoglobin, the corresponding nitric oxide
derivatives do not show an intermediate optimum oxygen pressure for
methemoglobin formation but oxidize more rapidly the higher the oxygen
pressure (Brooks, 1935; Urbain and Jensen, 1940). This might be ex-
pected, since dissociation of these derivatives would not be increased by
lowering the oxygen tension.
2. Light
Light accelerates all oxidative changes in meats, provided, of course,
that oxygen is available. Discolorations caused by exposure t o light
have become a particularly serious problem because of modern methods
of merchandising which require exposure of retail cuts in lighted display
cases. Cured meats are much more susceptible t o light discoloration than
fresh (Ramsbottom et al., 1951; Urbain and Ramsbottom, 1948). Appar-
ently light can bring about a dissociation of nitric oxide hemoglobin

similar t o its well-known effect on carbon monoxide hemoglobin. Display
case lighting does not significantly discolor fresh meats in periods u p t o
three days, but fading of cured, smoked, and table-ready meat is notice-
able in an hour under the same conditions. The several kinds of lighting
(incandescent, tungsten-filament, and fluorescent) all bring about equal
fading for the same time of exposure and light intensity. Display of meats
in the frozen state does not reduce their susceptibility t o fading. Ultra-
violet light appears t o have no greater effect than visible light of the same
intensity on oxidation of the cured meat pigments. On the other hand,
fresh meat pigments, which are not affected by visible light, discolor
when treated with ultraviolet light. Ultraviolet light is known t o bring
about protein denaturation (Haurowitz, 1950). This would account for
its oxidizing effect on hemoglobin and myoglobin.
The important facts concerning the effect of light on oxidation of fats
were for the most part established many years ago. The voluminous
literature has been critically evaluated by Lea (1939) and will not be
reviewed in detail here. Some of the more important conclusions t o be
drawn from this work are as follows: Shorter wavelengths of light are
more effective in causing oxidation than longer; ultraviolet light is a n
extremely potent accelerator of rancidity even in relatively stable fats.
For example, beef kidney fat which was not rancid after 1200 hours when
stored in the dark showed perceptible rancidity in 10 minutes when placed
in strong sunlight (containing a high proportion of ultraviolet rays). The
most noticeable effect of light is the elimination of the induction period,
although the rate of subsequent oxygen uptake can also be accelerated
by light of high intensity. Illumination of fats not only causes more rapid
peroxidation during the period when the fat is exposed t o light, but also
increases the rate of peroxide formation after the light is removed.
The use of ultraviolet light to retard growth of microorganisms in cold
storage rooms where meat is hung can cause rancidity, especially in pork.
Volz et al. (1949) found that skinned pork (but not beef) carcasses held
more than two days exposed t o ultraviolet light prior to freezing turned
rancid during subsequent freezer storage. This could be prevented b y
shading the carcasses or by leaving on the skins. On the other hand
Brady et al. (1949) found no direct relationship between irradiation of
pork and peroxide formation after freezing storage.
3. Temperature
As might be expected, the oxidation rate of both heme pigments and
fat of meat is accelerated with increasing temperature. Lea (1939) has
reviewed the earlier literature on temperature coefficients of fa t oxidation.
38 BETTY M . WATTS
In general, the oxidation rate of pure dry oils was approximately doubled
by a 10" C. (18" F.) rise in temperature in the absence of catalysts. In
the presence of light or metal catalysts the coefficient was much smaller.
There is no published information on the temperature coefficient of the
coupled reaction between hemoglobin and unsaturated fat. Numerous
more recent studies on frozen meats and poultry have emphasized the
importance of low storage temperatures in retarding rancidity (Cook and
White, 1939, 1941; Ramsbottom, 1947; Atkinson et al., 1947; Hall et al.,
1949; Klose et al., 1950; Palmer et al., 1953).
At the other extreme, high (cooking) temperatures undoubtedly
accelerate the oxidation of meat fats, but since peroxides and the various
aldehydes used to measure rancidity are unstable a t high temperatures,
results are likely t o be erratic. Chang and Watts (1952) found consider-
able peroxidation of the fat in meat and drippings when large cuts of
pork, beef, lamb, and poultry were roasted. On the other hand, the
peroxide values of body fat and drippings from frozen cuts of meat,
originally high, fell during the cooking period. Hanson et al. (1950) found
roasting to be much inferior to cooking in water as a preliminary treat-
ment for frozen poultry.
Although discolorations often accompany rancidity in frozen storage
studies such as those listed above, the lack of objective methods for
following such changes make it much more difficult to obtain even semi-
quantitative data on relative rates of discoloration a t various tempera-
tures. Cook and White (1941) measured reflected light from frozen pork
stored a t various temperatures. Although methemoglobin did not form
at the lower storage temperatures the meat was darker, possibly owing t o
desiccation. Urbain and Jensen (1940) have pointed out the high tem-
perature coefficient of nitric oxide hemoglobin oxidation. Solutions which
were completely oxidized in less than a day a t 37" C. (99" F.) required
thirteen days a t 10" C. (50' F.). On the other hand, Ramsbottom et al.
(1951) observed that frozen cured meats exposed t o light faded almost as
rapidly as refrigerated samples. It is possible that hemoglobin oxidation,
like fat oxidation, is less sensitive to temperature changes when light or
catalysts are present, but quantitative data are lacking.
4. Packaging Problems
The main considerations in the packaging of cured meat products are
the exclusion of oxygen and light. Any type of packaging which reduces
contact with oxygen retards both rancidity and discoloration. Whenever
tried, vacuum and gas packing have resulted in improved keeping
qualities. Storage of meat in atmospheres containing carbon dioxide has
been found in a number of studies to retard bacterial action as well as
OXID4TIVE RANCIDITY AND DISCOLORATION I N MEAT 39
oxidative changes (Brooks, 1933; Lea, 1939; Ogilvy and Ayres, 1951a, b ;
Kraft and Ayres, 1952). Ability of the packaging material to retain the
gas is, of course, an important consideration. Kraft and Ayres (1952)
found that frankfurters did not spoil until some time after the carbon
dioxide had been lost from the package. Urbain and Ramsbottom (1948)
and Ramsbottom et al. (1951), stress the desirability of excluding oxygen
in packaged cured meats.
Light exclusion interferes with the transparency of the package.
Transparent wrappings can be obtained which eliminate the ultraviolet
and shorter wavelengths of light mainly responsible for rancidity (Lea,
1939), but since fading of cured meats is accelerated by visible light rays,
the only wrappings which are very effective in retarding fading are so
deeply colored that they create sales resistance (Urbain and Ramsbottom,
1948; Ramsbottom et al., 1951). Since light accelerates oxidative changes
only in the presence of oxygen, vacuum or gas packing can eliminate the
detrimental effect of light.
With refrigerated fresh meats the problem is further complicated by
the fact that the purplish red of reduced hemoglobin is less desirable than
the oxygenated compound. Also lowered oxygen tension accelerates
oxidation to methemoglobin, and lowered p H caused by carbon dioxide
packaging further accelerates the oxidation. For these reasons i t has
generally been found preferable to use wrappings which allow some
passage of oxygen and to limit the use of carbon dioxide t o lower con-
centrations (Kraft and Ayres, 1952), even though these conditions arz not
optimum for preventing rancidity and other types of spoilage.
I n freezing preservation of both cured and fresh meats emphasis has
been placed on use of packages which are impermeable t o oxygen and
moisture. Various fresh meats remained palatable in freezer storage much
longer when packed in nitrogen (Steinberg e l al., 1949) or vacuum packed
(Hiner et al., 1951). Numerous workers have observed a correlation of
desiccation or ((freezerburn " with oxidative rancidity and discoloration
(Winkler, 1939b; Cook and White, 1939; Ramsbottom, 1947; and
many others). I n most of these studies it is impossible t o distinguish
between the effect of desiccation as such and the effect of oxidation. Stein-
berg et al. (1939) separated these two factors by storing a t different oxygen
levels with and without a desiccant. I n this study, color was adversely
affected by the desiccant, but palatability scores depended only upon
oxygen levels.
Impermeable wrappings can do much t o offset adverse effects of high
or fluctuating freezer temperatures. Winter et al. (1952) obtained highly
significant differences in frozen ground meat stored at - 18" C. (0"F.) as
compared t o fluctuating temperatures between this and - 10" C. (13" F.)
40 BETTY M. WATTS
when waxed freezer paper wrappings were used, but the fluctuating tem-
peratures had no adverse effect when the meat was wrapped in laminated
aluminum foil. Hanson et al. (1950) found the type of package t o be of
greater importance than the storage temperature for retention of quality
in precooked frozen poultry.
. VIII. SUMMARY
Measures for the control of oxidative rancidity in meats can begin
with the feeding of the meat animals. Rancidity of the fat in situ is
influenced by the characteristics of the fat itself and by the aqueous
medium with which it is intimately associated.
The inherent characteristics of the fat which have an effect on ran-
cidity have been established. They are: (1) the fatty acid composition,
particularly the number of active methylene groups between unsaturated
carbon atoms which occur in fatty acids having two or more double
bonds; and (2) the amount of natural antioxidant, specifically alpha
tocopherol, stored in the fat.
Within any one species of meat animal, these factors are determined
largely by ration. By greatly reducing the more highly unsaturated fats
in the ration, the body fat of hogs and poultry is rendered much less
susceptible t o rancidity. However, unsaturated fats are present in many
of the important stock feeds such as soybeans, alfalfa, and fish meal.
While it is not possible or desirable to eliminate such feeds, it might well
be feasible to select animals which have not received sources of highly
unsaturated fat when it is expected that their meat will be stored for a
period of time under conditions where rancidity might be a problem.
At present economic factors also preclude the feeding of tocophorol
concentrates to meat animals. Large doses of this vitamin in the feed do
increase fat stability, but most of it is excreted. Only a small fraction of
that fed is absorbed from the digestive tract and stored in the fat. It is
possible that deposition of either tocopherol or fatty acids or both may be
influenced by the feeding of substances which can affect fat metabolism,
but such studies are too few and in too early a stage to warrant any con-
clusions a t present.
Accelerators and inhibitors of rancidity in the aqueous muscle juice are
undoubtedly also of importance, but research has not advanced suffi-
ciently to illuminate more than the fringes of the field. Hemoglobin and
myoglobin are known t o bring about very rapid oxidation of fat along
with their own self-destruction. Thus contact of these pigments with
unsaturated fat in the presence of oxygen results in rancidity and dis-
coloration. This reaction is probably of particular importance in ground
meats preserved by freezing and in meats in cure, where the heme pig-
ments can dissolve in the brine. It probably does not occur in cooked
meats where the pigments are denatured. The reaction is much slower in
meats in the higher range of normal pH.
A number of normal constituents of muscle juice, including various
amino acids, biotin, niacin, and ascorbic acid, can inhibit rancidity when
added t o pure fats containing phenolic inhibitors. With the exception of
ascorbic acid, there is no information on the effect of these substances on
fats in close contact with an aqueous phase or on fat oxidations catalyzed
by hemoglobin. Nor is it known whether the limited amounts of these
substances which actually occur in muscle juice, in conjunction with the
limited amount of tocopherol naturally present in the fat, can exert a
protective influence.
Salt, as used in curing or when added t o frozen ground meats, defi-
nitely accelerates rancidity. Smoke constituents, including some artificial
smokes, and many spices, inhibit it. Newer phenolic inhibitors are very
effective in protecting fat even in the presence of a n aqueous phase con-
taining hemoglobin and could possibly effect the same protection in
meats provided that uniform distribution could be attained in the meat.
However, this is difficult, since their solubility is often too limited t o
allow them t o be incorporated readily in ground meats, curing brines, or
cooking waters. Various carriers and surface coatings designed t o assist
in getting the necessary antioxidant distribution have been described
but the problem is far from solution.
Fading and discoloration of meats is due to oxidation of the normal
pigments hemoglobin and myoglobin of fresh meat or their nitric oxide
derivatives in cured meat, all ferrous heme compounds. The oxidation
products are the same for fresh and cured meat pigments. They consist
usually of the ferric pigments, methemoglobin and metmyoglobin, brown
in color, and less frequently green or faded decomposition products of the
porphyrin ring.
Although both fresh and cured meat pigments oxidize along the same
pathways and t o the same brown or green end products, changes in the
physical and chemical environment affect the oxidation of these pigments
very differently. Some of these differences between the two pigments
may be traced t o variations in their dissociation. The available evidence
indicates that both the oxymyoglobin of fresh meat and the nitric oxide
myoglobin of cured meats must dissociate t o myoglobin before oxidation
to the brown ferric metmyoglobin takes place. The dissociation of OXY-
hemoglobin increases with decreasing oxygen tension. Fresh meat pig-
ments therefore turn brown more readily a t oxygen tensions considerably
below that of air a t atmospheric pressure. In contrast, dissociation of
nitric oxide hemoglobin does not increase a t lower oxygen tension, SO that
42 BETTY M. WATTS
rate of oxidation of cured meat pigment is progressively increased with

increasing oxygen supply. Again, visible light fades cured meats more
rapidly than fresh, probably because of a dissociating effect on nitric
oxide hemoglobin similar t o the known effect of light on carbon monoxide
hemoglobin.
Furthermore, any treatment which tends to denature the globin, even
partially or reversibly, serves to weaken the bonds between heme and
globin. Without these stabilizing bonds to native globin, heme loses the
ability t o combine reversibly with oxygen to form the bright red oxy-
genated compound and instead is oxidized very rapidly t o the brown
ferric form. On the other hand, the bright red color of nitric oxide hemo-
globin is not lost even by complete and irreversible denaturation of the
globin. I n fact, globin denaturation seems to accelerate the formation of
cured meat pigments.
Thus, as is well known from common experience, the heating of meat
to temperatures sufficiently high to denature the globin results in the
brown ferric hemochromogen of cooked meat. The heat-denatured nitric
oxide hemochromogen, on the other hand, retains the red color of the
ferrous compound, and some form of heat treatment is widely practiced
in the manufacture of cured meats. Likewise increased acidity, salt,
certain metals, freezing, and probably a variety of other environmental
conditions accelerate methemoglobin formation in fresh meat pigments,
but these factors either have no effect on cured meat pigments or actually
improve color fixation.
Direct attack on the porphyrin ring to produce green derivatives can
probably occur with any of the heme compounds, but again the optimum
conditions for producing such discolorations are not the same for fresh
and cured meat pigments. For example, free hydrogen peroxide (usually
of bacterial origin) has much less effect on fresh than on cured meat-
probably because of the catalase activity of the fresh meat-and hydro-
gen donors such as ascorbic acid protect nitric oxide myoglobin but accel-
erate oxidation of oxymyoglobin.
REFERENCES
Abramson, H. 1949. The oxidation of unsaturated fatty acids in normal and scorbutic
guinea pigs. J . Biol. Chem. 178, 179.
Atkinson, I., Cecil, S. R., Woodroof, J. G., and Shelor, E. 1947. Extending keeping
quality of frozen pork sausage. Food Inds. 19, 1198, 1367.
Bailey, A. E. 1951. Industrial Oil and Fat Products. Interscience Publishers, New
York.
Banks, A. 1944. Method for studying the effect of antioxidants on the oxidation of
aqueous suspensions of unsaturated fatty acids. J . SOC.Chem. Ind. 63, 8.
Barnard, R. D. 1937. Reactions of nitrite with hemoglobin derivatives. J . Biol. Chem.
120, 177.
Barnes, R. H., Lundberg, W. O., Hanson, H. T., and Burr, G. 0. 1943. Effect of certain
dietary ingredients on the keeping quality of body fat. J . Biol. Chem. 149, 313.
Barron, E.S. G., and Lyman, C. M. 1938. Studies on biological oxidations. X. Oxida-
tion of unsaturated fatty acids with blood hemin and hemochromogens as
catalysts. J. Biol. Chem. 123, 229.
Bate-Smith, E. C. 1948. The physiology and chemistry of rigor mortis, with special
reference t o the aging of beef. Advances in Food Research 1, 1.
Bauernfeind, J. C. 1953. The use of ascorbic acid in processing foods. Advances in
Food Research 4, 359.
Beadle, B. W., Wilder, 0 . H. M., and Kraybill, H. R. 1948. The deposition of trienoic
fatty acids in the fats of the pig and the rat. J. Biol. Chem. 176, 221.
Bentz, R.W., O’Grady, T. J., and Wright, S. B. 1952. Antioxidants and food preserva-
tion. Food Technol. 6, 302.
Bickoff, E. M., Coppinger, G. M., Livingston, A. L., and Campbell, T. W. 1952.
Pyrogallol derivatives as antioxidants for carotene. J . Am. Oil Chemists’ SOC.
29, 51.
Black, H. C. 1950. Gum guaiac treatment. U. S. patent 2,529,446.
Bodansky, 0 . 1951. Methemoglobinemia and methemoglobin producing compounds.
Pharmacol. Revs. 3, 144.
Boehm, E., and Williams, R. 1943. Study of the inhibiting actions of propyl gallate
and certain other trihydric phenols on the autoxidation of animal and vegetable
oils. Quart. J . Pharm. and Pharmacol. 16, 232,
Bowen, W. J. 1949. The absorption spectra and extinction coefficients of myoglobin.
J . Biol. Chem. 179, 235.
Brady, D.E., Smith, F. H., and Tucker, L. N. 1946. Control of rancidity in soybean-
fed pork. J . Animal Sci. 6, 358.
Brady, D. E., Smith, F. H., Tucker, L. N., and Blumer, T. N. 1949. Ultraviolet
irradiation of beef and pork for low temperature storage. Food Research 14,
183.
Bratzler, J. W., Loosli, J. K., Krukovsky, V. N., and Maynard, L. A. 1950. Effect of
the dietary level of tocopherols on their metabolism in swine. J . Nutrition 42, 50.
Brissey, G. E. 1952. Preparing cooked cured meats. U. S. patent 2,596,067.
Brooks, J. 1929. Post-mortem formation of methemoglobin in red muscle. Biochem. J .
23, 1391.
Brooks, J. 1933. The effect of carbon dioxide on the colour changes or bloom of lean
meat. J . SOC.Chem. Ind. 62, 17T.
Brooks, J. 1935. The oxidation of hemoglobin to methemoglobin of oxygen. 11. The
relation between the rate of oxidation and the partial pressure of oxygen. Proc.
Roy. Soc. (London) B118, 560.
Brooks, J. 1936. Oxygen uptake of pork and bacon. A factor in the production and
preservation of the colour of bacon. J . SOC.Chem. Ind. 66, 12T.
Brooks, J. 1938. Color of meat. Food Research 3, 75.
Bulman, C., and Ayres, J. C. 1952. Preservative effect of various concentrations of
curing salts in comminuted pork. Food Technol. 6, 255.
Burr, G.O., and Burr, M. M. 1929. A new deficiency disease produced by the rigid
exclusion of fat from the diet. J . Biol. Chem. 82, 345.
Burr, G. O., and Burr, M. M. 1930. On the nature and role of the fatty acids essential
in nutrition. J . Biol. Chem. 86, 587.
Calkins, V. P. 1947. The mechanism of the antioxygenic synergism of quinones and
quinols with phosphoric acid and other acids in fat systems. J . A m . Chem. SOC.
69, 384.
44 BETTY M. WATTS
Calkins, V. P., and Mattill, H. A. 1944. Kinetics of the antioxygenic synergism of

quinones with ascorbic acid in fat systems. J. Am. Chem. SOC.66,239.
Callow, E.H. 1947. The action of salts and other substances used in the curing of
bacon and ham. Brit. J. Nutrition 1, 269.
Chang, I., and Watts, B. M. 1949. Antioxidants in the hemoglobin catalyzed oxida-
tion of unsaturated fats. Food Technol. 3,332. .
Chang, I., and Watts, B. M. 1950. Some effects of salt and moisture on rancidity in
fats. Food Research 16, 1.
Chang, I., and Watts, B. M. 1952. The fatty acid content of meat and poultry before
and after cooking. J. Am. Oil. Chemists’ SOC.2B, 334.
Chipault, J. R., Mizuno, G. R., Hawkins, J. M., and Lundberg, W. 0. 1952. The anti-
oxidant properties of natural spices. Food Research 17, 46.
Clausen, D. F., Lundberg, W. O., and Burr, G. 0. 1947. Some effects of amino acids
and certain other substances on lard containing phenolic antioxidants. J. Am. Oil
Chemists’ SOC.24, 403.
Cohee, R. F., Jr., and Steffen, G. 1949. Takes heavy metals out of acid foods. Food
Inds. 21, 50.
Coleman, H. M. 1951. The mechanism of meat-pigment oxidation. The effect of
solutes on the hemoglobin-oxygen equilibrium. Food Research 16, 222.
Coleman, H. M., and Steffen, A. H. 1949. Process for treating animal materials. U. S.
patent 2,491,646.
Coleman, H. M., Steffen, A. H., and Hopkins, E. W. 1951. Process for treating animal
materials. U. S. patent 2,541,572.
Cook, W. H., and White, W. H. 1939. Frozen storage of poultry. 111. Peroxide oxygen
and free fatty acid formation. Food Research 4, 433.
Cook, W. H., and White, W. H. 1941. Effect of temperature and humidity on color of
lean and development of rancidity in fat of pork during frozen storage. Can. J.
Research lQD,53.
Cornwell, R. T. K. 1951. Protective coating composition for hams. U. S. patent
2,558,042.
Cox, W. W., and Wendel, W. B. 1942. The normal rate of reduction of methemoglobin
in dogs. J. Biol. Chem. 143, 331.
Criddle, J. E., and Morgan, A. F. 1951. Effect of tocopherol feeding on the coniposition
of turkey tissues. PTOC. SOC.Exptl. Biol. Med. 78, 41.
Cruickshank, E.M. 1934. Studies on fat metabolism in the fowl. Biochem. J . 28,265.
Davis, L. L., and Bywaters, J. H. 1951. Preliminary report on prolonging storage life
of frozen broilers. Unpublished data from the Virginia Agricultural Experiment
Station.
Doegey, J. L. 1943. Method for incorporating gum guaiac in fats. U. S. patent 2,308,-
912.
Dubois, C. W., and Tressler, D. K. 1943. Seasonings, their effect on maintenance
of quality in storage of frozen ground pork and beef. Proc. Inst. Food Technol.
p. 202.
Dugan, L. R., Hoffert, E., Blumer, G. P., Dabkiewicz, I., and Kraybill, H. R. 1951.
The antioxidant behavior of the isomers of butylated hydroxyanisole. J. Am. oil
Chemists’ SOC.28, 493.
Duisberg, P. C., and Miller, R. C. 1943. Relation of hydrogen ion concentration to
color developed in cured pork. Food Research 8, 78.
Dutton, H.J., Schwab, A. W., Moser, H. A., and Cowan, J. C. 1948. The flavor prob-
lem of soybean oil. IV. Structure of compounds counteracting the effect of pro-
oxidant metals. J. Am. Oil Chemists’ SOC.26, 385.
Eckey, E. W. 1934. Process for stabilizing the flavor and odor of fatty materials.
U. S. patent 1,982,907.
Eckey, E. W. 1935. Composition of fatty matter and process of stabilizing same.
U. S. patent 1,993,152.
Evans, E. I. 1935. Antioxidant properties of vegetable lecithin. J . Ind. Eng. Chem.
27, 329.
Fanelli, A. R. 1949. Simultaneous spectrophotometric determination of myoglobin
and hemoglobin. Bull. soc. chim. biol. 31, 457.
Filer, L. J., Mattil, K. F., and Longenecker, H. E. 1944. Antioxidant losses during the
induction period of fat oxidation. Oil & Soap 21, 289.
Fishberg, E. H. 1948. Excretion of benzoquinoneacetic acid in hypovitaminosis C.
J . Biol. Chem. 172, 155.
Fonyo, A. 1950. Treatment of unrendered fat with dihydronorguaiaretic acid. U. S.
patent 2,535,910.
Foulkes, E. C. and Lemberg, R. 1949. Formation of choleglobin and the role of catalase
in the erythrocyte. Proc. Roy. SOC.(London) B136, 435.
French, R. B., Olcott, H. S., and Mattill, H. A. 1935. Antioxidants and the autoxida-
tion of fats. 111. Ind. Eng. Chem. 27, 724.
Gaddis, A. M. 1952. Effect of pure salt on the oxidation of bacon in freezer storage.
Food Teehnol. 6, 294.
Gibbons, N. E., and Rose, D. 1950. Effect of ante-mortem treatment of pigs on the
quality of Wiltshire bacon. Can. J. Research 28F,438.
Gibson, Q. H. 1943. The reduction of methaemoglobin by ascorbic acid. Biochem. J .
37, 615.
Gibson, Q. H. 1948. The reduction of methaemoglobin in red blood cells andstudies
on the cause of idiopathic methaemoglobinaemia. Biochem. J . 42, 13.
Golumbic, C. 1941. Antioxidants and the autoxidation of Fats. XII. The antioxidant
properties of tocopherols, hydroxychromans, hydroxycoumaranes and related
compounds. J . Am. Chem. SOC.63, 1142.
Golumbic, C. 1942. Antioxidant properties of gallic acid and allied compounds. Oil &
Soap 19, 144.
Golumbic, C. 1943. Autoxidative behaviour of vegetable and animal fats. Oil & Soap
20, 105.
Golumbic, C., and Mattill, H. A. 1941. Antioxidants and the antoxidation oftats.
XIII. The antioxygenic action of ascorbic acid in association with tocopherols,
hydroquinones and related compounds. J. Am. Chem. SOC.63, 1279.
Granick, S. 1949. The chemistry and functioning of the mammalian erythrocyte.
Blood 4, 404.
Granick, S., and Gilder, H. 1947. Distribution, structure and properties of the tetra-
pyrroles. Advances i n Enzymology 7 , 305.
Grant, G. A., and White, W. H. 1949. Keeping quality enhanced by curing and
smoking poultry. Food in Canada 9 (5), 14; (8),27.
Greenberg, L. A., Lester, D., and Haggard, H. W. 1943. The reaction of hemoglobin
with nitrite. J. Biol. Chem. 161, 665.
Greenwood, D. A., Lewis, W. L., Urbain, W. M., and Jensen, L. B. 1940. The heme
pigments of cured meats. IV. Role of sugars in color of cured meats. Food Re-
search 6, 625.
Gunstone, F. D., and Hilditch, T. P. 1945. The union of gaseous oxygen with methyl
oleate, linoleate, and linolenate. J. Chem. SOC.2, 836.
Gutman, H. R., Jandorf, 13. J., and Bodansky, 0. 1947. The role of pyridine nucleo-
tides in the reduction of methemoglobin. J. Biol. Chem. 169, 145.
46 BETTY M. WATTS
Haldane, J. 1901. The red color of salted meat. J . Hyg. 1, 115.

Hall, G. 0. 1950. Curing of meat to avoid undesirable color change. U. S. patent
2,513,094.
Hall, J., Lowe, B., Kalen, J., Westerman, B. D., Mackintosh, D. L., and Vail, G. E.
1949. Keep the temperature low when storing pork. Refrig. Eng., 67,247.
Hankins, 0. G., Sulzbacher, W. L., Kauffman, W. R., and Mayo, M. E. 1950. Factors
affecting the keeping quality of bacon. Food Technol. 4, 33.
Hanson, H. L., Winegarden, H. M., Horton, M. B., and Lineweaver, H. 1950. Prepa-
ration and storage of frozen cooked poultry and vegetables. Food Technol. 4,430.
Hanson, H. T., Barnes, R. H., Lundberg, W. O., and Burr, G. 0. 1944. The deposition
of antioxidants in the abdominal fat depots. J . Biol. Chem. 156, 673.
Harper, R.H. 1953. Swift and Company, private communication.
Haurowitz, F., and Schwerin, P. 1941. Haminkatalysen in Grenzfliichenfilm zwischen
Wasser und Olphase. Enzymologia 9, 193.
Haurowitz, F., Schwerin, P., and Yenson, M. M. 1941. Destruction of hemin and
hemoglobin by the action of unsaturated fatty acids and oxygen. J . Biol. Chem.
140,353.
Haurowitz, F. 1950. “Chemistry and Biology of Proteins.” Academic Press, New
York.
Hilditch, T. P. 1944. Some factors affecting control of oxidative rancidity. Chemistry
d% Industry 8, 67.
Hills, G. L., and Conochie, J. 1946. The mechanism of the oxidant effects of commer-
cial salt and water in butterfat. J . Sci. Znd. Research (Australia) 19, 1.
Hiner, R. L., Caddis, A. M., and Hankins, 0. G. 1951. Effect of methods of protection
on palatability of freezer stored meat. Food Technol. 6,223.
Hite, J. P., Kloxin, S. E., and Kummerow, F. A. 1949. Fat rancidity in eviscerated
poultry. IV. The effect of variations in dietary fat, ethanolamine and choliric on
the characteristics of the fat extracted from turkeys. Poultry Sci. 28, 249.
Holden, H. F. 1936. Methemoglobin, a spectrophotometric study. Australian J . Ezptl.
Biol. Med. Sci. 14, 291.
Hopkins, E. W., Jendryaszek, L. J., and Coleman, H. W. 1950. Preparation of meat
products. U. S. patent 2,521,579.
Hostetler, E. H., and Halverson, J. 0. 1940. Feeding soybeans to pigs. N . Carolina
Agr. Expt. Sla. Tech. Bull. No. 63.
Hove, E. L., and Harris, P. L. 1951. Note on the linoleic acid and tocopherol relation-
ship in fats and oils. J . Am.. Oil Chemists’ SOC.28, 405.
Hove, E. L., and Hove, Z. 1944a. A method for estimating total fat-soluble anti-
oxidants based on the relation between fat peroxides and carotene destruction.
J . Biol. Chem. 166,611.
Hove, E. L. and Hove, Z. 1944b. Effect of temperature on relative antioxidant ac-
tivity of alpha, beta and gamma tocopherols~andof gossypol. J . Biol. Chem. 166,
623.
Hughes, A. H., and Rideal, E. K. 1933. On the rate of oxidation of monolayers of un-
saturated fatty acids. Proc. Roy. SOC.(London) A140, 253.
Husaini, S. A., Deatherage, F. E., and Kunkle, L. E. 1950. Studies on meat. 11.
Observations on relation of biochemical factors to changes in tenderness. Food
Technol. 4, 366.
Ingram, M. 1948. Muscular fatigue, pII and bacterial proliferation in meat. Ann.
Znst. Pasteur 76, 139.
Ingram, M.1949a. Salty flavor in bacon. J . SOC.Chem. Znd. 68,356.
Ingram, M. 194913. Curing of bacon with acid brines. Food Manuf. 24, 201.
Jensen, L. B. 1935. Meat treating method. U. S. patent 2,002,146.
Jensen, L. B. 1945. “ Microbiology of meats.” The Garrard Press, Champaign, Illinois,
2nd Ed.
Jensen, L. B. 1953. “Microbiology of Meats.” The Garrard Press, Champaign, Illinois,
3rd Ed. (In press).
Jensen, L. B. 1949. “Meat and Meat Foods.” The Ronald Press Company, New York.
Jensen, L. B., and Urbain, W. M. 1938a. Bacteriology of green discolorations in
meats and spectrophotometric characteristics of the pigments involved. Food
Research 1, 283.
Jensen, L. B., and Urbain, W. M. 193613. A delicate test for blood pigments. Food
Research 1, 275.
Johnson, 0.C. 1950. Methodof preparing and packaging bacon. U. S. patent 2,528,832.
Keilin, D., and Hartree, E. F. 1937. Reaction of nitric oxide with hemoglobin and
methemoglobin. Nature 139, 548.
Keilin, D., and Hartree, E. F. 1950. Reaction of methemoglobin with hydrogen per-
oxide. Nature 166, 513.
Kiese, M., and Kaeske, H. 1942. Verbindungen des Muskelhaernoglobins. Biochem. 2.
312, 121.
Kiese, M. 1943. Die Reduktion des Hamiglobins. Biochem. 2. 316, 264.
Kikuchi, G. 1950. Studies on the action of oxygen and hydrogen peroxide on hemin
derivatives. I. Process of formation of verdoheinochromogen. J . Japan. Biochem.
SOC.22, 213.
Klose, A. A., Hanson, H. L., and Lineweaver, H. 1950. The freezing preservation of
turkey meat steaks. Food Technol. 4, 71.
Klose, A. A., Mecchi, E. P. and Hanson, H. I,. 1952. Use of antioxidants in the frozen
storage of turkeys. Food Technol. 6, 308.
Klose, A. A,, Mecchi, E. P., Hanson, H. L., and Lineweaver, H. 1951. The role of
dietary fat in the quality of fresh and frozen storage turkeys. J . Am. Oil Chemists’
SOC.28, 162.
Komarik, S. L., and Hall, L. A. 1951. Curing process for bacon. U. S. patent 2,553,533.
Kraft, A. A., and Wanderstock, J. J. 1950. Meat color problem is closer to solution.
Food Inds. 22, 65.
Kraft, A. A., and Ayres, J. C. 1952. Post-mortem changes in stored meat. IV. Effect of
packaging materials on keeping quality of self-service meats. Food Technol. 6, 8.
Kraybill, H. R., and Beadle, B. W. 1948. Oxidation inhibitor for fats and oils. U. S.
patent 2,451,748.
Kraybill, H. R., Dugan, L. R., Beadle, B. W., Vibrans, F. C., Schwartz, V., and
Resabek, H. 1949. Butylated hydroxyanisole as an antioxidant for animal fats.
J . Am. Oil Chemists’ SOC.26, 449.
Krukovsky, V. N. 1949. The influence of tocopherols and cod liver oil on stability of
milk. J . Dairy Sci. 32, 198.
Kummerow, F. A., Vail, G. E., Conrad, R. M., and Avery, T. B. 1948. Fat rancidity
in eviscerated poultry. I. The effect of variations in diet on the cold storage life of
immature turkeys. Poultry Sci. 27, 635.
Kummerow, F. A,, Otto, E., Jacobson, G., and Randolph, P. 1949. The effect of anti-
oxidants on the metabolism of linolenic acid by acrodynic rats. 4th Conf. on Biol.
Antioxidants, Josiah Macy, Jr. Foundation, New York.
Lange, W. 1950. Cholesterol, phytosterol and tocopherol content of food products and
animal tissues. J . Am. Oil Chemists’ SOC. 27, 414.
Idea, C. H. 1933. The effect of smoking and influence of atmospheric humidity on
keeping properties of bacon. J . SOC.Chenz. I n d . 62, 57T.
48 BETTY M. WATTS
Lea, C. €I. 1937. Influence of tissue oxidases on rancidity. Oxidation of the fat of
bacon. J. SOC.Chern. Ind. 66 (Pt. 2), 3761’.
Lea, C. H. 1939. “Rancidity in Edible Fats.” Chemical Publishing Co., New York.
Lea, C. H. 1944. Experiments with antioxidants in dried pork. J . SOC.Chem. Ind. 63,55.
Lea, C. H. 1945. The determination of the peroxide value of edible fats and oils: the
influence of atmospheric oxygen in the Chapman and McFarlane method. J . SOC.
Chem. Ind. 64, 106.
Lea, C. H. 1940a. The determination of the peroxide values of edible fats and oils: the
iodometric method. J . SOC.Chem. Ind. 66, 286.
LeaA C. H. 1946b. A note on the effect of tinplate and of lacquered surfaces on the
oxidative deterioration of butterfat. J . SOC.Chem. Ind. 66, 136.
Lehmann, B. T., and Watts, B. M. 1951. Antioxidants in aqueous fat systems.
Lehmann, B. T., and Watts, B. M. 1952. Unpublished data. Syracuse University,
Syracuse, New York.
Lemberg, R., Legge, J. W., and Lockwood, W. H. 1939. Coupled oxidation of ascorbic
acid and haemoglobin. Biochem. J . 33, 754.
Lemberg, R., Legge, J. W., and Lockwood, W. H. 1941. Coupled oxidation of ascorbic
acid and haemoglobin. 11. Formation and properties of choleglobin. Biochem. J .
36, 328. 111. Quantitative studies on choleglobin formation. ibid. p. 339.
Lemberg, R., and Legge, J. W. 1950. Pyrrole Pigments. Ann. Rev. Biochem. 19, 431.
Lemberg, R., and Legge, J. W. 1949. “Hematin Compounds and Bile Pigments.”
Interscience Publishers, New York.
Lindsey, F. A., Jr., and Maxwell, W. T. 1949. Process for stabilizing edible fatty oil.
U. S. patent 2,486,424.
Lineweaver, H., Anderson, J. D., and Hanson, H. L. 1952. Effect of antioxidants on
rancidity development in frozen creamed turkey. Food Technol. 6 , 1.
Lips, A., Chapman, R. A., and McFarlane, W. D. 1943. Application of the ferric thio-
cyanate method to the determination of incipient rancidity in fats and oils. Oil &
Soap 20, 240.
Lovern, J. A. 1946. Some aspects of recent British studies on antioxidants. Oil & Soap
23, 40.
Lugg, J. W. H. 1950. Note on the effects of nitrate and nitrite upon ascorbic acid in
acid solutions. Med. J . Malaya 6 (2), 140.
Lundberg, W. 0. 1947. A survey of present knowledge, researches and practices in the
United States concerning the stabilization of fats. Hormel Inst. Univ. Minn.
Publ. No. 20.
Lundberg, W. O., Barnes, R. H., Clausen, M., and Burr, G. 0. 1944a. The deposition
and storage of alpha-tocopherol in abdominal fats. J . Biol. Chem. 163, 265.
Lundberg, W. O., Dockstader, W. B., and Halvorson, H. 0. 1947. The kinetics of the
oxidation of several antioxidants in oxidizing fats. J . A m . Oil Chemists’ SOC.24,89.
Lundberg, W. O., Halvorson, H. O., and Burr, G. 0. 1944b. The antioxidant properties
of nordihydroguaiaretic acid. Oil & Soap 21, 33.
Mackey, A. E., Oliver, A. W., and Fang, S. C. 1952. Chemical constituents, physical
properties and palatability of frozen pork. Food Research 17, 409.
Mahon, J. H., and Chapman, R. A. 1953. The relative rates of destruction of propyl
gallate and butylated hydroxyanisole in oxidizing lard. J . Am. Oil Chemists’ SOC.
30, 34.
Major, R., and Watts, B. M. 1948. The relation of fed and injected tocopherols to
development of rancidity in the stored meat and utilization of carotene by the
rabbit. J . Nutrition 35, 103.
Millikan, G. A. 1939. Muscle hemoglobin. Physiol. Revs. 19, 503.
Morgan, A. F., and Watts, B. M. 1948. Dehydrated meat scrapple and method of
making the same. U. S. patent 2,457,063.
Nagy, J. J., Vibrans, F. C., and Kraybill, H. R. 1944. An application of the Barcroft-
Warburg apparatus to the study of antioxidants in fats. Oil & Soap 21, 349.
Nagy, J. J., Beadle, B. W., and Kraybill, H. R. 1945. Use of dried air in the active
oxygen method of determining relative stabilities of fats. Oil & Soap 22, 123.
Newton, R. C., and Grettie, D. P. 1933. Antioxidant for fats and oils. U. S. patent
1,903,126.
Niell, J. M., and Hastings, A. B. 1925. The influence of the tension of molecular
oxygen upon certain oxidations of hemoglobin. J. Biol. Chem. 63, 479.
Niven, C.F., Jr. 1951. Influence of microbes upon the color of meats. Circ., Am. Meat
Inst. Found. No. 2.
Niven, C. F., Jr., Castallani, A. G., and Allanson, V. 1949. A study of the lactic acid
bacteria that cause surface discolorations of sausages. J . Bucteriol. 68, 633.
Norris, F. A. 1949. Stabilization of oleaginous materials. U. S. patent 2,462,664.
Ogilvy, W. S., and Ayres, J. C. 1951a. Post-mortem changes in stored meats. 11. The
effect of atmospheres containing rarbon dioxide in prolonging the storage life of
cut-up chicken. Food Technol. 6 , 97.
Ogilvy, W. S., and Ayres, J. C. 1951b. Post-mortem changes in stored meats. 111. The
effect of atmospheres containing carbon dioxide on prolonging the storage of
frankfurters. Food Technol. 6 , 300.
Olcott, H. S., and Emerson, 0. H. 1937. Antioxidant properties of the tocopherols.
J. Am. Chem. SOC.69, 1008.
Olcott, H. S., and Mattill, H. A. 1936. Antioxidants and the autoxidation of fats. Oil
& Soap 13, 98.
O'Leary, D.K. 1946. Stabilization of fats and oils. U. S. patent 2,397,976.
Palmer, A. Z., Brady, D. E., Naumann, H. D., and Tucker, L. N. 1953. Deterioration
in frozen pork as related to fat composition and storage treatments. Food Technol.
7 , 90.
Patton, S., Keeney, M., and Kurtz, G. W. 1951. Compounds producing the Kreis color
reaction with particular reference to oxidized milk fat. J. Am. Oil Chemists' SOC.
28, 391.
Pennell, R. B., and Smith, W. E. 1949. Preparation of stabilized solutions of hemo-
globin. Blood 4, 380.
Pool, M. F., and Prater, A. N. 1945. A modified Kreis test suitable for photocolor-
imetry. Oil & Soap 22, 215.
Pool, M. F., and Klose, A. A. 1951. Estimation of monocarbonyl compounds in rancid
foods. J . A m . Oil Chemists' SOC.28, 215.
Ramsay, W. N. M. 1944. Ferrihaemoglobinaemia in man and the horse. Biochem. J .
38, 470.
Ramsbottom, J. M. 1947. Freezer storage effect on fresh meat quality. Refrig. Eng.
63, 19.
Ramsbottom, J. M., Goeser, P. A., and Schultz, H. W. 1951. How light discolors meat:
What to do about it. Food Inds. 23, 120.
Reiser, R. 1949. Peroxidizing and carotene bleaching substances in bacon adipose
tissue. J . Am. Oil Chemists' SOC.26, 116.
Iiiedesel, M., and Watts, B. M. 1952. Unpublished data, Florida State University,
Tallahassee.
Iliemenschneider, R. W. 1947. Oxidative rancidity and the use of antioxidants. Trans.
Am. Assoc. Cereal Chemists 6 , 50.
50 BETTY M. WATTS
Robinson, M. E. 1924. Hemoglobin and methemoglobin as oxidative catalysts.

Biochem. J . 18, 255.
Rockwood, B. N., Ramsbottom, J. M., and Mehlenbacker, V. C. 1947. Preparation of
animal tissue fats for determination of peroxides and free fatty acids. Anal Chem.
19, 853.
Rosenwald, R. H., and Chenicek, J. A. 1951. Alkyl hydroxyanisoles as antioxidants.
Sair, L., and Cook, W. H. 1938. Relation of pH to drip formation in meat. Can. J .
Research 16D, 225.
Scarborough, D. A., and Watts, B. M. 1949. The prooxidant effect of ascorbic acid
and cysteine in aqueous fat systems. Food Technol. 3, 152.
Schreiber, M. L., Vail, G. E., Conrad, R. M., and Payne, L. F. 1947. The effect of
tissue fat stability on deterioration of frozen poultry. Poultry Sci. 26, 14.
Sethi, S. C., and Aggarwal, J. S. 1950. Stabilization of edible fats by condiments and
spices. Nature 166, 518.
Shenk, J. H., Hall, J. L., and King, H. H. 1934. Spectrophotometric characteristics of
hemoglobins. Beef blood and muscle hemoglobins. J . Biol. Chem. 106, 741.
Shrewsbury, C. L., Horne, L. W., Braun, W. Q., Jordan, R., Milligan, O . , Vestal,
C. M., and Westkamp, N. E. 1942. Chemical, histological and palatability changes
in pork during freezing and storage in the frozen state. Purdue Univ. Agr. Expl.
Sta. Bull. No. 472.
Simon, F. P., Howitt, M. K., and Gerard, R. W. 1944. Inhibition of catalyzed oxida-
tions by hemins. J . Biol. Chem. 164, 421.
Smith, F. H., Brady, D. E., and Comstock, R. F. 1945. Rancidity of bacon, effect of
antioxidants. Znd. Eng. Chem. 37, 1208.
Spicer, S. S., Hanna, C. H., and Clark, A. M. 1949. Studies in vitro on methemoglobin
reduction in dog erythrocytes. J . Biol. Chem. 177, 217.
Stansby, M. E. 1941. Determination of peroxide values for rancidity of fish oils. Znd.
Eng. Chem., Anal. Ed. 13, 627.
Steinberg, M. P., Winter, J. D., and Hustrulid, A. 1949. Palatability of beef stored
a t 0” F. as affected by moisture loss and oxygen availability. Food Technol. 3,367.
Stirton, A. J., Turer, J., and Riemenschneider, R. W. 1945. Oxygen absorption of
methyl esters of fat acids and the effect of antioxidants. Oil & Soap 22, 81.
Sumner, R. J. 1943. Lipoid oxidase studies. A method for the determination of lip-
oxidase activity. Znd. Eng. Chem., Anal. Ed. 16, 14.
Takeya, D. 1949. The reaction between hemoglobin and ascorbic acid. I. Formation
and decomposition of choleglobin. J . Japan. Biochem. SOC.21 134.
Tappel, A. L. 1952. Linoleate oxidation catalyzed by hog muscle and adipose tissue
extracts. Food Research 17, 550.
Taub, A., and Simone, R. 1947. Antioxidant stabilized material. U. S. patent 2,432,698.
Theorell, H. 1932. Kristallinischer myoglobin. Biochem. 2.262, 1.
Theorell, H. 1947. Heme-linked groups and mode of action of some hemoproteins.
Advances in Enzymol. 7, 265.
Trevor, J. S. 1949. Organic sequestering agents. Food, 345.
Urbain, W. M. 1951. Meat and meat products. Zn “The Chemistry and Technology
of Food and Food Products.” Interscience Publishers, New York, Vol. 11, p. 896.
Urbain, W. M., and Jensen, L. B. 1940. Heme pigments of cured meats. I. Preparation
of nitric oxide hemoglobin and stability of the compound. Food Research 6 , 593.
Urbain, W. M., and Ramsbottom, J. M. 1948. Controlling quality changes in cured
meats by packaging. Food Research 13, 432.
OXIDATIVE RANCIDITY AND DISCOLORATION IN MEAT 51
Vail, G. E., and Conrad, R. M. 1948. Determination of palatability changes occurring
in frozen poultry. Food Research 13,347.
Volz, F. E., and Gortner, W. A. 1947. A study of the determination of peroxides in
fats. J . A m . Oil Chemists’ SOC.24, 417.
Volz, F. E., Gortner, W. A., Pita, E. W., Jr., and Miller, J. I. 1949. The effect of ultra-
violet light in the meat cooler on the keeping quality of frozen pork. Food Technol.
3, 4.
Walters, W. P., Muers, M. M., and Anderson, E. B. 1938. An improved Kreis test.
J . SOC.Chem. Ind. 67,53.
Watson, R. H. 1935. Some observations on the estimation of muscle hemoglobin.
Biochem. J., 20, 2114.
Watt, D. B., Hall, J. L., Mackintosh, D. L., and Vail, G. E. 1949. Heat stability of
f a t peroxides. Food Technol. 3, 206.
Watts, B. M. 1950. Polyphosphates as synergistic antioxidants. J . A m . Oil Chemists’
SOC.27, 48.
Watts, B. M., Cunha, T. J., and Major, R. 1946. Effect of feeding and injecting hogs
with tocopherols on the susceptibility of pork fat to rancidity. Oil & Soap 23,254.
Watts, B. M., and Faulkner, M. 1953. Unpublished data. Florida State University,
Tallahassee.
Watts, B. M., and Major, R. 1946. Comparison of a simplified, quantitative Kreis
test with peroxide values of oxidizing fats. Oil & Soap 23, 222.
Watts, B. M., and Lehmann, B. T. 1952a. The effect of ascorbic acid on the oxidation
of hemoglobin and the formation of nitric oxide hemoglobin. Food Research 17.
100.
Watts, B. M., and Lehmann, B. T. 195213. Ascorbic acid and meat color. Food Technol.
6 , 194.
Watts, B. M., and Peng, D. 1947a. Lipoxidase activity of hog hemoglobin and muscle
extract. J. Biol. Chem. 170, 441.
Watts, U. M., and Peng, D. 1947b. Rancidity development in raw versus precooked
frozen pork sausage. J . Home Econ. 39, 88.
Watts, B. M., Peng, D., and Kline, E. A. 1948. Precooking pork for freezing storage.
J . Home Econ. 40, 579.
Watts, B. M., and Wong, R. 1951. Some factors affecting the antioxidant behavior of
ascorbic acid with unsaturated fats. Arch. Biochem. 30, 110.
Weiss, T. J., Green, R., and Watts, B. M. 1953. The effect of metal ions on the forma-
tion of nitric oxide hemoglobin. Food Research 18, 11.
White, W.H. 1941a. Methods for the investigation of rancidity, their interrelation and
application t o bacon fat. Can. J . Research 19D, 278.
Whitc, W.H. 1941b. Rancidity in pork f a t after frozen storage and conversion to
bacon. Can. J . Research 19D, 96.
White, W. H. 1944. Smoked meats. 11. Development of rancidity in smoked and
unsmoked Wiltshire bacon during storage. Can. J . Research 22F, 97.
Widmer, C., and Holman, R. T. 1950. Polyethenoid fatty acid metabolism. Deposition
of polyunsaturated fatty acids in fat-deficient rats upon single fatty acid supple-
mentation. Arch. Biochem. 26, 1.
Wiesman, C. K., and Ziemba, J. V. 1946. How to prevent rancidity in frozen pork
sausage. Food Inds. 12, 1863.
Wilbur, K. M., Bernheim, F., and Shapiro, 0. W. 1949. The thiobarbituric acid reagent
as a test for the oxidation of unsaturated fatty acids by various agents. Arch.
Biochem. 24, 305.
52 BETTY M. WATTS
Winkler, C. A. 1939a. Colour of Meat. I. Apparatus for its measurement, and relation
between p H and color. Can. J . Research 17D,1.
Winkler, C.A. 193913. Colour of Meat. 11. Effect of desiccation on the colour of cured
pork. Can. J . Research 17D,29.
Winkler, C. A., and Hopkins, J. W. 1940. Canadian Wiltshire bacon. XII. Effect of
heat treatment on the colour and colour stability of bacon. Can. J . Research 18D,
289.
Winkler, C. A., Cook, W. B., and Rooke, E. A. 1940. Colour of meat. 111. An im-
proved colour comparator for solids. Can. J . Research 18D,435.
Winter, J. D., Hustrulid, A., Noble, I., and Ross, E. S. 1952. The effect of fluctuating
storage temperature on the quality of stored frozen foods. Food Technol. 6, 311.
Woodcock, A. H., and White, W. H. 1943. Canadian Wiltshire bacon. XXIII. The
effect of concentration of curing salt8 on colour and colour stability. Can. J .
Research 21, 85.
Wyman, J., Jr. 1948. IIeme proteins. Advances in Protein Chem. 4, 410.
The Chemistry of the Sugar-SuMte Reaction and Its Relationship to
Food Problems
BY HARRY GEHMAN AND ELIZABETH M. OSMAN
Research Department, Chemical Division, Corn Prodnrts Refining Company,
Argo, Illi?Loin,
and
Departirient of Foods and Nutrition, Michigan Slate Collegp, East Lansing, Michigan
CONTENTS
Page
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
11. Nature of the Sugar-Bisulfite Addition Compounds. . . . . . . . . . . . . . . . 55
111. Analytical Procedures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
1. Gravimetric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2. Polarirrietric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3. Volumetric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
a. Direct Iodine Titration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
b. Indirect Iodine Titration.. . . . . . . . . . . . . . . . . . . . . . . . . 62
IV. Reaction Equilibrium Constant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
1. Influence of pH Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2. Influence of Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3. Influence of Concentration of R.eactant.9.. . . . . . . . . . . . . . . . . . . . . . 70
V. Reaction Velocity Constants. ................................ 71
VI. Application of the Sugar-Sul ction to Food Problems.. . . . . . . . . . 75
1. Inhibition of Fermentation by Sulfur Dioxide.. . . . . . . . . . . . . . . . . . . . . 77
2. Effect of Processing Conditions on the SOz-Combining Power of Sugar
Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3. Sulfur Dioxide-Combining Power of Citrus Juices.. . . . . . . . . . . . 83
a. Use of SO2 in Inhibiting Browning of Orange Juice.. . . . . . . . 85
4. Browning of Dehydrated Fruits and Vegetables.. . . . . . . . . . . . . . . . . 87
VII. Needed Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
................................. 91
I. INTRODUCTION
One of the oldest but as yet most poorly understood processes for the
preservation of fruits and dehydrated vegetables against undesirable color
changes involves treatment in some manner with sulfur dioxide. Despite
an extensive commercial application, the mechanism by which this com-
pound acts is no more clear than that through which the undesirable dis-
colorations themselves develop. However, there are many indications
that, a t least in the prevention of nonenzymatic browning, reducing
sugars, either naturally present or resulting from additions of sweeteners
during processing, somehow influence the action of the sulfur dioxide.
53
54 HARRY OEHMAN AND ELIZABETH M. OSMAN
Summarization of current knowledge of sugar-bisulfite addition com-

pounds, therefore, should be of assistance in studies of the general prob-
lem of browning.
Stadtman (1948) lists three modes of action which have been sug-
gested to explain the inhibition of nonenzymatic browning by sulfur
dioxide : (1) the ‘ I antioxidant theory” postulates that sulfur dioxide is
effective because of its reducing power, being oxidized in preference t o
other substances whose oxidation is accompanied by color formation; (2)
the “addition compound theory’’ is based on the fact that sulfur dioxide
adds readily to active carbonyl groups, thus presupposing that browning
involves condensation of an aldol or melanoidin type and that formation
of bisulfite addition compounds, by protecting the active carbonyl
groups, prevents such condensations; (3) the “bleaching theory ” suggests
that sulfur dioxide fails to inhibit the darkening reaction but acts t o
bleach the dark compounds as they are formed. As Stadtman points out,
there is evidence both for and against the first and third of these theories,
but there is very little direct evidence for the second theory. However, it
appears almost certain that active carbonyl groups are involved in some
browning reactions (Mohammad et al., 1949a, 1949b; Lewis et al., 1949).
Thus, the well-known binding of sulfur dioxide or bisulfite by such
groups seems t o offer an attractive and logical explanation for the estab-
lished efficacy of sulfur dioxide in protecting dried fruits and vegetables
as well as some fruit juices and fruit concentrates.
Although Joslyn and Marsh (1935) failed t o inhibit discoloration of
orange juice by removing sugar by fermentation, Stadtman (1948, p. 358)
found that fermentation of apricot sirups (which removed reducing sugars
almost completely) decreased the rate of browning to half. It was then
observed that the addition of fructose and glucose to fermented sirups in
amounts equivalent to the sugars originally present resulted in a restora-
tion of the normal browning rate.
Indirect evidence of the involvement of free carbonyl groups in the
browning of fruit juices is provided by the work of Markh (1950). H e
showed that apple and grape juices, when heated with added amino acids,
became colored, presumably owing to formation of melanoidins. Color
production could be prevented by fermentative removal of sugars or by
their inactivation by reaction #withsulfurous acid.
Lewis el al. (1949) have discussed the three theories of browning
presented by Stadtman (1948). They consider the theory that sulfur
dioxide is effective because of preferential oxidation t o be overruled by
the fact that the sulfur dioxide was more effective under their “anaerobic”
experimental conditions than when oxygen was present. Likewise, the
theory that sulfur dioxide bleaches dark compounds during their forma-
SUGAR-SULFITE REACTION A N D RELATIONSHIP TO FOOD PROBLEMS 55
tion was regarded as improbable. No color was ever observed in the mix-
tures treated with sulfur dioxide, and isolated pigment was not decolorized
by sulfur dioxide. The conclusion by Lewis et al. th a t the most likely
theory is that sulfur dioxide addition compounds are formed, thus re-
moving one or more of the reactants, seems most plausible in view of their
results. Although the quantity of sulfur dioxide used was far less than
enough t o react completely with either the sugar or amino acid present,
the possibility of the sulfur dioxide reacting with some intermediate
formed as a result of the reaction of the two initial reactants (thereby
preventing pigment formation) is not precluded. This would also account
for the delayed browning which often occurs after all the sulfur dioxide
has been consumed by reaction with these intermediates.
11. NATCJIZE OF THE SUGAR-BISULFITE ADDITIONCOMPOUNDS
Reaction between the carbonyl group of an aldehyde or a ketone and
bisulfite probably was first observed during work with taurine by Redten-
backer (1848).This report was followed closely by Bertagnini (1853), who
described a crystalline bisulfite addition compound of heptanal. Rocques
(1898) seems t o have been the first t o report the possible existence of a
glucose-bisulfite addition compound. He demonstrated that the aldehyde
group in the reducing sugar in wine combined with sulfur dioxide which
had been added as a protective agent.
Disagreement over whether the result of this addition reaction be-
tween aldehyde and bisulfite should be written as structure A or structure
B continued for many decades:
. 1"
R -H
OH
I
RC-H
I I
OSOzH SOaH
( A ) ~~-1iydroxy
sulfurous acid ester (B) a-hydroxy sulfonic acid
Among the proponents of the ester formula (A), Mendelejeff (1859)
and especially Muller (1873a) should be mentioned. Schiff (1881) and
Eibner (1901) argued for the a-hydroxy sulfonic acid formula. However,
much earlier Muller (1873b) had suggested that perhaps the ease with
which the aldehyde and sulfite could be separated was inconclusive
evidence for a n ester structure, since a structure having both sulfur and
hydroxyl attached to the same carbon atom possibly should be expected
to exhibit about the same stability as an ester. Despite this statement, the
ester structure gained favor from work by Reinking et al. (1905) and by
Schroeter (1918). Admitting later th at this position on the ester formula
had been proved untenable, Schroeter and Sulzbacher (1928) further
opposed the a-hydroxy sulfonic acid structure and proposed writing the
56 HARRY GEHMAN AND ELIZABETH M. OSMAN
bisulfite addition compounds with carbonyls as tripartite molecules with

such formulas as (R&: 0)(SO2)(HO.H). As late as 1934, Gibson (1934)
did not consider the question settled. Only a short time earlier, Pring-
sheim (1932) had even stated that there is actually no bisulfite addition
product formed by any of the reducing sugars.
Acetoxymethanesulfonic acid was prepared by Lauer and Lang-
kammerer (1935) both by acetylation of the formaldehyde bisulfite addi-
tion compound and by fusion of iodomethane sulfonic acid with potassium
acetate. Thus was obtained clear proof that, a t least in the case of formal-
dehyde, the addition compound which forms with bisulfite is a salt of an
a-hydroxy sulfonic acid. An interesting experiment by Backer and Mulder
(1934) gave further supporting evidence in that they were able to show a
C-S linkage in formaldehyde bisulfite. Treatment with ammonia gave
NHzCH~SO~H, which yielded chloromethane sulfonic acid on reaction
with nitrosyl chloride; a sulfonic acid structure in the original compound
appears to be a prerequisite for this result. These two research reports
were strong substantiation for this structure, which had become generally
accepted as a result of earlier work by Raschig and Prahl (1926) in which
an elaborate study of the reaction between formaldehyde and sodium
bisulfite and further reactions of this product had indicated the same
structural requirement.
Stelling (1928) found an absorption band in formaldehyde and acetone
bisulfite addition compounds at 4992.0 A? similar to that in sulfonic
acids at 4992.2 and differing from that of metal alkyl (4996.0) and dialkyl
sulfites (4997.7). He concluded from this that the sulfonic acid structure
must be present. Raman spectral examinations of several aldehyde and
ketone bisulfites by Caughlan and Tartar (1941) revealed the presence
of a carbon-sulfur bond, possibly a carbon-hydroxyl bond, but no carbon
doubly bonded to oxygen. This thus aided in discrediting both the tri-
partite molecule and the sulfurous acid ester structures. Sundman (1949)
believes that formation of a stable monomolecular complex of boric acid
and glucose bisulfite would be impossible if Schroeter’s tripartite molecu-
lar structure were correct. His examinations of this complex led him to
believe that its structure could be represented only by:
H
SUOAR-SULFITE REACTION A N D RELATIONSHIP TO FOOD PROBLEMS 57
With deuterium as the aldehyde hydrogen in the molecule, Thompson

and Cromwell (1939) found th at p-phenylbenzaldehyde-dl did not ex-
change aldehyde hydrogen with water but that the bisulfite addition
compound did. This was interpreted as possibly indicating a n enolization
of the bisulfite addition compound :
D
.. ..
:o:
:C
C~HKCRII,
D
..
:0:
.. ..
S : 0 : Na
.. .. ..
CeHoCGHl : C : S : 0 : Na
.. .. .. .. ..
: ..o : : ..o : :o: :o:
.. ..
H H
a condition thought to be best explained by way of the sulfonic acid struc-

ture, although no claim was made for its proof. Whitmore (1937) stated
that electronic concepts, as utilized in this aldehyde bisulfite addition
compound structure, are justified. An electronic mechanism for the reac-
tion between formaldehyde and sodium bisulfite has been written by
hlexander (1950) to provide the same final structure. Furthermore, this
concept is consistent with the known instability of the addition com-
pounds and the instability th at would be expected from a carbon atom
carrying both a hydroxyl and a sulfonic acid group.
Much research has been done in attempts to establish the fate of
sugars in pulping liquors, since the monosaccharides released during the
wood pulping process are of interest as possible consumers of bisulfite
during the actual cooking as well as a source of by-product fermentable
dry substance and of stream pollution problems. The relatively high tem-
peratures encountered in some of these operations appear to produce
basic structural changes. Marusawa and Uchida (1929) reported th a t
equimolar concentrations of reducing sugars and bisulfite heated a t
130" C. (266" F.) formed a t least two solid compounds entirely different
from those formed a t room temperature. Compound A formed first but
decomposed under heating and in the presence of hydrogen ion t o yield
thiosulfate, sulfate, and polythionate. At lower hydrogen ion concentra-
tions the supposedly cyclic Compound B was formed by rearrangement
of A. Hagglund (1929) and Hagglund and Urban (1929) proposed th a t
reactions between sugars and bisulfite, when strongly heated, should be
written as
2HSO3- + 2CeH120~+ Szo3- + 2CeI11207 + H20
without any requirement th at a bisulfite intermediate compound be
formed, as seems essential in the Marusawa and Uchida (1929) presenta-
tion. Hagglund el al. (1930) obtained calcium mannonate from a bisulfite-
mannose reaction at 130" C. (266" F.). The series of reactions was written
as possibly being :
The S-containing compounds produced by Marusawa and Uchida

(1929) and found to be rather stable to acids were not isolated, although
these were thought to possess cyclic structures. Hagglund et al. (1930) also
obtained an S08H-containing acid from heated glucose bisulfite addition
reactions which was stable to hot, dilute acid and cold alkali but unstable
to boiling alkali with loss of sulfite. A free carbonyl remained, since a
phenylhydrazone of the barium salt was crystallized. These results indi-
cated the presence of a carboxyl and possibly a sulfonic acid group as well
as carbonyl.
The sulfite pulping process represents a reaction condition such that
sugar-bisulfite addition compounds of the types reported from this high
temperature work should be expected; Hagglund et al. (1943) and Hei-
winkel (1944) observed that part of the dissolved sugar appears as aldonic
acid and part in S-containing compounds with the properties of sulfonic
acids, probably formed through a sugar-bisulfite addition compound and
a labile ketosulfonic acid. In addition to direct molecular interactions
between carbonyl functions and bisulfites and the apparent sulfonation
a t secondary hydroxyls of saccharide molecules or rearrangement under
more stringent conditions, account must be taken of reaction with possible
aldehydic and ketonic sugar degradation products.
Adler (1946) isolated two sulfonic acids from sulfite-bisulfite sugar
reactions a t pH 6 and confirmed the presence of sulfonic and carboxylic
functions in both of these as well as their stabilities to boiling hydro-
chloric acid and the stability of Compound B t o boiling alkaIi. Periodic
acid oxidations with formation of formaldehyde and no loss of sulfonic S
or introduction of new carboxyls permitted the interpretation that both *
A and B contain primary alcohol groups conjugated with secondary (or,

perhaps, tertiary) alcohol groups and that neither compound contains any
other pair of adjacent hydroxyls or any adjacent hydroxyl and aldehyde
or ketone carbonyl pair; however, the possibility of an a-hydroxy car-
boxylic acid being present is not eliminated. Although the structures
were not entirely elucidated, the evidence was indicative of an intra-
molecular rearrangement to permit location of a carboxyl on a branch.
The barium salt of pure Compound A analyzed as CeHsOsSBa and that
of Compound B as CeHloOsSBa. Adler concluded that perhaps as much
as 45% of sugar lost in a sulfite cook might be diverted into sulfur-con-
SUGAR-SULFITE REACTION AND RELATIONSHIP T O FOOD PROBLEMS 50
taining carbohydrate derivatives of this sort, although no claim was made

for the final validity of these results.
Bisulfites of amines have been made t o react with aldehydes and
ketones (Adams and Lipscomb, 1949; Adams and Garber, 1949). The
reaction products were generally like those obtained with sodium bisulfite,
although excess sulfur dioxide was quite commonly required to maintain
a stable condition. The only experiment with sugar reported, involving
glucose and dl-s-butylamine bisulfite, was unsuccessful. However, Tillit-
son (1942), as well as Friedman and Braitberg (1947), claims the produc-
tion of stable derivatives by reaction of amines with preformed glucose
bisulfite. I n these instances the substituted amine group replaced the
a-hydroxyl group (as expected; see Backer and Mulder, 1934), and the
-S03Na group remains as an integral part of the product. Such structures
as
CH~OH
were described.
The comprehensive investigatioiis by Kerp (1903), Kerp and Bauer
(190.2, 1907a, b, c), and Kerp and Wohler (1909a, b) stand today as the
classical background for all studies of sugar bisulfite and aldehyde
bisulfite reactions in general. Equilibrium constants were established a t
15-25" C. (59-77" F.) in neutral solution for the reaction between several
aldehydes and sodium bisulfite. These equilibrium constants increased
with temperature, indicating increasing decomposition, the rate of which
was found t o decrease in acid solutions. Kerp (1907) also obtained evi-
dence for the presence of glucose-bisulfite reaction products in sulfured,
dried fruits.
Despite the continuing unquestioned fundamental merit of the
research reported by Kerp, it rapidly became evident th a t his major con-
tributions were in laying foundations onto which later investigators could
anchor research plans; it is interesting to contemplate the probable com-
pleteness of this early work if more modern concepts of reaction kinetics
had been available.
111. ANALYTICAL PROCEDURES
1. Gravimetric
Quantitative precipitation procedures cannot be applied t o a general
study of the carbonyl-bisulfite addition reaction involving sugars because
of unfavorable solubility characteristics of the products which make their
separation from the unreacted sugars difficult or impossible.
60 HARRY GEHMAN A N D ELIZABETH M. OSMAN
1. Potarimetric
Kerp and Bauer (1904, 1907a) observed that the bisulfite addition
compound of glucose had an optical rotation of nearly zero (slightly
positive). They applied this observation by using a simple polarimetric
procedure to estimate the amount of glucose remaining unbound in
bisulfite reaction mixtures. However, this method has its greatest utility
only in reaction mixtures containing an excess of reducing sugar or in
equimolar solution of sugar and bisulfite; excess sodium bisulfite causes
a slight shift to the negative side (Kerp and Bauer, 1907b) so that it is
not applicable if highly accurate values are required. Nevertheless the
ease with which free sugars can be estimated by Kerp’s polarimetric
approach is attractive and has been applied only recently by Vas (1949)
in a study of the glucose-sulfurous acid equilibrium. A combination of
this procedure and changes in the reducing value of reaction mixtures was
used by Browne (1944) t o obtain evidence not only of the existence of
sugar-bisulfite complexes but also of the reversible nature of the reaction.
He found that, under conditions known t o favor decomposition of the
complexes, the rotation values shifted until a final value, always approxi-
mately equal to that of the total sugar present, was obtained; thus, a
substantially complete absence of any of the complex is indicated.
3. Volumetric
Although the reaction which produces sugar-bisulfite addition com-
pounds is typical of that present during the formation of many addition
compounds, in that a reversible equilibrium exists between the compo-
nents, the velocity of shift is so markedly affected by the reaction en-
vironment that conditions suitable for analysis of the reaction mixture
can be selected. For example, determination of excess or unbound bisulfite
by iodine is feasible because an iodine end point can be attained and read
without significant equilibrium disturbance, although certain well-
established principles must be utilized. Volumetric procedures have,
therefore, been applied generally in this type of investigation during
recent years. An excellent technical discussion relating specifically to the
carbonyl-bisulfite reaction is provided by Kolthoff and Stenger (1942).
Volumetric methods are applicable to the analysis of sugar-bisulfite
addition compounds, since, being products of a completely reversible
reaction, they can be separated into their components stoichiometrically
in a reaction system readily amenable to detection of the desired equiva-
lence point within an analytically practical period of time. Attainment of
acceptable precision is dependent on proper significance being attributed
t o factors which may affect the equilibrium rates: namely, temperature,
SUOAR-SULFITE REACTION A N D RELATIONSHIP TO FOOD PROBLEMS 61
concentration, hydrogen-ion content of the system, and specific char-

acteristics introduced by individual aldehydes or ketones. Actual deter-
mination as commonly practiced involves two titrations of the aqueous
solution containing the reaction components. The first titration, carried
out on the unaltered solution, gives a measure of the unreacted bisulfite
present. The second, carried out only after the conditions of the solution
have been so altered as to bring about substantially complete dissociation
of the addition compound, gives a measure of the sulfite which remained
bound in the reaction product during the first titration. This method was
originated by Ripper (1892), who used dilute potassium bisulfite aldehyde
solutions. He determined excess bisulfite tiy titration with a standard
iodine solution. Feinberg( 1913) conducted one of the early studies on this
general method for aldehyde determination, although he did not work
with sugars.
In a model system wherein the sugar content is known, it is possible
to determine equilibrium constants in this fashion. With total sugar un-
known, another analytical step is needed. For this Vas (1949) used the
polarimetric method of Kerp and Bauer (1904, 1907a) a s did Browne
(1944), who in addition used the conventional methods based on reducing
power. Ingram and Vas (1950a) used paper chromatography to estimate
sugars and other reducing compounds in work with orange juices.
A procedure satisfactory for the examination of a model system (pure
sugar and bisulfite) may not be suitable for examination of practical mix-
tures. This condition arises because of the possibility that such composi-
tions as are encountered in the food industry may be highly colored (and
thus render end-point observations invalid) or so complicated in com-
position that the direct approach useful for a model system is totally
inadequate. As a result, several general volumetric methods of variable
utility in food compositions have been proposed. These include :
a. Direct Iodine Titration. Iodine reacts with and oxidizes all un-
bound bisulfite present in an aqueous aldehyde and bisulfite reaction
system. Following the laws of mass action, the equilibrium then begins
to be re-established and additional free bisulfite becomes available for
oxidation. Therefore, the validity of direct titration t o determine un-
reacted bisulfite must be established on the basis of evidence that, with
the conditions used, the rate of decomposition of the addition compound
in question is relatively slow in comparison to the rate of the oxidant
reduction. Kolthoff and Stenger (1942) have discussed this feature of the
procedure and have calculated the analytical errors that may be expected
from application of the iodine titration method t o aldehyde-bisulfite
systems within broad ranges of concentration and containing addition
compounds with widely variant equilibrium constants. They conclude
62 HARRY QEHMAN AND ELIZABETH M. OSMAN
that compromises between accuracy and speed may be necessary but that
by working a t low temperature (wherein the rate of decomposition is
least) and in a pH range likewise favoring the greatest addition compound
stability, results of true analytical significance can be obtained.
This method was applied by Sundman (1949) for the determination
of a large number of sugar-bisulfite equilibrium constants at 19" C.
(66.2" F.) and at pH 4.65. In a typical experiment he pipetted 5 ml. of a
glucose solution (0.02-0.20 mole/l. of sugar), 2 ml. of acetate buffer
(pH = 4.65; 15 g. glacial acetic acid and 34 g. sodium acetate trihydrate/
100 ml.), 0.2-2.0 ml. of NaHS03 solution (60-65% SO2; about 10 g.
NaHS03/100 ml.) into a 10-ml. volumetric flask and filled it to the mark
with water. The solution was mixed thoroughly and then held in a
thermostat a t 19" If: 0.1" C. (66.2 If: 0.2" F.) for 4 hours. Analysis of the
reaction solution for free and bound bisulfite contents was then accom-
plished by the steps which follow: 5 ml. of the solution conditioned in the
thermostat was pipetted into a 200-ml. flask containing 50 ml. of water,
2 ml. of 25% sulfuric acid, and a few drops of starch indicator solution.
After rapid titration to the end point using 0.01 N iodine solution, 40 ml.
of 1 N KOH solution was added and the mixture was allowed to stand
10 minutes to release all bound bisulfite. The mixture was again titrated
with 0.01 N iodine solution after acidification with 15 ml. of 25% sulfuric
acid solution.
One of the difficulties with iodine titrations arises from the unsatis-
factory character of the starch indicator end point, and this is particularly
misleading in practical food systems. Ingram (1947a) introduced an
electrometric procedure and found errors only about 20% as great as
those encountered with starch indicators. This method had the added
advantage that it was just as useful in highly colored fruit compositions,
where in the use of a starch indicator is impossible. As finally evolved by
Ingram and Vas (1950b), total sulfur dioxide is determined by the modified
Monier-Williams distillation method and free sulfur dioxide is obtained
by iodometric titration using an electrometer.
b. Indirect Iodine Titration. Downer (1943), while studying the pre-
servative action of sulfurous acid on citrus juices, encountered unsatis-
factory variations in trying to apply the direct volumetric method useful
for simple sugar systems. A more time-consuming, but also somewhat
more accurate, method was then developed by adapting principles
established by Mapson (1942) for the determination of ascorbic acid in
the presence of sulfite. Total iodine-reducing substances in juice were
found by titration with 0.02 N iodine solution. An equal volume of juice
was mixed with 20% (by volume) of acetone; the mixture was allowed to
stand for 10 minutes and again titrated as above. The large excess of
SUQAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 63
acetone apparently combined with all the free sulfurous acid so that the
difference between the two titration values represented free sulfurous
acid. Ingram (1917b) made a very critical study of this procedure. His
excellent report shows that actually only 95% of the sulfur dioxide is
combined with the acetone after a holding time of as long as 30 minutes.
He found, furthermore, that about one orange juice sample in five yielded
definitely erroneous results by this method, since estimates of free sulfur
dioxide occasionally were actually greater than the total quantity found
by the distillation method.
Other carbonyl compounds have been used in the same manner as
acetone in this type of analysis. Ingram (1947b) tried acetaldehyde be-
cause of its greater affinity for sulfur dioxide. Results indicated, however,
that it combines with iodine-reducing substances in orange juice besides
sulfur dioxide-a difficulty also occasionally exhibited by acetone.
Potter and Hendel (1951), working with dehydrated white potatoes, con-
cluded that formaldehyde and glyoxal, especially the former, were
preferable to acetone. Although the average values they report indicate
this t o be true, the individual values obtained exhibit wide discrepancies.
In this work, comparisons were made with Monier-Williams distillation
results as obtained by the Thompson and Toy (1945) modification.
It must therefore be concluded that, although some method of this
type is indeed desirable, great care should be exercised in establishing
satisfactory precision and in interpreting the results.
IV. REACTION
EQUILIBRIUM CONSTANT
The reaction system consisting of reducing sugar and sodium bisulfite

produces a completely reversible addition product without loss of sub-
stance t o side reactions so that the components of the system should be
three in number. The possibility for the existence of an equilibrium con-
stant becomes of significance; this may be written as
wherein S represents total uncombined sugar, B total uncombined sul-

furous acid (in any form, since i t may be present as sulfite, bisulfite, or
sulfurous acid according to the hydrogen ion concentration of the system) ,
and S B reacted addition product (all expressed as moles).
Determination of [SB]can be referred to difference values obtained by
measurements of either uncombined sulfite or sugar with appropriate
subtractions from the known amounts added. Although Browne (1944)
indicated that there is more chance for error in the determination of
sulfite (owing to possible losses as volatile sulfur dioxide) than in the
determination of free sugar, the majority of investigators in this field have

used an iodometric method for free sulfite.
Simple aldehydes and ketones, such as are encountered outside the
carbohydrate field, probably react with SO2 t o form relatively simple
equilibrated systems. However, it should not be overlooked that aldoses
and ketoses are not strictly typical of aldehydes and ketones in many of
their reactions; for example, they do not give the specific red color with
Schiff’s reagents as would be expected from a true carbonyl function.
Therefore, since this sugar reacting group is only a potential carbonyl, the
equilibrium or equilibria on which i t must depend for release from the
hemiacetal structure must be taken into account in consideration of the
sugar-bisulfite equilibrium constant. Actually, it may supply a partial
reason for different equilibrium constants for various sugars, since the
mutarotation constants of sugars are not all the same. Thus, an aldose-
bisulfite system may \,e a complex series of equilibria:
a-Glycopyranose “ Aldehydrol” % p-glycopyranose
It
glyconaldehyde (Sbisulfite) $ sugar bisulfite
11
a-Glyrofuranose Aldehydrol ” % 8-glycofuranose
Reasoning from this basis, Sundman (1949, p. 43; 1951) has postulated
the existence of a functional relationship between the “slow” muta-
rotation constants (ml) of sugars and the equilibrium constants of their
bisulfite addition compounds. In a limited series of results, a relatively
constant value for the factor K190 X ilzl gave substantial evidence for
the validity of this postulate within related groups such as for glucose and
reducing disaccharides in which the reducing moiety is glucose. Another
factor with approximately one-third the numerical value of that for the
glucose series was obtained for a miscellaneous group which included
pentoses as well as other remaining hexoses (Table I ) .
Somewhat analogously, Lippich (1932) proposed to determine the
amount of free aldehyde in solutions of sugars by measurements of extent
of cyanhydrin formation, since this very rapid reaction should occur only
between cyanide and free carbonyl; it seems reasonable that a comparable
factorial relationship exists here. Greater reactivity of glucose 6-phosphate
with HCN than was experienced with glucose itself led Stepanov and
Stepanenko (1937) to postulate the presence of more open chains in the
derivative than in native glucose. These workers likewise explained
(Stepanov and Stepanenko, 1940) the muchmore rapid reaction of fructose
diphosphate than fructose monophosphate with HCN (30% vs. 13 %) on
the basis of a further shift t o the open carbonyl. Underivatized fructose
could not be induced to react under similar conditions. From their experi-
SUGAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 65
mental evidence of greater reactivity with HCN for the hexose phosphates
generally in comparison with the free hexoses, Stepanov and Stepanenko
postulated that the common conversion of hexoses to phosphoric acid
esters in biological systems may be traceable to the favorable sugar
labilities thus produced.
TABLE I
Lpparent Interrelationship between Sugar-Bisulfite Addition Compound Equilibrium
Constants and Sugar Miitrtrotation Constants*
Sugar K ~ ~) ) I , x~ 1 0 4 ~ K. ~ x ttL1
~ x ~1 0 4 ~ .
G1ucose 0.587 63 37
Cellobiose 0.530 46 24
Maltose 0.549 53 25
Lactose 0.532 47 29
Melibiose 0.353 86 30
Xylose 0.0688 203 14

Arabinose 0.0412 300 12
Ribose 0.0278 392 14
Lyxose 0.0187 568 11
Galactose 0.1040 80 8
Mannose 0.0688 173 11
Mannose.CaClz 0.0400 245 10
* Sundmen (1949).
Cantor and Peniston (1940) examined the dynamic equilibrium pre-
sumed t o be present in a reducing sugar solution by measuring polaro-
graphic wave heights of aldoses and interpreting them in terms of the
amount of “aldehydo” form present. It was possible to correlate their
results with mutarotation rates and also, where data permitted] to make
comparisons with results by Lippich (1932). Delahay and Strassrier
(1952), also using polarographic methods, showed that the ring --+ alde-
hydo transformation in aldoses increases rapidly with pH and tempera-
ture, the rate of change being greater for pentoses than for hexoses. This
is in agreement with Wiesner (1947), who claimed th a t the height of the
polarographic waves obtained with sugars is dependent on the rate of
the nonreducible aldose -+ reducible aldose reaction.
These investigations all point very strongly to the need for placing
greater emphasis on the equilibria present in mutarotating sugar solutions
when these solutions represent the sugar source in reactions normally
expected t o require free carbonyls for their accomplishment.
Equilibria known to exist in the sulfite and nonsugar aldehyde or
ketone reaction systems were examined critically b y Stewart and Don-
nally (1932a, b, c). For the reaction product between bensaldehyde and
sulfite ion, these workers showed that the hydroxysulfonic acid formed is
a strong monobasic acid' and that in the upper pH range practically all of
the addition reaction involves sulfite and not bisulfite ion. Within certain
general pH ranges the equilibria found were:
pH 0-1 CsHsCHO+ HzSOa CeHsCH(OH)S03- + H +
pH 3-7 CsHsCHO+ HSOa- % CsHsCH(0H)SOs-
+ SOs- G CsHsCH(O-)SOs
__
pH 10-12 CeHsCHO
903-
pH 12-13 CBHKCHO+ OH- k CeHs(O-)OH C ~ H K C ( O - ) S+
~ ~OH-
-
In their investigations with benzaldehyde and bisulfite, they also

showed a sharp rise in the equilibrium constant with rising temperature
in the range of about 3" t o 50" C. (37.4 t o 122' F.) ; also, the commonly
observed effects of pH changes on K were substantiated in that K was
minimal a t about pH 4 in coincidence with the range in which the bisulfite
form is predominant. Vas (1949) patterned certain phases of his work
with the sugar-bisulfite reaction after the studies of Stewart and Donnally
and concluded that their explanations held for systems containing glucose
as well as benxaldehyde.
A major difficulty in any examination of the literature in this field is
the extreme variation in reaction conditions used by various workers.
As mentioned previously, Kerp's data lacked significance in some respects
by reason of their antedating general acceptance of the pH concept. I n
addition, his solutions were apparently no stronger than l-molar (and
were usually much more dilute than this) and usually the sugar-to-sulfite
ratios were 1:l. In work with food products a ratio of this sort is quite
unusual. It becomes necessary, therefore, because of the specifically
different effects introduced by structures more or less peculiar to the
saccharides, to give separate attention t o the effects of various factors
on K for the sugar bisulfite compounds.
1. InJuence of p H LeJel
This represents the most effective agency for altering the position of
the equilibrium. Actually it is only in the general region of pH 4.5 that
a clear-cut end point between free and bound sulfite can be obtained.
Hopner (1944) and Sundman (1949, p. 35), for example, have shown that
the bisulfite compounds with glucose, galactose, mannose, and xylose
decompose least between pH 3 and pH 6. Hagglund et al. (1943) have
indicated that glucose-bisulfite is more stable in weakly acid solutions
than in alkaline or strongly acid solutions. This now well-established
I The first dissociation constant (Znt. Crit. Tables VI, 260) of sulfurous acid is 1.7 X
10-2, whereas Stewart and Donnally found 3.7 X for the or-hydroxy sulfonic
acid of benzaldehyde; a t higher pH levels they established a second dissociation con-
stant of 7 X 10-lo.
condition has made it evident to later workers th a t all testing should be

accompanied by rigid buffering. For example, Sundman (1949) carried
out equilibrium study titrations in an acetate buffer a t p H 4.65, or under
conditions such that the bisulfite form predominates; phosphate buffers
were used in some cases above this p H value. As an example of the effect
of pH on K , the data in Table I1 (modified from Sundman) are of interest.
TABLEI1
Effect of pH Level on Galactose and Glucose Bisulfite Equilibrium Constants*
Approximate Uncombined SO2 calculated Sugar-bisulfite
sugar: SO2 as sulfurous acid, addition compound, t
pH value ratio mole/l. X lo3 mole/l. X l o 3 KWO.
Galactose X 103 = 55.6mole/l.
11.0 1:1.2 64.9 1.8 1.94
10.8 1:l.l 56.5 2.3 1.31
10.6 1:l.O 54.3 1.2 2.46
10.1 1:1.3 67.9 2.6 1.38
7.8 1:1.2 64.9 3.0 1.14
7.5 1:1.3 67.6 4.1 0.849
7.0 1:l.O 52.8 7.8 0.324
6.7 1:1.3 60.7 12.0 0.221
Galactose X lo3 = 112 mole/l.
7.0 1:0.6 51.8 15.6 0.320
6.9 1:0.6 48.3 18.1 0.251
6.7 1:0.6 49.1 21.7 0.204
6.6 1:0.7 49.2 25.9 0.164
6.4 1:0.7 49.7 29.3 0.140
6.2 1:0.7 51.0 32.2 0.126
6.0 1:0.7 47.1 33.7 0.109
4.4 1:o.g 52.8 38.2 0.102
Glucose X 108 = 333 mole/l.
7.3 1:0.2 59.4 6.3 3.08
6.8 1:0.2 53.2 11.9 1.44
6.7 1:0.2 54.9 14.9 1.17
6.4 1:0.2 58.5 23.9 0.757
6.1 1:0.2 58.8 25.0 0.724
5.8 1:0.2 55.1 25.7 0.659
* From Sundman (1949): at 19" C. (66.2'' F.).
t Equivalent to combined SO%(in any form) in moles.
At 20" C. (68" F.) and with glucose concentrations of approximately

20 g. per 100 ml. (1.11 mole/l.) and with 320 mg./l. of added sulfur
dioxide (0.005 mole/l.), Vas (1949) obtained the data recalculated as
shown in Table 111. Below about p H 1, buffering was with HC1 and citric
acid, while N a 2 H P 0 4was used above this level. Free glucose and free and
combined sulfur dioxide contents were determined after standing a t
20" C. ( S S O F.) for one week; values after two weeks were substantially
unchanged.
TABLEI11
Effect of p H Level on the Apparent Equilibrium Constant of Glucose-Bisulfite at
20" C. (68" F.)*
uncombined SO2 calculated as Sugar-bisulfite addition
p H value sulfurous acid, mole/l. X lo3 compound, mole/l. X lo3 K Z PC.
10.61 4.955 0.045 110
9.25 4.915 0.085 64
9.07 4.900 0.100 53
7.34 3.855 1.145 3.7
6.61 2.805 2.195 1.4
5.82 1.925 3.075 0.69
5.71 1.920 3,080 0.69
4.84 1.800 3.200 0.62
4.76 1.770 3.230 0.60
3.91 1.795 3.205 0.61
3.04 1.890 3,110 0.66
2.20 2.340 2.660 0.96
1.14 3.290 1.710 2.1
1.04 3.780 1.220 3.4
0.20 4.690 0.310 17
0.06 4.900 0 . 100 53
* Vas (1949). Gluoose = 1.11 mole/l.; glucose: 802 ratio = 1:0.0045.
At 30' C. (86" F.) after a reaction time of 30 hours the (apparent)

equilibrium condition for 1 M solutions of glucose containing 1-molar
sulfur dioxide equivalent is shown in Table IV. I n this series, the desired
p H levels were attained by use of combinations of sodium sulfite, sodium
bisulfite, sulfurous acid, and HC1 as needed but without altering the 1: 1
molar ratio of available sulfur dioxide and glucose; the pH values shown
TABLEI V
Effect of Starting pH on the Apparent Equilibrium Constant of Glucose-Bisulfite a t
30" C. (86' F.)*
Uncombined SO2 calculated as Sugar bisulfite addition
pH value sulfurous acid, mole/l. X lo3 compound, molc/l. X lo3 h ' 3 p C.
7.95 942 58 15
7.15 849 151 4.8
6.53 774 226 2.7
6.00 606 394 1 .o
5.60 553 447 0.68
4.40 527 473 0.59
3.85 530 470 0.60
2.72 557 443 0.70
2.30 713 287 1.8
* Unpublished data froiti Corn Product,s Refining Laborat.ory. Glucone: 90%ratio = 1 : 1 : I-molar
solutions, unbuffrrrd.
in Table IV are those obtained immediately on mixing; the systems were

otherwise unbuffered.
Values for K in Tables 11, 111, and IV were derived by calculation of
all uncombined sulfite as free sulfurous acid. This approach lacks validity
to a certain extent since it assumes only HSOa-, free sugar, and bisulfite
addition compound in the equilibrium. Obviously, this is hardly possible
in systems the pH values of which are so high or so low that little or no
HSO3- ion is present. In fact, the addition compounds obtained a t the
upper ranges might actually be some type of aldehyde-sulfite rather than
aldehyde-bisulfite. There is an almost complete lack of quantitative
application of the concept of variant HS03- ion content in sugar work;
however, Sundman did extend some of his data (Table 11) by computing
K on the basis of HSOs- rather than total sulfur dioxide calculated as
bisulfite. The results are shown in Table V.
TABLEV
Effect of pH 1,cvel in Relation to Actual Bisulfite Ion Content on Galactose and
Glucose Bisulfites at 19' (66.2" F.)*
K l p c . based on total free sulfite K , p c . based on total free sulfite
pH value calculated as sulfurous acid probably in bisulfite formt
Galactose X 103 = 112 inole/l
7.0 0.320 0.107
6 . !J 0.251 0.099
6.7 0.204 0.106
6.6 0.164 0.094
6.4 0.140 0.095
6.2 0.126 0.098
6.0 0.109 0.091
4.4 0.102 0.102
Average 0.099
Glucose x 108 = 333 mole/l.
7.3 3.08 0.615t
6.8 1.44 0.600t
6.7 1.17 0.608
(5.4 0.757 0.515
6.1 0.724 0.608
5.8 0,659 0.606
Aiierage 0,591
$ Sundman (1049).
t We believe these values to be more correctly calculated than those in the original report.
The relative constancy of K calculated in this manner is strongly sug-

gestive of the need for the determination of a new family of constants for
use in ranges outside that in which HS03- predominates, or of the appli-
cation of a complex function designed t o take into account more correctly
the probable nature of the equilibria existing in the extreme pH ranges.
Rahn and Conn (1914) have calculated the relative contents of HzS03,
HSOa-, and 803- by use of Michaelis formulas at different pH values
(Table VI). Their results (which vary significantly from the values used
by Sundman in his calculations) show a maximum of HSOa- at pH 3.7,
with this ion representing a majority of contained sulfite between about
pH 2.0 and pH 5.0, with HzS03 disappearing a t about pH 6.0, and very
little besides SO3-- above pH 8.0
TABLEVI
Composition of Sulfite Solutions at Different pH Values*
pH value SOa-- ions HSOa- ions H2SOa undissociated
1.0 0 0.145 0.855
2.0 0 0.630 0.370
2.48 0.001 0.836 0.163
2.94 0.005 0.932 0.063
4.0-4.1 0.047 0.948 0.0056-0.0050
4.97 0.332 0.667 0.0004
5.9-5.82 0.804 0.196 10-6
6.08 0.838 0.142 10-6
6.12 0.870 0.130 0
6.40 0.926 0.074 0
6.80 0.969 0.031 0
6.75-7.00 0.968-0.970 0.032-0.020 0
7.05-7.25 0.975-0.980 0.019-0.014 0
7.6 0.996 0.004 0
7.6-7.8 0.997 0.0032 0
8.0 0.998 0.002 0
* Rahn and Conn (1944).
2. InJluence of Temperature
Temperatures above normal room conditions increase K (decrease in
contained addition product), whereas lower temperatures decrease K as
illustrated in Table VII. Since these data are derived from several sources,
differences in reactant concentrations and pH conditions result in lack of
uniformity, but the trends indicated are representative of effects similar
t o those established by many workers for nonsugar addition compounds.
Temperature effects on K values are real but are of much less signifi-
cance here than they are on rates of association and dissociation, as will
be shown later.
3. InJluence of Concentration of Reactants
As in the case of temperature, concentrations have a marked effect on
dissociation rates but the effect on the actual equilibrium constants of
sugar-bisulfite addition compounds is very small. Table VIII is an
adaptation from Sundman (1949, p. 25) illustrating this feature. Con-
centration effects were studied in a quite different range of conditions by

Vas (1949) with similar conclusions, as may be noted in Table I X (calcu-
lated from his data).
TABLEV I I
Effect of Temperature on Sugar-Bisulfitc Equilibrium Constants
Temperature, Sugar concentration,* pH
"C. mole/l. X los value K Source
Glucose
2-4 166.7 4.65 0.435 Sundman, 1949
17.5 166.7 4.65 0.594 Sundman, 1949
19 166.7 4.65 0.587 Sundman, 1949
26.2 166.7 4.65 0.665 Sundman, 1949
29.8 166.7 4.65 0.731 Sundman, 1949
35 166.7 4.65 0.784 Sundman, 1949
40 166.7 4.65 0.873 Sundman, 1949
10 1000 4.40 0.350 t'

20 1000 4.40 0.417 i
30 1000 4.40 0.587 t
22 1000 3.90 0.418 Hopner, 1944
22 1000 4.95 0.446 Hopner, 1944
49 1000 3.90 0.730 Hopner, 1944
49 1000 4.95 0.763 Hopner, 1944
Xylose
2 66.7 4.65 0.0432 Sundman, 1949
19 66.7 4.65 0.0688 Eundman, 1949
30.6 66.7 4.65 0.0908 Sundman, 1949
10 1000 4.40 0.0488 t

20 1000 4.40 0.0552 i
* Sugar:SOz ratios between about 1:0.5 and 1: 1.
t Unpublished data from Corn Products Refining Laboratory.
Data shown in Tables VIII and IX are from determinations with

ratios of glucose :sulfur dioxide ranging from about 1:0.00035 to about
1 : 1.6 and with glucose concentrations varying from about 2.2 to about
0.5 molar and sulfur dioxide from about 0.09 to about 0.0005 molar. It
therefore appears, despite indications to the contrary by Kerp, that K is
independent of concentration of both members of the reaction providing
temperature and pH conditions are under control.
V. REACTIONVELOCITYCONSTANTS
In the formal equation
Sugar + bisulfite kzkl sugar-bisulfite
kz represents the specific rate of the forward reaction (rate of formation)

and kl the specific rate of the reverse reaction (rate of decomposition).
TABLE VIII
Effect of Reactant Concentrations on the Apparent Equilibrium Constant of Monoaes*
Sugar Sulfurous acid
concentration Total Bound Kieo c.
&Galactose
22.2 71.6 8.5 0.102
44.4 69.8 15.2 0.105
55.6 17.1 5.5 0.106
55.6 27.4 8.5 0.105
55.6 34.9 10.6 0.103
55.6 64.5 17.7 0.100
55.6 76.5 21.4 0.104
55.6 145.0 29.0 0.106
d-Glucose
333.3 75.7 26.0 0.587
333.3 72.5 25.0 0.586
333.3 65.8 23.0 0.577
166.7 75.2 15.5 0.582
166.7 75.5 15.5 0.585
166.7 76.0 15.5 0.590
55.6 70.8 5.5 0.595
55.6 75.3 5.8 0.597
55.6 89.5 6.9 0.583
d-Mannose
54.3 65.7 22.0 0.0642
55.6 88.6 26.6 0.0692
54.5 43.4 16.2 0.0643
110.0 84.7 42.6 0.0676
l- Arabinose
37.8 56.0 18.1 0.0413
66.8 47.8 24.3 0.0411
67.1 63.6 30.1 0.0412
* Sundman (1949). Temperature 19O C. (86.2O F.) and pH 4.65; concentrations in mole/l. X 108.
These constants are derived from an expression of the net change with
time in the concentration of reactants and products for the case of a
bimolecular forward reaction opposed by a monomolecular reverse
reaction. kl and kz are constant at fixed temperature and hydrogen ion
concentration; they vary significantly (increase) with decreasing hydro-
gen ion concentration above approximately pH 2.
The ratio of kl to Icz is equal to the equilibrium constant ( K ) of the
reaction. Direct experiments on rate of addition (kz) give a function of
SUGAR-SULFITE REACTION AND R E L A T I O N S H I P T O FOOD PROBLEMS 73
two constants and calculation of their separate values is possible from

experimentally determined equilibrium constants. Also, rate of decom-
position (kl) can be determined readily by simply noting the unreacted
iodine present a t any given time following the addition of an excess t o a n
equilibrated bisulfite reaction system. It is thus possible definitely t o
establish lil, since every vestige of free bisulfite is oxidized by iodine a t
the moment of release so the determination is not complicated by recom-
bination to any measurable extent.
TABLEl X
Effect of Variations in Sulfur Dioxide and Sugar Contents on the Equilibrium
Constant of Glucose Bisulfite*
pH Glucose Sulfurous acid
value concentration Total Bound Free: total SO2 K w C.
5.98 62.5 5.000 4.454 11 0.65
5.91 280.5 5.000 4.464 11 0.73
5.88 550.1 5.000 2.140 57 0.73
5.86 1101.1 5.000 3.070 39 0.69
5.83 1664.6 5.000 3.565 29 0.67
5.81 2218.9 5.000 3.875 23 0.64
5.75 1400.0 0.482 0.353 27 0.52
5.73 1400.0 9.710 6.447 33 0.71
5.68 1400.0 1.963 1.353 31 0.64
5.68 1400.0 4.970 3.370 32 0.68
5.63 1400.0 19.600 12.995 34 0.71
5.61 1400.0 55.030 36.980 33 0.68
3.97 56.6 21.000 1.617 92 0.66
3.97 277.1 21.000 6.132 71 0.66
3.97 553.6 21.000 9.618 54 0.64
3.89 2170.4 21.000 16.380 22 0.61
3.89 1100.0 0.475 0.313 34 0.52
3.84 1100.0 2.070 1.331 35 0.60
3.84 1100.0 5.070 3.209 37 0.62
3.84 1100.0 10.110 6.369 37 0.63
3.84 1100.0 20.660 12.933 37 0.63
3.82 1100.0 51.750 32.344 38 0.63
Average of all K values = 0 . 6 1
* Calculated froiii Vas (1949). Temperature 20° C. (68" F . ) ; concentrations in rnole,'l. X los;
buffer 0.2 M.
Vas (1949) studied the kinetics of the addition of glucose and bisulfite
under controlled conditions probably best suited for conclusions in fruit
juice treatments with sulfur dioxide, using concentrations of glucose far
in excess of the sulfur dioxide contents. For such conditions the system
would be expected t o behave in a pseudomonomolecular manner, since
the proportion of change of glucose from beginning t o end of t,he reaction
would be very small, However, values for kl a t 20" C. and kr a t 20" C.
were affected by pH changes in the same general manner for the glucose +
bisulfite system as Stewart and Donnally (1932a, b, c) found for benzalde-
hyde + bisulfite, in that Vas found the apparent velocity constant of
decomposition t o be almost exclusively a function of pH and independent
of either glucose or sulfur dioxide concentration. The same conclusion
(that the exact state of the SO2, H2S03,HS03-, and SOa-- equilibrium
itself affects the specific velocity constants) is, therefore, considered valid
for glucose-bisulfite systems. However, Shalaiken (1940) claimed that the
velocity of the reaction between glucose and H2S03is dependent on the
concentration of the components and that the ratio of bound t o free
bisulfite varied with concentration changes and reaction time. Sundman
(1949, p. 58) varied sugar and bisulfite concentrations in both directions
from the equimolar but less widely than Vas varied glucose and sulfur
dioxide. Separate experiments enabled the calculation of independent
values for k2, the addition reaction; these did not agree entirely with k 2
values arrived a t by working from k, and K determinations as shown in
Table X.
TABLEX
Sugar Risulfite Addition Compound Decomposition and Reaction Velocities**t.$
Bisulfite addition
compound of ki X 106 ki X 10'
Glucose 139 240

Galactose 158 1520
Arabinose 170 4130
Xylose 158 2300
Galacturonic acid 155 4730
Mannose 58 880
Lyxose 59 3160
Rhamnose 64 500
Ribose 86 3090
* Sundman (1949). Temperature 19' C. (66.2" F); pII 4.65.
t During the first 10-15 minutes.
k, = k d K .
An interesting classification of sugars appears in Table X. Those with

a cis-trans configurational relationship a t the hydroxyls on C2 and C3 all
exhibit numerically larger k~ values than the other sugars. Sundman sug-
gests that the wide spread between these factors might be of value in
interpretation of questionable sugar structures, and this seems to be a
reasonable conclusion, although insufficiently investigated.
Sundman's direct determinations of k 2 were made by time-controlled
analyses of reacting solutions (pH 4.65 and 19" C.) of 0.1 molar sugar
starting 60 seconds after mixing. It was found that k2 is dependent on
bisulfite concentration and time of reaction for five of the seven sugars
examined, but glucose and rhamnose k2 values showed independence
(Table X I ) . The effect of reaction time variations may not be unexpected

because of the theoretical dependence of the bisulfite addition on the
mutarotation pattern of the various sugars, but the unique character of
the results with glucose and rhamnose is not explained.
Ingram and Vas (1950b) also have reported differences between k z
values for glucose-bisulfite obtained by direct determinations of addition
rates and by calculations from the equilibrium constants and k, values
directly determined. They emphasized the exacting requirement for
precise pH description and control. Although it has been well established
that the equilibrium constant itself is not markedly different within the
general range of pH 3-5.5 or G , the rate at which equilibrium is reached is
affected critically. For each 0.3 pH unit in this range, the rate is approxi-
mately doubled so that for each p H unit it is increased approximately 8
times. Data from Vas (1949) indicate this characteristic which is more
nearly quantitative for kz values (Table XII). However, the ratedecrease
with increasing acidity is not of indefinite nature, since there is a slacken-
ing of rate change a t some point between pH 3.0 and 2.0. In the case of
furfuraldehyde-bisulfite, Hopner (1917) noted a sharp minimum in the
rate of dissociation between p H 1.0 and 2.0.
A distinct temperature effect on equilibrium constants is t o be noted
in Table VII. Velocity constants, likewise, are influenced very consider-
ably by temperature. Ingram and Vas (1950b) calculated kz a t 20" C.
(4.85 X and lcz a t 37" C. (1.55 X thus showing a n approxi-
mately 3-fold increase in the value of the addition rate constant in going
from 20" t o 37" C.
Table XI11 compares Sundman's kz values with Cantor and Pennis-
ton's (1940) per cent open aldehyde forms (as determined polarigraphi-
cally) and also with Delahay and Strassner's (1952) approximate rate
constants and entropies of activation for the ring-to-aldehyde opening,
also determined polarigraphically. In their study, Delahay and Strassner
found marked temperature eff ects-an observation common t o investiga-
tions of the aldehyde-bisulfite reaction.
VI. APPLICATION OF THE SUGAR-SULFITE REACTION TO FOOD PROBLEMS
It is apparent that information on the reaction between sulfur dioxide
and sugars in well-characterized systems is incomplete. Even less under-
stood is the interaction of sulfur dioxide with sugars and other food in-
gredients in the vastly more complex systems represented by fruit and
vegetable products. The intricate nature of these commodities makes it
understandable that most studies have, of necessity, been empirical.
Another complication results from the apparent variety of reactions
induced in foods by sulfur dioxide. For example, in dehydrated foods and
76 HARRY QEHMAN A N D ELIZABETH M. OSMAN
TABLE XI
Sugar-Bisulfite Addition Compound Forination Rate Constants ( k d **t*$
4 a1 SBt, SB,
Sugar seconds mole/l. X 103 inolc/l. X 103 inole/l. x 103 kz X 1 0 6
Galactose 60 18.0 1.25 8.4 1257
Galactose 60 26.3 1.70 12.3 1163
Galactose 60 68.0 3.65 27.9 965
Galactose 60 165.0 5.70 52.1 615
Galactose 60 327.0 7.75 71.1 419
Galactose 300 17.5 4.05 8.2 1078
Galactose 300 66.1 11.15 27.2 742
Galactose 300 160.0 16.00 51.1 41 1
Galactose 300 322.0 21.40 70.7 270
Galactose 600 44.0 12.40 19.3 775
Galactose 660 73.1 18.60 33.6 572
Galactose 720 177.0 29.90 54.2 370
Glucose 5 60 34.4 0.70 11.4 117

Glucose 60 147.0 3.10 44.6 122
Glucose GO 305.0 5.70 82.5 108
Glucose 60 407.0 7.20 102.5 103
Glucose 300 70.0 2.00 9.4 107
Mann ose 60 46.8 2.30 24.9 868

Mannose 60 70.5 3.30 35.0 828
Mannose 60 138.0 5.60 55.6 670
Mannose 60 277.0 9.30 75.4 60 1
Mannose 60 558.0 12.80 87.7 417
Mannose 365 70.2 13.60 34.9 698
Ribose 60 40.3 7.75 29.0 3842

Ribose 60 127.0 17.90 68.0 2872
Ribose 60 259.0 28.40 85.4 2340
Ribose 60 519.0 37.90 93.9 1619
Ribose 300 70.9 31.50 46.6 2826
Arabinose 60 44.0 5.2 28.0 2216

Arabinose 60 131.0 10.3 62.5 14G8
Arabinose 60 259.0 13.7 81.2 987
Arabinose 60 527.0 17.0 91.3 604
Arabinose 300 72.8 26.2 42.4 2053
Arabinose 600 72.8 35.0 42.4 1956
Xylose 60 34.8 2.6 18.8 1321

Xylose 60 147.0 7.2 56.8 885
Xylose 60 282.0 11.0 75.1 714
Xylose 60 422,O 12.3 83.1 532
Xylose 60 555.0 13.1 87.2 431
Xylose 317 68.8 19.4 33.7 1400
TABLEXI (Continued)
4 a, SBt, SB,
Sugar seconds mole/l. X 103 mole/l. X 103 inole/l. X 103 kz x 105
Rhamnose 60 68. 8 1.9 25.5 479
Rhamnose 60 132.0 3.3 41.6 436
Rhamnose 60 272.0 7.0 62.3 459
Rhaninose 60 560.0 13 7 79.1 452
* Sundinan (1049).
t Sugar concentration, 0.1 niole/l.; temp. 10‘ C . ; pH 4.65.
kn derived by
SB SB(SBISB - aS)
kn = t w ) aS(SBt - S B )
wherein
S = siignr concentration (mole/l.) at start of test;
a = bisulfite Concentration (mole/l.) at start of test;
I = reaction time in seconds;
S B I = sugar bisulfite concentration (mole/l.) after time L ;
S B = sugar bisulfite concentration (niole/l.) at equilibrium.
I Glucose conccntration in these tests 0.3 mole/l.
in juices, its function is generally regarded as the prevention of non-

enzymatic browning (Stadtman, 1948), whereas in frozen foods it is
thought t o act as an inhibitor of enzymes responsible for, or instrumental
in, deterioration processes (Joslyn and Ponting, 1951). A third action of
sulfur dioxide important in food preservation is its toxic effect on micro-
organisms (Tanner, 1944). Although a given set of conditions may permit
a single one of these functions t o assume prime importance, there is no
reason t o suppose th at the others are always excluded.
The greater homogeneity of fruit juices as compared with solid
products such as frozen or dehydrated fruits and vegetables, may account
for the more thorough investigations that have been made with these
products. The effects of sulfur dioxide addition to orange juice have been
reported more completely than on any other food product, especially by
13ritish workers investigating imported citrus juices ; the original purpose
of such additions was to prevent fermentation. Experience has shown
that only a portion of the added sulfur dioxide remains in the “free”
state, and it is generally believed that only this fraction is active in pre-
venting the development of microorganisms.
1. Inhibition of Fernzedation by Sulfur Dioxide

The need for sulfur dioxide in the free state was pointed out by Hailer
(1911), who examined the growth-inhibiting and lethal effects of sulfurous
acid, its sodium and potassium salts, and several of its addition com-
pounds, on microorganisms such as those commonly encountered in food
products. The bacteria used were found to be very sensitive t o the pres-
enre of sulfurous acid in the nutrient medium, the yeasts were consider-
78 HARRY GEHMAN AND ELIZABETH M. OYMAN
ably more resistant, and mold fungi were capable of withstanding rela-
tively high concentrations. The concentrations necessary to kill were in
the ratio of 1:4:5, respectively.
TABLEXI1
Values of the Apparent Equilibrium and Velocity Constants as a Function of pH at
20" C. (68" F.)*
PH K 2 0 0 C. k 1 2 0 0 c . x 108 k 2 2 0 0 C . x 103
1 2.9 0.36 0.12

2 1.0 0.44 0.44
3 0.67 2.0 3.0
4 0.61 16 26
5 0.61 130 210
6 0.79 1000 1300
*Calculated from Vas (1940).
TABLE XI11
Comparative Order of Rate Constants for Bisulfite Addition Compound Formation
and Free Aldehyde -4ppearance for Aldoses
Delahay and Strassner
Cantor and (1952) free aldehyde
Sundnian (1949) Penniston (1940) Rate Entropies
Aldose kz X 106* free aldehyde, % constants of activation
Arabinose 1550 0.13 65.5 5.7
Xylose 880 0.10 52.0 4.7
Galactose 760 0.070 23.2 3.4
Mannose 670 0.040 14.7 2.3
Glucose 125 0.012 6.4 0.6
* Averages of kl values determined directly at differing times and with variants in SO? concentration.
Sulfurous acid had a greater effect on all the organisms tested than an
equimolar solution of phenol, and the sulfur dioxide was more effective a t
37" C. (98.6" F.) than at 22" C. (71.6"F.). Hailer observed that when the
sulfur dioxide was replaced by sodium bisulfite, the germicidal power was
decreased; sodium and potassium sulfites had no effect. This observation
has been substantiated in a recent report by Schelhorn (1951). Complex
formation with carbonyl compounds decreased the activity of the bi.
sulfite to various degrees, apparently depending upon the dissociatiori
of the specific carbonyl complex involved. Addition of glucose to the
glucose-sodium bisulfite complex lowered its germicidal power, probably
by decreasing the degree of dissociation. Hailer concluded that the
quantity of sulfur dioxide needed for preservation is, therefore, dependent
primarily on the chemical nature of the medium in which it is to be used
and will thus be influenced by small fluctuations in chemical composition,
with the result that the amount of sulfur dioxide required for preservation
must be determined separately for each food product.
SUQAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 79
Muller-Thurgau and Osterwalder (1914) suggested that the behavior
of grape, pear, or apple juice could be explained by combination of the
sulfur dioxide with some product of the fermentation. They presumed
that this product was acetaldehyde, which since has been established as
an intermediate in the metabolism of sugar by yeasts (Gortner, 1949).
More recent studies by Downer (1943) are in general agreement with
these reports of Muller-Thurgau and Osterwalder. Working with sulfited
citrus juices, he found a much higher incidence of yeast infections among
samples of concentrated (3.5-5-fold) orange and grapefruit juice (104 out
of 1236) compared to single-strength juices (1 out of 688). Those juices
giving positive fermentation tests invariably showed a free sulfurous acid
content lower than the average value for the juices as a whole. Concen-
trated juices, both orange and grapefruit, gave higher reducing values
t,han the corresponding natural, unconcentrated juices. Although the total
sulfur dioxide present was greater in the concentrated juices, the per-
rentage in the free form was lower. Presumably a larger portion was in-
activated by combination with the reducing substances present, thus
accounting for t,he lower resistance of the concentrated juices t o fermenta-
tion. Fermentation of citrus juices, once it had become vigorous, usually
could not be arrested by addition of large amounts of sulfur dioxide. The
explanation suggested by Muller-Thurgau and Osterwalder (1914) for
similar behavior by grape, pear, or apple juice, that the sulfur dioxide was
inactivated through bisulfite combination with aldehydes (presumably
acetaldehyde) formed during the fermentation, appears to be applicable
here.
Because of the complexity of fruit juice as a medium and the variety
of yeasts which might be involved in its fermentation, Ingram (1948)
chose t o work with a more simple, synthetic system. I n this way it was
possible t o vary the quantity of free sulfur dioxide without simultaneously
changing any other factor which might influence the results. The yeast
(similar to but not identical with Zygosacc. prioriamus) grew at the same
rate in nutrient solutions with 401,as with 40% of glucose. TOsuch solu-
tions equal quantities of sulfur dioxide were added. Analyses showed
that much more of the sulfur dioxide remained free in the 4 % glucose
solutions. Since the total sulfur dioxide was the same in both cases, and
the differing sugar concentrations used induced the same rate of growth
when free of sulfur dioxide, differences in the rates of growth in the two
media could be attributed to differences in the proportions of free and
combined sulfur dioxide. When the total concentration was varied be-
tween approximately 0.025 and 0.00035 molar, the p H levels of the
nutrient solutions remained practically constant. This was interpreted
as indicating that variations in yeast growth in these media were de-
pendent only on available sulfur dioxide content. It was clearly shown

also that the rates of growth were comparable in media containing
different quantities of sugar but the same quantity of free sulfur dioxide.
However, the rates were entirely diflerent in solutions with the same total
quantity of sulfur dioxide but with different sugar concentrations. Thus
the conclusions drawn by others from less well-defined systems that the
effect of sulfur dioxide on yeast growth results only from that portion
which remains free were confirmed. The control of bacterial fermentation
in silages by additions of liquid sulfur dioxide (3-7 lb. per ton) has been
shown by Skaggs and Knodt (1952) to provide a means t o retain high
reducing sugar levels and to prevent the development of high lactic acid
values which, in untreated silages, provide the high acidities essential t o
stopping fermentation.
Although Ingram's experiments with glucose effectively showed the
dependence of yeast growth upon the relative absence of free (iodine-
titratable) sulfur dioxide, he observed that there was a striking difference
in the toxicity of equal quantities of free sulfur dioxide in pure glucose
solutions and in concentrated fruit juice. With pure solutions, 10 p.p.m.
of free sulfur dioxide or less were effective in killing quite large numbers
of yeast, whereas five times that concentration in concentrated fruit juice
permitted slow growth, and a t times even larger quantities were ineffec-
tive. ,The difference was not one of acidity, since it was practically the
8ame in all cases. It was concluded that in concentrated fruit juices not
only the reducing sugars combine with iodine-titratable "free" sulfur
dioxide but other constituents which form loose combinations (the sulfur
dioxide remains titratable with iodine) also contribute to decreasing its
preserving action.
Yang and Wiegand (1950) have shown that both free and total sulfur
dioxide contents of light sweet wines decrease t o definitely lower levels
when stored a t 37" C. (98.6' F.) than a t 30" C. (86" F.). The decrease in
the free component was distinctly more rapid in wine than in a 10%
glucose solution. However, loss of free sulfur dioxide was so rapid during
the first day a t each temperature that retention of preserving quantities
was insufficient to prevent fermentation. They proposed a novel system
for controlling this difficulty which consists of suspending into the wine a
slightly permeable (polyethylene) bag containing potassium metabisulfite.
2. E$ect of Processing Conditions o n the SOz-Combining Power of Sugar
Solutions
Ingram and Vas (19504 studied this reaction with several compounds
other than glucose which might be present in orange juice. These included
other sugars, other aldehydic and ketonic compounds, and pectin. It was
concluded on the basis of a paper chromatographic separation that, in

addition to sucrose, glucose, and fructose, a small amount of raffinose was
present. Galactose, arabinose, and pentoses in general were absent. Since
there was no evidence of the presence of another reducing sugar, i t was
assumed that glucose was the only sugar likely to play a n important
direct role. However, the SOz-combining power of the juice was con-
siderably higher than could be explained by the presence of glucose alone.
The pectin investigated by Ingram and Vas was a grade 100 powdered
citrus product. Examination for sugars by paper chromatography re-
vealed the presence of only glucose, which comprised about one third of
the total weight of the sample. However, the pectin appeared to have
about twice the ability to combine with sulfur dioxide th a t could he
attributed t o its glucose content.
The effect of temperature on the equilibrium constants of sugar
bisulfites is such that extent of combination is decreased with increasing
temperature (Table VII) . However, strongly heated solutions of sugars
undergo extensive degradation and some of the resultant products
themselves can combine with sulfur dioxide. Ingram and Vas (1950a)
found that, whereas unheated fructose and sucrose solutions do not com-
bine a t all, the heated solutions, particularly that of fructose, exhibit a
high combining power. The combining power of the glucose solutions not
only increases, but the percentage combining a t equilibrium is no longer
independent of the total quantity of sulfur dioxide added. With large
additions of sulfur dioxide t o glucose solutions previously heated t o a
just-perceptible change in color, the proportion combined is nearly the
same as in unheated samples; however, with little added sulfur dioxide
the percentage combined is considerably higher (Table XIV). Such be-
havior suggests that heating of sugar solutions results in the production
of substances with a high affinity for sulfur dioxide.
TABLEXIV
Effect of Strong Heating on SOz-Combining Power of Sugars**t
Sugar Heating SO2, mole/l. X lo3 Free SO2
Kind Molc/l. X l o 3 conditions Total Bound Total SO?
c:1ucose 1077 None 2.73 1.72 37
Glucose 1077 131" C.-0.5 hr. 2.73 2.15 21
Glucose 1077 Nonc 2.97 1.92 35
Glucose 1077 134" c.-1 . o 11r.t: 2.97 2.73 8
(:lucoYct I1 I I 131° c.--n.5 111.. 2.80 2.47 12
Fructose IIII 131' C.-0.5 hr. 2.80 2.72 3
Sucrose 1111 131' (3-0.5 hr. 2.80 0 . 97 05
* Calculated from Ingrani and Vns (1860a).
t Solutions heated to just-perceptible color and then buffered at pH 3.17-3.19 prior to SO2 addition.
t Heated to n light brown color.
Kerp and Bauer (1907b) found that if glucose is mixed with a com-
pound (e.g., acetaldehyde) with more affinity for sulfur dioxide, it is the
latter which combines first with the added sulfur dioxide. If acetaldehyde
is added to a glucose-bisulfite solution, it replaces glucose in the com-
pound. Therefore, if to a heated glucose solution containing such aldehyde
(or ketone) substances, the sulfur dioxide added is only sufficient t o react
with these substances, the glucose will remain largely uncombined and
the percentage of added sulfur dioxide combined at equilibrium will be
appreciably higher than in the case of a solution of pure glucose. On the
other hand, if such reactive carbonyl compounds are present only in
small amounts and the amount of sulfur dioxide added is relatively large,
these compounds may exert comparatively little apparent effect.
Ingram and Vas (1950a) indicate the possible formation of a number of
reactive carbonyl compounds in heated solutions containing sugars. The
importance of such heat breakdown products in the browning reaction
seems a significant possibility. That such color development is accom-
panied by increased SO2-combining power is indicated by data in Table
XIV, although there is no obvious simple relationship between the two
phenomena.
Nonenzymatic browning has generally been considered as originating
in melanoidin condensations involving amino acids and reducing sugars or
ascorbic acid or possible active aldehydes derived by decomposition of
these or related carbohydrate types. Because of the previously known
effect of amino acids in increasing browning, Ingram and Vas (1950a)
examined the effect on the SOz-combining power of glucose solutions
(buffered at p H 4.73) produced by heating with glycine and arginine, the
latter being the predominant amino acid in orange juice. Visual and
optical evidence of increased browning were obtained ; these were accom-
panied by increases in combined sulfur dioxide, although to a much lesser
degree.
Lewis et al. (1949) have shown that increased browning of glucose solu-
tions is not restricted to the effect of amino acids but is brought about by
the presence of carboxylic acids in general. Color formation was more
rapid with citrate than with glycine, which is one of the most reactive of
the amino acids as far as color development is concerned. With acetate,
oxalate, fumarate, tartrate, and lactate, color formation was also con-
siderable, although not as great as with glycine or citrate. Since nitrogen-
free carboxylic acids are common ingredients of many foodstuffs, they
may play a very important role in the browning reaction. Because the
inhibitory effect of sulfur dioxide on the Maillard reaction has been known
for some time, its effect in inhibiting browning brought about by these
other carboxylic acids was studied by Lewis and co-workers. I n the
SUGAR-SULFITE REACTION A N D RELATIONSHIP TO FOOD PROBLEMS 83
absence of oxygen, browning of glucose solutions with glycine, citrate, or

acetate a t pH 5.5 was completely inhibited by sulfur dioxide; at p H 6.6
color formation with glucose and citrate was only partially inhibited. In
the presence of oxygen, sulfur dioxide inhibited but did not entirely
prevent browning. The unbuffered mixtures showed, in general, a greater
drop in pH during the reaction when sulfur dioxide was present than when
it was absent. This would be expected if aldehyde-bisulfite addition
products are formed, since the free acids of these compounds appear to
be strong electrolytes (Kerp and Bauer, 1907a).
The significance of the Strecker degradation of amino acids in relation
to certain nonenzymatic browning reactions may be considerable and
may, likewise, merit attention in studying suspected sulfur dioxide reac-
tions in food systems. As noted in a review of this degradation by Schon-
berg and Moubacher (1952), no a-amino acid degradation due to inter-
action with glucose occurs in the absence of oxygen even a t 100" C.
(212' F.). IJnder conditions wherein this degradation can occur, a n
ultimate amino acid breakdown product is a n aldehyde and such non-
sugar aldehydes are definitely better binding agents for sulfur dioxide
than are the reducing sugars. Hodge and Rist (1953) have shown that the
deoxyamirioketoses formed in the Amadori rearrangement of the N-gly-
cosidic primary reaction products (or glycosylamines) of amine groups
and reducing sugars exhibit the typical characteristics of natural non-
enzymatic browning systems when in the presence of amino acids. As
shown in model experiments, the deoxyaminoketoses liberate carbon
dioxide from amino acids through a Strecker degradation. Even in the
absence of air, browning is accelerated by increases in pH and tempera-
ture but i t is significantly hindered by relatively small amounts of bisulfite
and substantially eliminated by high concentrations.
3. S d f u r Dioxide-Combining Power of Citrus Juices
Data by Vas (Table IX) indicate that the proportion of free to total
sulfur dioxide would not be expected to change a t a fixed glucose concen-
tration even though the total sulfur dioxide is increased a hundredfold.
Irigram and Vas (1950a) found this proportion t o decrease with increasing
sulfur dioxide (Table XV) when a concentrated orange juice was used
as the medium. In this respect the result was the same as with heated
glucose solutions. This agrees qualitatively with Downer (1943) th a t
with total sulfur dioxide contents in citrus juices increasing from 500 t o
2000 p.p.m., the proportion of combined sulfur dioxide tended to de-
crease, the maximum difference being 6%. Traces of sulfur dioxide appear
t o be almost completely combined, but, with increasing quantity, the
proportion combining falls and, with large additions, approaches a con-
84 HARRY QEHMAN A N D ELIZABETH M . OSMAN
stant value approximating that to be expected from combination with

the glucose in the juice. It was suggested that carbonyl-containing sub-
stances which combine preferentially with sulfur dioxide might be present
in the concentrated juice, probably as a result of heating during concen-
tration and pasteurization. Formation of substances which combine
might be augmented by heating in the presence of large quantities of
citric acid, as suggested by the work of Lewis et ul. (1949), or in the pres-
ence of substantial amounts of amino acids, ascorbic acid, and possibly
pectin, as suggested by Ingram and Vas (1950a). Volatile carbonyl com-
pounds in raw juice are probably removed to a large extent during con-
centration i n vucuo. Their presence may, however, explain the SO2-com-
bining power of fresh orange juice in excess of that accounted for by the
reducing sugars in the studies of Downer (1943), thus making the fresh
juice behave more nearly like the concentrated juice than might other-
wise be expected.
TABLEXV
Relation between Quantity of SO2 Added and Proportion Combined in Orange Juice+, t
Ratio
Equilibrium SO2, mg./l. Free SO2
Total Free Total 802
35 8 75
76 25 65
155 60 60
295 165 45
* Ingram and Vas (19BOa).
t G>'i-fold
Sharnouti concentrate diluted 1 : 1 ; glucose concentration 7.5 g./100 tnl. or about 40% com-
bined Sot.
Bisulfite addition products tend to lose sulfur dioxide a t increased

temperatures. Glucose solutions, whether heated previously or not, typi-
cally exhibit decreased power to combine with sulfur dioxide a t increased
temperature. With concentrated orange juices, however, moderate
increases in temperature increase the proportion combined. Although the
temperatures used in studying the behavior of the juice (Ingram and
Vas, 1950a) were less than those used in studying the effects of amino
acids and other carboxylic acids on browning and SO2-combining power
of pure glucose solutions, it seems probable th a t related ingredients of the
juice assist formation of carbonyl compounds a t increasing rate as the
temperature rises. The juices became perceptibly more brown, as would
be expected with such reactions.
Different concentrated orange juices were found by these investigators
to combine with sulfur dioxide a t very different rates. This can probably
be explained, in part, by differences in the p H of the juices, which com-
monly vary from 3.1 to 3.6 according to season and variety. However,
cveii a t pH 3.0 the reaction of glucosc with bisulfite reaches equilibrium

in at least 10 hours and other aldehydes react more rapidly, whereas with
orange juice equilibrium is not attained for several days, even a t tem-
peratures a t which breakdown of the juice would be expected to be
negligible. No explanation for this observation is available.
Just as the percentage of sulfur dioxide combined at equilibrium may
be related t o the presence of more reactive aldehydes, so likewise may the
increased speed of addition. Not only was the rate of reaction greater than
could be accounted for by the glucose content of the juice but no con-
sistent velocity constant values could be derived, indicating that a more
complicated reaction had taken place in which two or more substances
reacted with sulfur dioxide. It was thought that the more reactive alde-
hydic substances were probably compounds such as methyl glyoxal or
5-hydroxymethylfurfuraldehyde. Although their presence in concen-
trated orange juice has not been reported, they are formed when sugar
solutions are heated and both furfuraldehyde and 5-hydroxymethyl-
furfuraldehyde have been identified as constituents of extracts of dark-
ened apricots (Stadtman, 1948, p. 357). Fluctuations in the proportion
of combined sulfur dioxide after the first day or two, however, do not
seem t o be explained satisfactorily even by assuming the presence of such
compounds. This effect may be in some way related to the losses in total
sulfur dioxide during storage observed by Ingram (1949). Such losses
cannot be explained adequately by oxidation to sulfate on the basis of
available data and their cause remains unexplained.
a. Use of SO2 in Inhibiting Browning of Orange Juice. It has already
been pointed out that the power of heated sugar solutions t o combine
with sulfur dioxide is broadly related t o the degree of browning which has
occurred. Ingram (1949) likewise has shown that the proportion of free
to total sulfur dioxide in concentrated orange juice diminishes on storage
and is similar whether present throughout the storage period or only
added finally. This change is greatly accelerated by a moderate rise in
temperature, so that in concentrates stored for several weeks under warm
conditions, the amount of free sulfur dioxide may be reduced to negligible
proportions. Under similar conditions, with no sulfur dioxide present, the
juice turns brown. An inverse relation was found between browning due
t o heating and the proportion of subsequently added sulfur dioxide which
remains free.
Experience has shown that the presence of sulfur dioxide retards the
development of brown color in orange juice a t high temperature. Ingram
believes that if there is a relation between browning and loss of free sulfur
dioxide, more should remain unbound if it is added before rather than
after storage. His data are indicative only and certainly are not an ade-
86 HARRY GEWMAN AND ELIZABETH M. OSMAN
quate basis for definite conclusions. Ingram’s position appears to require

an assumption that the more active SOz-combining compounds and the
brown pigment are one and the same, although he suggests that com-
pounds like 5-hydroxymethylfurfuraldehyde, methyl glyoxal, and fur-
furaldehyde (compounds identified in heated sugar solutions) are respon-
sible for the increased SOS-combining power. These compounds in the
pure form are colorless. Wolfrom (1952), in a summary of his work on the
browning reaction, has proposed that an amino acid such as glycine may
catalyze the formation of furan bodies from reducing sugars and that it is
interaction between these and the amino acids which produce a pigment.
On the other hand, Guss (1952) does not consider that either furfural or
5-hydroxymethylfurfural are essential pigment precursors. However,
these furan bodies do turn brown very readily if isolated; thus the result-
ing color may not be due to the compounds themselves but to products of
further reaction. Whether these resulting colored products retain the
ability to combine with sulfur dioxide does not appear to have been
examined. It has been stated (Lewis et al., 1949) that the addition of
sulfur dioxide to isolated pigments has no effect on the color of these com-
pounds. Furthermore, darkened fruits cannot be lightened in color by
exposure to sulfur dioxide (Long et al., 1940).
The action of sulfur dioxide in preventing nonenzymatic browning
appears to involve, in part a t least, its addition to the active aldehydic
intermediates blocking their further reaction to form the brown pigments.
Existing data do not offer conclusive evidence that the presence of sulfur
dioxide in the solution slows down the formation of these “active”
aldehyde intermediates. In fact, the delayed browning reaction often
observed after addition of sulfur dioxide may result from the steady
formation of these compounds beyond the point where sulfur dioxide is
present in sufficient quantity to combine with them, at which time the
uncombined aldehydes are free to proceed to the formation of the brown
pigment. Therefore, unless the pigment retains the ability to combine
with sulfur dioxide, its formation would actually lower the SOz-combining
power of the solution by removing “active ” aldehyde compounds, con-
trary to the apparent assumption of Ingram. Hurd (1952) believes that,
after reaction of the aldehyde group of a reducing sugar with an amino
group, the first intermediate product rearranges to a ketose with the
nitrogen-bound group remaining on C1.No bisulfite addition compound
should be expected with this hypothetical ketose. However, the next step
required by this hypothesis involves the cyclization of the ketose to
form a furfural derivative (which might, combine with sulfur dioxide),
or a pyrrolecarbinal derivative by further reaction with more of the
amine.
In the somewhat unrelated case of enzymatic browning (encountered
in many frozen fruits and vegetables) the polyphenolase system has been
claimed to oxidize sulfur dioxide in a competitive reaction while the sulfur
dioxide is inactivating the enzyme (Ponting and Johnson, 1945). As the
enzyme is inactivated, ascorbic acid destruction by oxidation or reaction
to form pigment bodies is retarded. Ascorbic acid was thought to be
oxidized preferentially by the polyphenolase system which protects the
sulfur dioxide from oxidation and leaves it free to inactivate the enzyme.
Conversely, Hohl et al. (1949) reported that a combination of citric acid
and SO2 exerted a slight protective action on ascorbic acid in frozen fruit
pur6es. Bergner (1947) found that ascorbic acid in pumpkin was entirely
protected against oxidation by sulfur dioxide. Preformed glucose-
bisulfite has been reported by Cohee (1951) to be more effective in pre-
venting discoloration of frozen peaches than a citric and ascorbic acid
mixture. Preserves prepared from fruit frozen under the two methods of
treatment proved equally acceptable in taste tests.
Although it seems highly probable that the sugar of citrus juices plays
a major role in the formation of the brown color, the possibility exists that
ascorbic acid is also a factor, since an increase in browning is associated
with a decrease in ascorbic acid. Two explanations for the loss of ascorbic
acid during browning have been suggested (Stadtman 1948): (1) It acts
as an antioxidant, being oxidized in preference to other substances in the
juice (which upon oxidation yield dark pigments or precursors of dark
pigments) ;and (2) the oxidation products of ascorbic acid may themselves
be the actual precursors of dark compounds. Possible support for the
second hypothesis lies in the observation that only the oxidized form
(dehydroascorbic acid) reacts with a-amino acids to form a reddish color
(Koppanyi et al., 1945).
Hamburger and Joslyn (1941) concluded that darkening of citrus
juices starts only after all the ascorbic acid is in the dehydro form. When
there is no more oxidizable ascorbic acid, darkening occurs very rapidly.
They believe that sulfur dioxide delays the oxidation of ascorbic acid,
apparently by competing with it for the available oxygen. The process
appears to be similar in orange and lemon juices, except that with the
latter the oxidation is more rapid.
4. Browning of Dehydrated Fruits and Vegetables
The most extensive use of sulfur dioxide in food preservation is in
connection with the dehydration of fruits and vegetables; it is the most
effective inhibitor of deterioration yet found. Browning during drying and
storage, believed to be primarily nonenzymatic in nature, has long been
recognized as visible evidence of changes which affect both palatability
and nutritive values. However, its measurement in solid products is

a real problem. Earlier measurement of color was made merely by
visual inspection of the product, the storage life being taken as the time
required for the fruit to darken to the point described as the “poorest
acceptable commercially’’ (Nichols and Reed, 1931).
I n an effort to obtain a more sensitive and accurate method for deter-
mining color change in dried apricots, Stadtman et al. (1946a) measured
the light transmission through 50% alcohol extracts of the fruit, using a
photoelectric colorimeter with blue filter #420. Although this method was
reported t o be satisfactory, provided fruit of a given sulfur dioxide con-
centration was used, extracts of fruits which had been sulfured to different
sulfur dioxide levels prior to storage revealed marked differences in ab-
sorption characteristics. A high sulfur dioxide level resulted in increased
absorption in the blue relative to the green, as compared with a low sulfur
dioxide level. Consequently, photometric measurements of samples given
different sulfur dioxide treatments did not provide values consistent
with the visual appearances of the fruit. Therefore, a visual comparison
of 50% alcoholic extracts of dried fruits with standard reference solutions
under standard conditions was adopted. Measurement of browning by
this method was said to be consistent with the visual appearance of the
fruit, irrespective of the sulfur dioxide treatment.
This method was apparently quite satisfactory for the studies made
on dried apricots, but its general applicability seems questionable. The
extraction was made by 24-hour contact of the fruit with the 50% alcohol.
Whether this resulted in complete extraction of the color is not clear, but
it seems possible that with some fruits this might not be the case. Unless
all pigments formed are equally soluble, incomplete extraction might
result in different, ratios of pigments being recovered for measurement.
Thus it seems that the utility of the method depends upon complete
extraction of all unnatural pigments as well as upon the extraction of
none or a constant quantity of the natural pigments.
A very careful study of extracts of dried vegetables by Legault et al.
(1947) making use of this same general technique revealed that the in-
crease in browning pigments in unsulfited dried vegetables is linear a t
least to the point of unpalatibility over the temperature range of 24’-
49” C. (75.2-120.2” F.). However, there was deviation from linearity if
the vegetables were sulfited. Legault et al. (1949) then studied the rates
of sulfite disappearance in dehydrated vegetables and concluded that
(1) similarities between the rates of change with moisture increases and
temperature rise and (2) apparent activation energies involved in browning
sfid in sulfite disappearance indicated a common rate-determining step
for the two processes, Since sulfite disappearance may originate largely
from oxidation by atmospheric oxygen and binding in complexes, i t is not

surprising t ha t the rate of disappearance is only about half as great in
nitrogen as in air and that prolonged soaking in water before dehydration
(which reduced the total sugar content by half) markedly decreased the
sulfite disappearance. Working with potatoes, Ross et al. (1945) showed
a definite relationship between high reducing sugar content and dis-
coloration during the heating stages of dehydration. Ross (1948) has
shown that potatoes dried and stored optimally can be handled a t about
70% higher reducing sugar contents if sulfited. Burton (1945) noted th a t
the browning reaction in potatoes can be inhibited by reducing agents and
compounds which react with carbonyl groups; sulfur dioxide definitely
belongs in this classification. Ross (1948, p. 278) found that rate of loss
of sulfur dioxide (part of which may be presumed to have been combined
chemically) was seven times as great a t 48.9" C. (120" F.) as a t 37.8" C.
(100" F.) with potatoes containing 2.5% or more reducing sugar. The
rate was only three to four times as great with lesser amounts of sugar.
Sulfured and dried apricots stored in friction top jars for 10.5 years
were found by Sorber (1944) t o have an unchanged total sulfur content
but only 10% remained which could be accounted for as sulfur dioxide.
Furthermore, about half could not be found in the sulfite and sulfate total,
which indicated that much of the original sulfur was now combined with
fruit constituents in other than volatile or barium-precipitable forms.
Observations of earlier investigators on the effect of temperature on
rate of browning were related by Stadtman et al. (1946b) to sulfur dioxide
disappearance, carbon dioxide production, and oxygen consumption.
The rates of all these reactions were found to be in agreement with the
Arrhenius equation, since the logarithms of the rates were a linear func-
tion of the reciprocal of this equation. They found that the temperature
coefficient of the browning reaction of apricots is quite independent of
moderate changes in sulfur dioxide, moisture, oxygen concentration, and
age of the fruit. For this reason the storage life a t low temperature can be
predicted from data obtained in accelerated storage tests a t 49" C .
(120.2" F.). They concluded from the linear relation of temperature t o
rate of browning that either the same reaction controls the rate a t all
temperatures or that, if several reactions are involved, they must have
nearly identical temperature coefficients. This work, along with th a t of
Stadtman et al. (1946a, c), led to the general conclusion th a t although
sulfur dioxide retards, i t does not prevent, deterioration. The storage life
of apricots was found t o be directly proportional to initial sulfur dioxide
content within the range of 1500-8000 p.p.m.
Quinn (1926) has stated that sulfur dioxide absorption is more rapid
by fruits that are fully mature, which seems reasonable if reducing sugars
assist iii binding it within the fruit. Perry et al. (1946) have indicated that
the more immature the fruits the more rapid the loss of sulfur dioxide
during the process of drying. This might be predictable, since the relative
absence of reducing sugars in immature fruits provides a less effective
SOz-binding system.
Fixation of sulfur dioxide by fruits is not directly proportional to
absorption (Long et al., 1940). Quickly sulfured fruit of high initial sulfur
dioxide content loses it a t the fastest rate; this has been attributed to a
shorter "fixation-time," which might well be described as being too short
for establishing adequate reaction equilibria with the sugars and other
aldehydes of the fruit.
Increases in temperature during or immediately after sulfuring appear
to cause an increased retention of sulfur dioxide, and this, too, may be
explainable by a resultant increase in the rate of formation of bisulfite
addition products. Blanching in steam or hot water prior to sulfuring has
been reported (Nichols and Christie, 1930a; Chace et al., 1933) to have no
effect on the sulfur dioxide absorption or retention capacity of the fruit.
Nonetheless Long et al. (1940) found that fruit blanched after sulfuring
retained 50% more sulfur dioxide than the unblanched fruit. Likewise,
Nichols and Christie (1930b) state that an increase in the temperature
of the sulfur chamber increases the rate of combination of sulfur dioxide
with other substances in the fruit, as well as increasing penetration of it
into the fruit tissues. Fisher et al. (1942) report that fruit exposed to sulfur
dioxide in the high-temperature areas of a sulfur chamber retains more
than fruit from cooler areas. Long et al. (1940) found that sulfuring fruit
at relatively high temperatures (100"-120" F.) tends to decrease absorp-
tion but increase retention.
Temperature during drying also seems to affect the sulfur dioxide
retention possibly through increased formation of bisulfite addition
products. Long et al. (1940) found that heating of the fruit by solar
radiation and drying of the surface appeared to increase the fixation of
sulfur dioxide. They advised spreading the fruit in the drying-yard
immediately after sulfuring to secure full exposure to the sun. Nichols
et al. (1939) had previously reported that complete shade drying of
apricots reduced the retention of sulfur dioxide and that direct exposure
of the fruit to sunlight for one day was sufficient to prevent undue losses.
Rapid drying in a dehydrator caused increased retention, sometimes as
great as 375% of that observed in sun-dried control samples (Long et al.,
1940). This increase may likewise be attributable, at least in part, to the
increased temperature and the accompanying increased rate of SOrbind-
ing through bisulfite addition compound formation. The sulfur dioxide
retention was further increased by preheating the fruit for 2 to 4 hours
in the dehydrator with the vents closed, thus raising the fruit to de-
hydrating temperature more rapidly.
RESEARCH
VII. NEEDED
Further studies of the relation of the sugar-bisulfite reaction to the
complex and economically important problem of browning in foods seems
amply justified. Despite the fact that the “addition compound theory”
alternate of Stadtman (1948) has not been established as a definite
chemical mechanism in nonenzymatic browning, accumulating secondary
implications continue t o lend credence to this concept. Kinetic informa-
tion derived from model systems has provided a reasonable background
for application of sugar-bisulfite studies to practical systems. Although
a direct study of the formation of these compounds in fruits and vege-
tables dried after sulfuring, for example, has not been made, many of the
conditions which are known t o favor sulfur dioxide retention are those
which would be expected to increase the rate of formation of bisulfite
addition products.
Improvement and simplification of analytical procedures for deter-
mination of total contained sulfur dioxide, regardless of form, and for an
unquestionable distinction between bound and unbound states in com-
plex food systems are prime needs. This is particularly true for composi-
tions involving both ascorbic acid and sulfur dioxide.
s
REFERENCE
Zdams, R., and Garber, J. D. 1949. Amine bisulfites. 11. Their use as resolving agents
for aldehydes and ketones. J . A m . Chern. SOC.71, 522.
i2dams, R., and Lipscomb, R. D. 1949. Amino bisulfite addition products of aldehydes
and ketones. I. J . A m . Chena. Soc 71, 519.
’idler, E. 1946. On sockersulfonsyror, deras bildning vid sulfitprocessen och deras
konstitution. Sriensk Papperstzdn. 49, 339.
Alexander, E. R. 1950. “Principles of Ionic Reactions.” John Wiley & Sons, New
York, p. 161.
Backer, H. J., and Mulder, H. 1934. Acetylation of some 1,l-aminosulfonic acids and
1,l-hydrazinesulfonic acids. Rec. trav. chirn. 63, 1120.
Ilergner, K. G. 1947. The behavior of L-ascorbic acid-oxidizing enzymes and copper
ions toward sulfur dioxide. Deut. Lebensm.-Rundschau 43, 95; C . A . 44, 7363
(1950).
I h t a g n i n i , C. 1853. Ueber die Verbindungen einiger Oele mit den zweifachschweflig-
sauren Alkalien. Ann. 86, 268.
Ikowne, H. H. 1944. Evidence for the existence of bisulfite compounds of sugars.
J . Org. Chem. 9, 477.
Uurton, W. G. 1945. Mashed potato powder. 111. High temperature browning of
mashed-potato powder. J . SOC.Chem. Znd. 64, 215.
Cantor, S. M. and Peniston, Q. P. 1940. Reaction of aldoses a t the dropping mercury
electrode. J . Am. (‘hem. Sor. 62, 2113.
Caughlan, C. N. and Tartar, H. V. 1941. Raman spectra of aldehyde and ketone

bisulfites. J. Am. Chem. SOC.63, 1265.
Chace, E. M., Church, C. G., and Sorber, D. G. 1933. Large scale experiments in
sulfuring apricots. 11. Effect, of dehydrating, shade, drying, and blanching. Ind.
Eng. Chem. 26, 1366.
Cohee, R. F., Jr. 1951. Glucose-sulfur helps hold color in peach preserves. Food Pucker
32 (8), 26.
Delahay, A., and Strassner, J. E. 1952. A polarographic study of the kinetics of the
ring -+ aldehydo transformation for various aldoses. J. Am. Chem. SOC.74, 893.
Downer, A. W. E. 1943. The preservation of citrus juices with sulphurous acid.
J. SOC.Chem. Ind. 62, 124.
Eibner, A. 1901. Ueber Verbindungen dcr schwefligen Siiure und der Alkalibisulfite
mit Schiff’schen Basen. Ann. 316,89.
Feinberg, B. G. 1913. A quantitative study of some aldehyde reactions. Am. Chem. J.
49, 87.
Fisher, C. D., Mrak, E. M., and Long, J. D. 1942. Effect of time and temperature of
sulfuring on absorption and retention of sulfur dioxide by fruits. Fruit Products J.
21, 75.
Friedman, H. L., and Braitberg, I,. D. 1947. Glucose bisulfite coinpounds of alkoxy
amino pyridine. U. S. patent 2,426,011.
Gibson, D. T. 1934. Significant studies in the organic chemistry of sulfur. Chem. Revs.
14, 435.
Gortner, R . A 1949. “Outlines of Biochemistry.” John Wiley & Sons, New York,
3rd Ed., p. 707.
Cuss, C. 0. 1952. Contributions of browning research to ration item stability. Pub-
lished by the Research and Development Associates, Food and Container In-
stitute of the Armed Forces, Inc., Chicago, p. 29.
Hiigglund, E. 1929. Uber den Einfluss von Bisulfit-Losungen auf Zuckerarten bei
hoherer Temperatur. (I. Mitteil.). Ber. 62,437.
Hagglund, E., Heiwinkel, H., and Bergek, T. 1943. Uber die Einwirkung von Sulfit-
kochsauren a d Zuckerarten und die Vergarbarkeit von Sulfitablaugen ver-
schiedener Herkunft. J. prukt. Chem. 162, 2 .
Hagglund, E., Johnson, T., and Urban, H. 1930. Uber den Einfluss von Bisulfit-
Losungen auf Zuckerarten bei hoherer Temperatur. (I11 Mitteil.). Ber. 63B,1387.
Hagglund, E., and Urban, H. 1929. Uber den Einfluss von Bisulfit-Losungen auf
Zuckerarten bei hoherer Temperatur. (I1 Mitteil.). Ber. 62B,2046.
Hailer, E. 1911. Experiments on the properties of free sulfurous acid of sulfites, and a
few complex compounds of sulfurous acid in killing germs and retarding their
development. Arb. kaiserl. Gesundh. 36, 297; Chem. Abstr. 6, 1805.
Hamburger, J. J., and Joslyn, M. A. 1941. Auto-oxidation of filtered citrus juices.
Food Research 6,599-619.
Heiwinkel, H. 1944. S-containing carbohydrate compounds in sulfite waste liquor.
Svensk Papperstidn. 47, 265; C . A . 38, 5082 (1944).
Hodge, J. E., and Rist, C. E. 1953. The Amadori rearrangement under new conditions
and its significance for noneneymatic browning reactions. J . Am. Chem. Soc. 7 6 ,
316.
Hohl, L. A., Buck, P., and Rosoff, H. 1949. Further studies on frozen fruit and vege-
table purees. Food Technol. 3, 100.
Hopner, T. 1944. Uber die loslichen Spaltungsprodukte des Fichtenholaes bei der
Allfitz~llst,offkorhung.Cellicloaerh~eniis.22, 72.
Hopner, T. 1947. uber die Beziehungen zwichen Furfurol und schwefliger Saure und
die analytische Bestimmung des Furfurol. Das Papier 1, 102.
Hurd, C. D. 1952. Contributions of browning research to ration item stability. Pub-
lished by the Research and Development Associates, Food and Container In-
stitute, of the Armed Forces, Inc., Chicago, p. 23.
Ingram, M. 1947a. An electrometric indicator to replace starch in iodine titrations of
sulphurous acid in fruit juices. J . SOC.Chem. Ind. 66, 50.
Ingram, M. 1947b. Investigation of errors arising in iodometric determination of free
sulphurous acid by the acetone procedure. J . SOC.Chem. Ind. 66, 105.
Ingram, M. 1948. The germicidal effects of free and combined sulphur dioxide. J . SOC.
Chem. Ind. 67, 18.
Ingram, M. 1949. Behavior of sulphur dioxide in concentrated orange juice. Food
Research 14, 54.
Ingram, M., and Vas, K. 1950a. Combination of sulphur dioxide with concentrated
orange juice. I. Equilibrium states. J . Sci. Food Agr. 1, 21.
Ingram, M., and Vas, K. 1950b. Combination of sulphur dioxide with concentrated
orange juice. 11. The rate of reaction. J . Sci. Food Agr. 1, 63.
Joslyn, M. A., and Marsh, G. L. 1935. Browning of orange juice. Ind. Eng. Chem. 27,
186- 189.
Joslyn, M.A., and Ponting, J. D. 1951. Advances in Food Research 3, 1-44.
Kerp, W. 1903. u b e r organisch gebundene schweflige Saure in Nahrungsmitteln. 2.
Untersuch. Nahr.-u. Genussm. 6 , 66.
Kerp, W. 1907. Sulfurous acid in foods. Chem. Ztg. 31, 85.
Iierp, W., and Bauer, E. 1904. Zur Kenntnis der gebundenen schwefligen Sauren. I.
Arb. kaiserl. Gesundh. 21, 180.
Kerp, W., and Bauer, E. 1907% Zur Kenntnis der gebundenen schwefligen Sauren.
11. Arb. kaiserl. Gesundh. 26,231.
Kerp, W., and Bauer, E. 1907b. Zur Kenntnis der gebundenen schwefligen Sauren.
111. Arb, kaiserl. Gesundh. 26, 269.
Kerp, W., and Bauer, E. 1907c. Die elektrolytische Dissoziationskonstante der
schwefligen Saure. Arb. kaiserl. Gesundh. 26, 297.
Kerp, W., and Wohler, P. 1909a. Zur Kenntnis der gebundenen schwefligen Sauren.
IV. Arb. kaiserl. Gesundh. 32, 99.
Kerp, W.,and Wohler, P. 1909b. Zur Kenntnis der gebundenen schwefligen Sauren.
V. Arb. kaiserl. Gesundh. 32, 120.
Kolthoff, I. M., and Stenger, V. A. 1942. “Volumetric Analysis.” Interscience Pub-
lishers, New York, 2nd Revised Ed., Vol. 1, p. 213.
Koppanyi, T., Vivino, A. E., and Veitch, F. P., Jr. 1945. A reaction of ascorbic acid
with a-amino acids. Science 101, 541.
Lauer, W. M., and Langkammerer, C. M. 1935. The constitution of the bisulfite
addition compounds of aldehydes and ketones. J . A m . Chem. SOC.67, 2360.
Legault, R. R., Hendel, C. E., Talburt, W. F., and Rasmussen, L. B. 1949. Sulfite
disappearance in dehydrated vegetables during storage. Ind. Eng. Chem. 41, 1447.
Legault, R. R., Talburt, W. F., Mylne, A. M., and Bryan, L. A. 1947. Browning of
dehydrated vegetables during storage. Ind. Eng. Chem. 39, 1294.
Lewis, V. M., Esselen, W. B., Jr., and Fellers, C. R. 1949. Nitrogen-free carboxylic
acids in the browning reaction. Znd. Eng. Chem. 41, 2591.
Lippich, Fritz. 1932. Die Reaktion zwischen Zucker und Cyankalium in ihren Be-
ziehungen sum Problem der Zuckermodifikation in wasseriger Losung. Biochem.
2.248, 280.
Long, J. D., Mrak, E. M., and Fisher, C. D. 1940. Investigations in the sulfuring of
fruits for drying. Calif. Agr. Expt. Sta. Bull. No. 636, 5.
Mapson, L. W. 1942. The determination of ascorbic acid in fruits and vegetables in
the presence of SO2. Biochem. J . 36, 196.
Markh, A. T. 1950. Color changes in fruit products during heat treatment. Biokhimiya
16, 107; C . A . 44, 6986 (1950).
Marusawa, Den-Ichi Naito, and Uchida, Jun-Ichi. 1929. Contribution on the knowl-
edge of the sulfite process. I. The action of bisulfite liquor on various sugars.
Mem. Ryojun Coll. Eng. 1, 351; C . A . 23, 4340.
Mendelejeff, D. 1859. ttber die onanthol-schweflige Saure. Ann. 110, 241.
Mohammad, A., Fraenkel-Conrat, H., and Olcott, H. S. 1949a. The “browning”
reaction of proteins with glucose. Arch. Biochem. 24, 157.
Mohammad, A., Olcott, H. S., and Fraenkel-Conrat, H. 1949b. The reaction of pro-
teins with acetaldehyde. Arch. Biochem. 24, 270.
Miiller, M. 1873a. Ueber Oxymethanesulfonsaure und Oxymethandisulfonsaure. Ber.
6, 1031.
Mtiller, M. 187313. Ueber Oxypropansulfonsaure und die Verbindung des Acroleins
mit den sauren schwefligsauren Alkalien. Ber. 6, 1441.
Muller-Thurgau, H., and Osterwalder, A. 1914. Einfluss der schwefligen Saure auf die
durch Hefen und Bakterien verursachten Garungsvorgange im Wein und Obst-
wein. Landwirtsch. Jahrb. Schweiz 28, 480-548.
Nichols, P. F., and Christie, A. W. 1930a. Dehydration of grapes. Calif. Agr. Expt.
Sta. Bull. No. 500.
Nichols, P. F., and Christie, A. W. 1930b. Drying cut fruits. Calif. Agr. Expt. Sta.
Bull. No. 485.
Nichols, P. F., Mrak, E. M., and Bethel, R. 1939. Effect of drying and storage condi-
tions on color and SO2 retention of dried apricots. Food Research 4, 67.
Nichols, P. F., and Reed, H. M. 1931. What happens in the tropics. Western Canner
and Packer 23, 11-15.
Perry, R. L., Mrak, E. M., Phaff, H. J., Marsh, G. L., and Fisher, C. D. 1946. Fruit
dehydration. I. Principles and equipment. Calif. Agr. Expt. Sta. Bull. No. 698.
Ponting, J. D., and Johnson, G. 1945. Determination of sulfur dioxide in fruits. Znd.
Eng. Chena., Anal. Ed. 17, 682.
Potter, E. F., and Hendel, C. E. 1951. Determination of sulfite in dehydrated white
potatoes by direct titration. Food Technol. 6, 473.
Pringsheim, Hans. 1932. “Chemistry of the Monosaccharides and the Polysao-
charides.” McGraw Hill Book Co. New York, p. 17.
Quinn, G. 1926. Notes on sulfuring fruits prior to drying. J . Dept. Agr. S. Australia
SO, 500.
Rahn, O., and Conn, J. E. 1944. Effect of increase in acidity on antiseptic efficiency.
Ind. Eng. Chem. 36, 185.
Raschig, F., and Prahl, W. 1926. Die Konstitution der Aldehyde- und Keton-Bisulfite.
Ann. 448, 265.
Redtenbacker, J. 1848. Ueber die Constitution des Taurins und einen damit isomeren
Korper. Ann. 66, 37.
Reinking, K., Dehnel, E., and Labhardt, H. 1905. Zur Constitution der Aldehyd-
schwefligsauren Salze und der hydroschwefligen Saure. Ber. 38, 1069.
Ripper, M. 1892. Estimation of sulphurous acid in wine. J . prakt. Chem. (2) 46, 428;
1893. J . Chem. SOC.Abst. 11, 189.
Rocques, M. X. 1898. L’acide sulfureux dans les vins. J . pharm. chim. 7, 605.
Ross, A. F. 1948. Deterioration of processed potatoes. Advances in Food Research 1,

284.
Ross, A. F., Hilborn M. T., and Jenness, L. C. 1945. Discoloration can be avoided.
Food Packer 26 (lo), 38, 40, 42, 78.
Schelhorn, M. V. 1951. Untersuchungen uber Konservierungsmittel VII. Wirksamkeit
und Wirkungsbereich der schwefligen Saure. Deut. Lebensm.-Rundschau 47, 170.
Schiff, H. 1881. Ueber Aldehydodisulfite von Aminsauren und von Aminen. Ann. 210,
123.
Schonberg, A., and Moubacher, R. 1962. The Strecker degradation of a-amino acids.
Chem. Revs. 60, 261.
Schroeter, G. W. 1918. Ueber Methinonsaure und deren Verwendung zu Synthesen.
Ann. 418, 161.
Schroeter, G. W., and Sulzbacher, M. 1928. u b er die chemische Konstitution der
Aldehyde- und Keton-Bisulfite. Ber. 61B, 1616.
Shalaiken, F. 1’. 1940. The reaction of sulfurous acid with glucose. J . Appl. Chem.
(U.S . S. R.) 13, 263; C . A . 36, 370 (1941).
Skaggs, S. R., and Knodt, C. B. 1952. Sulfur dioxide preservation of forage crops.
J . Dairy Sci. 36, 329.
Sorber, D. G. 1944. The relation of sulfur dioxide and total S content of dried apricots
to color change during storage. Fruit Products J . 23, 234.
Stadtman, E. R. 1948. Nonenxymatic browning in fruit products. Advances i n Food
Research 1, 325-372.
Stadtman, E. R., Barker, H. A., Mrak, E. M., and Mackinney, G. 1946a. Storage of
dried fruit. Influence of moisture and sulfur dioxide on deterioration of apricots.
I d . Eng. Chem. 38,99.
Stadtman, E. R., Barker, H. A. Haas, V., and Mrak, E. M. 1946b. Storage of dried
fruit. Influence of temperature on deterioration of apricots. Znd. Eng. Chem. 38,
541.
Stadtman, E. R., Barker, H. A., Haas, V., Mrak, E. M., and Mackinney, G. 1946c.
Storage of dried fruit. Gas changes during storage of dried apricots and influence
of oxygen on rate of deterioration. Znd. Eng. Chem. 38, 324.
Stelling, 0. 1928. Constitution of hydrogen sulfite compounds of aldehydes and
ketones. Cellulosechemie 9, 100; C. A . 23, 5465 (1929).
Stepanov, A. V., and Stepanenko, B. N. 1937. Active form of simple sugars IV. Com-
parative study of the addition of hydrocyanic acid to glucose 6-phosphate and to
glucose. Biokhimiya 2, 917; C . A . 32, 2513 (1938).
Stepanov, A. V., and Stepanenko, B. N. 1940. Active form of simple sugars VII.
Reactivity of fructose 1,g-diphosphate. Biokhimiya 6, 567; C . A . 36, 4742
(1941).
Stewart, T. D., and Donnally, L. H. 1932a. The aldehyde bisulfite compounds. I.
The rate of dissociation of benzaldehyde sodium bisulfite as measured by its first
order reaction with iodine. J . A m . Chem. SOC.64, 2333.
Stewart, T. D., and Donnally, L. H. 1932b. The aldehyde bisulfite compounds. 11.
The effect of varying hydrogen ion and of varying temperature upon the equi-
librium between benzaldehyde and bisulfite ion. J. Am. Ckem. SOC.64, 3555.
Stewart, T. D., and Donnaliy, I,. H. 1 9 3 2 ~The
. aldehyde bisulfites. 111. The effect of
changing hydrogen ion concentration upon the specific reaction rate of the addi-
tion of sodium bisulfite to benzaldehyde. IV. Discussion of the equilibrium rates,
and temperature coefficients as affected by hydrogen ion concentration. J . Am.
Chem. SOC.64, 3559.
Sundman, J. 1949. Sockerbisulfiternas Egenskaper och Kvantitativ bestamning av

Olika Sockerslag, Sarskilt i Sulfitavlut, med Tillhjalp av Na-Bisulfit. Disserta-
tion, Finnish Institute of Technology.
Sundman, J. 1951. Sokeribisulfiiteista. Suomen Kemistilehti 24, 3.
Tanner, F. W. 1944. “The Microbiology of Foods.” Garrard Press, Champaign,
Illinois, p. 633.
Thompson, A. F., and Cromwell, N. H. 1939. The preparation and properties of alde-
hydes containing deuterium in the functional group. J . Am. Chem. SOC.61, 1374.
Thompson, J. B., and Toy, E. 1945. Determination of sulfur dioxide. Improved
Monier-Williams Method. Znd. Eng. Chem., Anal. Ed. 17, 612.
Tillitson, E. W. 1942. Organic amine derivatives and method of obtaining same.
U. S. patent 2,287,071.
Vas, K. 1949. The equilibrium between glucose and sulphurous acid. J . SOC.Chem. Znd.
68, 340.
Whitmore, F. C. 1937. “Organic Chemistry.” D. Van Nostrand Co., New York, p. 229.
Wiesner K. 1947. Polarographic investigation of the electroreduction of sugars.
Collection Czech Chem. Communs. 12, 64;C . A . 41, 5459.
Wolfrom, M. L. 1952. Contributions of browning research to ration item stability.
Published by the Research and Development Associates, Food and Container
Institute of the Armed Forces, Inc., Chicago, p. 39.
Yang, H. Y., and Wiegand, E. H. 1950. A method for maintaining free sulfur dioxide
content in wine. Food Technol. 6 , 103.
The Chemistry and Technology of the Pretreatment and Preservation of
Fruit and Vegetable Products with Sulfur Dioxide and Sulfites
BY M. A. JOSLYN AND J. €3. S. BRAVERMAN*
Ijepurtnaent of Food Technology, University of California, Berkeley, California
CONTENTS
Page
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
11. Chemistry of Sulfur Dioxide, Sulfites, and Their Organic Compounds. . . . 99
111. Determination of Sulfur Dioxide in Fruit and Vegetable Products. . . . . . . 113
IV. Sulfur Dioxide as Preservative and Sanitizing Agent.. . . . . . . . . . . . . . . . . . 123
V. Sulfur Dioxide as an Inhibitor of Enzymic and Nonenzymic Browning.. . 128
VI. Source and Application of Sulfur Dioxide.. . . . . . . . ...130
VII. Sulfur Dioxide in Fruit Juices, Syrups, Concentrates and PurBes . . . . . . . . 134
VIII. Sulfur Dioxide in Wine and Vinegar Maki .......... 136
IX. Sulfur Dioxide in Dehydrated and Dried F Products.. 141
X. Sulfur Dioxide in “Brining” of Cherries and “Barrelling” of Fruit. . . . . . 143
XI. Sulfur Dioxide in Transportation and Storage of Grapes and in Other
Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
I. INTRODUCTION
Sulfur dioxide, in its gaseous or liquid form, or dissolved in water t o
form sulfurous acid, or in the form of its neutral or acid salts (sulfites,
bisulfites, metabisulfites) is used widely as a chemical preservative t o
reduce or prevent spoilage by microorganisms, and as a selective inhibitor
of undesirable organisms in the fermentation industries. It is also used as
an antioxidant and inhibitor of enzyme-catalyzed oxidative discoloration
and of nonenzymic browning during preparation, storage, or distribution
of many food products. As an antioxidant it is useful also in improving
the retention of ascorbic acid, carotene, and other oxidizable biologically
active components. The preservative value of sulfurous acid for color
retention and its inhibition of microbial growth and activity have long
been recognized, and i t has long been used for the control of undesirable
changes in color and flavor during processing and as a relatively cheap
and readily available preservative for a wide variety of products-juices,
* Dr. Braverman now is Professor of Food Technology at Technion, Israel Institute of
Technology, Haifa, Israel. The authors gratefully acknowledge the assistance of
Dr. C. J. B. Smit, now a t Stellenbosch University, South Africa, for preparing the
graphs used in the text and for assisting in compiling the literature.
97
98 M . A. JOSLYN AND J. B. S. BRAVERMAN
syrups, concentrates, purBes, fermented beverages, dehydrated and sun-

dried foods, in preserves, jams, and marmalades, in brined fruits and pulps
for subsequent processing, and in the storage and distribution of perish-
ables (fresh fruits, vegetables, meat, and fish), Its pronounced bleaching
action has been employed in the preparation of specialty products such
as Maraschino cherries, citron and citrus peel, watermelon rind, and
Zucca melon. The convenience with which it can be used and the ease
with which its concentration can be reduced to a level low enough to be
tolerated in certain products make it peculiarly attractive.
Sulfur dioxide has been used in food preservation since ancient times.
There is evidence that the use of the fumes of burning sulfur in wine
making was known to the early Egyptians and Romans (Bioletti, 1911).
The old practice of sulfuring was strictly empirical and based on secular
experiences and customs. Its correct use was possible under these condi-
tions only in a few cases and in the hands of wine makers of long experi-
ence. Its abuses were many and were recognized as early as 1738 in
France, according to Brhmond (1937). It was not until the beginning of
the last decennial of the nineteenth century that serious study of the
effects of sulfurous acid on must and wine were undertaken in France.
Shortly afterwards, in 1902, the use of sulfur dioxide in wine making in
France was regulated (BrBmond, 1937). Sulfuring of fruits for drying was
an established practice in the late nineteenth century.
Although the practice of sulfuring fruit for drying was widely used in
California in the late nineteenth century, it apparently was not recognized
abroad. In 1902, Beythien and Bohrisch discovered that practically all
the dried fruit imported into Germany from America was heavily sulfured
and this observation was followed by publication of numerous papers in
which the admissibility of sulfur dioxide as a preservative was actively
debated (Beythien and Bohrisch, 1902, 1903).
It was used for preserving meats in the United States as early as 1813
and somewhat later for fish. So widespread was the use of sulfur dioxide
in foods in the nineteenth century that Wiley (1907) warned against its
promiscuous use as a preservative (Anon., 1907).
Sulfurous acid was used almost from the beginning in the purification
of sugar-beet juices in the European beet sugar industry (Maxwell, 1916;
McGinnis, 1951). I t was introduced by Proust in 1810 in the form of
calcium sulfite, but since 1858 sulfur dioxide gas has been used for
decolorization of cane and beet juices. It was first tested with cane juice
in Mauritius in 1865 and introduced into Java in 1895 (Marches, 1953).
In the period 1935-1940, 40% of the white sugar produced annually in
Java was sulfitation sugar. In the purification of sugar-containing juices
it is used for neutralization of excess alkalinity, for the decolorization of
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 99
the extracted juices by reduction of naturally occurring pigments and

colored decomposition products, and for prevention or inhibition of color
formation in later stages of processing, evaporation, and crystallization.
It has been claimed t o facilitate crystallization. The sulfitation process in
cane-sugar and beet-sugar technology is described in detail in the texts
on sugar technology edited by Honig (1953) and McGinnis (1951).
Investigations of the scientific basis of the use of sulfur dioxide and
sulfites in the preparation and preservation of foods have been carried
out on wines early in the twentieth century and are still being conducted
both in the United States and in Europe. Extensive investigations of
sulfuring practices and problems were conducted by Long et al. (1940),
during the period of 1936-1939. The widespread use of sulfur dioxide in
the dehydration of vegetables during World War I1 in Britain and British
Dominions and in the United States focused attention anew upon the
problems involved. The intensive investigation of the fundamental
nature of browning and discoloration in processed foods (Mitchell and
Peterson, 1952) also resulted in support of investigations on the chemistry
of the interaction of sulfite with sugar. In spite of its long history of use
and the many factors influencing its applicability our knowledge of the
chemistry and technology of sulfurous acid and sulfite pretreatment and
preservation of foods is still incomplete. The chemistry of the products
formed and the state of sulfurous acid in foods, the reactions involved in
their formation, the changes in the initial products formed during storage,
and the actual nature of the preservative and antibrowning effect still
have t o be established. The present status of our knowledge in this field
is summarized in the following review, which is restricted to foods of plant
origin. Only the more important literature in the field is reviewed and the
references cited are not complete. Where they are available, reference is
made t o review articles and all phases of the field are not reviewed in the
same detail.
OF SULFUR
11. CHEMISTRY SULFITES,
DIOXIDE, AND THEIRORGANIC
COMPOUNDS
1. Sulfur Dioxide and the Sulfites
Sulfur dioxide formed by burning sulfur in air also contains between
6-8 % of sulfur trioxide, whose presence accounts for the ‘(foggy” appear-
ance of the gas, the fog being due t o droplets of water condensed about
SO3. The sulfur trioxide present in sulfur dioxide may be determined
readily by suitable filtration (Eckman, 1927). The formattion of sulfur
trioxide is promoted by high oxygen content, low temperature of the air
stream in sulfur furnaces, and presence of catalysts such as iron oxide.
100 M. A. JOSLYN A N D J. B. S. BRAVERMAN
Complete combustion of sulfur in sulfur furnaces used in the sugar indus-

try or in the sulfur houses used in the dried fruit industry which would
theoretically produce gas with 21 volume per cent of SO2is not obtained.
According to Marches (1953) the gas produced in the sulfur furnaces used
in Java contains 6-14 volume per cent of SO2 and small amounts of SOa.
PH
FIG.1. Distribution of various constituents of sulfurous acid at various pH values.

After Vas and Ingram (1949).
In more efficiently designed and operated sulfur burners such as those

used in the paper industry and in the manufacture of sulfuric acid
(Shreve, 1944), the sulfur is melted, atomized in compressed air, and
burnt in a separate combustion chamber. In these burners gas of a con-
stant composition, 19-20 volume per cent SO2, without any sublimation
and with only 0.14%of the sulfur transformed into SOi, can be obtained.
Sulfur dioxide is a colorless gas having a characteristic odor, a normal
molecular volume of 21.89 l., and a molecular weight of 64.06 g. It is
soluble to the extent of 36.4 volumes in one volume of water at 20" C. Its
solubility in water decreases from 8.6% by weight a t 20" C. t o 0.1% at
100" C. At atmospheric pressure SO2liquefies a t - 10" C.; a t 20" C. liquid
SO2 exerts a pressure of 3.25 atm. or 40.6 p.s.i. The solubility of sulfur
dioxide in water has been determined accurately by Beuschlein and
Simenson (1940), and the thermodynamic data for the system S02-Hn0
are given by Plummer (1950).* In water sulfur dioxide exists as the dis-
solved gas, as undissociated sulfurous acid, H2S03,as the bisulfite ion,
HSOa-, and as the sulfite ion, so3--. Sulfurous acid is a rather weak
dibasic acid; its first ionization constant is 1.7 X and the second is
5 X loF6.At 25" C. the corresponding pK' values are 1.8 and 5.3, respec-
tively. The calculated distribution of the various ionic species in aqueous
solution of sulfurous acid in the range of pH of 0 to 8 is shown in Fig. 1.
At pH's over 9.5 only SO,- ions exist, in the range of pH 9.5 to 4.5 both
SO3-- and HS03- occur, and at pH 4.5 and lower, SO3-- no longer exists
in appreciable concentrations.
The alkali sulfites are but slightly hydrolyzed. The sulfites of many of
the heavy metals are insoluble. The sulfites occur as the normal salts,
NazS03, as the acid sulfites, NaHS03, and as the metabisulfites, Na2S206.
The latter is the anhydride of the acid sulfite:
The various sulfites are particularly adaptable as sources of sulfur

dioxide because of the convenience of handling as dry chemicals and the
ease with which their solutions can be prepared. They are best when
added to acid products of fairly high acid content so that possible reduc-
tion in total acidity or increase in pH as a result of liberation of sulfurous
acid is avoided. This may be avoided also when they are used as solutions
by the addition of a slight excess of inorganic acid such as HC1 or HzS04.
The theoretically available sulfur dioxide content and the usual range of
SO2 content for various preparations is shown in Table I. During storage,
particularly when exposed to moist conditions, the sulfites decrease in
available SO2 content as a result of oxidation. In the dry state stored
* Plummer (1950) made a complete survey of the literature on the equilibrium

vapor pressure exerted by the system SOZ-H~O,and statistically correlated such
data for SO2 concentrations of 0.03 g. to 10.0 g. of SO2 per 100 g. H20, at tempera-
tures of 0" to 130" C. The correlated results were presented as straight lines on a n
Othmer-Cox-type chart by modifying the ordinate. He found that the total pressure
+
P, = (P, P,) could be presented as a linear equation log,, P , = b log,, P , a +
and tabulated values of the over-all slope b and the over-all intercept a for various
concentrations of SO*.
Whitney and Vivian (1949) and Whitney et al. (1953) reported data on the
mechanism of absorption of sulfur dioxide gas in water, sodium carbonate, and
sodium sulfite solutions. This is basic not only to the manufacture of sulfite cooking
liquors used in the paper industry, but also in other industries where sulfur dioxide
gas is absorbed in liquids. I n the absorption of sulfur dioxide gas in water, both gas
and liquid film resistances are involved. I n strongly alkaline solution the resistance
is primarily in the gas film, and in sulfite liquors it, is primarily in the liquid film.
102 M. A. JOSLYN AND J. B. S. BRAVERMAN
exposed to air in order of increasing stability they are: sulfite > bisulfite
> metabisulfite (Mason, 1928; Phillips, 1928).
TABLBI
Available Sulfur Dioxide Content of Various Sources
Available per cent of SOa
Compound Formula Theoretical Actual

Liquid sulfur dioxide SO2 100.0 100.0
Acid sulfurous, 6% HaSOa 6.0 6.4-6.8
Calcium sulfite CaSOa.1 f 5 H z 0 23.0 43-45
Potassium sulfite KzSOa 33.0 36
Sodium sulfite Na2SO3 50.84 48
Potassium bisulfite KHSOa 53.31 -
Sodium bisulfite NaHSOa 61.59 55
Potassium metabisulfite KzSz06 67.43 52
Sodium metabisulfite Na2Sz06 57.65 61
Antioxidant and Reducing Properties. Sulfurous acid and sulfites in
solution are oxidized by air to sulfates. The rate of oxidation of sulfite
ions by oxygen has been shown by a number of investigators to be a chain
reaction of unusual length for condensed systems and, as is frequently the
situation in such cases, to be very sensitive to both positive and negative
catalysis. The early investigations on oxidizability of sulfite solutions are
discussed by Kolthoff and Menzel (1928). Fuller and Crist (1941) re-
ported the reaction of sodium sulfite solutions saturated with oxygen at
one atmosphere pressure to be strictly first order with respect to sulfite
ion concentration with a specific reaction rate of 0.013 see.-' at 25" C.
The rate of oxidation is independent of the pH between 8.8 and 8.2 but
decreases in a complicated manner between 5.9 and 3.2. The rate is
directly dependent upon the cupric ion concentration when this exceeds
M and the catalytic constant is 2.5 X lo6 liters moles-l set.-' at
25" C. The inhibitory effect of mannitol, however, is uniform over a
106-foldchange in mannitol concentration. Mitchell et al. (1933) reported
the oxidation of sulfurous acid and sulfite t o be reduced by decreasing pH
and by addition of various antioxidants.
The inhibitory action of sulfite in the auto-oxidation of several phenols
has been established. Branch and Joslyn (1935) reported that the addition
of 0.001 M of potassium sulfite to 0.1 M catechol a t pH 8.3, reduced the
rate of oxidation from 3.3 to about 0.2 initially, but after about 4 minutes
the rate increased t o that normal for the conditions used. James and
Weissberger (1939) reported that, in the auto-oxidation of hydroquinone
and its homologs in the presence of excess sulfite, these substances react
with oxygen to form quinones and hydrogen peroxide. The latter then
oxidizes the sulfite to sulfate. If the quinone formed has a t least one
nuclear hydrogen it reacts with sulfite to form the hydroquinone mono-

sulfonate which auto-oxidizes and finally forms the disulfonate. The
occurrence of a sulfonic acid of this type is quite significant in view of the
finding by Stadtman (1948) that the sulfur dioxide which disappears
during the storage of sulfured dried apricots in the absence of oxygen is
converted into a relatively strong polyhydroxy acid containing 10 %
sulfur. Although this compound has not been identified, Stadtman
postulated it t o be a sulfonic acid of partly unsaturated sugars.
Sulfurous acid and sulfites also inhibit the auto-oxidation of ascorbic
acid. Some of the earlier investigators, e.g., Williams and Corran (1930),
were of the opinion that sulfur dioxide, when used as a preservative for
fruit juices, had a definite destructive action on ascorbic acid a t normal
temperatures (15 t o 18" C,).Bennett and Tarbert (1933) reported that
sulfur dioxide exerted only a slight destructive effect on ascorbic acid in
comparison with the marked destruction occurring in benzoated juice
which they wrongly ascribed to the benzoate itself. Hamburger and
Joslyn (1941), however, found no evidence for an induced oxidation of
ascorbic acid in citrus juices by sulfur dioxide. They found, on the con-
trary, that sulfur dioxide inhibits the oxidation of ascorbic acid in citrus
juices and that this inhibition was more pronounced a t 500 p.p.m. than
250 p.p.m. of SOz. Joslyn (1941) and Downer (1942) reported that sulfur
dioxide definitely inhibited the auto-oxidation of ascorbic acid in citrus
juices. Busing and Raabe (1938) reported that in grape juice, also,
ascorbic acid was preserved by sulfur dioxide. Morgan et al. (1929, 1931,
1935) found that sulfuring improved retention of ascorbic acid in dried
fruits.
Sulfurous acid and sulfites, although they serve as anti-oxidants and
are useful in inhibiting oxidation of ascorbic acid, have an adverse affect
upon thiamine. Sulfurous acid has long been known t o destroy the anti-
neuritic activity of rice polish extracts (Williams et al., 1935b) and of dried
fruits, grape juice, and wine. Williams et al. (1935a), Buchman (1936))and
Buchman et al. (1935), showed that this destruction of vitamin B1 activity
was due to the cleavage of thiamine into 4-methyl-5 hydroxy ethyl
thiazole and the sulfonic acid of 2,5-dimethyl-4-amino-pyrimidine. This
cleavage is complete a t room temperature a t pH 5 in 24 to 48 hours, but
is slower at lower pH values. I n saturated aqueous solution of sulfurous
acid, 50 % yields of cleavage products were obtained after storage a t room
temperature for three months. Morgan et al. (1931, 1935, 1939) showed
that sulfiting of fruit for drying and sulfiting of grape juice and wine was
destructive to vitamin B1.
Bleaching Action. Sulfur dioxide (sulfurous acid and the acid sulfites)
reduces many colored compounds to colorless derivatives and is fre-
quently used as a bleaching agent. Sulfur dioxide has a pronounced

bleaching effect on the anthocyanin pigments, and red- and blue-colored
fruits are rapidly bleached in brines containing sulfites. It has a slight
bleaching action on yellow fruits. Sulfurous acid, however, does not
bleach chlorophyll but catalyzes the conversion of chlorophyll to pheo-
phytin. This conversion, however, is considerably slower with sulfurous
acid than with sulfuric or oxalic acids of the same concentration (Joslyn
and Mackinney, 1938; Mackinney and Joslyn, 1941). Mackinney (1937)
found that chlorophyll was present in raisins made from sulfured grapes.
Dried cut fruit, such as dried apricots, which have darkened only slightly
can be almost completely restored to their original color by treating with
SO2 (Jewell, 1937). The nature of this bleaching action is unknown, but it
is probably due to the formation of colorless compounds as a result of
combination of SO2 with some constituent. The decolorization of basic
fuchsin by sulphur dioxide has been fairly well elucidated and is believed
to be due to the formation of colorless sulfonates.
In the sugar industry it has long been known that sulfur dioxide
bleaches not only the naturally occurring pigments such as anthocyanins
and other colored nonsugars but also the nonsugars and sugars which
develop color in sugar manufacture (Gillett, 1953; Zerban, 1947). Lewis
et al. (1949a, b) reported that colored decomposition products formed by
heating glucose are decolorized by NaHS03 a t pH 5.5. Bleached syrups
exposed to air for some time darken again because of oxidation of sulfite.
Sulfur dioxide not only bleaches coloring matter already formed but
reduces darkening during evaporation and crystallization. Zerban (1947)
suggested that this protective effect was due to combination of sulfur
dioxide with reducing sugars, thus blocking the carbonyl function essen-
tial for caramel and melanoidin formation. The inhibition of browning by
sulfurous acid and bisulfite may be due not only to its reaction with the
carbonyl group in sugars but also to its action as reducing agent in keep-
ing reductones in the inactive reduced form rather than in the active
dehydro form. Hodge (1953) in his recent review of the chemistry of
browning reactions in nitrogenous model systems has stressed the impor-
tant role of dehydrogenated reductones in both enzymatic and non-
enzymatic browning reactions.
2. Bisulfite Addition Products: a-Hydroxysulfonic Acids
Sulfurous acid and the alkali bisulfites have long been known to react
with aldehydes and ketones. The sodium and potassium bisulfites form
crystalline addition products which are useful in purification but seldom
used for the identification purposes, since their melting or decomposition
point is not sharp and they are essentially insoluble in organic solvents.
The bisulfites react with unsaturated organic compounds, their addition
\
to -C=C- bonds being almost as rapid as to the C=O groups.
/
Amines, particularly tertiary amines, in the presence of water, with sulfur
dioxide readily form amine bisulfites or complexes of amine bisulfites with
sulfur dioxide. The amine bisulfites form addition products with alde-
hydes and ketones which are useful for their characterization as well as
their resolution (Adams and Lipscomb, 1949; Adams and Garber, 1949).
\
In the reaction of amines with aldehydes Schiff bases containing ‘C=N-
/
groups are formed as intermediates and these could also react with
bisulfites. It has been shown recently that glucose sodium bisulfite reacts
with amines to form substituted a-amino sulfonates. This reaction, as
pointed out by Danehy and Pigman (1951), may be of significance in
accounting for the inhibitive effect of sulfur dioxide in the discoloration
of food products. Certain alkaloids as well as the primary, secondary, and
tertiary amines form bisulfite addition compounds.
Only comparatively recently has it been unequivocally established for
a few carbonyl compounds that the bisulfite addition products are
hydroxysulfonic acid derivatives rather than sulfites (Suter, 1944). The
evidence for the hydroxysulfonate structure of aldehyde and ketone
bisulfite compounds, the sulfonic groups of which are mobilized by the
adjacent hydroxyl group, is now quite convincing. While all aldehydes
form hydroxysulfonates with bisulfites, many ketones do not react. With
the exception of ethyl ketone, which reacts slowly and to a limited extent,
only ketones containing either a methyl group attached to the carbonyl
or having the carbonyl part of a ring system of four to seven carbon
atoms combine appreciably. The addition of an alkali bisulfite to an
unsaturated aldehyde or ketone in which the double bond is conjugated
with the carbonyl group is more complex. Citral, for example, forms, in
addition to the normal hydroxysulfonate, also a compound two molecules
of bisulfite. With unsaturated aldehydes three types of bisulfite addition
compounds are possible: addition of -SOaNa to the C of the C=O group,
to one of the C= C carbons, or to both positions.
The combination of bisulfites with sugars is much slower than with
aldehydes and ketones and the products formed are relatively more
unstable. Only the glucose and arabinose bisulfite addition compounds
have been prepared in pure or relatively pure form. The combination with
other sugars is largely inferred from measurements of decrease in “free ”
sulfur dioxide content. The early work of Kerp, Schmidt, and others in
the laboratories of the German Public Health Department during 1904-
106 M. A. JOSLYN AND J. B. 9. BRAVERMAN
1909 (Kerp, 1903, 1904,1913), is summarized by Monier-Williams (1927))

Browne (1944), Berg (1946)) and Ingram and Vas (1950a, b). The results
obtained by Ingram and Vas (1950a) confirmed those obtained previously
by Kerp (1904) and by Farnsteiner (1904)) that of the sugars tested
galactose, mannose, and arabinose combined with bisulfites rapidly and
formed compounds which dissociated less; maltose, lactose, and glucose
were less active; raffinose was but slightly active and fructose and sucrose
were probably inactive.
Ingram and Vas (1950a) also supported the early inference of Paris
(1920) that pectinic acids combine with bisulfite, but their observations
were made on 100 grade citrus pectin containing about one-third its
weight of glucose. Sulfurous acid has been reported to combine with
dextrins, cellulose, and lignin; proteins and possibly gelatins (Monier-
Williams, 1927). The lignin sulfonic acid in sulfite pulp liquor is particu-
larly stable and gives off sulfur dioxide with extreme slowness on distilla-
tion with strong acids (Stutzer, 1910, cited by Monier-Williams, 1927).
Solutions of glucose, mannose, and xylose when heated a t 135” C. in the
presence of sodium-bisulfite-sulfurous acid solutions containing sulfuric
acid were found by Hagglund et al. (1929, 1930) t o yield gluconic, man-
nonic, and xylonic acid by the transformation of the primary formed
bisulfite addition complexes. Under the same conditions fructose yielded
a stable sulfur compound which was believed to be fructose sulfonic acid.
Sugar solutions on heating may yield decomposition products which can
combine readily with sulfur dioxide. Ingram and Vas (1950a) list among
such substances reported t o occur during the decomposition of sugar solu-
tions heated alone or in presence of added alkalis or amino acids: formal-
dehyde, glyceraldehyde, 5-hydroxymethyl furfural, furfuralic and other
aldehydes, methylglyoxal, pryuvaldehyde, dihydroxyacetone, acetal,
diacetyl, and “reductone.”
Convincing proof that the bisulfite addition product of formaldehyde
is a hydroxysulfonic acid, containing carbon-sulfur linkage, is based on
both chemical and physical evidence. The condensation product of the
formaldehyde bisulfite addition product and ethyl acetoacetate has been
shown to be a true sulfonate; the amino compound derived from formal-
dehyde, sodium bisulfite, and ammonia is converted by nitrosyl chloride
into sodium chloromethanesulfonic acid ; and acetylation of the formalde-
hyde addition product was found to give an acetate identical with that
obtained from potassium iodomethanesulfonate. The physical properties
of the hydroxysulfonates are in accord with this structure : their absorp-
tion spectra contain a band a t 4992 A. characteristic of sulfonic acids and
differing from that of sulfites; the conductivity of solutions of the free
acids is high, indicating the presence of typical sulfonic acids; the second
hydrogen is weakly acidic having a dissociation constali t of about

7 x 10-1O.Proof of the structure of bisulfite addition products other than
those of aldehydes is not as, strong, but Braverman (1953) has presented
good evidence for the structure of glucose and arabiriose sulfonates and
for the possible mechanism of the interaction involved.
The a-hydroxysulfonic acids decompose in either acid or alkaline solu-
tions t o give the original carbonyl compound. In acid solutions a t room
temperature the reaction is slow but on heating it is more rapid and the
decomposition during distillation from strongly acid solutions is almost
complete. It may be represented as follows:
H
R- :$I/8
- -0Na + HC1-t R- 8+
H
I NaCl + S 0 2 t + HzOT
In alkaline solutions the decomposition of the hydroxysulfonates is

quite rapid even at room temperature, largely owing to the fact that
neutral sulfites do not combine with aldehydes or ketones. The decom-
position in alkali may be represented as follows:
H
0 0 0
II
€d-i!!-ONa + NaOH + R-C + NaO-S-ONa + H20
II
Adl H
I
0
The a-hydroxysulfonates are not oxidized by iodine in neutral or acid

solutions, but this does occur in alkaline solutions where it is preceded by
dissociation into the free sulfite. They are also not readily oxidizable by
oxygen.
The formation of the a-hydroxysulfonic acids is influenced by concen-
tration of the reactants (carbonyl compound and bisulfite), temperature,
and pH. The rate of association as influenced by the relative concentra-
tion of bisulfite, aldehyde, and sugars was investigated early by Kerp
(1904) and his collaborators and others, and more recently by Ingram and
Vas (1950b) and by the Corn Products Refining Company.' The effect of
pH and temperature on dissociation of the sulfonic acids has been in-
vestigated also for some of the compounds (see the review by Suter,
1 The Chemical Division of the Corn Products Refining Company investigated the
effect of temperature, concentration, alcohol, and free sulfurous acid on the degree of
association of glucose and sodium bisulfite in aqueous solution in connection with the
research sponsored by the Subsistence Section of the QMC Research and Develop-
ment Branch during World War I1 on browning.
1944). Most of the publications in the field, however, have dealt with
reactions in solution or with partially purified sulfonates. Only very
recently have pure sulfonates been prepared and their properties in-
vestigated. The characteristics of two typical compounds, the acetalde-
hyde sulfonic acid and the glucose sulfonic acid, are discussed briefly
below. The literature on the sugar sulfonates is reviewed in more detail
by Gehman and Osman (1954).
Acetaldehyde Sulfonate. Kerp and his collaborators prepared acetalde-
hyde sodium hydroxysulfonate essentially by the method of Bunte
(1873). This consists in the slow addition of acetaldehyde to a concen-
trated solution of sodium or potassium bisulfite, with continuous cooling
until excess of aldehyde is evident to the nose and no further temperature
rise occurs. The sulfonate is then obtained by evaporation in a desiccator
over concentrated sulfuric acid, or by storing the reaction mixture over
night and precipitating the complex with methanol. The impurities
present, chiefly the alkali sulfates, are eliminated by recrystallization
from warm methyl alcohol at 40” C. (104’ F.). With sulfonates so pre-
pared, Kerp reported the following percentages dissociated into free
bisulfite titratable with iodine: 0.17% in 1 N solution; 0.45% with 0.1 N
solution; and 0.71 % with $60 N solution. The association of bisulfite with
acetaldehyde was investigated by titrating the residual “free ” bisulfite
with iodine. The degree of association increases as the concentration of
acetaldehyde increases, but from 70 t o 95% of the bisulfite was found to
be bound after 2 minutes and after 1 hour, over 99% of the bisulfite was
bound. The acetaldehyde bisulfite compound was more stable than that
of glucose, and the equilibrium constants for the association reaction
were such that in a mixture of acetaldehyde and glucose, the former com-
bines first with added bisulfite. If acetaldehyde is added to a glucose
bisulfite solution it replaces glucose in combination. Bianconi and Bianchi
(1932) investigated the extent of association of potassium bisulfite with
acetaldehyde and reported complete binding after 20 minutes in 0.05 N
solutions. In the presence of tartaric acid, the rate of association was
slower, and this was ascribed to inhibition by tartrate. Neither Kerp nor
Bianconi and Bianchi, however, determined the effect of pH on associa-
tion and dissociation. This was done first by Tomoda (1927)) who found
that a t pH 6 to 8 the dissociation of the acetaldehyde sulfonate was less
than 5%; a t pH 8 t o 10.5, it was 50%; and above pH 12 practically com-
plete dissociation occurred. Tomoda calculated the degree of dissociation
of the aldehyde sulfonate from the mass action law to be:
where a = degree of dissociation of the acetaldehyde sulfonate;

K = dissociation constant for the acetaldehyde sulfonate, which
according t o Kerp is 2.8 X
K I = first dissociation constant for sulfurous acid, 1.7 X
K O= second dissociation constant for sulfurous acid, 5 x
The degree of dissociation, expressed as percentage of bisulfite or
acetaldehyde sulfonate at various pH values, as calculated by Tomoda
is shown in Fig. 2. Tomoda (1928) later pointed out that the pH during
FIG.2. Dissociation of acetaldehyde sulfonate a t various pH values. After Tomoda

(1927).
glycerine fermentation (normally conducted a t not above pH 8.3) should

not be much below pH 6 if undesirable concentration of bisulfite ion is to
be avoided.
In the course of our investigations2 on the chemistry of acetaldehyde
and other bisulfite addition compounds begun in 1950, several attempts
were made to prepare pure crystalline acetaldehyde hydroxysulfonate
using the procedures published in the literature, but all attempts failed.
With the best preparation of acetaldehyde-sulfonate available, which
according to analysis was 70% pure, the effect of pH on dissociation was
determined. The results obtained, yhown as curve 3 in Fig. 3, indicate the
* These investigations were begun in our laboratories with the assistance of C. J. Smit,
without knowledge of the previous unpublished data obtained by the Corn Products
Refining Company research workers, but were continued after these data became
available and led to Rraverman’s successful preparation of the pure glucose sulfonate
(Rravcrman, 1953).
total iodine titratable sulfite liberated in 1 minute a t the pH range of

2 to 12. At pH 2 with a preparation equivalent to 160 p.p.m. of SO2 only
2.7% was dissociated; with a preparation equivalent to 600 p.p.m. of
SOz only 1.3% was dissociated in 1 minute. As the pH increased, there
was a t first an increase in dissociation which had a maximum a t about
3.2 to 3.3, which then dropped to a minimum a t pH of 4.5, increased
again, and was relatively constant in the range of pH 5 to 9. Above pH 9
the dissociation increased rapidly reaching a maximum a t pH 11.5.
FIG.3. The dissociation of sulfonates expressed as total free SO2 a t various p H

values in presence of 0.1 M tartrate. Data for curves 1 and 3 are from Joslyn, Braver-
man, and Smit (1950); for curve 2 from Ponting and Johnson (1945).
Glucose Hydroxysulfonate. The preparation of sodium glucose sulfonate
originally described by Kerp (1904) was modified by Neuberg (1929) and
more recently by Gehman and Osman (1954). Even the latter procedure
resulted in our hands in only 70% yield of 90% pure compound. These
procedures have the disadvantage that the methanol used in separating
the glucose sulfonate will precipitate unchanged glucose and sodium
bisulfite as well as the sulfonate. Braverman (1953) successfully prepared
crystalline sodium glucose hydroxysulfonate of high purity as follows :
Dissolve 144 g. (0.8 mole) of anhydrous glucose in 60 ml. of distilled water and cool
to 40" C. I n another beaker dissolve 88 g. of NaHSOJ (0.8 m. + 10% excess) in 100 ml.
H 2 0 and also cool to 40' C. At this temperature the two solutions are mixed slowly
with constant slow stirring. The mixture is then placed in an incubator or in a constant
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEIGETABLE PRODUCTS 111
temperature bath at 30-35" C. and constantly stirred with a slow-moving stirrer. After
an hour or so, the solution becomes quite clear. After 50 to 60 hours, the liquid becomes
turbid with suspended crystals. Continue stirring for 3 to 5 days in all, after which an
abundant crop of beautiful crystals is obtained. Filter on a Buchner funnel with suction
to separate as much as possible the mother liquor. Wash with small portions of 75%
methanol. Rinse with 99% methanol. Dry in vacuum dessicator. Yield obtained: 66%
of the theoretical. (See Figure 4.)M.P. 92" C.
FIQ.4. Photograph of crystals of sodium glucose hydroxysulfonate.
The equilibrium constants for the equilibrium and velocity of the

reaction of glucose with sulfurous acid were first measured by Kerp and
his collaborators some forty years ago, but their results are difficult to
interpret and apply because pH data are lacking and because most of
their data relate to rather dilute solutions of glucose (18% or less) and
bisulfite (800 p.p.m. or less). Tomoda and Taguchi (1930), Bianconi and
Bianchi (1932), and more recently Vas (1949) investigated this reaction.
112 M. A. JOSLYN AND J. B. 6 . BRAVERMAN
Vas (1949) reported that the equilibrium constant was mainly a function
of pH, being least in the range of pH 3 to 5.5 and increasing rapidly in
more acid and more alkaline regions. Changes in concentration of glucose
in the range of 1 to 40 g. per 100 ml. and of SO2 in the range of 130 to
3500 mg. per liter had only a minor influence on the equilibrium constant.
12
I I I I I
PH
FIQ. 5. Effect of glucose on the dissociation of a solution of glucose hydroxy-

sulfonate at various pH’s.
1. Glucose hydroxysulfonate 2.2175 g./l.
2. Glucose hydroxysulfonate 2.2175 g!J. + 10% glucose.
3. Glucose hydroxysulfonate 2.2175 g./l. + 20% glucose.
+
4. Glucose hydroxysulfonate 2.2175 g./L 50% glucose.
From Braverman (1953).
The apparent velocity constants of decomposition and addition were
found to be functions of pH only. Braverman (1953) reported that at
pH 5, over 4001, of the SOz present was bound after the first 30 minutes.
In a more acid medium (pH 2) the rate of combination was slow and
amounted to only 32% after prolonged storage. As the concentration of
glucose was increased the percentage of SO2 bound first increased and
then decreased. At pH 7 he found that the pure glucose sulfonate disso-
ciates completely. The effect of pH on the dissociation of the glucose
sulfonate measured as total free SO2 is shown in Fig. 3, in comparison
with that of frozen apples and acetaldehyde sulfonate. It is evident that
the constituent combining sulfur dioxide in apples is glucose and that the
dissociation of the glucose sulfonate increases rapidly with pH and is
practically complete a t pH 6. The dissociation of glucose sulfonate in
absence and presence of glucose is shown in Fig. 5. As the excess glucose
present is increased in concentration, the amount of SOz dissociated a t
various pH levels decreases at first, rises to a distinct maximum in the
range of pH 4 to 4.5, drops to a minimum in the range of 5.5 to 6.5, and

then rises again.
111. DETERMINATION
OF SULFUR IN FRUIT
DIOXIDE
AND VEGETABLEPRODUCTS
It is obvious from the evidence summarized above that sulfur dioxide
added to fruit and vegetable products as liquid sulfur dioxide, as gaseous
sulfur dioxide, as solution of sulfmous acid in water, or as dry or dissolved
sulfites will exist as the undissociated sulfurous acid, as the free bisulfite
ion, as the free sulfite ion, and as combined sulfur dioxide in the form of
the hydroxysulfonates. The pH, temperature, composition of the food
product, conditions of sulfuring or sulfiting, aiid the subsequent treatment
and storage conditions will determine the equilibrium between the various
forms of sulfur dioxide present. Oxidation of the sulfurous acid and sulfites
to sulfate ions and formation of organic sulfur compounds other than the
hydroxysulfonates which may occur during processing and storage will
also be involved. I t is obvious that the analyses for these constituents at
any given time would be most useful in determining the relative efficiency
of a particular type of sulfuring or sulfiting practice and in interpretation
of the mechanism of sulfite inhibition of undesirable changes. Unfor-
tunately the methods of analysis so far available are not sufficient for this
and for the most part these have been based on empirically developed
procedures tested either by maximum recovery of the sulfur dioxide
present or in a few instances by recovery of sulfur dioxide added as such.
The recovery of sulfurous acid added as such or as bisulfite is not suffi-
cient measure of the accuracy of a given method for total sulfur dioxide.
Even in cases where the bound sulfur dioxide is likely t o be present largely
as hydroxysulfonate investigation of the adequacy of the procedure
selected has not been checked by addition of known amounts of the
sulfonate. High results, reproducibility of results, and avoidance of
possible errors in determination have been relied upon generally in de-
velopment of analytical methods. The available methods of analysis can
be segregated into those designed to measure the free sulfur dioxide and
those for total sulfur dioxide. The latter can be subdivided into two
groups: (1) those in which the “bound” sulfur dioxide is liberated by
distillation from acid; and (2) those in which the combined sulfur dioxide
is liberated by treatment of the liquid product or extract with excess
alkali a t room temperature and subsequent acidification t o prevent
recombination. The free or total sulfur dioxide may be determined
volumetrically, gravimetrically, or colorimetrically. Monier-Williams
(1927) has reviewed critically and thoroughly the earlier investigations in
this field and has reported in detail the basis of the procedure he de-
veloped to overcome several of the difficulties with previous methods. He

stressed particularly the fact that ((determinations of free and combined
sulfurous acid in foods are not likely to give reliable results unless the
conditions under which the analysis is conducted are such that no appre-
ciable alteration of equilibrium can occur during the determination.”
“Free ” Sulfur Dioxide: The determination of “free” sulfur dioxide in
beer, white wines, and lightly colored juices, concentrates, and beverages
is based on the direct iodine titration method developed by Ripper (1892)
for wine.
In this method a 50-ml. sample, after acidification with 5 ml. of dilute sulfuric acid
+
(1 3), and expelling of the air present either by displacement of air present in the
flask with carbon dioxide gas as suggested by Ripper or by addition of 0.5 g. of sodium
carbonate (Assoc. Offic. Agric. Chemists, 1950) as in the official A.O.A.C. method, is
rapidly titrated directly with 0.02 N iodine solution to a starch end point which per-
sists for a few minutes. The added acid is relied upon to reduce the rate of dissociation
of the “bound” sulfur dioxide. Kerp and Bain (see Kerp, 1913) showed that for their
preparation of the acetaldehyde sulfonate the addition of N/30 hydrochloric acid
markedly decreased the rate of dissociation, although it increased the actual extent of
dissociation at equilibrium. With their preparation of glucose sulfonate the addition
of hydrochloric acid also markedly decreased the rate of dissociation but did not
influence the actual extent of dissociation. The effect of pH and temperature on the
rate of dissociation of a-hydroxysulfonates in presence of other constituents likely to
influence the rate of dissociation, however, has not been determined. Ripper’s method
has been applied to red wines (Benvegnin and Capt, 1931; Sumuleanu et al., 1937); to
beer (Butlestone, 1928); to sweet musts (Fischler and Kretzdom, 1938); to apple juices
and ciders (Warcollier and Le Moal, 1929); to concentrates and syrups; and also to
extracts prepared from dehydrated vegetables, dried fruits, etc. In the analysis of
diluted concentrates and extracts of dried or dehydrated foods the possible change in
the equilibrium between bound and free sulfur dioxide as a result of dilution during
preparation of the extracts has not been controlled. Hydrochloric acid has been
favored by some over sulfuric acid for acidifying the solution, e.g., Jaulmes and
+
Espeael (1935) used a dilute hydrochloric acid (1 4). Hydrochloric acid was pre-
ferred also by Bennett and Donovan (1943), Prater et al. (1944), Ponting and Johnson
(1945), Reifer and Mangan (1945), and others. In the preparation of extracts for
direct titration losses of the sulfur dioxide present by oxidation as well as changes in
the distribution of sulfur dioxide occur. These losses are significant in cold water
extraction and in the alkali extraction procedures of Prater et al. (1944) and Potter and
Hendel (1951). The oxidation losses were reduced by Reifer and Mangan (1945) by
extraction of the finely ground material in boiling water containing sugar and a
phosphate buffer of pH 7.6. They claimed no loss of sulfur dioxide due to evaporation
or oxidation under these conditions. Ponting and Johnson (1945) controlled oxidation
by blending fruit in a solution of salt and tartrate buffer a t pH 4.5.
Even when errors in true free sulfur dioxide content due to dilution or oxidation
are avoided, the direct iodometric determination of sulfurous acid and sulfites is
subject to error. Kolthoff and Menael (1929) point out that an appreciable error occurs
when sulfite solutions are titrated with iodine owing to oxidation by air and point out
that correct results can be obtained only when the sulfurous acid or sulfite solution is
allowed to flow into a solution of iodine. The titration of sulfurous acid or bisulfites is
subject also to possible reduction of SO2by hydriodic acid to sulfur and iodine, and to
loss of SO2 by evaporation during the titration. In strongly acid solutions volatilization
of Sot, particularly when its concentration in the aliquot taken for analysis is over
0.04%, is more important as a source of error than oxidation. Mason and Walsh (1928)
have investigated the magnitude of the errors occurring during the titration of dilute
sulfite solutions with iodine, and concluded that loss of SO2 due to volatilization is
more important than by oxidation. Reduction of sulfur dioxide by H I was not found
by them.
In the titration of wines, beers, juices, and other liquid food products, Monier-
Williams (1927) lists as additional errors: (1)action of iodine on substances other than
sulfur dioxide and (2) recombination of sulfur dioxide and acetaldehyde or other
carbonyl compound. Ingram (1947a) pointed out that, in the direct iodine titration of
citrus juices by iodine using starch indicator, the end point tends to drift either be-
cause excess of iodine combines with other reducing substances present or because of
slow decomposition of combined SOz. He suggested potentiometric titration using a
bright platinum electrode in conjunction with an Ag-AgC1 reference electrode and an
electrometer. This is particularly desirable with colored juices or beverages. Using the
electrometric titration, Ingram (1947b) carefully estimated the range of the errors
involved in the determination of free SO2in diluted sulfited orange juice concentrate.
If the concentrated juice was diluted before titrating, extra free SOZwas formed a t
the rate of 1 p.p.m. per minute in a juice containing about 100 p.p.m. of free SOZ. If
delay occurred during the titration at any point, more free SO1 was formed and this
error was greater the greater the degree to which the titration had progressed before
delay occurred. With delay a t the end point there was a relatively large increase in the
titration.
To correct for iodine reducing substances other than sulfur dioxide which may be
present, the iodine titration is carried out both on the original wine, juice, or extract
to measure the total reducing power including SO, and on an aliquot to which is added
an excess of a bisulfite binding agent to measure the reducing power of the juice or
extract itself. The difference in titration is taken as a measure of sulfur dioxide present.
Acetone was suggested as such a binding agent by Downer (1943) and Bennett and
Donovan (1943), and used by Prater et al. (1944). It was not found satisfactory by
Ponting and Johnson (1945) for fruit extracts, by Reifer and Mangan (1945) for de-
hydrated cabbage, by Ingram (1947b) for citrus concentrates and citrus juices, and by
Potter and Hendel (1951) for dehydrated white potatoes. Ponting and Johnson (1945),
Reifer and Mangan (1945), and Potter and Hendel (1951) found that the rapid fading
of the end point in the acetone-treated sample, as a result of rapid dissociation of the
acetone-sulfite complex, can be largely avoided by the use of formaldehyde as a binding
agent, since it forms a more stable complex with sulfite than does acetone.
In addition to the iodometric determination, direct precipitation as barium sulfate
before and after treatment with bromine was suggested for both quantitative and
quaIitative test for sulfur dioxide in wine (see Monier-Williams, 1927). Precipitation
of SO2 after oxidation with HeOl as the benzidine sulfate was proposed by Rothen-
fusser (1929); reduction of the molybdenum in phosphomolybdic acid by the sulfite
ion present in an aqueous solution of the food was proposed by Sasaki (1928) as a
colorimetric method; formation of a blue color from a solution of 1%methylene blue
and 5 % iodine in potassium iodide was proposed by Svershkov (1939). Mathers (1949)
proposed a turbidimetric method based on the distillation of wine into a dilute solution
of lead acetate to form a colloidal suspension of lead sulfite whose spectral transmit-
tance in the range of 400 to 700 mp could be used as a measure of sulfur diaxide. This is
similar to turbidimetric methods based on turbidity produced by adding BaClr to a
116 M. A, JOSLYN AND J. B. 5. BRAVERMAN
sulfite oxidized by HzOr but is more rapid and more sensitive in the range of 0 to 100
p.p.m. Francis el al. (1938) proposed a conductometric method based on the change in
electrical conductivity of neutral HZOZwhich results from passage of SO? for the
estimation of SO2 in industrial gases and a similar method. Conductometric methods
were used previously by Thomas (1932), Thomas and Abersold (1929), Thomas el al.
(1943, 1946) for the determination of small concentrations of SO2 in air and photo-
electric determination employing dilute starch-iodine solutions was proposed by
Katz (1950).
The determination of sulfurous acid and sulfites by oxidation to sulfates and
precipitation as barium sulfate because of its high degree of specificity in acid solu-
tions, high degree of insolubility, and favorable gravimetric factor, commands wide
acceptance. Gravimetric determination of sulfate by precipitation as BaSOr is still
suggested as optional in the official Monier-Williams method of the Association of
Official Agricultural Chemists (1950). To utilize the advantages of barium sulfate in
the microgram range of microanalysis, titrimetric methods using tetrahydroxy-
quinone or rhodizonates as indicators have been introduced (see Little, 1953; Toennies
and Bakay, 1953). With tetrahydroxyquinone as indicator a practical sensitivity of
k 3 micrograms of sulfur per determination can be achieved in the titration of sulfate
with barium chloride. Toennies and Bakay (1953) recently investigated the determina-
tion of barium sulfate in suspension by nephelometry and developed a sensitive photo-
nephelometric micro method based on the use of a glycerol-alcohol-water system to
obtain reproducible and stable suspensions of barium sulfate. Their procedure with
slight modification waa successfully applied in our laboratory (Lukton and Joslyn,
1954) to the determination of sulfur dioxide in solutions of sulfurous acid and sulfites
but could not be applied directly to white wines. Recovery of added sulfite and sulfate
to wine was high and variable. Even when it was applied to wines after dry or wet
ashing, the results were not satisfactory.
Acid bleached fuchsin to which formaldehyde is added has been developed as a
sensitive method for the determination of sulfur dioxide. As developed first by Grant
(1947)the acidified fuchsin formaldehyde solution is added in excess to a n aliquot
containing sulfite. Hoffpauir and O’Connor (1949) suggested sensitization by addition
of a ketone such as acetone. Steigmann (1950) proposed that the acidified basic fuchsin
and formaldehyde reagents be prepared separately and that the former be filtered
after storage for 3 days. La Rosa (1950) suggested that the sulfuric acid-basic fuchsin-
formaldehyde reagent be decolorised with carbon to remove colored impurities in the
dye solution and filtered before use. This reagent was applied by Atkin (1950) to the
determination of sulfur dioxide in presence of sulfur trioxide, by Urone and Boggs
(1951)to determination of sulfur dioxide in the atmosphere, and by Dupaigne (1951)
as a micromethod for sulfur dioxide in grape juice and was proposed also for use in
wine and beer analysis.
The above substitutes for the iodometric determination with one or two exceptions
have not been tested for specificity for free SOZ.Mathers (1949) on the basis of the fact
that “free” sulfur dioxide is removed early in the distillation of wine proposed that
10 ml. of the distillate from the alcohol determination be mixed with 0.5 ml. of 5%
neutral lead acetate and the turbidity of the suspension formed be used to correct
volatile acidity for sulfur dioxide. La Rosa (1950)assumed that the color formed on
mixing the fuchsin-sulfuric acid-formaldehyde reagent with white wine was a measure
of free SOz but did not give any data to confirm this. The fuchsin procedure and the
lead sulfite procedures are very sensitive, the former more so, and can be applied only
to very small aliquots or to dilute solutions.
Joslyn (1953) found t h at Grant’s reagent (Grant, 1947) could not be used with the
Klett-Summerson colorirneter a t SO2 levels ahove 10 p.p.m. because the reading was
off the scale. When 5 rnl. of the fuchsine reagent and 1 ml. of test solution were mixed
and diluted to 25 ml. before transfer to t h e colorimetcr tube, fairly reproducible read-
ings were obtained in the range of (r50 p.p.m. A high blank reading with even the
freshly mixed reagent and increase in this blank and decrease in sensitivity of the
reagent with storage was observed. Higher sensitivity and greater stability was
obtained with the Steigmmn (1950) modification, but the free SO2 content of wine
determined colorimetrically was considerably higher than given by iodometric titra-
tion, the difference increasing with increase in SO2 level.
Total Sulfur Dioxide: After Alkali l’reatrnenl: The fact that neutral sodium sulfite
does not combine with carbonyl compounds and t h a t the hydroxysulfonic acid com-
pounds are rapidly decomposed on treatment with alkali was used by Ripper (1892)
as the basis for the determination of total sulfur dioxide in wine b y direct iodine
titration. I n his method, 50 ml. of wine were pipetted into a 200-ml. flask containing
25 ml. of 1 N KOH. The mixture was shaken and allowed to stand for 10 to 15 minutes.
+
Then 10 ml. of dilute sulfuric acid (1 3) were added, and the solution titrated
rapidly with 0.02 N iodine solution t o a starch end point which persisted for some time.
This method was used as the official direct titration method for wine in the first edition
(1919) of the A.O.A.C. Methods of Analysis; in the third (1930) edition it was ex-
tended t o white grape juice, wine, and similar products (1 N NaOH or KOH was used
and the solution during standing for 15 minutes was occasionally agitated); but it was
dropped from the fourth (1935) and succeeding editions. Ripper compared his method
with the Haas distillation method on ten wines whose SO2 content varied from 42 to
1488 mg. per liter and found the difference between the two to vary from 0 to 5 mg.
TABLEI1
Collaborative Results on Sulfur Dioxide in Wine*
Total Sulfur Dioxide as p.p.m.
Analyst Distillation Ripper titration

MAJ 435 418
OSN 486 433
WHP 518 379
JHF 380 240
RWM 433 420
JLB 408 336
DB 433 424
MM 400 388
MAA 424 437
LQ 425 408
FQ 43 1 413
AGP 448 420
T 426 405
HP 438 370
Average 435 392
Maximum 518 433
Minimum 380 240
* Joslyn (1930).
118 M. A. JOSLYN AND J. B. 8 . BRAVERMAN
per liter higher or lower. A comparison between the Ripper method and the Nichols
and Reed (1932) distillation method by fourteen collaborators with the same sample
of white wine gave the results shown in Table 11. As pointed out by Monier-Williams
(1927), the Ripper method for total sulfur dioxide is subject, in addition to the errors
inherent in the iodometric determination of sulfurous acid, to errors due to oxidation
of sulfite in alkaline solution, to possible recombination of liberated sulfur dioxide with
the aldehyde, and to reduction of iodine by substances other than sulfur dioxide.
Under certain conditions by a compensation of errors results closely approximating
those obtained by the distillation method may be obtained. Mills and Wiegand (1942)
reported close agreement between the Nichols and Reed (1932) distillation method and
the Ripper method for sweet wines when the end point of the iodine titration was
taken as the first blue lasting approximately one minute rather than a blue color lasting
several minutes. Contrary results were reported by Archinard (1937), Casanave
(1910), Ferr6 (1915), Mathieu (1910), and Taylor et al. (1937).
The Ripper method is considered acceptable for white wines containing less than
300 p.p.m. of SO2 by Fabre (1936), Hennig (1946), RibCrau-Gayon and Peynaud
(1947), and Jaulmes (1951). Amerine (1952) modified it by decreasing the volume of
wine used from 50 ml. to 20 ml. Joslyn (1953) found better correlation between the
total SO2 content of wine determined iodometrically and that determined by Monier-
Williams reflux distillation when the amount of alkali present was sufficient to bring
the wine to p H 12.5 and the period of standing was reduced to 2 minutes. (Ten milli-
liters of wine pipetted into 3 ml. 1 N NaOH titrated after acidifying with 1 N HCl
before and after addition of 1 ml. of formaldehyde to correct for iodine reducing sub-
stances other than SO2.) Comparative values for several wines determined by modi-
fied acid-fuchsine colorimetric method were considerably higher than by the iodo-
metric or Monier-Williams method.
The Ripper method has been subjected to several modifications largely directed to
improving precision and to obtaining results in better agreement with distillation pro-
cedures accepted as being more accurate. Little or no attention, however, has been
given to determination of its relative accuracy. The fundamental data on rates of
dissociation of the aldehyde and sugar sulfonic acids have not been applied, nor has
recovery of added pure sulfonates been used as a measure of its accuracy. Recovery of
added sulfurous acid or bisulfites, as pointed out above, can not be accepted as suffi-
cient proof of its accuracy. Advantage of the data on effect of p H on rates of association
and dissociation and position of equilibrium has been used, however, in the methods
developed for determination of acetaldehyde, such as that developed by Jaulmes and
Espezel (1935) for acetaldehyde in wine, which was found satisfactory by Joslyn and
Comar (1938).
To avoid errors due to reduction of iodine by substances other than sulfur dioxide,
sulfite-binding compounds (acetone or formaldehyde) were added in excess after
acidification. Bennett and Donovan (1943) and Downer (1943) used acetone to com-
bine with the liberated or free sulfur dioxide, in order to determine the iodine-reducing
power of the juice itself. Prater et al. (1944) in extending the method to dehydrated
vegetables determined the effect of p H on stability of the acetone complex formed
with extracts of dehydrated cabbage, carrots, and potatoes in presence of large excess
of acetone and found this t o be in the range of p H 2 to 3, in which the starch-iodine
end point was sharper than a t p H 4. The difficulties with acetone as a binding agent
for citrus juices were pointed out by Ingram (194713) and others.
The Ripper method in modified form was extended to the determination of total
sulfur dioxide in frozen and dehydrated sulfited fruit and vegetable products. Jensen
(1928) was one of the first to suggest its application to determination of total sulfur
dioxide in foods after alkali extraction or digestion, and Brooks (1928) applied it to
dried fruit. I n Brooks’s procedure a 50-g. aliquot of finely ground dried fruit was
macerated with 100 ml. of distilled water and then treated with 50 ml. of a 5.6%solu-
tion of KOH, thoroughly mixed, allowed to stand for 15 minutes, acidified with 25 ml.
(1 + 3) sulfuric acid, and then titrated with 0.1 N iodine to a starch-iodine end point.
Prater et al. (1944) extracted and liberated the sulfur dioxide present in dehydrated
vegetables by suspending 8 g. in 400 ml. of water, adding 5 ml. of 5 N NaOH, and
allowing the mixture to stand 20 minutes. For cabbage, carrots, and potatoes the total
sulfur dioxide liberated increased during the first 10 minutes of treatment and then
remained essentially constant up to 60 minutes. A similar extraction of sulfite was used
by Potter and Hendel (1951) for dehydrated white potatoes. Reifer and Mangan
(1945) extracted the sulfite in dehydrated fruit or vegetable by adding 2-g. portions of
finely ground material to 350 ml. of boiling water containing 5 ml. of phosphate buffer
a t pH 7.6 and 10 g. of sugar. The flasks containing the mixture were then stoppered
and allowed to cool. Ponting and Johnson (1945) blended a 100-g. sample of fresh or
frozen fruit in a Waring Blendor with 10 ml. of 0.5 M tartrate buffer at p H 4.5 and
490 ml. of 20% NaCl solution for 3 to 5 minutes. For dried fruits they used a 20-g.
sample.
I n the Ponting and Johnson method, the extract was then filtered and the sulfite
in the filtrate was liberated by addition of 2 ml. of 1 N NaOH to 50 ml. aliquot and
allowing to stand for 30 seconds. In frozen apples they found that the liberation of
sulfur dioxide was complete in 30 seconds or less a t p H of 9 (see curve 2, Fig. 3). By
reducing the concentration of alkali added and the p H and time of alkaline treatment,
losses due to oxidation were minimized. After alkali treatment they as well as Reifer
and Mangan (1945), Prater el al. (1944), and Potter and Hendel (1951) acidified the
solution with hydrochloric acid in preference to sulfuric acid used b y Ripper. In the
Ponting and Johnson method, 2 ml. of 6 N HC1 were added to a 50-ml. aliquot and in
the Potter and Hendel (1951) procedure 7 ml. of 5 N HCl were added to 400 ml. of
extract. To determine reducing material other than sulfite, 1 ml. of 40% formaldehyde
and 10 ml. of 36% formaldehyde, respectively, were added to separate acidified
aliquots, and after standing for 10 minutes these were titrated with 0.02 N iodine to a
definite blue starch-iodine end point. Recovery of sulfur dioxide added as sulfurous
acid or bisulfite and agreement with results obtained b y the Monier-Williams (1927)
procedure were used as indices of accuracy.
Total Sulfur Dioxide: Distillation Procedures: The early development and appli-
cation of distillation procedures to various foods, beginning with Haas’s method of
distilling wine in a current of carbon dioxide into iodine solution contained in a Pelegot
tube immersed in water proposed in 1882, are critically reviewed b y Monier-Williams
(1927). The official distillation method of the A.O.A.C. in 1919 waa the distillation of
20 to 100 g. of sample in a current of carbon dioxide after addition of 5 ml. of 20%
phosphoric acid into bromine water and the determination of sulfate formed by pre-
cipitation with BaCL after expelling excess of bromine. This waa still the official
method in 1925, and in 1930 but with a cautionary note that i t was not applicable if
volatile organic sulfur compounds may be set free. I n 1930 Monier-Williams method
both in its volumetric and gravimetric modifications was listed as tentative. I n 1935
and subsequent editions of Methods of Analysis, the Monier-Williams method was
adopted as official, and the old bromine procedure (May, 1927) is no longer cited. The
iodine distillation method has been investigated from time to time since its proposal
by Haas (Anon., 1928; Black, 1928; Black and Warren, 1928; Nichols and Reed, 1932,
1933; H a n d and VoFigek, 1937), but because of errors involved (volatilization of
iodine solution in the receiving flask, distilhtjon of iodine-reducing substances other
120 M. -4. JOSLYN A N D J. B. 8. BRAVERMAN
than SO*, and interference by naturally occurring volatile organic sulfur, compounds
in products like onions, garlic, and horse-radish, see Dyer et al., 1941), it has not been
considered desirable for other than control tests. Iodine distillation, however, has been
used in analysis of wine and dried fruits and has been developed as a micromethod for
sulfurous acid in wine and fruit juices by Woidich (1930) and others. The method
proposed by Monier-Williams (1927) is based upon the selective oxidation of sulfur
dioxide by hydrogen peroxide at room temperature. When iodine or bromine are used,
oxidntion of hydrogen sulfide and of volatile organic sulfur compounds to sulfuric acid
occurs. Fitelson (1929) early compared the Monier-Williams method with the A.O.A.C.
method and found it to be superior to the latter even when cadmium chloride or copper
salts were used to remove sulfides from the distillates.
The Monier-Williams method consists in distilling for 1 hour with a reflux con-
denser and sweeping the sulfurous acid into a cold, sulfate-free, acetanalide-free,
neutral hydrogen peroxide by a current of'carbon dioxide. As a rule, volatile acids and
organic sulfur compounds do not distill over when a reflux condenser is med, although
when larger quantities of volatile organic sulfur compounds are present a great part of
them may be carried over. The retention of the volatile organic acids in the distilling
flask permits a volumetric determination of the sulfurous acid. Because of the use of a
reflux, however, a longer distillation is necessary to insure sweeping of all of the
sulfurous acid into the receiver. Also, the combined effects of the use of carbon dioxide
and a reflux condenser make frothing a minor problem in the actual manipulation.
Neutral hydrogen peroxide is used, thereby permitting titration of the sulfuric acid
with 0.1 N NaOH and subsequent gravimetric determination aa BaSO4, if desired.
All manipulation of the H202 is carried out in the cold, since on heating any H B or
organic sulfur compounds will be oxidized. Precipitation of the sulfuric acid as Bas04
is carried out in the cold. Although the method is time-consuming, it has the following
advantages:
1. The sulfurous acid is more completely liberated, owing to the use of HC1 instead
of HaP04.
2. Errors due to volatile S compounds are avoided.
3. Organic acids do not pass over with the distillate, thereby allowing a volumetric
determination.
4. Frothing is kept to a minimum.
Since its introduction in 1927, the Monier-Williams method has been modified
from time to time, chiefly in connection with design of apparatus and conditions of
refluxing and distillation (Nissen and Petersen, 1943; Thompson and Toy, 1945; and
others). It has been applied to a wide variety of foods, dried and dehydrated fruits
(Miller, 19271, dehydrated vegetables, juices, concentrates, wines, beers, etc. (Veisman
and Svereva, 1937).
I n general, it is agreed that distillation is the most reliable method of determining
sulfur dioxide in foods, but opinions differ considerably as to the way in which the
distillation should be carried out and as to the best method of determining sulfur
dioxide in the distillate. Some authors favor distillation in steam, others under reduced
pressure, and the majority distill in carbon dioxide or in some other inert gas. The
sulfur dioxide in the distillate may be oxidized to sulfuric acid by bromine, hydrogen
peroxide, or possibly other oxidizing agents, and determined either volumetrically or
gravimetrically.
I n the distillation method, precautions have to be taken to assure complete sepa-
ration of SO2 from the fruit product, to avoid oxidation of the liberated SO2 prior to
its absorption in the receiving flask, to eliminate the effect of other reducing substrtnces,
and to eliminate the effect of other sulfur compounds. Titration of the sulfur dioxide
in the distillate, either as such or after oxidation to sulfuric acid, gives acvxrate results
if precautions are taken to climinate volatile organic compounds. Gravimetric deter-
mination as Bas04 gives accurate results in nearly all cases, except, when the distillate
contains appreciable amounts of volatile sulfur compounds.
Most authors have used phosphoric acid for acidifying the food prior t o distillation.
In certain fruit products, such as dried fruits and molasses, where the sulfurous acid is
in close combination with sugars, i t is difficult to effect a complete liberation of the
sulfurous acid by using phosphoric acid. The use of hydrochloric acid insures a faster
and more complete liberation of sulfurous acid from its combinations. In the case of
apricots, even when relatively large quantities of hydrochloric acid are used, it takes
more than 1 hour for complete distillation of sulfurous acid. Preliminary treatment
of the product with sodium bicarbonate does not yield appreciably different results.
Increasing the acid concentrations hastens hydrolysis of the food constituents and
reduces frothing, but it may favor production of volatile sulfur compounds from
protein-containing foods.
Dried fruit or other sugary materials change to a dark or chocolate brown color
when boiled with hydrochloric acid, probably owing t o decomposition of sugars, etc.
With phosphoric acid, the amount of decomposition is slight. With HC1, the change in
color occurs more markedly toward the end of the distillation; some authors claim that
this change, which is more or less abrupt, denotes complete evolution of SOZ,and take
it as an indication of the end point. Monier-Williams has shown that reduction of
sulfates owing to this decomposition does not take place, whereas others have claimed
this to be the case and use it to explain the continued slow evolution of SO2 from the
product.
To avoid oxidation of the evolved SOzon its way to the receiving flasks, distillation
in a current of COz, or other inert gas, is resorted to. However, this oxidation seems to
be rather small, and similar results have been obtained with and without the use of a
stream of COz. It is optional whether the system be swept out first with a stream of
COzfrom an outside source and the food then distilled in this atmosphere, or whether
the COz is generated in the distilling flask itself by the addition of chalk, NazCOa, or
NaHC03, or a solution of NaaC03, even though in the latter case the system is not
swept as clear of air as in the former case. When carbonates or bicarbonates are used to
generate COz in the distilling flask, an excess of acid must be added to compensate for
the acid used up in generating COz.
NaHCOs + HCI + NaCl -t HzO -I- coz
or
NaaCOa + 2 H C 1 4 2NaCl + HzO 4-COz
The carbonates are less efficient in the production of COz than the bicarbonates, for
which reason, and because the latter can be obtained more readily in the pure state,
they are used to a greater extent than the normal carbonates.
In the distillation of solid foods, it is necessary to add water to the material in the
flask. It is optional whether the water is tap, distilled, recently boiled distilled, or
carbonated distilled water. The error due to the oxygen present in the water is negli-
gible when compared to other possible errors. To avoid foaming during the initial
period of distillation, it is desirable to add a small quantity of tannin or a few drops
of mineral oil. Paraffin will distill over and gum up the condenser and therefore should
not be used.
Distillation of sulfur dioxide into standard iodine solution and titration of the
excess iodine has been used with modifications by several chemists. It i s satisfactory
only when volatile organic compounds capable of reducing iodine are absent, and also when
necessary precautions are taken to avoid loss of iodine by volatilization i n the current of
carbon dioxide.
+ +
SO2 I2 2H20 ---t 2HI +H2SO4
+
IZ 2NazSzOa + Na&Oa +2NaI
The excess iodine is generally titrated with standard thiosulfate solution. To minimize
the loss of Izin the receiving flask, the flask should be shielded from heat reflected by
the burners used in distilling and should be equipped with a trap of some sort. Traps
containing K I or NapS2Oa have been used. Cooling the receiving flask in ice water is
recommended (see Fig. 6 ) .
Total Sulfur Dioxide-Other Methods: The acid bleached fuchsin-formaldehyde
procedure has been proposed for the determination of total sulfur dioxide in white
wine and beer after alkali treatment and acidification. For red wines and colored juices
this procedure can be used by distilling the sulfur dioxide with steam in a micro-
Kjeldahl still. Particular attention, however, has to be given to using an aliquot which
does not contain too much sulfur dioxide. In the procedure proposed by Grant (1947),
4 ml. of the fuchsin reagent are added to 1ml. of sulfite solution and after 5 minutes
the intensity of color formed is measured in a photoelectric colorirneter using a green
filter. The transmittancy increases with SO2 concentration over most of the range of
0 to 10 micrograms of S02. The bound SO1 in white wine is liberated by La Rosa
(1950) by treating 0.5 ml. of wine diluted with 15 ml. water with one drop of 10%
NaOH and allowing the mixture to stand 1 minute. With red wine 0.5-ml. sample and
2 ml. of 25% phosphoric acid is still distilled in a micro-Kjeldahl still into 1 ml. of
water containing one drop of 10% NaOH. One ml. of beer or its equivalent of distillate
may be used. Mathers (1949) found that distilling 50-ml. samples of wine after addition
of 50 ml. of 5% sulfuric acid and determining the sulfite in an aliquot of the distillate
with neutral lead acetate gave results as low as one-half that obtained by iodometric
titration or the Monier-Williams method. The neutral lead acetate solution, however,
was shown to be a good absorption medium for sulfur dioxide. Lead sulfite suspensions
could be acidified and the sulfur dioxide determined by iodometric titration. Low
recovery in the distillation procedure was ascribed to oxidation of sulfur dioxide
during distillation, since the same wines when distilled under reflux after addition of
oxalic acid or sodium arsenite gave photometric values closely approximating those
by iodometric titration and the Monier-Williams titrimetric or gravimetric procedure.
The photometric lead sulfite method is limited to 10-ml. aliquot of distillates contain-
ing 5 to 100 mg. per liter of sulfur dioxide. The fuchsin-formaldehyde procedure is
limited to determination of 0 to 10 micrograms of sulfur dioxide in 1ml. of sample with
an error of less than 0.1 microgram for pure solutions (Grant, 1947). Joslyn (1953)
obtained higher reproducibility3 with the Steigmann (1950) modification. Acid-
bleached fuchsine prepared by adding 100 ml. of concentrated sulfuric acid to about
800 ml. water, cooling the solution, and then adding 0.5 g. of National Aniline Com-
pany’s basic fuchsine (NF60) moistened with 20 ml. of alcohol, gave a stable clear
solution. This was mixed with dilute formaldehyde solution (1/20) before use in the
volume ratio of 10 parts of the fuchsine solution to 1 part of formaldehyde solution.
To 1 ml. of wine pipetted into a 100 ml. glass-stoppered flask were added 20 ml. of
mixed reagent and after mixing diluted to volume and transferred to colorimeter tube
for free S02. For total SO2 in white wines, to 1 ml. of wine were added 3 ml. of 0.1 N
NaOH, and after 2 minutes exactly 3 ml. of 0.1 N NaOH and 20 ml. of reagent and
brought to volume. If the alkaline wine were only partially neutralized before addition
of color reagent, the readings were lower than with wine that was not acidified or wine
that was over acidified. Similar results were obtained with pure sulfurous acid solu-
tions and with distillates from red wine. The values for total SOz content by this
method, however, did not agree with iodometric distillation or titration values.
IV. SULFURDIOXIDEAS PRESERVATIVE AND SANITIZING AGENT
The preservative value of sulfur dioxide in fruit and vegetable prod-
ucts and other foods and beverages is discussed by Abdulov (1938),
Cruess (1948), Borgstrom (1953), Tanner (1944), von Schelhorn (1951),
and Woodroof and Cecil (1945). The actual mechanism of the preserva-
tive value of sulfurous acid and its salts, however, is not known. It is
believed that its strong reducing power may beinvolved either by reducing
the oxygen tension in the food tissues and in beverages to a point below
that a t which aerobic organisms can grow or by maintaining some enzyme
system necessary for growth in a reduced state. Sulfur dioxide is only a
temporary preservative when used in moderate amounts because of loss
of preservative value on oxidation to sulfate, on volatilization, and on
combination with chemical constituents capable of forming a-hydroxy-
sulfonic acid or other addition products. Because of the readiness with
which it can be removed to a large extent by heating under vacuum it has
been favored in Europe as a temporary preservative, particularly for the
bulk storage of juices and pur6es. There is some evidence that more sulfur
dioxide is required to prevent fermentative activity than to inhibit
growth (Bioletti and Cruess, 1912). Sulfur dioxide also has a selective
antiseptic action (Cruess, 1911). Acetic acid bacteria, lactic acid bacteria,
and many varieties of molds are more sensitive to sulfur dioxide than
yeasts. Among the yeasts, the more strongly aerobic species are generally
more sensitive than the more fermentative species. The fermentative
yeasts, particularly those strains selected as being more desirable indus-
trially, can be adapted to sulfur dioxide, according to Porchet (1931).
Bioletti and Cruess (1912) did not believe that adaptation occurred but
the evidence presented by Porchet was convincing. The antiseptic action
of sulfur dioxide towards microorganisms, particularly yeasts, varies with
stage of growth or development, microbial population, temperature, pH,
and composition of product treated (acetaldehyde content, sugar content,
alcohol content, etc.)
Effect of p H : Muller-Thurgau and Osterwalder (1914) were the first to
observe that sulfurous acid and its salts were effective as preservatives
only in acid media. Perry and Beal (1920) confirmed this by finding that
with more acid solutions the lower the concentration required to inhibit
fermentation and mold growth. Bioletti and Cruess (1912) did not find
that acidity or sugar content was involved in the decrease in antiseptic
effect of SO2in ripe as compared with unripe grapes. Subsequently how-
ever, Cruess and Irish (1932), Cruess (1932), and Cruess et al. (1931) in
their investigations of the effect of pH on toxicity of sulfurous acid and
other preservatives found that at pH 3.5, two to four times as much SO2
was required to inhibit growth as a t pH 2.5, whereas at pH 7 sulfite was
without effect on yeast and molds and as much as 1000 p.p.m. were
required to inhibit growth of bacteria in apple juice. Cruess and his
collaborators suggested that antiseptic properties might be confined only
to undissociated sulfurous acid or to molecular SO2 and be absent in the
neutral sulfate ions. This was also suggested by Rahn and Conn (1944),
who concluded that yeast growth was checked only by undissociated
sulfurous acid and not by bisulfite or sulfite ions. They reported that
4 p.p.m. of undissociated sulfurous acid inhibited growth of Saccharomyces
ellipsoideus, whereas 100 p.p.rn. of bisulfite ions were required to inhibit
bacterial growth. Gillespy (1946) also considered un-ionized sulfurous
acid as the lethal agent. He found that the effect of SO2in increasing the
heat sensitivity of the asci of Byssochlamys fulva was a function of pH.
Vas and Ingram (1949) pointed out on the basis of the calculated
distribution between H2S03, HSO,-, and sos- that even slight changes
in pH, in the region of pH 3.5 and above, would markedly affect the pro-
portion of undissociated sulfurous acid present. Thus the proportion of
free sulfur dioxide present in the undissociated form increases from 0.5%
at pH 4 to 5.5% at pH 3. With an osmophilic Zygosaccharomyces strain
isolated from orange concentrate, as little as 1.5 mg. of SO. per liter com-
pletely inhibited it. They suggested addition of acid to lower the pH
value as a means of obtaining better preservation with less sulfur dioxide.
A t the lower pH, combination of sulfur dioxide with glucose is delayed so
that this allows a longer time for a greater quantity of free sulfur dioxide
to act on the microorganisms present as well as increases the proportion
of the antiseptic sulfurous acid present.
Bound Sulfite: It has been known for a long time that sulfurous acid
when combined with aldehydes or sugars exercises practically no anti-
septic action. This was first observed by Ripper in 1892 and was later
observed by Harter in Germany in 1911, by Muller-Thurgau in Switzer-
land in 1914, and by Laborde in France in 1916 (see Monier-Williams,
1927). Bioletti and Cruess (1912) reported that free SO2 has more than
thirty times the disinfecting power of bound SO2 and is sixty times as
effective as bound SO2 in inhibiting fermentation. Yeasts transferred to
fresh must from water suspensions containing 350 p.p.m. of free SO2 did
not ferment, whereas those from grape musts containing 2400 p.p.m. of
total SO2of which only 277 p.p.m. were free did ferment. Downer (1943)
explained the greater susceptibility of citrus juice concentrates to fermen-
tation at the same level of total SO2 than dilute juices, as being due to
their binding a greater portion of the added SO2. Downer divided a sul-
fited citrus juice inoculated with yeast into two lots, to one of which
acetone was added to bind the free SO2. This lot having a much lower
concentration of free S O r fermented sooner than the untreated lot.
Although i t is generally accepted that in using SOz to preserve wines,
juices, and concentrates, the free SO2 is the best guide to its preservative
action, the first unequivocal proof of this was made by Ingram (1918).
Ingram selected a Zygosaccharomyces species which grew a t the same rate
in nutrient solutions with 4 or 40% glucose, and added a total quantity
of SO2 such that in the 4 % glucose most of the SO2 was free, while in the
40% glucose only a small proportion of it was free. I n this way he varied
the quantity of free SOz without, also changing any other factor which
might affect growth. Under these conditions the total viable yeast count
was related t o the concentration of free SO2 and not to the amounts of
bound SOz present. A similar correlation was found between the growth
of yeast in concentrated orange juice and the concentration of free, but
not the combined or total, SOZ. Ingram concluded therefore th a t the
bound SO2 has little if any germicidal value.
Other aspects of preservative value are discussed by Lochhead and
Farrell (1936), Hamman (1951), Souci (1951), and von Schelhorn (1953).
Sanitizing Action: Sulfur dioxide has long been used in winery plant
sanitation for destroying undesirable microorganisms on surfaces of wood
or concrete equipment, conveyors, fermenters, storage containers, wall,
and floors, and for disinfecting wood and concrete platforms or crushing
equipment or for spraying on grounds and pomace piles outdoors. It is
also used in sanitizing bottles and corks. It has been used abroad in
sanitizing equipment and bottles and closures used in sterilization-filtra-
tion of wines, fruit juices, and vinegars. I t s corrosiveness reduces its
application in the fermentation industries to surfaces more resistant than
the ordinary metal surfaces and its reducing action and pronounced odor
and taste limit its application to other food products. It is particularly
adapted t o the sterilization of wooden surfaces because of its ready pene-
tration into the pores and its resistance to dissipation by attack upon or
combination with wood. Its application in wine manufacture is described
by Amerine and Joslyn (1951). Sulfurous acid and sulfite solutions may be
applied continuously by sprays or by immersion for the continuous con-
trol of microorganisms on conveyors and food-processing equipment in
the dried fruit industry (Vaughn et al., 1948).
Efect of Sulfur Dioxide on Heat Processing: Sulfur dioxide can be
combined with heat processing in preservation of juices and pur6es. In the
presence of sulfur dioxide a more rapid destruction of microorganisms
occurs so that lower pasteurizing temperatures and times may be used. A
more rapid destruction of Byssochlamys fulva in the presence of sulfurous
acid was observed by Gillespy (1946). Pederson and Tressler (1938)
126 M. A. JOSLYN AND J. B. 6. BHAVERMAN
observed more rapid destruction of yeast in apple juice in the presence

of sulfurous acid. This effect has long been used in the practice of pre-
serving fruit pulps, particularly berries, by sulfiting after preheating, but
it has not been investigated. Sulfur dioxide is used in small amounts t o
shorten the heat treatment in the repeated two-stage pasteurization of
grape juice in France (Borgstrom, 1953). Heating, however, may alter the
ratio of free to total sulfur dioxide. Under some conditions it may increase
the sulfur dioxide binding power, as Ingram (1949) reported for orange
juice concentrate.
In addition to increasing the thermal death rate of microorganisms
present sulfur dioxide also increases the resistance of fruit juices and other
products to change in color and flavor on heat treatment. This increase
in resistance to heat damage is quite pronounced in pasteurized orange
juice and is also directly observable in dehydrated vegetables. Cruess
et al. (1944a, 1944b) reported that the temperature of dehydration for a
number of vegetables could be increased after sulfiting by at least 18" C.
(10" F.) without appreciably affecting color and flavor. The nature of
this effect on reducing heat damage, however, has not been thoroughly
investigated as yet.
DesulJiting: The possibility of removing sulfur dioxide from fruit
products preserved with sulfur dioxide has long been recognized and used
in the bulk storage of various fruit products for subsequent processing.
Maraschino cherries and other fruit are preserved in a sulfite brine and
then, after removal of excess sulfur dioxide by a combination of leaching
with water and boiling, used in preparing candied and glaced products.
Cruess and Nouty (1927) were among the first to investigate the problems
involved in storage and leaching of sulfur dioxide from a wide variety of
fruits. It has long been a practice in British Columbia to store small fruits
and fruit pulps with added sulfite for subsequent production of jams, and
this method was intensively investigated by Atkinson and Strachan
(1941a) in British Columbia and by Charley (1934) in England. This prac-
tice is discussed by Morris (1933) and by Cruess (1948), and will be dis-
cussed later in this review. The removal of sulfur dioxide from whole small
fruits and fruit pulps, however, is never complete. The quantities of sulfur
dioxide remaining are usually too small and the subsequent processing
(candying, dyeing, or jam making) is such that alteration in flavor as a
result of sulfuring and desulfuring is not serious. Incomplete removal,
however, is a problem where the product is canned, because of interaction
of the residual SO2 with tin plate.
Fruit juice beverages are prepared extensively in Europe, England,
Australia, New Zealand, India, Israel, and elsewhere from sulfited juices
and concentrates by dilution and sweetening which tend to mask the
undesirable sulfite odor and taste. Such products, however, are not as
desirable as the fresh, frozen, or flash pasteurized juices.
The possibility of removing the sulfite added as a preservative by
heating to decompose the combined sulfur dioxide and remove the free
sulfur dioxide by volatilization has been investigated from time to time.
In the earlier methods, heating under vacuum in steam-jacketed vacuum
pans was used. Cruess and Berg (1925) investigated this as a means of
removal of sulfurous acid from grape syrup but did not find it to result
in complete removal or to be without effect on flavor. More recently
specially devised apparatus to increase removal of sulfur dioxide by
mechanically agitating the juice, bubbling an inert gas through it, or
by spraying the juice while under vacuum against baffles have been
introduced (Fabre, 1947). I n a recently developed French process heating
under vacuum is combined with a reflux condenser to recover volatile
flavoring constituents. Although some French investigators claim that
grape juice so desulfited is the equal in quality of the fresh, the Swiss have
not accepted these claims and desulfited juice cannot be sold as fresh in
Switzerland. Liithi (1950) particularly objected to the preservation of
fruit juices with sulfurous acid, now used widely in many countries,
particularly in warm climates, and their distribution after desulfiting. In
his opinion, although sulfur dioxide is a very convenient method of
preventing fermentation and changes due to oxidation, these results can
be obtained without difficulty by other means and with less effect on
quality.
In addition to desulfiting by physical methods, chemical methods
ranging from treatment with oxidizing agents like hydrogen peroxide to
addition of substances to combine with free sulfur dioxide such as acet-
aldehyde have been proposed for the treatment of juices and dried fruits.
Vacuum drying in a shelf drier proposed originally as a method of re-
moving sulfur dioxide from dried fruits now is used for the production of
apple nuggets and similar dried fruit products.
V. SULFURDIOXIDEAS AN OF ENZYMIC
INHIBITOR AND NONENZYMIC
BROWNING
The enzyme-catalyzed oxidative browning of fruit and fruit products
was reviewed by Joslyn and Ponting (1951) and the nonenzymatic
browning by Stadtman (1948). As pointed out by Joslyn and Ponting,
sulfurous acid and sulfites have long been known to inhibit the enzyme-
catalyzed oxidative discoloration of fruits as a result of mechanical or
physiological injury during the ’ preparation for canning, drying, or
freezing. The application of sulfurous acid and sulfites to the preparation
of apples for baker’s use as freshly peeled, cored, and sliced apples or as
128 M. A. JOSLYN AND J. 13. S. BRAVERMAN
froaen apples is described by Joalyn and Mrak (1933), Anon. (1944a,

1945), MacArthur (1945), Joslyn and Hohl(l948) , and Tressler and Evers
(1947). The preparation of prepeeled potatoes and the control of their
discoloration by sulfites and other antioxidants is described by Olson and
Treadway (1949). Under certain conditions when the structure of the fruit
tissue and the residual oxygen content of the interior cells permits, it is
sufficient to protect the exposed surfaces of the cut fruit against oxidation
by dipping them into, or spraying them with, dilute sulfurous acid or
sulfites. I n other cases it is necessary for the sulfur dioxide to penetrate
the fruit tissue completely, and therefore the concentration of sulfite solu-
tion and duration of treatment are important (Ponting, 1944). Synergistic
effects were obtained by Johnson and Johnson (1952) by using a combina-
tion of sodium chloride, ascorbic acid, and ‘sodium bisulfite by means of
which it was possible to inhibit enzymatic browning by using quantities
of sulfite too small t o be tasted and still completely inactivate the poly-
phenol oxidase in Jonathan apples. In this connection it is important to
realize that in the taste appraisal of sulfited foods special attention should
be taken not to offer too many samples at one time and to develop suitable
scoring techniques (Boggs and Ward, 1950).
Joslyn and Ponting (1951) suggested that polyphenol oxidases were
the chief if not the only enzymes involved in browning of plant tissues.
The role of sulfur dioxide in inhibiting this browning, however, is not
known. Enzyme-catalyzed oxidation of sulfur dioxide was observed by
Ponting and Johnson (1945) but not investigated. Sulfur dioxide con-
ceivably could act by reducing oxygen and making it unavailable for
oxidation or by reacting with the quinones or other intermediates in
polyphenol oxidation. All naturally occurring melanins are conjugated to
proteins, and Mason (1953) favors interaction of enzymically produced
quinones with proteins rather than oxidation of tyrosine residues within
the polypeptide chains. If interaction with proteins is also involved in
enzymatic browning, then sulfur dioxide could act either upon the enzyme
protein or on intermediate products of oxidation.
The nonenzymatic browning of fruit and vegetable products appar-
ently involves the interaction of sugars, organic acids, and amino acids
and proteins (Stadtman, 1948). Whether decomposition of sugars t o
furfural and other substances precedes condensation with nitrogenous
constituents or not is not known. There is a strong possibility that the
inhibition by sulfur dioxide of browning reactions involving interaction
of sugars with amino acids and protein\s may be due to the stabilization
of the intermediate formed. Neither the mechanism of browning in a
particular instance nor the effect of conditions of sulfuring or sulfiting are
known. As pointed out previously, there are little if any data on the actual
SULFUR DIOXIDE TREATMENT OF FRUIT h N D VEGETABLE PRODUCTS 129
distribution of sulfur dioxide into its various inorganic and organic states
of combination. If for example inhibition of browning in dried apricots by
sulfur dioxide were due to the fact that hydroxymethyl furfural and
similar substances did not form from glucose hydroxysulfonic acid or that,
the latter did not combine with amino acids and proteins, then control
of browning would be based on conditions of sulfuring which would favor
the formation and accumulation of glucose sulfonate. On the other hand,
if the inhibition were due to the formation of a relatively stable sulfur
dioxide compound of the intermediate amine or Schiff base then better
protection would be afforded by allowing this t o be formed before adding
sulfite. A third possibility is the reaction between glucose hydroxysul-
fonate and amines referred t o by Danehy and Pigman (1951). The in-
ability of fructose t o form a-hydroxysulfonates and its reactivity in
browning, however, would indicate that the formation of sugar bisulfite
addition compounds is not the chief cause of inhibition of browning by
sulfur dioxide.
Hodge (1953) in his recent review of the chemistry of browning reac-
tions in model systems pointed out some relationships between the many
different types of reactions leading to the production of brown pigments
a t moderate temperatures: carbonyl-amino, non-amino, oxidative, etc.
He proposed a mechanism for browning in sugar-amine systems based on
the Amadori rearrangement in the Maillard reaction and stressed the
importance of dehydrogenated reductones in both enzymatic and non-
enzymatic browning reactions. The known inhibitors for browning,
cyanide, dimedon, hydroxylamine, hydrazines, mercaptans, and bisulfite,
are chiefly carbonyl reagents. However, mercaptans and bisulfites, which
are the best of the above inhibitors from the practical standpoint, are also
reducing agents. Hodge suggested that their functions as inhibitors may
be related to this property, i.e., their ability to keep reductones involved
in browning in the inactive reduced form rather than the active dehydro
form.
The elucidation of the mechanism of inhibition of browning by sulfur
dioxide thus still remains for the future and upon it will depend the more
rational use of sulfur dioxide in the industry.
VI. SOURCE
AND APPLICATION
OF SULFUR
DIOXIDE
In the pretreatment and preservation of foods with sulfur dioxide, it
may be applied by burning flowers of sulfur, obtained directly from under-
ground or underwater deposits as is done in Louisiana or from iron
pyrites or similar sources, in pans, specially devised burners, or as sulfur
wicks or matches. Salts of sulfurous acid, particularly the alkali or acid
salts (sodium or potassium bisulfite or metabisulfite), the alkali neutral
130 M. A . JOSLYN AND J. B. 8. BRAVERMAN
salts (sodium or potassium sulfite) may be added as the dry chemicals t o

liquid products or after solution in water. More recently liquid sulfur
dioxide has been introduced commercially and this has largely supplanted
other forms of sulfur dioxide.
Sulfur Fumes: Sulfur wicks or matches were at one time widely used
for sulfuring barrels or casks in wineries by suspending by means of a
sulfur bung and burning in the cask. The undesirable melting and drop-
ping of the sulfur from such wicks on the bottom and walls of the con-
tainer was minimized by the use of sulfur cages or cups. Various devices
for sulfuring empty containers of must and wine by burning the sulfur
outside the cask and passing the fumes into the cask or into the wine are
described in the old literature on wine making (for example, Bioletti and
Cruess, 1912). The introduction of melted sulfur into the fermenter either
by direct melting and deposition or by volatilization, the introduction of
undesirable products of combustion from the supporting cloth, and the
difficulties in measuring and regulating the amount of sulfur dioxide
introduced strongly limited this method. It was believed by Bioletti and
Cruess (1912) that hydrogen sulfide formed during alcoholic fermenta-
tion was derived by reduction of the sublimed or melted sulfur introduced
during sulfuring.
Fumes of burning sulfur were considered by Bioletti and Cruess (1912)
to be the cheapest source of SO2and the best for disinfecting purposes but
unsuitable for control of fermentation because of uncertainty in applica-
tion and difficulty in regulating.
The occurrence of hydrogen sulfide in the gases evolved during fer-
mentation of grape and other fruit juices and of cereal beverages has been
widely reported since Nessler showed in 1869 that free sulfur in a ferment-
ing must causes the formation of hydrogen sulfide. The formation of
hydrogen sulfide from sulfites and sulfates as well as from sulfur has been
observed. Autolysis of yeast and the reduction of the sulfur-containing
amino acids is also known to be involved. The extensive literature on
hydrogen sulfide formation during fermentation has been reviewed
recently by Ricketts and Coutts (1951), who have also published the
results of their investigations on hydrogen sulfide formation during the
fermentation of wort. They confirmed the early investigations of Oster-
walder, who showed that hydrogen sulfide formation was a characteristic
of certain strains of fermenting yeasts independent of the presence of free
sulfur or of decomposition products, All bottom fermenting yeasts were
found by Ricketts and Coutts (1951) to produce H2S during fermentation
of malt worts or of sugar solutions, but the top fermenting yeasts either
did not produce H2S or only in traces after a period of storage. Sulfites
and sulfates stimulated H2S production. As a result of their investigations
on the effect of various inhibitors and other substances on H2S production,

they concluded that it is catalyzed by some dehydrogenase, either the
triose phosphate or the alcohol dehydrogenase or both.
Although sulfur fumes are used at present to but a limited extent in
wine making and other alcoholic beverage industries, except for the
sulfuring of small casks and barrels, the use of sulfur fumes in the dried
fruit industry is still common. Cut fruits and to a limited extent grapes
are treated before sun drying or dehydration with fumes of burning sulfur
in specially constructed suIfur houses. Long el al. (1940) investigated the
design factors involved in the construction of a suitable sulfur house to
permit rapid and uniform sulfuring and the factors influencing absorption
and retention of sulfur dioxide by cut fruit sulfured with fumes of burning
sulfur. They reported that sulfur contaminated with petroleum oils and
similar organic materials or burnt in dirty pans did not burn completely.
Bisson et al. (1942) investigated this factor in more detail. Sulfur house
construction and operation is described also by Mrak and Long (1941)
and Phaff and Mrak (1948). The effect of time and temperature of sulfur-
ing on absorption of sulfur dioxide by cut fruits is discussed by Fisher
et al. (1942). Mrak et al. (1942) discussed the effect of certain substances
and pretreatments on retention of sulfur dioxide.
In beet-sugar and cane-sugar factories situated far from industrial
centers where liquid SOz is produced, sulfur dioxide produced in sulfur
burners still is used owing to high transport costs for the heavy steel
containers necessary to store liquid SO2. Marches (1953) described the
various types of sulfur furnaces used in the cane-sugar factories of Java
for the generation of sulfur dioxide by burning commercial sulfur in a
stream of air. He discusses in detail Blekkingh’s analysis of the process
and the effect of air volume, temperature of sulfur, air, and gas, and other
factors on the SO2 and SO3 content of the gas. It was found early in
Java that the quality of sulfur used affected performance of the sulfur
furnace. Sulfur containing impurities such as bitumen which form a slag
on the burning surface hinder the evaporation of the sulfur underneath
and thus decrease furnace capacity. Stirring devices by which the sulfur
surface can be skimmed, thus eliminating the slag film, were introduced
in Java to facilitate burning of poorer qualities of sulfur.
Alkali Salts of Sulfurous Acid: The ease with which they could be used
either in the dry form or as solutions has long made the alkali sulfites and
bisulfites particularly attractive in sulfiting a wide variety of products.
Potassium metabisulfite has long been used in wine making for disinfec-
tion, control of fermentation, and preservation. Stability, freedom from
heavy metal and arsenic impurities, and cost are the chief factors that
determine selection of salts. In addition their possible effect in reducing
total acidity and introducing metallic cations has to be considered in Borne

products. The sulfites and bisulfites differ in the ease with which they are
absorbed and penetrate into fruit and vegetable tissues. In the dehydra-
tion of vegetables, the neutral sulfites or a mixture of neutral sulfites and
bisulfites is preferred and this is best applied as a spray one-third of the
way down a continuous blancher (Cruess and Mackinney, 1943; Mac-
kinney, 1945) or as a series blanch in batch blanching. Sulfite solutions
also are better absorbed by apples and other fruit that is treated by
dipping or immersion before freezing (Joslyn and Mrak, 1933). Sulfite
solutions are useful also in sulfiting apples and other cut fruits before
drying (Mrak et al., 1942). Commercial sulfiting procedures are discussed
by Beavens and Bourne (1945), Woodroof and Cecil (1945), and Cruess
(1948).
Liquid Sulfur Dioxide: The advantages of liquid sulfur dioxide were
apparent to Bioletti and Cruess (1912) even before it became available
commercially. They cite the accuracy with which its doses can be meas-
ured and the absence of impurities and prefer it to solutions of sulfur
dioxide in water, which they found to be quite variable in strength,
corrosive, and bulky and inconvenient to handle. The use of liquid sulfur
dioxide in wineries has many advantages: its stability and constancy of
composition, the fact that it introduces no undesirable residues (S, HzS,
impurities, etc.), and its adaptability to treating wines of low fixed
acidity. Its apparently greater cost is more than offset by the ease with
which it can be used and its freedom from harmful impurities. Its present
wide adoption in the wine and other food industries, however, depended
upon the development of methods of handling, transferring, and meas-
uring. At first it was used in larger wineries by taking off directly from a
cylinder on a platform scale through suitable connections to a distributer
in the fermenting or storage tank and measured by weighing. As soon as
pressure vessels fitted with calibrated gage glasses for introducing meas-
ured volumes of liquid sulfur dioxide became available, its use even in the
smaller plants became widespread. The use of liquid sulfur dioxide in
various food industries is described by Willson et al. (1943a, b).
The effect of type of sulfur dioxide compound used and method of
application on the changes occurring in color and flavor during subse-
quent treatment and storage is still largely unknown. Such investigations
as have been made have been limited by unavailability of methods of
analysis which would determine the distribution of sulfurous acid in the
product. It has long been believed that fumes of burning sulfur are par-
ticularly efficacious in the sulfuring of cut fruits (Nichols and Christie,
1930). It was believed a t one time that the presence of SO3 as well as SOz
improved penetration and retention, but it is now known that tempera-
ture and concentration of SOz are the more important factors influencing
absorption and retention (Long et al., 1940). In the industry it was felt
that it was desirable to sulfur the fruit initially to a sufficiently high level
and that fruit that was resulfured did not store as well. The data reported
by Stadtman et al. (1946) for commercially dried apricots which were
resulfured to various levels with liquid SO, or by adding charcoal satu-
rated with SO2do not indicate that resulfuring had any effect. The storage
life at a given moisture content increased linearly with initial SO2 content
in the range of 1500 to 8000 p.p.m. Whether this would be true for other
fruits and other conditions is not known. A difference in palatability has
been observed between dehydrated white potatoes sulfured by sulfite
sprays during blanching and those sulfured during dehydration by the
sulfur dioxide present in the hot air as a result of sulphur impurities in
the oil being used as a fuel. This field still needs t o be critically and
objectively investigated.
VII. SULFUR DIOXIDE IN FRUITJUICES, SYRUPS,
CONCENTRATES, AND PUREES
Fruit juices, particularly citrus juices, are commonly preserved with

added sulfurous acid or sulfites, particularly in warm countries. They
may be preserved with sulfurous acid in bulk for subsequent marketing
after sweetening with sugar as beverage bases or as diluted still or car-
bonated beverages. With proper precautions when the initial microbial
infection is low the acid juices may be preserved by the addition of SUE-
cient sulfite or liquid SOz to bring the total SO2 content to between 350
to 600 p.p.m. When the free SO2 content is low and the infection is high
or when SOz-tolerant microorganisms are present, this may not be suffi-
cient to preserve the juices. Under these conditions flash pasteurization,
or the use of a combination of sodium benzoate and sulfites, may be prac-
ticed. To avoid the clearing of citrus juices due to the activity of the
naturally occurring pectic enzymes, flash pasteurization before sulfiting is
desirable. The practice of sulfite preservation of citrus juices is discussed
in detail by Braverman (1949), Downer (1943), Feigenbaum and Israel-
shvili (1949), Tressler et al. (1939), and others. Sulfur dioxide is used both
to inhibit deterioration due to yeast and mold activity and to inhibit
discoloration. The latter can be accomplished usually with much lower
levels of SO, than the former. In less acid juices, such as apple juice,
more SO2is required (see Section IV, p. 124, and Yamada and Okumura,
1950).
The concentration of SO2 required for preservation depends on the
extent and type of infection, source of SOz, and composition of the juice
(sugar content and pH). There is some evidence that the bisulfites and
metabisulfites are somewhat more efficient in preserving citrus juices

than is sulfurous acid (Feigenbaum and Israelshvili, 1949). The addition
of a small initial dose followed by storage until clarification occurs,
siphoning off, and resulfiting has been found superior to preservation by
a single large dose of sulfur dioxide for lime juice.
Fruit juice concentrates, particularly citrus juice concentrates, are
also preserved with SO2. Large quantities of lemon juice concentrated t o
not less than 40" but not more than 45" Brix were packed in California
for shipment to Britain in World War I1 in fir barrels preserved with
500 p.p.m. of sulfur dioxide. Orange juice concentrate, testing a minimum
of 65" Brix, was also prepared in California and Florida for the Lend Lease
Program but this was usually flash-pasteurized in No. 10 cans. Some was
preserved with sulfur dioxide. During the War and after large quantities
of orange concentrate preserved with sulfur dioxide were prepared in
Palestine for shipment t o Great Britain, and this is still an important
outlet for Israeli citrus concentrates (orange and grapefruit). The con-
centrated juices, however, because of the fact that only a small part of the
added sulfur dioxide is free, require much higher concentration of total
S o t , usually ranging from 1000 t o 1500 p.p.m. or more. The combination
of sulfur dioxide with concentrated orange juice was investigated ex-
tensively by Ingram (1949) and Ingram and Vas (1950a, b).
The rate of association and equilibrium between SO2 and glucose was
investigated early by Kerp (1904, 1913) and more recently by Vas (1949)
and Braverman (1953). The rate of association and extent of association
both vary with concentration of sugar.
In a food product containing only 2% of glucose the percentage of
bound SO2 will increase only to about 15% of the total SO2 added, but in
a product containing 45% glucose more than three quarters of the total
sulfur dioxide will combine, leaving only about 25% of free SO2. Ingram
and Vas (1950a, b) compared the relation between the combining power
of SO2 and glucose concentrations in pure solution and in various citrus
juices. These authors confirm an essential feature, previously noted by
Downer (1943), that the combining power in various citrus juices and
concentrates is always greater, and often very much greater, than that of
the sugar present. They do point out, however, that these data are some-
what unreliable because the concentrations of glucose present were cal-
culated as half the total monosaccharides (presumed t o be invert sugar) ,
the temperature a t which combination occurred was not known, and the
quantity of added SOz varied from one set of experiments to another. In
making this comparison, however, the authors took into account (as
shown in Fig. 6) only the amount of glucose present in the various juices
tested, and paid no attention to the total concentration of the soluble solids.
These experiments led Ingram and Vas t o conclude that the presence
of glucose is not the only factor responsible for the binding of SO2 and
that other aldehyde and ketone-like substances and possibly pectin are
the cause of the higher combining power of SOz in these juices. However,
9
I I I
9
R\@ ___ - - - - -- -- - -0
*--5
l @ l
I I
a"
ul
P
i
3
I I I
25 50 75
X GLUCOSE OR TOTAL SOL. SOLIDS
FIG.6. The relation between the power of combining with SO2 and the glucose
concentration in pure solutions and various juices. The solid black line and the num-
bers are those of Ingram and Vas (1950a1b). The dotted line and the encircled num-
bers are those of Braverman (1953).
1. 7-fold Rhodesian orange.
2. 6 34-fold U. S. and Palestine orange (250 samples within rectangle).
5. 4-fold orange.
6. 4-fold orange.
7. 4-fold grapefruit.
8. 4-fold lemon.
9. %fold u. s. lemon.
10. 3-fold Jamaica grapefruit.
13. Natural grapefruit.
14. Natural orange.
15. Natural lemon.
by replotting the results obtained by these authors, not against glucose

contents but against the total concentration of soluble solids, it can be
shown that all the points lie on the same curve as that of pure glucose.
This clearly indicates that although glucose is largely responsible for
136 M. .4. JOSLYN AND J. B. S. BRAVERMAN
binding SO2 the degree of combination is entirely due to the total concen-
tration of soluble solids in the food. (See encircled numbers in Fig. 6 . )
To prove that the combining power of SO2 is dependent largely on the
total solids concentration, Braverman (1953) compared five samples each
containing the same amount of glucose with varied amounts of sucrose,
which does not bind SO2 a t all and which was used in this case solely t o
raise the total solids in the respective solutions. The results showed that
although they each contained the same amount of glucose, the percentage
of combined SO2 increased progressively with increase in total sugar
content.
Fruit pulps and purees as well as juices and concentrates may be
preserved with sulfur dioxide. This practice is quite widespread abroad,
as was mentioned before, although it is not used in the United States,
since preservation freezing of fruits and fruit pulps is preferable to sulfiting
for subsequent use by jam and preserve industries. The preservation of
fruit pulps with sulfur dioxide is described by Atkinson (1941), Atkinson
and Strachan (1941), and Charley (1934). Both cold fruit pulps and hot
fruit pulps may be barrelled with SO2; the latter usually require a lower
concentration of SO2 for preservation.
Cooked fruit pulp, preserved in barrels with sulfur dioxide, has long
been used successfully by British jam manufacturers. This method was
varied by research workers at the Long Ashton Research Station by the
development in 1924 of the prcservation of raw fruit with a solution of
sulfur dioxide in sealed containers (see Wallace and Marsh, 1953). This
cold process method, however, had the disadvantage of toughening the
skins of some of the fruits, progressive loss of pectin during storage, and
of being limited only to the preservation of more acid fruits. In 1940 the
cold process method was reinvestigated in England for the preservation
of a glut of plums. For domestic use prepared tablets of sodium and
potassium bisulfite were introduced in 1941.
VIII. SULFURDIOXIDEIN WINE AND VINEGARMAKING
Sulfur dioxide is used in wine making as a sanitizing agent to eliminate
undesirable microorganisms from all surfaces of equipment and from
fermenters and storage tanks which come into contact with the grape
must or fruit juice that is to be fermented. It is also used to eliminate or
inhibit the development of undesirable bacteria and yeast present in
crushed fruit or fruit juice that is t o be fermented and to preserve the
wine during storage and aging against bacterial spoilage. Bioletti and
Cruess (1912) reported that very small amounts of SO2 (50 to 75 p.p.m.)
are sufficient to prevent growth of mold, undesirable yeast, and acetic
acid bacteria in musts and to insure pure fermentation with a starter
SULFUR DIOXIDE THEATMENT OF FRUIT AND VEGETABLE PRODUCTS 137
of wine yeast. One hundred p.p.m. of SO2was reported by them to elimi-

nate over 99.9% of the undesirable aerobic yeasts, Penicillium and Asper-
gillus spores, and acetic acid bacteria from musts. Wines made with sulfur
dioxide were found to be lower in volatile acidity and of better storage
quality. This observation of Cruess (191 1) confirming the European
practice of sulfuring musts particularly for making dry wines in hot
countries (Bioletti, 1905; Dupont and Ventre, 1906; Martinaud, 1908)
was confirmed by Cruess (1935a, b). The use of sulfur dioxide in wine
making was established by extensive investigations in Switzerland
(Muller-Thurgau and Osterwalder, 1914) ; in France (Ribereau-Gayon,
1947; Mathieu, 1913; Bailly, 1924); in Italy (Casale, 1938; Mensio, 1917,
1930); and elsewhere (Amerine and Joslyn, 1951).
The utility of sulfur dioxide in controlling growth and activity of
microorganisms is based on the greater sensitivity of these organisms to
SO2 and the peculiar power of adaptability possessed by yeasts. Generally
yeasts are seldom destroyed by chemical preservatives; they are usually
only temporarily inactivated or dormant under unfavorable conditions,
awaiting a chance to become active again. Some strains of yeast can
easily be “trained” to develop in the presence of SO2, especially if the
amounts used are slowly increased. This important fact is used success-
fully in the wine industry when it is desired to ferment the must by a
specially selected strain of yeast, and to avoid undesirable fermentation
by other yeasts.
The selected strain of yeast is “trained” to ferment in a starter con-
taining a very small amount of S O z ;when the fermentation of the starter
is in full progress, it is mixed with further quantities of must containing
progressively increasing amounts of SO2. The yeasts in the starter thus
become SO2-resistant and can be employed for the fermentation of the
bulk of the must, to which a fair amount of SO2 has been added, and in
which no microorganisms or undesirable yeast, other than the selected
strain which has become SO2-resistant, will develop.
Bioletti and Cruess (1912) did not believe that culture yeasts were
acclimatized to SO2; on the contrary, they favored postponing their addi-
tion to a sulfited must until after the full effect of the maximum amount
of free SO2 was exerted on the undesirable organisms and when the yeast
starter was exposed only to the minimum amount of free SOZ.
Sulfur dioxide is also used in wine making for the most important
operation known as the “defecation” of must. It is customary when
making white wines to allow the must to clear by settling before fer-
mentation, and then t o draw off the supernatant clear must into the
fermentation vats. The use of sulfur dioxide in this process, which may
last from two t,o three days, is beneficial in several ways:
1. It delays the free fermentation of the must until after it has cleared,
2. it assists in the hydrolysis of the pectins, which are largely respon-
sible for maintaining the colloids and grape particles in suspension; and
3. it causes the removal of a greater quantity of living cells of wild
yeast, molds, and other microorganisms, which are dragged down with
the grape pulp and other particles held in suspension, in the process of
defecation.
Bettoli in 1911 (see Bioletti and Cruess, 1912) found that the original
must, before defecation, contained over 0.5 % of suspended matter-a
quantity large enough to cause injurious effect on the odor, taste, and
fragrance of the wine. He also demonstrated that, by defecation with
100 p.p.m. of SO2, the number of active cells in a must can be reduced
from over a million per milliliter to just a few hundred per milliliter at
22" C. (72" F.).
The observations made previously on the relative preservative action
of the various forms of sulfur dioxide, free and bound, apply equally well
to the use of SO2 in wine making. As shown by Neuberg (1929), bound
SO2has no toxic effect upon yeasts, and, therefore, it is only the remain-
ing free part of the SO2 which should be taken into account.
In the case of musts and wines, however, additional compounds
besides sugars, particularly acetaldehyde, help to bind an appreciable
part of the SO2.For a long time, ever since SO2came to be used generally
in wine technology, the problem of bound SO2 occupied the minds of
research workers. Rocques in 1897 maintained that SO2in musts and wine
was bound by glucose, but Ripper (1892) and Schmidt (see Kerp, 1904)
thought that sulfur dioxide in wines existed in the greater part as a com-
bination with aldehydes.
Bianconi and Bianchi (1932) tested the SO2 content before, during,
and after the fermentation of grape musts, and found that during fer-
mentation a great part of SO2 is set free (the current of C 0 2 evolved
during the process assisting this liberation) , while the remaining SO2
remained combined with aldehydes. This part, in the opinion of the
authors, remains quasiconstant in the must-wine substrate during the
fermentation.
Farnsteiner (1904), who examined the application of SO2 to agri-
cultural products, came to the conclusion that constituents of the vege-
table tissues, other than the sugars, can also interact with sulfurous acid.
He thought that such complexes in wines are created not only with com-
pounds containing an aldehydic group, but also with cellulose, proteins,
organic acids, and tannin. Paris (1920), on the other hand, showed that
no combinations are possible between tartaric or malic acids and SO2.
Bianconi and Bianchi (1932) refuted the idea that tannin could form any
possible combination with SO2. Paris, however, in work on white cherries,

asserted that after cherry juice had fermented until the sugar content
had been removed, it still contained 0.019 to 0.016% of combined SOz,
or 25 to 27% of the total, and attributed this combination to pectic
substances.
Recent investigations with pure substances have showed that, of the
sugars contained in must, only glucose combines with SO2, and that the
existing aldehydes, mainly acetaldehyde, readily combine with SO2 into
even more stable compounds than the glucose sulfonate. During alco-
holic fermentation, however, acetaldehyde is formed in quantity, and
this compound combines immediately with SO2. When SO2 is added
gradually to a must in active fermentation, it is converted almost in-
stantly into an acetaldehyde-bisulfite complex and the fermentation is not
hindered. Bertin (1924) described this phenomenon as the auto-desulfiting
of musts.
In the course of his studies on wine, Paris observed that when in-
creased quantities of sulfur dioxide are added to the grape musts, the
percentage of free SO2 also increases. I n musts containing 18% sugar, he
reported the following levels 5 minutes after the addition of the bisulfite:
SO2 added, Free SOZ, Combined
g.11. % so2
1.6 61 39
1.7 70 30
2.0 73 27
2.5 74 26
11.5 88 12
Paris does not mention what sugars constitute the 18%. However, if
one supposes about 7.5% of the above t o be glucose, Paris’s observations
agree with those of other workers on different fruit juices and on pure
sugar solutions.
Bianconi and Bianchi (1932), on the other hand, found that 32 hours
after adding 0.24 g./l. SO2 to an 18% grape must, 50.30% of the SO2
was bound, and this remained the same after 96 hours. These authors
asserted that 36.2% of the SO2 was bound with glucose and the rest with
fructose. These results must definitely be regarded with reserve, for fruc-
tose has been shown to be incapable of combining with SO2a t all.
Moreau and Vinet (1937a, b) reported that the true antiseptic power
of sulfurous acid is dependent on the proportion of free undissociated acid
present and showed this to vary with acidity. During the period of 1928
to 1938 they studied the conditions in wine and must that determine the
equilibrium between free and combined sulfur dioxide and developed an
iodine index for determining this (Vineau and Moreau, 1937; Amerine
and Joslyn, 1951; Ribereau-Gayon, 1947). Oversulfiting of wine as a

result of absorption of SO2 from sulfured casks and vats was observed by
Wanner (1938). The sulfur dioxide fixing power of wine was investigated
most recently by Procopio (1953).
In controlling fermentation, the usual practice is to add all the SOz
required at one time. In special fermentations, however, such as that of
sweet table wines like Sauternes, several small additions of SOz are used
to check the fermentation while fermentable sugar still remains. Quinn
(1940) has recommended frequent small additions of sulfur dioxide to
wine rather than an occasional large addition to take advantage of the
germicidal power of free sulfur dioxide.
'
Peynaud and Lafourcade (1952) recommend against the use of high
amounts of SO2, either in settling the must or addition prior to or during
fermentation, because under these conditions acetaldehyde accumulates
and a larger percentage of the sulfur dioxide added is fixed. When possible
sulfur dioxide should not be added in the presence of yeast. I n adding
sulfur dioxide to new wines they recommend a large initial addition to
cause precipitation of as much of the yeasts as possible rather than
several smaller additions.
In the bulk storage of wine following fermentation, it is necessary to
maintain the sulfur dioxide level a t 50 to 75 p.p.m. t o prevent bacterial
spoilage. Here too it is the free sulfur dioxide rather than the total that
determines keeping quality, but unfortunately data on the changes in
equilibrium states are too limited to allow for adequate control of the
level of free SO2 present.
In addition to grape wines, sulfur dioxide is used in fruit wine making.
Its use in preserved cider was reported by Lindet (1922) and Durham
(1909). The use of sulfur dioxide in preparing and preserving light sweet
wines was investigated by Mills and Wiegand (1942) and Yang and
Wiegand (1949). Yang and Wiegand (1951) described a method of main-
taining free sulfur dioxide content in wine. Yang (1953) cautioned against
excessive use of sulfur dioxide in fruit wines and recommended only
100 p.p.m.
In spite of the extensive investigations on wine making, there are
surprisingly few complete data on the changes in free and bound sulfur
dioxide during fermentation and storage. The most extensive data pub-
lished previously were those of Bioletti and Cruess (1912), but even these
are insufficient to determine changes due to volatilization and oxidation
during fermentation and storage. Complete analyses of the distribution
of SO2 in wine8 and of changes in the sulfur compounds during storage
are not available. Mills and Wiegand (1942) reported data on total SO2
before and after storage for six months in the bottle of 210 samples of
different wines. The sulfur dioxide lost during storage in full sealed con-
tainers in wines containing initially 116 to 398 p.p.m. of SO2 varied from
37 to 176 p.p.m., or from 14 to 53% of the initial SO2 content. Alcohol
content (11.8 to 20.0%) had no influence on SOz loss, but this loss was
affected by the initial concentration and by the sugar content.
The utility of sulfur dioxide in cider vinegar manufacture was pointed
out early by Cruess et al. (1915), and it is now widely used to reduce losses
by incomplete alcoholic fermentation. In modern vinegar generator prac-
tices considerable quantities of SO2 are tolerated in the vinegar stock and
the sulfiting of vinegar stock to 50 p.p.m. or over has been found useful
in obtaining better color retention in red wine vinegar.
IX. SULFUR IN DEHYDRATED
DIOXIDE FRUITAND VEGETABLE
AND DRIED
PRODUCTS
In the preparation of fruits and vegetables for drying or dehydration,
sulfur dioxide is added for the preservation of color and flavor during
processing and subsequent storage. Sulfuring or sulfiting is used to pre-
vent enzyme-catalyzed oxidative changes during preparation and also to
prevent microbial deterioration and facilitate drying by plasmolyzing the
cells. In the dehydration and drying of fruits, however, it is used pri-
marily to inhibit the lionenzymatic browning reaction occurring during
storage a t room temperatures and above. I n the dehydration of vege-
tables it is used to preserve both color and flavor.
Dried and Dehydrated Fruit: To maintain the desired qualities in
dried fruit it has been found necessary to incorporate in it an excess of
sulfur dioxide to allow for losses occurring during drying, processing, and
storage. The actual sulfur dioxide content needed will vary with the type
of fruit, the final moisture content, and storage conditions. Long et al.
(1940) recommend the following as a guide, in parts per million, of SO2
at the drying yard : apricots-2000; peaches and nectarines-2000;
pears--1000 ; golden bleach raisins-800; sulfur bleach raisins-1500; and
apples--800. Extensive data have been obtained on the various factors
involved in SO2 absorption and retention and on the factors that deter-
mine the role of SO2in inhibiting browning. The earlier investigations in
California are summarized by Nichols and Christie (1930), Nichols and
Cruess (1932), Roleson and Nichols (1933), Nichols ef al. (1938), and
Long et al. (1940). The earlier investigations in Australia are discussed
by Jewel1 (1927, 1937) and Quinn (1926); in South Africa, by Anderssen
(1929).
The general principles of sulfuring cut and whole fruit are discussed by
Nichols et al. (1925), Cameron et al. (1929), Chaw ct al. (1941), von
h e s e c k e ( 19-23), Perry ct nl. (1 9 M ) , Morris (1 947), C'riirss (1 948), and
142 M. A. JOSLYN AND J. B . 8. BRAVERMAN
Mrak and Mackinney (1951). Instructions for sulfur house operation

are given by Phaff and Mrak (1948), for sun drying fruits by Mrak and
Phaff (1949), and for dehydrating freestone peaches by Mrak and Perry
(1948).
The relation of total sulfur dioxide content and other factors to color
changes during storage of apricots is discussed by Nichols and Reed
(1931), Nichols et al. (1938), Chace et al. (1930, 1933), and Sorber (1944).
The role of sulfuring in dehydration of cherries and small fruits is dis-
cussed by Wiegand et al. (1945), and the effects of methods of sulfuring,
dehydration, and temperature of storage on ascorbic acid content and
carotene content of dehydrated peaches by Eheart and Sholes (1946), as
well as others mentioned previously.
In previous investigations on the influence of sulfur dioxide and other
factors on the storage deterioration of dried apricots, the quality of the
fruit was poorly described. Stadtman et al. (1946) developed a visual
index of darkening based on comparison of a 50% alcohol extract of the
fruit with the color of a standard solution (containing cobaltous sulfate,
potassium dichromate, and cupric sulfate) representing the color of the
fruit at the limit of edibility. Using this darkening index they investigated
the effect of moisture content, sulfur dioxide content, and storage condi-
tions (temperature, oxygen content, etc.) on the storage deterioration of
dried apricots. The results obtained have been reviewed already (Stadt-
man, 1948).
Dehydrated Vegetables: The earlier investigation on the methods of
dehydration of vegetables were reviewed by Cruess and Mrak (1940,
1942a, b), Cruess and Mackinney (1943), and Anon. (1944b). The treat-
ment of certain vegetables (particularly cabbage and white potatoes)
with sulfite solutions was used and recommended in England, Australia,
and Canada a t the start of World War I1 but was not applied commer-
cially in the United States until some two years after its use abroad.
Dehydrated cabbage is usually sulfited to 750 t o 1500 p.p.m., and de-
hydrated white potatoes and carrots to not over 500 p.p.m. Sulfiting not
only improves the retention of color, flavor, and carotene and ascorbic
acid content but makes possible the use of higher finishing temperatures
and consequently shorter drying times. With cabbage, finishing tem-
peratures 5 to 8" C. (10 to 15' F.), higher can be employed safely without
scorching (Mackinney et al., 1943; Mackinney and Howard, 1944; Friar
and Van Holten, 1945). Vegetables before dehydration may be sulfured
and blanched by immersion in hot water containing sodium sulfite, as
was first proposed in England and also used in Canada, or by applying
a sulfite solution as a spray on cabbage as it is conveyed through the
steam blancher, as was done commercially in Australia, New Zealand,
and the United States. In the United States sulfite spray was applied to
partly blanched vegetables and in Australia to vegetables either a t the
entrance or exit of the blancher. The sulfite solution used varied in con-
centration and composition depending on the method of application.
Neutral salts or a combination of neutral sulfite with the metabisulfite
have been used.
The high retention of vitamins and palatability of sulfited dehydrated
cabbage in large-scale food service was pointed out by Fenton et al. (1946).
The use of SO2 in the dehydration of eastern potatoes and other vege-
tables was discussed by Green et al. (1946).
A considerable amount of work has been carried out on the changes
in sulfur dioxide content of vegetables during dehydration and storage
and the relation of sulfite content and sulfite disappearance to changes in
color, flavor, and nutritive value during storage. Many of these data,
however, have not been published as yet and appear only in reports for
restricted distribution. The behavior of sulfur dioxide in dehydrated
vegetables was investigated by Mangan and Doak (1949) in New Zealand.
The most extensive investigations available on the effect of sulfuring on
susceptibility t o browning and on sulfite disappearance and browning of
dehydrated sulfited vegetables are those of Legault and his collaborators
at the Western Regional Research Laboratories (Legault et al., 1947,
1949, 1951 ; Hendel el al., 1953). Although extensive investigations have
been conducted in this field (both published and unpublished), the mecha-
nism of the protective effect of sulfite in inhibiting changes in color and
palatability still remains to be elucidated. Ross (1948) has reviewed the
present status of our knowledge of deterioration of processed potatoes
(including dehydrated white potatoes).
x. SULFUR DIOXIDEI N “BRINING” OF CHERRIES .4ND “BARRELLING”
OF FRUIT
For many years cherries were prepared for subsequent dyeing and
marketing in syrup by storage in barrels in a sea-water brine containing
sulfur dioxide, in Italy and France. These were imported by eastern
Maraschino cherry processors. In the 1930’s, as a result of the investiga-
tions of Wiegand and his collaborators in Oregon and of Cruess in Cali-
fornia, successful methods were developed for the production of a bleached
cherry of uniform light color, firm and free from surface checks, cracks,
and blemishes from Pacific Coast grown sweet cherries of Royal Anne or
Napoleon variety. As a result of this the tonnage of cherries barrelled on
the Pacific Coast increased from less than 2000 tons in 1925 to over 27,000
tons in 1946. In addition to the white-fleshed cherries, Bings, Lamberts,
Republicans, and other varieties are barrelled. The present status of the
brined cherry industry was discussed by Wiegand (1946). A standardized

procedure of brining cherries was developed by Bullis and Wiegand
(1931) which involved filling xtandard 50-gallon paraffin-lined fir barrels
with from 240 to 250 pounds of fresh cherries, covering them with dilute
solution of sulfur dioxide (1.5%) and calcium carbonate or calcium
hydroxide (0.9%), and with some agitation allowing the barrels to stand
for four to six weeks or until the cherries are cured. The storage of
cherries in brine was investigated by Cruess and Henriques (1932) and
Cruess (1937). The processing of cherries in British Columbia was investi-
gated by Atkinson and Strachan (1935a, b, 1941a, b), and more recently
Weast (1940) has described improved methods of preparing the solution
to be used in brining cherries from liquid SO2 and lime.
The cracking and splitting of cherries in brine, which at one time was a
serious problem in the industry, was investigated by Cruess and Hen-
riques (1932), Cruess (1933, 1935c), and Wiegand et al. (1939). It was
found by Cruess that this was related to excessively high acidity and
could be prevented by the addition of the proper amount of slaked lime
and even better by the use of calcium sulfite instead of sulfurous acid
and lime. Wiegand et al. (1939) compared the effect of storing cherries in
a 1.18% sulfur dioxide “brine” with and without the addition of various
alkalis (calcium carbonate, calcium hydroxide, sodium hydroxide, and
potassium hydroxide). The original brine had a pH value of 1.56, whereas
that with added alkali ranged initially in pH from 1.69 to 2.2 as the
percentage of SO2 neutralized increased from 12.5 to 100%. After six
weeks of storage the final pH of the cherry brines ranged from 2.6 to 5.6,
depending on the type of alkali used and on the extent to which the SO2
was neutralized. The maximum per cent (87.6) of perfect cherries was
obtained with CaC03 in brines having a pH of 1.96. With Ca(OH)2 a
maximum of 50% perfect cherries was reached a t pH 1.94; with NaOH
a maximum of 55% perfect cherries at pH 2.1; with KOH the per cent of
perfect cherries increased progressively as the amount of KOH added
increased until 77% was achieved a t 100% neutralization of SO2 a t
pH 2.1.
Orange, lemon, grapefruit, and citron peel is improved in quality if it
is stored prior to candying in a salt solution containing added SO2.
Uniformity and translucency of color and texture is markedly improved
by sulfite-brine storage in comparison with fermentation in a dilute brine.
This was found particularly true of citron (Cruess and Glickson, 1932;
Fellers and Smith, 1937).
I n the barrelling of fruit for use in jams and preserves, both small
whole fruit and halved large fruit is preserved by storage in a “brine”
containing sulfurous acid and lime. The barrelling of peaches was success-
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PILODUCTS 145
fully carried out by Mrak and Henrique; (1932) and extended to other
fruit by Mrak et al. (1934) and Weast (1915). The use of sulfur dioxide
in the processing of watermelon rinds for food is described by Woodroof
and Cecil (1942b). They also investigated the factors influencing the
quality of preserves made from sulfited fruit (Woodroof and Cecil, 1943).
The varietal adaptability of strawberries to preservation in sulfur dioxide-
calcium solutions was investigated extensively by Culpepper and Caldwell
(1943). Charley ct al. (1943) reported on a large-scale preservation of
plums by sulfur dioxide.
The general aspects of preserving fruit for subsequent, preserve and
jam making or candying is discussed by Woodroof and Cecil (1942a,
1945). The ease with which the microorganisms normally causing spoilage
(yeasts, molds, and acetic acid bacteria) may be inhibited by sulfite (the
concentration of sulfite as SO2 in the fruit required t o preserve the fruit
varies from 1500 to 2000 p.p.m. to as low as 350 p.p.m.); the close
similarity of pH of fruit tissues to th at of the sulfite solutions used as
preservative, which Woodroof and Cecil believe to be a factor in texture
preservation; the presence of tannins and organic acids which tend to
mask the taste of the small quantities of residual sulfur dioxide remaining
in the finished product; and the ease with which the sulfur dioxide ab-
sorbed by the fruit during sulfiting is lost during storage (by volatilization
or oxidation) and removed by soaking arid cooking, render fruits par-
ticularly suitable for preservation with sulfite solutions. With soft fruits
such as peaches or strawberries a firming agent (calcium hydroxide or
calcium carbonate) must be added, but with firmer fruit such as black-
berries or black currants no firming agent is needed. Woodroof and Cecil
(1945) give directions for preparation of stock solution of preservative
sulfur dioxide solution from the salts of sulfurous acid and its use in
barrelling various fruits.
The British jam manufacturers use considerable amounts of straw-
berries, raspberries, and similar soft fruit preserved in sulfurous acid
solutions without heat treatment and have experienced considerable
difficulty owing to varying degrees of softening and even complete
mushiness of such fruit. Pandhi (1953) critically reviewed the various
speculations and observations as to the cause of this condition. Among the
factors suggested as being involved were: variety, maturity, and growing
conditions of fruit; concentration and type of sulfurous acid solution;
manner of packing the fruit into the barrels and method of adding the
preservative ; character of the water used ; time elapsed between harvest-
ing and preserving; and storage temperature. Pandhi concluded that the
softening of strawberries is probably due to pectic enzymes secreted by
contaminating microorganisms (chiefly molds) and not to the activity
of the naturally occurring pectic enzymes present in the fruit or the action
of sulfurous acid itself.
XI. SULFURDIOXIDEIN TRANSPORTATION AND STORAGE OF GRAPES
AND IN OTHERPRODUCTS
Grapes: California table grapes are widely fumigated with sulfur

dioxide before transportation to Eastern markets. This fumigation
reduces molding in transit and also saves on the number of icings re-
quired en route. The concentration of SO2added, however, must be care-
fully controlled as damage a t the stem end of the berry is likely to occur.
The sensitivity t o damage varies with the maturity and variety of grapes
shipped. The utilization of sulfur dioxide in the marketing of grapes was
investigated by Winkler and Jacob (1925), Jacob (1929), Asbury and
Pentzer (1931), Pentzer et al. (1932), and Rose and Pentzer (1932). Dunn
(1940) reported that mixing 5 g. of potassium or sodium metabisulfite per
case (30 t o 32 pounds) with granulated cork packing reduces wastage of
grapes by mold. Insertion of sodium or potassium bisulfite tablets or
spraying the packing with bisulfite solution was suggested in South
Africa. Pentzer, however, preferred addition of the metabisulfite as a
separate package to adding it to the packing. Van der Plank (1939) de-
veloped a device for regulated release of SO2 in packages of stored table
grapes. Although Winkler and Jacob (1925) reported that 50 p.p.m. of
SO2 would approximately double the keeping quality of grapes and up t o
100 p.p.m. would not injure the color, flavor, and quality, the amount of
SO2 now used is much lower, ranging from 10 p.p.m. for varieties most
susceptible t o injury t o not over 50 p.p.m.
Other Perishables: I n addition to grapes, prepared fresh apples for
baker’s use (Joslyn and Mrak, 1930) and prepeeled potatoes (Olson and
Treadway, 1949) are treated with sulfites or SO2 to prevent discoloration
during preparation, storage, and distribution. Small amounts of SO2 are
added also t o prepared horse-radish to prevent discoloration (Blumenthal,
1937), cut or shredded fresh vegetables for salad use are sulfited in some
packs, and coconuts were a t one time preserved with SO2 (Dybowski,
1908).
Use in Manufacturing: Sulfur dioxide has been used for many years
as a bleaching agent and browning inhibitor in the extraction and refining
of sugar from sugar beets and in some cases sugar cane (Reed, 1918). It
is also used in improving the keeping quality of syrups prepared from
corn starch hydrolyzates. Although the presence of small quantities of
SOz (usually less than 50 p.p.m.) in sugars and syrups used in confec-
tionery and similar products is not objectionable, refined sugars bleached
by SO2,or corn syrups containing it, cannot be used in canning. I n canned
foods internal corrosion is increased and objectionable flavors due to

formation of H,S may occur.
Sulfur dioxide is used in the extraction of pectin from citrus peel
(Wilson, 1925) as a depolymerizing agent and is also used in the prepara-
tion of liquid pectin extracts t o produce a light-colored extract. The SO2
added t o the water extract before filtration is easily removed during con-
centration in oacuo. It is also used in Europe as a preservative for liquid
pectin preparations to which i t is added in amounts ranging from 500 to
2000 p.p.m., usually about 1250 p.p.m. I n the United States liquid pectin
preparations are preserved by heat treatment. Sulfur dioxide has also
been recommended for preservation of apple pomace t o be used in pectin
manufacture, Charley et al. (1942), Burroughs et al. (1953), and Mehlitz
(1941) reported rapid loss in pectin content in wet apple pomace during
storage and recommended the use of sulfur dioxide to prevent or reduce
loss in jelly grade. Dryden et al. (1952) more recently reported data on
effect of freedom from rot, cold storage of apples, and of allowing pomace
to stand before drying on pectin losses. They reported excellent retention
of jelly grade in apple pomace dried in a laboratory drier a t 90” C.
(194” F.).
REFERENCES
Abdulov, A. 1938. Preserving foods with sulfurous acid. AzerbaZdzhan. Med. Zhur.
1938 (2), 146-149; Chem. Zentr. 1939 (pt. 11),2389; 1941. C. A . 36, 4857.
Adams, R., and Garber, J. D. 1949. Amine bisulfites. 11. Their use as resolving agents
for aldehydes and ketones. J . A m . Chem. SOC.71, 522-526.
Adams, R., and Lipscomb, R. D. 1949. Amine bisulfite addition products of aldehydes
and ketones. I . J . A m . Chem. Sor. 71, 519-522.
Amerine, M. A. 1952. “Laboratory procedures for enology,” Syllabus, Univ. of Calif.,
Dept. of Viticulture, Davis, see pp. 33-36.
Amerine, M. A., and Joslyn, M. A. 1951. “Table Wines. The Technology of Their
Production in California.” University of California Press, Berkeley.
Anderssen, F. G. 1929. Sulfur dioxide in dried fruit. Union S . Africa, Dept. Agr. Bull.
NO.84, 3-14.
Anon. 1907. General results of the investigations showing the effect of sulphurous acid
and sulphites upon digestion and health. 77. S. Dept. Agr., Bur. Chem., Circ. No.
37, 1-18.
Anon. 1928. Determination of sulfur dioxide in sugars, golden syrup, glucose, molasses.
Intern. Sugar J. 30, 259-265.
Anon. 1944a. Sulfur dip prevents browning of frozen sliced apples. Food Znds. 16,
(lo), 80-81, 137-138.
Anon. 1944b. Vegetable and fruit dehydration. A manual for plant operators. U . S.
Dept. Agr. Mise. Publ. 640, 1-218; see pp. 42-46.
Anon. 1945. Commercial preparation and freezing preservation of sliced apples. U.S.
Dept. Agr., Bur. Agr. and Ind. Chem., Information Sheet AIC-57.
Archinard, P. 1937. Sur le dosage de l’anhydride sulfureux et des aciditks fixes e t
volatiles dans les vins sulfit6s. Rev. viticult. 87, 257-263.
148 M. A. JOSLYN AND J. B . S. BRAVERMAN
Asbury, C. E., and Pentzer, W. T. 1931. Investigations on the transportation of

California table grapes. Calif. Grower 3 (6), 6.
Assoc. Offic. Agr. Chemists 1950. Official Methods of Analysis. Published by the
A.O.A.C., Washington, D. C., 7th Ed., pp. 145-146, 176, 470-472.
Atkin, S. 1950. Determination of sulfur dioxide in presence of sulfur trioxide. Anal.
Chem. 22, 947-948.
Atkinson, F. E. 1941. Preservation of fruit pulps with sulfur dioxide. Canner 94, (2),
14-15.
Atkinson, F. E., and Strachan, C. C. 1935a. Cherry processing. I. Sulphur dioxide
treatment with special reference to a useful acid generator. Fruit Products J. 14
(5), 136-137.
Atkinson, F. E., and Strachan, C. C. 1935b. Cherry processing. 11. Leaching of
sulphured stock. Fruit Products J . 14 (6), 174-176.
Atkinson, F. E., and Strachan, C. C . 1941a. Preservation of fruits with sulphur dioxide
in British Columbia. Fruit Products J. 21, 5-8, 43-45, 60, 110-112, 141-144, 153.
Atkinson, F. E., and Strachan, C. C. 1941b. Candying of fruit in British Columbia
with special reference to cherries. Fruit Products J. 20 (5-lo), 132-135, 16f3-169,
185, 199-201, 217, 219, 229-232, 262-264, 289, 291, 310-312, 323, 324.
Bailly, M. 1924. The usefulness of sulfurous acid (in wine making). Rev. viticult. 61
145- 147.
Beavens, E. A., and Bourne, J. A. 1945. Commercial sulfiting practices. Food Znds.
17 (9), 1044-1045.
Bennett, A. H., and Donovan, F. K. 1943. Determination of SO$ in citrus juices.
Analyst 88, 140-142.
Bennett, A. H., and Tarbert, D. S. 1933. Vitamin C in citrus juices. Biochem. J. 27,
1294-1301.
Benvegnin, L., and Capt, E. 1931. Du dosage de l’acide sulfureux libre et total dana
les vins rouges. Mitt. Gebiete Lebensm. u. Hyg. 22, 257-263, 365-368.
Berg, V. A. 1946. Binding of sulfurous acid by sugars. Vinodelie i Vinogradarstuo
S.S.S.R. 6 (2), 15-18.
Bertin, C . 1924. Auto-desulfiting of musts. Bull. SOC.ind. Rouen 62, 427-434.
Beuschlein, W. L., and Simenson, L. 0. 1940. Solubility of sulfur dioxide in water.
J. Am. Chem. SOC.62, 610-612.
Beythien, A. 1904. tfber die Verwendung der schwefligen Siiure als Konservierungs-
mittel, insbesondere den jetzigen Stand der Beurteilung geschwefelten Dorrobstes.
2. Untersuch. Nahr.-u. Genussm. 8, 36-53.
Beythien, A., and Bohrisch, P. 1902. uber amerikanisches, getrocknetes Obst. 2.
Untersuch. Nahr.-u. Genussm. 6, 401-409.
Beythien, A., and Bohrisch, P. 1903. uber gesehwefeltes Dijrrobst. 2. Untersuch.
Nahr.-u. Genussm. 6, 356-360.
Bianconi, A., and Bianchi, A. 1932. Sui rapporti di indole chimica fra l’acido sulforoso
ed i composti a funzione aldeidice e chetonica. Ann. chim. appl. 22, 291-316.
Bioletti, F. T. 1905. The manufacture of dry wines in hot countries. Calif. Agr. Ezpt.
Sta. Bull. 167, 1-66.
Bioletti, F. T. 1911. Sulfurous acid in wine making. OGg. Corn. 8th Intern. Congr. Appl.
Chem. 14, 31-59.
Bioletti, F. T., and Cruess, W. V. 1912. Enological investigations. Calif. Agr. Expt.
Sta. Bull. No. 230, 25-64.
Risson, C. S., Atlinger, H. W., and Young, H. A. 1942. Some factors affecting the
burning of sulfurs used in sulfuring fruits. Hilgardia 14 (6), 361-372.
Black, J. W.et al. lY28. Determination of SO2. Analyst 63, 118-129.
Black, J. W., and Warren, B. J. W. 1928. Notes on the effect of other reducing sub-
stances on the determination of sulfur dioxide. Analyst 63, 130-132.
Blumenthal, S. 1937. Relationship between organic sulfur and sulfites in the prepared
horseradish. Fruit Products J . 16, 239-240.
Boggs, M. M., and Ward, A. C. 1950. Scoring techniques for sulfited foods. Food
Technol. 4, 282-284.
Borgstrom, G. 1953. Preservation of fruit juices by sterilization-filtration, chemical
preservatives, etc. I n “Chemistry and Technology of Fruit and Vegetable Juice
Production” (D. K. Tressler and M. A. Joslyn, eds.). Avi Publishing Co., New
York, Chapter 6.
Branch, G. E. K., and Joslyn, M. A. 1935. The kinetics of the auto-oxidation of
catechol in the presence of several foreign substances. J. Am. Chem. SOC.67,
2388-2394.
Braverman, J. B. S. 1949. “Citrus Products. Chemical Composition and Chemical
Technology.” Interscience Publishers, New York, see pp. 249-257.
Braverman, J. B. 8. 1953. The mechanism of the interaction of sulphur dioxide with
certain sugars. J . Sci. Food Agr. 4, 540-547.
BrBmond, E. 1937. L’anhydride sulfureux en oenologie. Rev. agr. ( N . Africa) 37, 518-
523, 533-537, 548-552.
Brooks, R. 0. 1928. A new rapid method for SO2 in dried fruits. Chemist Analyst 17
(11, 3.
Browne, H. H. 1944. Evidence of the existence of bisulfite compounds of sugar. J .
Org. Chem. 9, 477-483.
Buchman, E. R. 1936. Studies of crystalline vitamin BI. XIV. Sulfite cleavage. IV.
The thiazole half. J . Am. Chem. SOC.68, 1803-1805.
Buchman, E. R., Williams, R. R., and Keresztesy, J. C. 1935. Studies of crystalline
vitamin B1. X. Sulfite cleavage. 11. Chemistry of the basic product. J . Am.
Chem. SOC. 67, 1849-1851.
Bullis, D. E., and Wiegand, E. H. 1931. Bleaching and dyeing Royal Ann cherries for
Maraschino or fruit salad use. Oregon Agr. Expt. Sta. Bull. No. 276, 1-30.
Bunte, H. 1873. Ueber Aethylaldehyd-schweflige Salze und die Einwirkung des
schwefligsauren Natrons Aethylidenchlorin. Ann. 170, 305-310.
Burroughs, L. F., Kieser, M. E., Pollard, A., and Steedman, J. 1943. The treatment of
apple pomace prior to drying for subsequent pectin extraction. Ann. Rep!., Agr.
and Hort. Research Sta., Long Ashton, Bristol 1948, 136-140; Fruit Products J .
24, 4-6. 1944.
Busing, K. H., and Raabe, W. 1938. The vitamin C content of grape juice preserved
with sulfur dioxide. Klin. Wochschr. 17, 1766.
Butleetone, N. C. 1928. The determination of sulfur dioxide in beer. Brewers’ J.
(London) 64, 143-147.
Cameron, 5. S., Guinn, G., Savage, C. G., Jewell, W. R., and Lyon, A. V. 1929. The
sulfuring of dried fruits. J . Council Sci. Ind. Research 2, 151-155.
Casale, L. 1938. L’acide sulfureux en vinification. 6th Cmgr. Intern. Vigne et Vin,
Lisbon 2, 86-94.
Casanave, M. P. 1910. Determination of sulfurous anhydride and a new method for
the estimation of the same in free and combined conditions. Ann. fals. et fraudes
3, 154-160.
Chace, E. M., Church, C. G., and Sorber, D. G. 1930. Large-scale experiments in
sulfuring apricots. I. Ind. Eng. Chem. 22, 1317-1320.
Chace, E. M., Church, C. G., and Sorber, D. G. 1933. Large-scale experiments in

sulfuring apricots. 11. Effect of dehydrating, shade drying and blanching. Znd.
Eng. Chem. 26, 1366-1370.
Chace, E. M., Noel, W. A., and Pease, V. A. 1941. Preservation of fruits and vege-
tables by commercial dehydration. U.S. Dept. Agr. Circ. No. 619, 1-40.
Charley, V. L. S. 1934. The preservation of fruit with 902. Effect of hot and cold
methods on residual SO2 content of jam. Ann. Rept. Long Ashton Research Sta.
1984, 255-260. .
Charley, V. L. S., Burroughs, L. F., Kieser, M. E., and Steedman, J. 1942. The treae
ment of apple pomace prior to drying for subsequent pectin extraction. Ann.
Rept., Agr. and Hort. Research Sta., Long Ashton, Bristol 1942, 89-100.
Charley, V. L. S., Hopkins, D. P., Kieser, M. E., and Sills, V. E. 1943. The large
scale preservation of plums by the sulfur dioxide cold process. Fruit Products J .
22,362-366.
Cruess, W. V. 1911. The effect of sulfurous acid on fermentation organisms. J . Znd.
Eng. Chem. 4, 581-585.
Cruess, W. V. 1932. Hydrogen-ion concentration in preservative action. Znd. Eng.
Chem. 24, 648-649.
Cruess, W. V. 1933. Uee of calcium sulfite for the storage of cherries. Fruit Products J .
12 (8), 230.
Cruess, W. V. 1935a. Notes on producing and keeping wines low in volatile acidity.
Fruit Products J . 16, 76-77, 108-109.
Cruess, W. V. 1935b. Further data on the effect of SO2 in preventing high volatile
acidity in wines. Fruit Products J . 16, 324-327, 345.
Cruess, W. V. 1935c. Splitting of cherries in brine. Fruit Products J . 14, 271-272.
Cruess, W. V. 1937. Maraschino cherries. Fruit Products J . 16 (9), 263-265, 277, 279,
281, 283.
Cruess, W. V. 1948. Commercial Fruit and Vegetable Products. McGraw-Hill Book
Company, New York, 3rd Ed., see pp. 314-315, 472-596, 597-617, 69&761,
762-80 1.
Cruess, W. V., Balog, E. G., Friar, H. F., and Lew, M. 1944a. Effect of sulfiting on
dehydration temperature. Canner 98 (5), 18.
Cruess, W. V., Balog, E. G., Friar, H. F., and Lew, M. 1944b. Sulfiting to improve
vegetables for dehydration. Food Packer 26 ( l ) , 31, 62.
Cruess, W. V., and Berg, E. W. 1925. Removal of sulfurous acid from grape sirup.
Znd. Eng. Chem. 17, 849-850.
Cruess, W. V., and Glickson, D. 1932. Observations on brining and candying of citron.
Fruit Products J . 12, 17-18, 25.
Cruess, W. V., and Henriques, V. de F. 1932. Experiments on the storage of cherries
in “brine.” Fruit Products J . 11, 231-234.
Cruess, W. V., and Irish, J. H. 1932. Further observations on the relation of pH to
toxicity of preservatives to micro-organisms. J . Bacteriol. 2S, 163-166.
Cruess, W. V., and Mackinney, G. 1943. The dehydration of vegetables. Calif. Agr.
E q t . Sta. BUZZ. NO. 680, 1-76.
Cruess, W. V., and Mrak, E. M. 1940. The dehydration of vegetables. Fruit Products
J . 20, 100-103.
Cruess, W. V., and Mrak, E. M. 1942a. The dehydration of vegetables. Fruit Products
J . 21, 201-204, 241-242, 269-272, 302-307, 337-340.
Cruess, W. V., and Mrak, E. M. 1942b. What’s known today about dehydrating vege-
tables. Food Znds. 14 ( l ) , 57-60; (2), 41-43, 96-97.
Cruess, W. V., and Nouty, A. H. 1927. Studies on the preservation of fruits in sulfurous
acid solutions. Fruit Products J . 6 ( l l ) , 18-19.
Cruess, W. V., Richert, P. H., and Irish, J. H. 1931. The effect of pH on the toxicity
of several preservatives to microorganisms. Hilgardia 6 (lo), 295-314.
Cruess, W. V., Zion, J. R., and Sifredi, A. V. 1915. The utility of sulfurous acid and
pure yeast in cider vinegar manufacture. J . Znd. Eng. Chem. 7 , 324-325.
Culpepper, C. W., and Caldwell, J. S. 1943. Varietal adaptability of strawberries to
preservation in sulfur dioxide-calcium solution. Fruit Products J . 23 (1 and 2),
5-9, 27, 29, 46-51.
Danehy, J. P., and Pigman, W. W. 1951. Reactions between sugars and nitrogenous
compounds and their relationship to certain food problems. Advances i n Food
Davis, M. B., and staff, Division of Horticulture, Dominion Dept. Agr. (Canada).
1943. Dehydration of Vegetables in Canada. No. 3, 43 p., and No. 5, 27 p., of a
series of booklets reprinting articles which have appeared in “Food in Canada.”
Consolidated Press Ltd.
Dept. of Scientific and Industrial Research and Ministry of Food. 1943. Dehydration.
United Kingdom Progress Reports, printed by H. M. Stationery Office, London
(Restricted Circulation). Sections 1-11. (See particularly Section VI. Vegetables;
Section IX. Bacteriological Control.)
Division of Food Preservation, C.S.I.R. (Australia) 1942. The Dehydration of Vege-
tables. Mimeo. 42 p.
Downer, A. W. E. 1942. The stability of ascorbic acid in citrus fruit juice products.
J. SOC.Chem. Znd. 61, 80-82.
Downer, A. W. E. 1943. The preservation of citrus juices with sulfurous acid. J. SOC.
Chem. Ind. 62, 124-127.
Dryden, E. C., Willaman, J. J., and Hills, C. H. 1952. Some factors affecting pectin
grade of apple pomace. Food Technol. 6, 474-478.
Dunn, T. C. 1940. Sulfite preservation of table grapes. J . Dept. Agr. W . Australia
NO.17, 439-443.
Dupaigne, P. 1951. Micro-determination of sulfur dioxide in grape juice. Ann. fals.
el fraudes 44, 111-121.
Dupont, E., and Ventre, J. 1906. L’acide sulfureux en vinification. Progr. agr. et vit.
46, 157-161.
Durham, H. E. 1909. Studies on ciders and Perry. I. Sulfite preservation. J . Hyg. 9,
17-20.
Dybowski, J. 1908. The preservation of coconuts. Compt. rend. 147, 756-757.
Dyer, B., Taylor, G., and Hamence, J. H. 1941. The occurrence of “apparent” sulphur
dioxide in dried onions. Analyst 66, 9-12.
Eckman, J. R. 1927. Determination of sulphur trioxide in the presence of sulphur
dioxide, together with some analysis of commercial liquid sulphur dioxide. Natl.
Bur. Standards ( U . S . ) Sci. Paper 664, 277-285.
Eheart, M. S., and Sholes, M. L. 1946. Effects of methods of sulfuring, dehydration,
and temperature of storage on ascorbic acid and carotene content of dehydrated
peaches. Food Research 11, 332-340.
Fabre, J.-Henri. 1936. “ Analyse de Vins. et interprbtation des resultats analytiques.
ProcedEs modernes de vinificetion,” Vol. 2, 2nd ed. La Typo-Litho and Jules
Carbonel, Alger, see pp. 157-168.
Fabre, P. 1947. Nouvelle technique de deaulfitation. Intern. Congr. Grapes, Grape
Juice, Wine, Istambul, Oct. 2-7.
Farnsteiner, K. Z. 1904. uber organisch gebuedene schweflige siiure in ”shrung-

smitteln. 2. Untersuch. Lebensm. 7, 449.
Feigenbaum, J., and Israelshvili, S. 1949. Sulfurous acid in citrus juices. Ind. Eng.
Chem. 41, 797-799.
Fellers, C. R., and Smith, E. G. 1937.Chemical composition and fermentation studies
on citron. J. Agr. Research SS, 859-867.
Fenton, F., Gleim, E., Amason, A., Thompson, J. F., Albury, M., and Phillips, M.
1946.Retention of vitamins and palatability in large-scale food service of sulfited,
dehydrated cabbage. Food Research 11, 475-484.
F e d , L. 1915.Determination of sulfurous acid in wine. Ann. chim. anal. 20, 151-155.
Fischler, M.,and Kretzdom, H. 1938.Determination of sulfurous acid in sweet musts.
2. Untersuch. Lebensm. 76, 38-39; C . A . S2, 3502.
Fisher, C. D.,Mrak, E. M., and Long, J. D. 1942. Effect of time and temperature of
sulfuring on absorption and retention of sulfur dioxide in fruits. Fruit Products J.
21, 175-176, 199-200, 237-238.
Fitelson, J. 1929. Comparison of Monier-Williams and the A.O.A.C. methods for
determination of sulfurous acid in food products. J. Assoc. Oflc. Agr. Chemists 12,
120-129.
Francis, W., Morley, M. C., and Gillham, E. W. F. 1938. Automatic estimation of
sulfur dioxide in industrial gases. J. SOC.Chem. Ind. 67, 419-424; C. A . 33, 1626.
Friar, H. F.,and Van Holten, P. 1945. Effect of sulfiting on maximum drying tem-
perature of vegetables. Fruit Products J . 24, 337-339.
Fuller, E. C., and Crist, R. H. 1941. The rate of oxidation of sulfite ions by oxygen.
J. Am. Chem. SOC.68, 1644-1650.
Gehman, H., and Osman, E. M. 1954. The chemistry of the sugar-sulfite reaction and
its relationship to food problems. Advances in Food Research 6, 53-96.
Gillespy, T. G. 1916. Studies on the mould Bissochlamys fulva. Ann. Rept. Fruit
Vegetable Preserv. Research Sta. Campden Ilniv. Bristol, pp. 31-39.
Gillett, T. R. 1953.Color and colored nonsugars. In “Principles of Sugar Technology”
(P. Honig, ed.). Elsevier Publishing Co., New York, pp. 214-290.
Grant, W. M. 1947. Colorimetric determination of sulfur dioxide. Anal. Chem. 10,
345-346.
Green, E. L., Culpepper, C. W., Caldwell, J. S., and Hutchins, M. C. 1946.The use of
SOa in dehydration of eastern potatoes and other vegetables. Fruit Products J . 26,
16-20, 26, 39-44, 81-86.
Guadagni, D. G. 1949. Sirup treatment of apple slices for freezing preservation. Food
Technol. 8, 404-408.
Hagglund, E.,Johnson, T., and Silander, S. 1929. Influence of sugars on the stability
of bisulfite solutions. Ber. 62B,84-90.
Hagglund, E.,Johnson, T., and Urban, H. 1930.Influence of sulfite and bisulfite solu-
tions on sugars a t high temperatures. 111. Ber. 63B, 1387-1395.
Hamann, V. 1951.Die Verwendung chemischer Konservierungsmittel in Lebensmittel-
verkehr. 2.Lebensm-Untersuch. u.-Forschung 05, 93-101.
Hamburger, J. J., and Joslyn, M. A. 1941. Auto-oxidation of filtered citrus juices.
Food Research 6 , 599-619.
H a n d , J., and VoHgek, J. 1937. The determination of sulfur dioxide as a preserving
agent in dried fruit. Chem. Listy S1, 408-413; C . A . 32, 3501.
Hcndel, C. E., Legault, R. R., and Talburt, W. F. 1953.Effect of processing variables
on leaching loss, carotene content and susceptibility to browning in dehydrated
carrots. Food Technol. 7 , 160-162.
Hennig, K. 1946. “Chemische Untersuchungsmethoden fiir Weinbereiter und Slies-
mosthersteller.” Verlag Eugen Ulmer, Stuttgart, 2nd ed., see pp. 39-42.
Hodge, J. E. 1953. Dehydrated foods. Chemistry of browning reactions in model
systems. Agr. Food Chem. 1, 928-943.
Hoffpauir, C. L., and O’Connor, R. T. 1949. Sensitization of fuchsin-sulfurous reagent
for formaldehyde by addition of kctoncs. Anal. Chem. 21, 420.
Honig, P. (ed.) 1953. “Principles of Sugar Technology.” Elsevier Publishing Co.,
New York, See pp. 240-241, 582-654.
Ingram, M. 1947a. An electrometric indicator to replace starch in iodine titrations of
sulfurous acid in fruit juices. J. SOC.Chem. Ind. 66, 50-55.
Ingram, M. 1947b. Investigation of errors arising in iodometric determination of free
sulphurous acid by the acetone procedure. J . SOC.Chem. Ind. 66, 105-115.
Ingram, M. 1948. Germicidal effects of free and combined sulphur dioxide. J. Sac.
Chem. Ind. 67, 18-21.
Ingram, M. 1949. Behavior of sulphur dioxide in concentrated orange juice. Food
Research 14, 54-71.
Ingram, M., and Vas, K. 1950a. Combination of sulphur dioxide with concentrated
orange juice. 1. Equilibrium states. J. Sci. Food Agr. 1, 21-27.
Ingram, M., and Vas, K. 1950b. Combination of sulphur dioxide with concentrated
orange juice. 11. The rate of the reaction. J. Sci. Food Agr. 1, 63-66.
Jacob, H. E. 1929. The use of sulfur dioxide in shipping grapes. Caalif. Agr. Expt. Sla.
Bull. NO.471, 1-24.
James, T. H., and Weissberger, A. 1939. Oxidation processes. XIII. The inhibitory
action of sulfite and other compounds in the auto-oxidation of hydroquinone and
its homologs. J. Am. Chem. SOC.61, 442-450.
Jaulmes, P. 1951. “Andyse des vins” 2nd ed. Librairie Poulain, Montpellier, see
pp. 423-438.
Jaulmes, P., and Espezel, R. 1935. The determination of acetaldehyde in wines and
spirits. Ann. fals. et fraudes 28, 325-335.
Jensen, H. R. 1928. Rapid determination of sulfites by alkaline liberation, or extraction
and titration. Analyst 63, 133-135.
Jewell, W. R. 1927. Sulfuring dried fruit. J. Dept. Agr. Victoria 26, 457-462.
Jewell, W. R. 1937. “Moist Pack” processing of dried apricots. J. Dept. Agr. Victoria
36, 498-500.
Johnson, G., and Johnson, D. K. 1952. Natural flavor retained in new frozen uncooked
apple pulp. Food Technol. 6, 242-245.
Joslyn, M. A. 1936. Unpublished data.
.Joslyn, M.A. 1941. Color retention in fruit products. Znd. Eng. Chem. S3, 308-314.
Joslyn, M. A. 1953. Unpublished observations.
Joslyn, M. A., Braverman, J. B. S., and Smit, C. J. B. 1950. Unpublished observations.
.Joslyn, M. A., and Comar, C. L. 1938. Determination of acetaldehyde in wines.
Ind. Eng. Chem., Anal. Ed. 10, 364-366.
Joslyn, M. A., and Hohl, L. A. 1948. The commercial freezing of fruit products. Calif.
Agr. Expt. &a. Bull. No. 703, 1-108.
.Joslyn, M. A., and Mackinney, G. 1938. The rate of conversion of chlorophyll to
pheophytin. J. Am. Chem. SOC.60, 1132-1136.
Joslyn, M. A,, and Mrak, E. M. 1930. Prepared fresh apples for baker’s use. Fruit
Products J . 9, 309-311.
Joslyn, M. A., and Mrak, E. M. 1933. Investigations on the use of sulfurous acid and
sulfites in the preparation of fresh and frozen fruit for baker’s use. Fruit Products
J . 12, 135-140.
Joslyn, M. A., and Ponting, J. D. 1951. Enzyme-catalyzed oxidative browning of fruit
products. Advances i n Food Research 8, 1-44.
154 M. A. JOSLYN AND J. B. B . BRAVERMAN
Katz, M. 1950. Photoelectric determination of atmospheric sulfur dioxide. Employing

dilute starch-iodine solutions. Anal. Chem. 22, 1040-1047.
Kerp, W. 1903. Uber organisch gebundene schweflige Saure in Nahrungsmitteln.
2. Untersuch. Lebensm. 0, 66-68.
Kerp, W. 1904. “Die schweflige Saure und ihre Verbindungen mit Aldehyden und
Ketonen. Chemische und pharmakologische Untersuchungen.” (Sonderabdruck
aus den Arbeiten des Kaiserlichen Gesundheitsamtes, Band XXI, Heft 2). Julius
Springer, Berlin, Part 1, 236 p.
Kerp, W. 1913. “Die schweflige Saure und ihre Verbindungen mit Aldehyden und
Ketonen.” (Sonderabdruck aus den Arbeiten des Kaiserlichen Gesundheitsamtes).
Julius Springer, Berlin, Part 2, 369 p.
Kolthoff, I. M., and Menzel, H. (Translated by N. H. Furman.) 1928. “Volumetric
Analysis.” “The Theoretical Principles of Volumetric Analysis,” Vol. I, pp. 177-
188, 235-238. John Wiley & Sons, New York.
Kolthoff, I. M., and Menzel, H. (Translated by N. H. Furman.) 1929. “Volumetric
Analysis.” “Practical Volumetric Analysis,” Vol. 11, pp. 399-400, 45e452. John
Wiley & Sons, New York.
La Rosa, W. V. 1950. Unpublished observations.
Legault, R. R., Hendel, C. E., Talburt, W. F., and Pool, M. F. 1951. Browning of
dehydrated sulfited vegetables during storage. Food Technol. 6 , 417-423.
Legault, R. R., Hendel, C. E., Talburt, W. F., and Rasmussen, L. B. 1949. Sulfite
disappearance in dehydrated vegetables during storage. Znd. Eng. Chem. 41,
1447-1451.
Legault, R. R., Talburt, W. F., Mylne, A. M., and Bryan, L. A. 1947. Browning of
dehydrated vegetables during storage. Ind. Eng. Chem. 39, 1294-1299.
Lewis, V. M., Esselen, W. B., Jr., and Fellers, C. R. 1949a. Non-enzymatic browning
of foods. Production of carbon dioxide. Ind. Eng. Chem. 41, 2587-2591.
Lewis, V. M., Esselen, W. B., Jr., and Fellers, C. R. 1949b. Nitrogen-free carboxylic
acids in the browning reaction. Ind. Ens. Chem. 41, 2591-2594.
Lindet, L. 1922. Sulfurous acid in preserved cider. Compt. rend. acad. agr. France 8,325.
Lochhead, A. G., and Farrell, L. 1936. Effect of preservatives on fermentation and
viability of sugar-tolerant yeasts. Food Research 1, 517-524.
Long, J. D., Mrak, E. M., and Fisher, C. D. 1940. Investigations on the sulfuring of
cut fruits for drying. Calif. Agr. Expt. Sta. Bull. No. 030, 1-56.
Little, R. C. 1953. Sulphur in soils. I. Determination of readily soluble sulfates of Soils.
J. Isci. Food Agr. 4, 336-345.
Liithi, H. 1950. Address in PTOC.2nd Intern. Congr. Fruit Juice Producers, Zurich,
Switzerland, p. 93-96.
Lukton, A., and Joslyn, M. A. 1954. Unpublished observations.
MacArthur, M. 1945. Freezing apples for bakers-further investigations. Can. Food
Packer 10 (8), 13-14.
McGinnis, R. A. (ed.) 1951. “ Beet-Sugar Technology.” Reinhold Publishing Corp.,
New York, see pp. 196-197, 245-247.
Mackinney, G. 1937. Chlorophyll in sultanina grapes and raisins. Plant Physiol. 12,
1001-1004.
Mackinney, G . 1945. The sulfiting of vegetables. Fruit Products J. 24, 300-301.
Mackinney, G., Friar, H. F., and Balog, E. 1943. Sulfured dehydrated vegetables.
Fruit Products J . 22, 294, 315.
Mackinney, G., and Howard, L. B. 1944. Sulfite retards deterioration of dehydrated
cabbage shreds. Food Ind8. 10, 355-356.
Mackinney, G., and Joslyn, M. A. 1941. Chlorophyll-pheophytin: Temperature coeffi-

cient of the rate of pheophytin formation. J . Am. Chem. SOC.63,2530-2531.
Mangan, J. L., and Doak, B. W. 1949. Behavior of sulfur dioxide in dehydrated vege-
tables. New Zealand J. Sci. Technol. BSO (4), 232-236.
Marches, J. 1953. Clarification of cane juices by means of the sulphitation process.
Zn “Principles of Sugar Technology” (P. Honig, ed.). Elsevier Publishing CO.,
New York, pp. 582-654.
Martinaud, P. 1908. Alcoholic fermentation in the presence of sulfurous acid. Compt.
rend. 149,465-469.
Mason, H. M. 1928. Note on the oxidation of sulfites by air. Analyst 63, 142-144.
Mason, H. M., and Walsh, G. 1928. Note on the titration of dilute sulfite solutions
with standard iodine solutions. Analyst 63, 144-149.
Mason, H. S. 1953. The structure of melanins. Zn Pigment Cell Growth, Proceedings
of the Third Conference on the Biology of Normal and Atypical Pigment Cell
Growth. Academic Press, New York, pp. 277-303.
Mathers, A. P. 1949. Rapid determination of sulfur dioxide in wine. J . Assoc. Ofic.
Agr. Chemists 32, 745-751.
Mathieu, L. 1910. The determination of sulfurous acid. Ann. jaEs. et fraudes 3,410-415.
Mathieu, L. 1913. Liquide sulfure dioxide in fermentation. Rev. viticulte 41, 421-428.
Maxwell, F. 1916. “Sulphitation in White Sugar Manufacture.” N. Roger, London.
May, P. 1927. The determination of sulphur dioxide in dried fruit. Analyst 62 (614),
271-273.
Mehlitz, A. 1941. tfber Herstellung und Lagerung von Obstpulpe und Obstmark unter
besonderer Beruksichtigung der Vitamin-und Pektinerhaltung. VorratspJlege
u. Lebensmitteljorsch. 4, 572-623.
Mensio, C. 1917. Wine making experiments with sulfur dioxide. Staz. sper. agrar. Ztal.
60, 300-314; J . SOC.Chem. Znd. 37, 102-103A.
Mensio, C. 1930. Azione della anidride salforosa sulla materia colorante dei vini.
Giorn. Vinicolo Ital. 9, 3-13.
Miller, J. 1927. Determination of sulfur dioxide in dried fruits. Analyst 62, 338.
Mills, D. R., and Wiegand, E. H. 1942. Effect of storage on sulfur dioxide in wine.
Fruit Products J . 22, 5-9.
Mitchell, J. H., Jr., and Peterson, M. S. (editors) 1952. Contributions of browning
research to ration item stability. Research and Development Associates Food and
Container Institute, 1-56.
Mitchell, J. S., Pitman, G. A., and Nichols, P. F. 1933. Inhibiting effect of certain
substances upon oxidation of sulfurous acid. Znd. Eng. Chem., Anal. Ed. 6, 415-
416.
Monier-Williams, G. W. 1927. Determination of sulfur dioxide in foods. Repts. Public
Health and Med. Subjects No. 43, 1-56, Great Britain Ministry of Health, London.
Moreau, L., and Vinet, E. 1937a. Determination of the real antiseptic power of
sulfurous acid in musts and wines by the method of the iodine index. Compt. rend.
acad. agr. France 23, 570-576.
Moreau, L., and Vinet, E. 1937b. D6termination du pouvoir antiseptique riel de
I’acide sulfureux dans les moiits et les vins. Ses variations suivant l’acidit6 du
milieu. Rev. viticult. 87, 25-28, 61-67; Compt. rend. acad. agr. France 23, 599-607.
Morgan, A. F., and Field, A. 1929. The effect of drying and sulfur dioxide upon the
antiscorbutic property of fruits. J . Biol. Chem. 82, 579-686.
Morgan, A. F., Field, A., and Nichols, P. F. 1931. Effect of drying and sulphuring on
vitamin content of prunes and apricots. J . Agr. Research 42, 35-45.
Morgan, A. F., Kimmel, L., Field, A., and Nichols, P. F. 1935. The vitamin content of
sultanina (Thompson seedless) grapes and raisins. J . Nutrition 9, 369-382.
Morgan, A. F., Nobles, H. L., Wiens, A., Marsh, G . L., and Winkler, A. J. 1939. The
B vitamins in California grape juices and wines. Food Research 4, 217-229.
Morris, T. N. 1933. “Principles of Fruit Preservation.” Chapman & Hall, London, see
pp. 213-221.
Morris, T. N. 1947. “The Dehydration of Food.” Chapman & Hall, London, see p.
143 el seq.
Mrak, E. M., Campbell, S. V., and Rayer, H. P. 1934. Barreling fruits in sulfur
dioxide solution for export. Food Znds. 6, 507-509.
Mrak, E. M., Fisher, C. D., and Bornstein, B. 1942. The effect of certain substances
and pre-treatments on the retention of color and sulfur dioxide by cut fruits.
Fruit Products J. 21, 297-299.
Mrak, E. M., and Henriques, V. 1932. Trial shipment of barreled peaches to Germany.
Fruit Products J. 11, 142-144.
Mrak, E. M., and Long, G. D. 1941. Methods and equipment for the sun-drying of
fruits. Calif. Agr. Ezpt. Sta. Circ. No. 360.
Mrak, E. M., and Mackinney, G. 1951. The dehydration of foods. Zn “The Chemistry
and Technology of Food and Food Products” (M. B. Jacobs, ed.). Interscience
Publishers, New York, Vol. 111, pp. 1773-1821.
Mrak, E. M., and Perry, R. L. 1948. Dehydrating freestone peaches. Calif. Agr. Ezpt.
Sta. Circ. No. 381, 1-11.
Mrak, E. M., and Phaff, H. J. 1949. Sun-drying fruits. Calif. Agr. Ezpt. Sta. Circ. No.
392, 1-19.
Miiller-Thurgau, H., and Osterwalder, A. 1914. Influence of sulfurous acid on the
fermentation processes due to yeasts and bacteria in wine, perry and cider.
Landwirtsch. Jahrb. Schweiz 28, 480-528.
Miiller-Thurgau, H., and Osterwalder, A. 1915. Prevention by sulfur dioxide of alco-
holic fermentation in fruit and grape juice. Landwirtsch. Jahrb. Schweiz 1916,
421-432.
Neuberg, C. 1929. tfber das Verhalten des glucose-schwefligsaurenNatriums zu Hefe
und damit zusammenhiingende Fragen. Biochem. 2.212, 477-517.
Nichols, P. F., and Christie, A. W. 1930. Drying cut fruits. Calif. Agr. Ezpt. Sta. Bull.
NO. 486, 1-46.
Nichols, P. F., and Cruess, W. V. 1932. Sulfur dioxide as dried fruit preservative. Znd.
Eng. Chem. 24, 649-650.
Nichols, P. F., Mrak, E. M., and Bethel, R. 1938. Effect of drying and storage condi-
tions on color and SO2 retention of dried apricots. Food Research 4, 67-74.
Nichols, P. F., Powers, R., Gross, C. R., and Noel, W. A. 1925. Commercial dehydra-
tion of fruits and vegetables. U.8. Dept. Agr. Bull. No. 1336.
Nichols, P. F., and Reed, H. M. 1931. What happens in the tropics. Western Canner
and Packer 23, 11-15.
Nichols, P. F., and Reed, H. M. 1932. Distillation methods for the determination of
sulfur dioxide. Znd. Eng. Chem., Anal. Ed. 4, 79-84.
Nichols, P. F., and Reed, H. M. 1933. Flask oxidation in the determination of sul-
furous acid by distillation. Znd. Eng. Chem., Anal. Ed. 6, 398-399.
Nissen, B. H., and Petersen, R. B. 1943. Determination of sulfur dioxide in beer.
A modification of Monier-Williams method. Ind. Eng. Chem., Anal. Ed. 16, 129-
130.
Olson, R. L., and Treadway, R. H. 1949. Pre-peeled potatoes for commercial use.
U . S . Dept. Agr., Bur. Agr. Ind. Chem., Mimeo. Publ. A1C-246, 1-14.
Pandhi, P. N. 1953. The softening and breakdown of vegetables in brine and of fruits
in solutions of sulphur dioxide. British Food Manufacturing Industries Research
Association Sci. and Tech. Survey No. 22, 1-23.
Paris, G. 1920. L’industria della piliegl bianchi. Staz. sper. agrar. Ztal. 63, 187-231.
Pederson, C. S., and Tressler, D. K. 1938. Flash pasteurization of apple juice. Znd.
Eng. Chem. 30,954-959.
Pentrer, W. T., Asbury, C. E., and Hamner, K. C. 1932. Effects of fumigation of
different varieties of Vinifera grapes with sulphur dioxide gas. Proc. A m . SOC.
Hort. Sn‘. 29, 339-344.
Perry, M. C., and Beal, G. D. 1920. The quantities of preservatives necessary to
inhibit and prevent alcoholic fermentation and the growth of molds. J . Znd. Eng.
Chem. 12, 253-255.
Perry, R. L., Mrak, E. M., Phaff, H. J., Marsh, G. L., and Fisher, C. D. 1946. Fruit
dehydration. I. Principles and equipment. Calif. Agr. Expt. Sta. Bull. No. 698,
1-68.
Peynaud, E., and Lafourcade, S. 1952. Sur les conditions d’emploi de l’anhydride
sulfureux dans les vins liquoreux. Bull. Ofice Intern. V i n 26:252, 110-120. (See
also Bull. Inst. Nat. Appl. Orig. Vins el Eaux-de Vie 42, 20-31. 1952.)
Phaff, H. J., and Mrak, E. M. 1948. Sulfur-house operation. Calif. Agr. Expt. Sta.
Circ. 382, 1-10.
Phillips, R. J. 1928. Note on the stability of solutions of potassium metabisulfite.
Analyst 63, 160.
Plummer, A. W. 1950. Thermodynamic data for the system sulfur dioxide-water.
Bibliography and critical analysis. Chem. Eng. Prog. 46, 369-374.
Ponting, J. D. 1944. Catechol test for frozen fruits. Quick Frozen Foods 7 (5), 31, 46.
Ponting, J. D., and Johnson, G. 1945. Determination of sulfur dioxide in fruits. Znd.
Eng. Chem., Anal. Ed. 17, 682-686.
Porchet, B. 1931. Contribution b1’6tude de l’adaptation des levures & I’acide sulfureux.
Ann. Agr. Suisse 32 (2), 135-154.
Potter, E. F., and Hendel, C. E. 1951. Determination of sulfite in dehydrated white
potatoes by direct titration. Food Technol. 6, 473-475.
Prater, A. N., Johnson, C. M., Pool, M. F. and Mackinney, G. 1944. Determination
of sulphur dioxide in dehydrated foods. Znd. Eng. Chem., Anal. Ed. 16, 153-157.
Procopio, M. 1953. Sulfur dioxide fixing power of wine. Riv. viticolt. e enol. (ConegZiano)
6, 9-20.
Quinn, G. 1926. Notes on sulfuring of fruits prior to drying. J . Dept. Agr. S . Australia
30, 500-510.
Quinn, D. G. 1940. Sulfur dioxide, its use in the wine industry. J . Dept. Agr. Victoria
38, 200-204.
Rahn, O., and Conn, J. E. 1944. Effect of increase in acidity on antiseptic efficiency.
Znd. Eng. Chem. 36, 185-187.
Reifer, I., and Mangan, J. L. 1945. A rapid determination of sulfur dioxide in de-
hydrated foods. New Zealand J . Sci. Technol. 27, 57-64.
Reed, I. W. 1918. The use of sulfur dioxide in sugar factones. Sugar 20, 95-96.
RibBreau-Gayon, J . 1917. “Trait6 d’Oenologie.” Librairie Polytechnique Charles
BBranger, Paris.
RibBreau-Gayon, J., and Peynaud, E. 1947. “Analyse et controle des vins.” Librairie
Polytechnique Ch. Bbranger, Paris, see pp. 374-389.
Ricketts, J., and Coutts, M. W. 1951. Hydrogen sulphide in fermentation gas. Parts
I-IV. Am. Brewer 84 (Aug.), 27-30, 71; (Sept.) 27-30, 74-75; (Oct.) 33-36,
100-101.
158 M. A. JOSLYN AND J. B . 5. BBAVERMAN
Ripper, M. 1892. Die schweflige Siiure im Weine und deren Bestimmung. J . Prakt.
Chem. (N. F.) 46, 428; 40, 428.
Roleson, E. P., and Nichols, P. F. 1933. Pointers for sulfur dioxide. Western Canner
and Packer 26 (No. 3), 7.
Rose, D. H., and Pentzer, W. T. 1932. Commercial methods of fumigating grapes with
sulphur dioxide. U . S. Dept. Agr., Bur. Plant Ind., Mimeo., 7 p.
Ross, A. F. 1948. Deterioration of processed potatoes. Advances in Food Research 1,
257-290.
Rothenfusser, S. 1929. Detection and determination of sulfur dioxide. 2. Untersuch.
Lebensm. 68, 98-109.
Sasaki, R. 1928. A new colorimetric method for the determination of sulfurous acid
in the food products. Bull. Agr. Chem. SOC.Japan 4, 38-40.
Shreve, R. N. 1944. “The Chemical Process Industries.” McGraw-Hill Book Co.,
New York, pp. 1-970.
Sorber, D. G. 1944. The relation of the sulfur dioxide and total sulfur content of dried
apricots to color changes during storage. Fruit Products J . 23, 234-237.
Souci, S. W. 1951. Zur Frage der chemischen Konservierung der Lebensmittel. 2.
Lebensm.-Untersuch. u.-Forsch. 93, 37-51, 61-72.
Stadtman, E. R., Barker, H. A., Mrak, E. M., and Mackinney, G. 1946. Storage of
dried fruit. Influence of moisture and sulfur dioxide on deterioration of apricots.
Znd. Eng. Chem. 38, 99-104.
Stadtman, E. R. 1948. Non-enzymatic browning in fruit products. Advances in Food
Steigmann, A. 1950. Acid-bleached fuchsin solution as analytical reagent. Anal.
Chem. 22, 492-493.
Sumuleanu, C., Botezatu, M., and Nicolau, T. 1937. New micromethod for free sulfur
dioxide applicable to all wines. Bull. intern. vin 10, (108), 27-34.
Svershkov, I. V. 1939. New reactions for the detection of sulfurous acid in fruit and
berry juices. Lab. Prakt. (U.S.S.R.) 1939 (2-3), 30; C. A . 33, 4689.
Suter, C. M. 1944. “The Organic Chemistry of Sulfur. Tetracovalent Sulfur Com-
pounds.” John Wiley & Sons, New York, see pp. 126-136, 136-141.
Tanner, F. W. 1944. “The Microbiology of Foods.” Garrard Press, Champaign,
Illinois, 2nd Ed., see pp. 11-12, 633, 634.
Taylor, L. V., Jr., Lueck, H., and Beardsley, C. L. 1937. A study of manufacturing
control methods for estimation of sulfur dioxide in alcoholic beverages. J . Assoc.
O&. Agr. Chemists 20, 610-617.
Thomas, M. D. 1932. Automatic apparatus for the determination of small concentra-
tions of sulfur dioxide in air. 111. Znd. Eng. Chem., Anal. Ed. 4, 253-256.
Thomas, M. D., and Abersold, J. N. 1929. Automatic apparatus for the determination
of small concentrations of sulfur dioxide in air. 11. Znd. Eng. Chem., Anal. Ed.
1, 14-15.
Thomas, M. D., Ivie, J. O., Abersold, J. N., and Hendricks, R. H. 1943. Automatic
apparatus for determination of small concentrations of sulfur dioxide in air.
Application to hydrogen sulfide, mercaptan and other sulfur and chlorine com-
pounds. Ind. Eng. Chem., Anal. Ed. 16, 287-290.
Thomas, M. D., Ivie, J. O., and Fitt, T. C. 1946. Automatic apparatus for determina-
tion of small concentrations of sulfur dioxide in air. Znd. Eng. Chem., Anal. Ed.
18,383-387.
Thompson, J. B., and Toy, E. 1945. Determination of sulfur dioxide. Improved
Monier-Williams method. Znd. Eng. Chem., Anal. Ed. 17, 612-615.
Toennies, G., and Bakay, B. 1953. Photonephelometric microdetermination of sulfate

and organic sulfur. Anal. Chem. 26, 160-165.
Tomoda, Y. 1927. On the production of glycerine by fermentation. IV. Dissociation
of acetaldehyde-bisulfite complex in alkaline solution. J. SOC.Chem. Ind. (Japan)
30, 747-759.
Tomoda, Y. 1928. Uber die Gewinnung von Glycerin durch Garung. VI. Einfluss der
Zuckerkonzentration auf die Ausbeute an Glycerin. J. Sac. Chem. Ind. (Japan)
31, 151-152.
Tomoda, Y., and Taguchi, T. 1930. Effect of sodium bisulfite upon the polarizing
power of sugars and the use of NaHS08 in sugar analysis. J. SOC.Chem. Ind.
(Japan) 33, 434-441.
Tressler, D. K., and Evers, C. F. 1947. “The Freezing Preservation of Foods.” The
Avi Publishing Company, New York, 2nd Ed., see pp. 390-394.
Tressler, D. K., Joslyn, M. A., and Marsh, G. I,. 1939. “Fruit and Vegetable Juices.”
Avi Publishing Co., New York, pp. 35, 190, 218, 238.
Urone, P. F., and Boggs, W. E. 1951. Acid-bleached fuchsin in determination of sulfur
dioxide in the atmosphere. Anal. Chem. 23, 1517-1519.
Van der Plank, J. E. 1939. A device for regulated release of SO2 in packages of stored
table grapes. Union S. Africa. Dept. Agr. Forestry, Rept. Low Temp. Research
Lab. 1939, 76.
Vas, K. 1949. The equilibrium between glucose and sulfurous acid. J. SOC.Chem. Ind.
68, 340-343.
Vas, K., and Ingram, M. 1949. Preservation of fruit juices with less SO2. Food Manuf.
24, 414-416.
Vaughn, R. H., Winter, F. H., and Smith, E. E. 1948. Continuous sanitizing of equip-
ment in the dried fruit industry. Food Technol. 2, 292-296.
Veisman, S., and Svereva, E. 1937. Determination of the total amount of sulfurous
acid in foodstuffs. Konservnaya i Plodoovoschch Prom. 1937 (2), 27-28; Chem.
Zentr 1937 (11), 1909; C. A . 33, 4685.
von Loesecke, H. W. 1943. “Drying and Dehydration of Food.’’ Reinhold Publishing
Corp., New York.
von Schelhorn, M. 1951. Control of micro-organims causing spoilage in fruit and
vegetable products. Advances i n Food Research, 3, 429-482.
von Schelhorn, M. 1953. Efficacy and specificity of chemical food preservatives. Food
Technol. 7, 47-101.
Wallace, T., and Marsh, R. W. (eds.) 1953. “Science and Fruit,” University of Bristol,
England, see pp. 78-98, 282-283.
Wanner, E. 1938. Beitrage zur Frage der Ueberschwefelung von Wein. Wein und
Rebe 20 (9, lo), 267-292.
Warcollier, G., and Le Moal, A. 1929. The determination of sulfurous acid in apple
musts and ciders. Ann. jals. et jraudes 22, (246), 333-340.
Weast, C. A. 1940. Preparation of solution to be used in brining cherries. Vestern
Canner and Packer 32, 26-27.
Weast, C. A. 1945. Barreling of fruit for use in jams and preserves. Canner 100 (25),
34, 36, 38.
Whitney, R. P., Han, S. T., and Davis, J. L. 1953. The mechanism of sulfur dioxide
absorption in aqueous media. Tappi 36, 172-175.
Whitney, R. P., and Vivian, J. E. 1949. Absorption of sulfur dioxide in water. Chem.
Ens. Progr. 46, 323-337.
160 M. A . JOSLYN AND J. B. S. BRAVERMAN
Wiegand, E. H. 1946a. The brined cherry industry. Western Canner and Packer. 88 (3),
60-63.
Wiegand, E. H. 194613. Oxidation and its control. Quick Frozen Foods 8 (9), 81-83, 92.
Wiegand, E. H., Litwiller, E. M., and Hatch, M. B. 1945. Dehydration of cherries and
small fruits. Fruit Products. J. 26, 9-14, 23.
Wiegand, E. H., Norton, C. E., and Pentzer, D. J. 1939. Investigations of the cracking
problem in brining of sweet cherries. Food Research 4, 93-100.
Wiley, H. W. 1907. Influence of food preservatives and artificial colors on digestion
and health. 111. Sulphurous acid and sulphites. U . S. Dept. Agr., Bur. Chem. Bull.
No. 84 (Part 3).
Williams, J., and Corran, J. W. 1930. The preservation of the antiscorbutic vitamin in
lemon juice. Biochem. J. 24, 37-40.
Williams, R. R., Buchman, F,. Re., and Reuhle, A. E. 1935a. Studies of crystalline
vitamin BI. 11.Chemistry of the acidic product. J. Am. Chem. Sac. 67,1093-1097.
Williams, R. R., Waterman, R. E., Keresztesy, J. C., and Buchman, E. R. 193513.
Studies of crystalline vitamin B1. 111. Cleavage of vitamin with sulfite. J. Am.
Chem. Sac. 67, 536.
FVillson, K. S., Walker, W. O., Mars, C. V., and Rinelli, W. R. 1943a. Liquid sulfur
dioxide in the fruit industries. Fruit Products. J. 28 (2), 72-82.
Willson, K. S., Walker, W. O., Rinelli, W. R., and Mars, C. V. 1943b. Liauid sulfur
dioxide. Chem. Inde. 68, 176-186.
Wilson, C. P. 1925. The manufacture of pectin. J. Ind. Eng. Chem. 17, 1065-1068.
Winkler, A. J., and Jacob, H. E. 1925. The utilization of sulfur dioxide in the market-
ing of grapes. Hilgardia 1, 107-
Woidich, K. 1930. Microchemical determination of sulfurous acid in wine and fruit
juices. Microchemie (N. S.) 2, 147-150.
Woodroof, J. G., and Cecil, S. R. 1942a. Preserving fruits with sulfur dioxide. Fruit
Products J. 22, 132, 202, 237.
Woodroof, J. G., and Cecil, S. R. 194213. Processing watermelon rinds for food. Fruit
Products J. 22, 73.
Woodroof, J. G., and Cecil, S. R., and Thompson, H. H. 1943. Improving the quality
of preserves made from sulfited peaches. Fruit Products J. 22, 269.
Woodroof, J. G., and Cecil, S. R. 1945. Sulfur dioxide solution as preservative for
fruits and vegetables. Georgia Agr. Expt. Sta. Bull. No. 288, 1-32.
Yamada, M., and Okumura, E. 1950. The prevention of the discoloration of apple
juice. Nippon Joz8 Ky8kai Zasshi 46J, 57-58.
Yang, H. Y. 1953. Fruit wine. Requisites for successful fermentation. J. Agr. Food
Chem. 1,331-333.
Yang, H. Y., and Wiegand, E. H. 1949. Possibilities of secondary fermentation in
light sweet wines. Fruit Products J. 28, 134138, 157.
Yang, H. Y., and Wiegand, E. H. 1951. A method for maintaining free sulfur dioxide
content in wine. Food Technol. 6, 103-106.
Zerban, F. W. 1947. The color problem in sucrose manufacture. Sugar Research
Foundation Technol. Rept. Ser. No. 2, 1-31.
Statistical Methods in Food Research
BY B. OSTLE AND ROBERT G. TISCHER

Montana State College, Bozeman, Montana,
and
Iowa State College, Ames, Iowa
CONTENTS
Page
I. Introduction, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
11. Statistics and Research.. . .................................... 162
111. Basic Statistical Concepts. . . . . . . . . . . . . . . . . . . . . . .
1. Descriptive Statistics .................................... 164
2. Estimation Problems .................................... 170
3. Tests of Hypotheses.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IV. Analysis of Variance.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Completely Randomized Design. . . . . . . . . . . . . . . . . . . . . .
a. Simple Treatments: No Subsampling. . . . . . . . . . . . . .
b. Simple Treatments: Subsampling. . . . . . . . . . . . . . . . . . . . . 182
c. Factorial Treatments: No Subsampling. . . . . . . . . . . . . . . . . . . . . . 193
d. Factorial Treatments: Subsampling, . . . .
2. Randomized Complete Block, Latin Square,
Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
V. Regression Techniques.. . . .
1. Linear Regression, . . . .
a. One Independent
b. Two or More Ind
2. Nonlinear Regression.
a. One Independent
b. Two or More Ind
3. Biological Assays. . . . .
VI. Correlation.. . . . . . . . . . . . . .
VII. Analysis of Covariance.. . . . . . . . . . . . . . 246
VIII. Other Statistical Techniyws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
IX. Needs for Future Research.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Refercnces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
I. INTRODUCTION
There exists in biological science an increasing interest in the use of
formal statistical methods for the design of experimental procedures and
the analysis of results. One apparent motivation for this increase is the
promise of added efficiency and economy in research which is carefully
161
162 B. OSTLE AND ROBERT 0. TISCHER
planned and executed taking full account of the statistical principles of

sampling, design, and analysis.
The application of statistical methods to research dealing with foods
has included determinations of the significance of differences, making of
confidence statements concerning estimates, and tracing trends through
the use of analysis of variance and/or regression techniques. At the pres-
ent time a considerable portion of the literature on foods contains some
use of statistical methods. The possibilities of more extensive use of sta-
tistical methods in food research are vast. However, still wider use is
needed if we are to make the most of this valuable tool.
I n this contribution, it is the intention of the authors t o discuss
statistical design and techniques in the light of their use in food research;
to point out the applications which are considered to be especially effec-
tive; t o suggest additional techniques which enhance the value of the
research; and finally, t o explore fields in which further research is needed.
AND RESEARCH
11. STATISTICS
Before attempting to discuss the role of statistics in food research, it
is necessary to define just what is meant by the words statistics and
research. Research is an inquiry into the nature of, the reasons for, and the
consequences of, any particular set of circumstances, whether these be
experimentally controlled or just observed as they occur. Further, re-
search implies that the researcher is interested in more than just these
particular results-he is also interested in the reproducibility of the
results and in their extension to more complicated and general situations.
It will be observed that the above “definition” of research is a very broad
and flexible concept. Scientific research is essentially compounded of two
elements: observation, by which knowledge of certain facts is obtained
through sense perception; and reasoning, by which the meaning of these
facts, their interrelation, and their relation to the existing body of knowl-
edge are ascertained as far as the present state of knowledge and the
investigator’s ability permit.
In general, it is through experimentation or observation that statistics
enters into the scientific method. It may readily be seen that any investi-
gation, be it an experiment or a survey, is only a means towards an end;
that is, it is a device for testing some stated hypothesis or for gaining
some amount of knowledge, however small, from which some conclusion
may be drawn. It is well known that most statements resulting from
scientific investigations are only inferences, that is, they are uncertain in
character, and it is the measurement of this uncertainty which suggests
the need for statistics.
STATISTICAL METHODS IN FOOD RESEARCH 163
Some writers refer to statistics as the technology of the scientific

method. Since it has been noted that experimental and survey investiga-
tions are integral parts of the scientific method and further, that these
procedures lead to the use of statistical techniques, such a conclusion is
without a doubt a reasonable one. If this is so, then it is quite obvious
that the research worker should become more intimately acquainted with
the basic concepts and procedures which make up the body of statistical
theory and methodology. More and more articles published in the various
scientific journals contain statistical analyses in support of their argu-
ments; if the reader is not aware of the principles underlying these tech-
niques, he is in no position to appreciate fully the validity of the conclu-
sions drawn.
Statistics has frequently been classified as a method of research along
with, or in opposition to, such methods as: (1) case studies, (2) the his-
torical approach, and (3) the experimental method. Generally, this is not
a wise classification; it frequently leads to confused and incorrect think-
ing. Instead of adopting the above classification, one should think of
statistics as supplying a kit of tools which can be extremely valuable in
research. Only when one knows which tool to use, how to use it, and how to
interpret results can the most productive research be undertaken. It is
contended that the science of statistics has much to offer the food research
worker in planning, analyzing, and interpreting the results of investiga-
tions. An attempt to review some of the more important methods and
their application to food research will be made in this paper.
Statistics deals with exact probabilities applied to uncertain infer-
ences. It is concerned with: (1) the design of experiments and surveys;
(2) the collection and summarization of data; (3) measuring the magni-
tude of the variations found in the data; (4) estimating population
parameters and providing various measures of the accuracy and precision
of these estimates; ( 5 ) the testing of hypotheses about populations; and
(6) the study of relationships among two or more variables. Knowledge
of these concepts is of great importance to all research workers; here
especially we wish to emphasize their application t o research on foods.
Up to this point, the discussion has been in quite general terms,
although such technical phrases as design of experiments’’ have been
((
introduced. Designing an experiment means planning it so that the data

collected will be relevant to the problem under investigation. All too often
data are collected which turn out to be of little or no value in the solution
of the stated problem. First and foremost we must be certain that the
hypothesis to be tested is fully understood. This implies that the person
designing the experiment must insist on stating the hypothesis in a clear
and concise manner (if possible, a mathematical formulation of the
164 B. OSTLE AND ROBERT G . TISCHER
hypothesis) so that everyone concerned has the same (and, of course, the
correct) interpretation of the problem.
Experimental design is, then, the plan to be used in conducting the
investigation. It involves the assignment of treatments to the experi-
mental units and a thorough understanding of the analysis to be per-
formed when the data become available. Statistics enters into experi-
mental design because, even in the best planned experiments, all factors
cannot be controlled. Inferences will be made which will have to be based
on the observed sample data. Being uncertain, these inferences must, to
be of any practical use, be accompanied by a statement of probability
which expresses the degree of confidence which we can place in them.
The purpose of any experimental design is, of course, to provide the
maximum amount of information about the effects to be studied. It is
desirable, naturally, to keep the design as simple as possible while still
performing an efficient job within the limitations of the budget of time
and money provided for the study a t hand. Fortunately, the most effi-
cient designs are usually capable of simple analyses. This is one very
important reason why the statistician should be consulted in the early
stages of any proposed research project; he can often supply a fairly
simple design which is both efficient and economical.
We have said that the purpose of any experimental design is to provide
the maximum amount of information at minimum cost; from this it is
evident that the design of experiments is a subject which involves not
only statistical methodology but also economic considerations. If a most
efficient design is defined as one which provides the greatest amount of
information, every experiment should incorporate both efficiency and
economy into the design; failure to do so may mean a poor design, and
this means wasted time, effort, or money, perhaps all three.
111. BASICSTATISTICAL CONCEPTS
Before presenting examples of some of the statistical techniques com-
monly employed in food research, it is necessary t o comment on the
fundamental concepts and to outline briefly the methods of calculation.
To do this, we will consider three major subdivisions of statistics: (1)
descriptive statistics; (2) problems of estimation; and (3) testing of
hypotheses. The last two of these constitute what is generally referred
to as problems of statistical inference, and it is with them that the
statistician is usually concerned.
1. Descriptiue Statistics
Perhaps the simplest descriptive measure used in research is some
sort of average, The term “some sort” of average is used because more
than just the common arithmetic average (or m e a n ) of a set of values is

often encountered; the geometric mean, the harmonic mean, the median,
and the mode are also in frequent use (Table I). An average, however, is
seldom considered sufficient by itself to represent the data from which it
was calculated. It is desirable to obtain some measure of the variability
(or dispersion) among the observations which produced the computed
TABLEI
Types of Averages Commonly Used in Statistics
Type For t h e population as a whole For a sample of the population
C
N n
Arithmetic mean P = Yi/N

i=l
Geometric mean A. G M , = (nN
i=l
Yi) '" A. GM = (n
i=l
n
Yi)'ln
or or
n
N
Harmonic mean H M , = N / C (l/Yi)
i=
1 i=l
Median Med, = the value which sub- Med = the value which sub-
divides the population divides the sample into
into two groups, each two groupa, each group
group containing one containing one half the
half the items in the items in the sample. The
population. The items items in one group are
in one group are all less all less in value than
in value than those in those in the other group.
the other group.
Modr M o p = the most frequently M o = the most frequently oc-
occurring value in the curring value in the
population. sample.
average. As with averages, there are several measures of this character-

istic, the most common being the range, the average (or mean) deviation,
the standard deviation, the variance, the standard error of the mearl, and
the variance of a mean (Table 11).
The definitions may be further subdivided according to whether they
refer to the population as a whole or to some sample obtained from this
population. It should be kept in mind that this distinction will be of great
help in understanding the techniques and examples to be presented.
Thew are, of course, other statistical quantities and terms which the
c
TABLE
I1 a
ua
Measures of Variability Commonly Used in Statistics
Type For the population as a whole For a sample drawn from the population
Range R = Y(n) Y(1) - R = Y(n) - Y ( u
(largest value in the population) - (smallest (largest value in the sample) - (smallest
value in the population) value in the sample)
Average deviation
A. A. H
AD, = 2
i=l
IYi -pI/N AD = 2
i =1
Iyi -
or or
B. B.
N
Variance
N N
Standard deviation
Variance of mean
Standard error of a mean
Coefficient of variation
Per cent variability
researcher will encounter; however, these will be introduced in the de-

velopment of the discussions.
To illustrate the way in which some of the above quantities are used,
consider two cases: (1) a normal distribution and (2) an asymmetric
(positively skewed in our example) distribution (Figs. la, lb).
Actual examples of sample data which approximate these two types
of distribution are provided in Figs. 2 and 3.
For the normal population, the mean ( p ) and variance (u2) are ex-
tremely helpful in describing the population. It may be observed that the
y-38 JJ-2a v-a V pta yteo vt3u
FIG.la. Hypothetical distribution curve for normal population (mean and standard
deviation intervals shown).
FIG.lb. Hypothetical distribution curve for positively skewed population (mean,

median, and mode shown).
area under the normal distribution curve from p - u t o p +

u includes
approximately 68.26% of the total area under the curve; for 2a this per
cent is 95.44% and for 3u1 99.74%. In terms of probability this means
that 68.26% of the values obtained by a random sampling would be
expected to fall between p - a and p +
a ; 95.44% within the 2a limits;
and 99.74% within the 3u limits. I n the case of the skewed population,
interpretations are more involved, but here too the mean and variance
provide exact information about the nature of the population under
study. The corresponding sample values, namely 7 and s2, are useful in
describing a sample of a population in the same way that p and u2describe
the population as a whole. Not only do they convey information about
the sample but also they provide estimates about the true nature of the
population as a whole.
168 B. OSTLE AND ROBERT G. TISCHER
4c
9 3a
f
UJ
k
a
I
fz 20
10
0
r-f
I
5. I
I
5.3 5.5
I
s7
PH
5.9 6.1
I
6.3
FIG.2. Distribution of 205 pH measurements on nearly maximally grown, green

pimientos, 1948 season (from Sane et al., 1950).
0
a 60
W
I
a
2 40
20
0 ~ 8 7
10 9
FIG.3. Distribution of 438 duplicate half tests over the tenderness scale (Deather-
age and Reiman, 1946).
To illustrate the calculation of the above quantities (we shall examine

only the values associated with a sample, the corresponding values for the
population as a whole being calculated in a like manner), consider the
following sample values recorded for shear force when testing a certain
type of meat (5 lb., 6 lb., 7 lb., 1’1lb., 11 lb.) :
(5 + + + +
6 7 11 11)/5 = 40/5 = 8
(5 X 6 X 7 X 11 X 11)” = (25, 410)’* s 7.603l
(log 5+ + +
log 6 + log 7 log 11 log 11)/5
(0.6990+ + +
0.7782+ 0.8451 1.0414 1.0414)/5
(4.4051)/5 = 0.8810
5/(1/5 + + + ++
1/6 1/7 1/11 ++
1/11)
5/(0.2000 0.1667 0.1428 0.0909 +
0.0909)
5/0.6913 E 7.233
7 (The reader should note, however, that the median will
not always be one of the recorded sample values.)
11
11-5=6
(15 - 81
(3 + + + +
2
+ 16 - 81
1 3
+ 17 - 81 +111 - 81
3)/5 = 12/5 = 2.4
+
(11- 8/)/5
(by definition A.)
(2
+
(15 - 71
+ + + +
1 0
+ +
16 - 71
4
+ 17 - 71 111- 71 111- 71)/5
4)/5 = 11/5 = 2.2 (by definition B.)
+ +
{ (5 - 8)’ (6 - 8)’ + (7 - 8)’ (11 - 8)’
19 + + + +
4 1 9
+
9 ) / 4 = 32/4 = 8
(11 - 8)2)/(5 - 1)
+ + + +
( 5 2 62 7’ 11’ 11’
+ + + +
- (5 6 7 11 11)’/5)/(5 - 1)
{25 + + + +
36 49 121 121 - 1600/5]/4
(352 - 320)/4 = 32/4 = 8 (This is the “machine
method’’ and is preferred
when calculating the vari-
ance.)
z/ss 2.828
8/5 = 1.6
Per cent variability = lOO(0.3535) = 35.35%
1 Found by using the equation: log (GM) = (1/5) log 25, 410
= (4.40500)/5 = 0.88100
which tells us (on looking up the antilog of 0.88100) that Gill 2 7.603.
170 B. OSTLE AND ROBERT a. TISCHER
2. Estimation Problems
Perhaps the most obvious way in which a statistical inference is made
is by calculating some representative value from the observed sample
values and using this as an estimate (i.e., a calculated guess) about the
sampled population.
Now, let us take a more or less typical situation in food research
where an estimation problem has been solved. Skok (1951) has reported
on the effect of on-vine vs. off-vine ripening on the chemical characteristics
of tomatoes. Table 111, taken from this paper, shows some of the results.
The mean values not only describe the sample analyzed but also esti-
mate the true average situation in the population from which it was
drawn. However, a simple point estimate (in this case the sample mean)
of a population quantity is not always this satisfactory. It is usually
desirable to have some confidence interval estimate of the population
quantity. This confidence interval is the one within which we are fairly
certain that the true population quantity will be included. Employing
both the sample mean and the standard error, an interval may be con-
structed (q., 21.27 f 0.20 for the juice from the vine-ripened tomatoes).
If it is assumed that the population sampled was normal, the above
interval is one of approximately 68% confidence for estimating the popu-
lation mean. It has become customary to calculate intervals of 95% and
99% confidence (y = 0.95 or y = 0.99, where y is known as the confidence
coefficient). If a 1007% confidence interval is desired forestimatingp (the
mean of the population), and the population is assumed t o be of normal
form, one calculates two limits, L1 and L z (L1 < L z ) ,specifying the inter-
val by means of the following equation:
where i&.,)(,+l) is found in the standard t table for n - 1 degrees of

freedom (Snedecor, 1948).
From the data presented in Table 111, the 95% confidence interval
for the ascorbic acid content of vine-ripened tomatoes would be:
= 21.27 T 2.002(0.20) =
Other confidence intervals may be computed for various choices of t , that

is, for other values of y.
Many times an estimate of the true difference between two population
means is needed. This may be calculated easily, for two independent
TABLEI11
Comparative Composition of Tomato Juice Made from Vine-Ripened and Room,-Ripened Tomatoes'
Total pigments
Bscorbic acid, Total Total as lycopene, Carotene,
mg./100 g. solids, % sugars, % mg./100 g. mg./100 g. PH
Type of ripening Vine Room Vine Room Vine Room Vine Room Vine Room Vine Room
Mean values4 )('f 21.27 17.97 5.93 5.94 3.43 3.27 6.02 5.63 0.285 0.311 4.2 4.2
Standard error of mean
1
bi-1 0 . 2 0 0.29 0.04 0.05 0 . 0 6 0.06 0 . 1 6 0.17 0.005 0.002 0.012 0.015 0
0
% varia.bility 7 . 3 4 11.11 5.39 6.12 13.18 13.41 19.92 20.93 12.62 3.59 2.15 2 . 4 5 U
1 From Skok (1951).
2 For vine-ripened fruita n = 58; for room-ripened fruits n = 48.
samples of sizes nl and nz, respectively, from normal populations as

follows:
Point estimate = El - P,
where
Interval estimate =
“.I
Lz
= (PI - Pz)+
-
t(l-r)(nl+na-2)~I,-P,
S$,-?, = s?(l/n1 + 1//1.2) .

and
/zl + n2 - 2
= pooled estimat.e of u2
Again using the example given in Table 11, the true difference between
the ascorbic acid contents of vine-ripened and room-ripened tomatoes
may be estimated from the point estimate and 99% confidence interval
estimates:
1. P I - Pz = 21.27 - 17.97
= 3.30 mg./100 g.
’ = 3.30 T 2.626(0.34) = (::ti] mg./100 g.
According to these calculations, the true difference between the ascorbic

acid contents of vine-ripened and room-ripened tomatoes is estimated as
being somewhere between 2.41 mg./100 g. and 4.19 mg./100 g. The
researcher is 99% confident that these values are correct.
3. Tests of Hypotheses
Now let us examine the concept of a “test of a hypothesis.” Just how
does one go about testing (statistically) a hypothesis? The procedure is
as follows:
1. Specify the hypothesis in mathematical form. This is often some
type of null hypothesis.2
2. Decide what risk (i.e., what probability) of rejecting the hypothesis
we are willing t o allow when it is really true. We call this probability a ;
1OOa is often referred t o as the signijicance level.
* By a null hypothesis we just mean that some population quantity has been hypothe-
sized to be zero.
3. Decide what test procedure will be followed, that is, what test
criterion (statistic) will be calculated.
4. Split all possible values of the test statistic into two groups, one
group being known as the rejection (critical) region and the other the
region of nonrejection for the stated hypothesis. These two groups are
formed so that the size of the rejection region ( i . e . , the probability of
observing a sample value in the rejection region) will be a. This division
could, of course, be made in many ways, but it is customarily done in
accordance with the principle of minimizing the probability of accepting
the hypothesis when it is really false.
5. If the value of the test statistic calculated from the sample falls
in that group of values we have designated as the rejection region, we
reject the hypothesis.
Consider once again the tomato data (Table 111). Let us make the
hypothesis that there is no difference between the mean ascorbic acid
contents of vine-ripened and room-ripened tomatoes (21.27 vs. 17.97
mg./100 g). The appropriate test procedure is to calculate:
SY,-Yz
and compare the value obtained with those tabulated for various levels
of significance. At a 1 yosignificance level (a = O.Ol), this hypothesis will
have to be rejected if the calculated value of t exceeds
~0.01(nl+n2--2-104) = 2.626
or is less than -t0.01(104) = -2.626. Since, for the data under discussion,
t = 3.30/0.34 = 9.71, the hypothesis under test must be rejected, that is,
the difference is significant. The t-test has been frequently used in food
research; however, there are numerous occasions where it could be
profitably employed and yet has not been. Research workers should
familiarize themselves with this useful test; it is especially valuable for
determining the significance of differences obtained in simple comparisons.
IV. ANALYSISOF VARIANCE
Analysis of variance is, perhaps, the most powerful statistical tech-
nique available to the research worker in view of its wide applicability
and ease of application. It will therefore be fitting to spend considerable
time discussing not only the mechanics of this technique but also the type
of problem to which it has legitimately been applied.
Arithmetically, analysis of variance is nothing more than a device for
subdividing the total variation in a set of observations (as measured by
the slim of the squares of the deviations of each observation from the
mean of all the observations) into two or more pa&. These are then
associated with ‘(sources of variation” which have been recognized as
being present and pertinent to the experiment being conducted. Once
this subdivision of the total variation has been accomplished, and certain
assumptions made, the researcher is in a position to assess the effects of
the different factors being examined. This assessment may take one or
more of several forms :
1. The testing of one or more hypotheses concerning the factors.
2. The estimation of the true average effect of the various factors.
3. The estimation of the amount of variation to be expected among
repeated observations of a similar type and, eventually, the estimation
of the proportion of the total variability that may be ascribed to each
factor.
4. The estimation (or discovery) of relationships among the different
factors involved.
5. A study of the efficiency of the experimental design with a view t o
finding ways of making future experiments more efficient.
1 . Completely Randomized Design
a. Simple Treatments: No Subsampling. To introduce the analysis of
variance technique, let us consider the simplest design, namely, a com-
pletely randomized design (or an “among and within groups” design).
This is one in which each experimental unit has been subjected to one
treatment, the choice of which treatment was applied to each experi-
TABLEIV
Illustration of an Experimental Design with Simple Treatments and No Subsampling
Treatment no. 2 ... t Total
1’1 1 Yfl Yt1

YIP YPZ y6 2
YI,, YP,,, YtW
Total 1‘1 = 2
j=l
Yli !r2 = 2
j=l
nz
YZi rr; = 2
j-1
= Yt”, I ’ =
1
I T i
i=l
No. of
observations nl n nt $
i= 1
ni
Mean Y,= T l / n l Y P= 2’2/nz Yt= That Y = T/ 2

i-1
1
ni
mental unit having been made entirely by chance (at random). If we
denote the observation recorded from the experimental unit subjected
jt'l
to the ithtreatment by Y;j,the resulting data might appear as in Table IV.

Now it may be shown that the total sum of squares may be sub-
divided into two parts, one showing the variation among the treatment
means and the other showing the variation among experimental units
treated alike. The various sums of squares are found as follows:
2 y2 = total sum of squares
t n
if each ni = n.
T,, = among treatments sum of squares
2 Ti2/n,
t 1
= - T2/ ni
i=l i-1
= Ti2/n - T2/tn
i=l
if each n, = n.
E,, = experimental error sum of squares
= among experimental units treated alike sum of squares
= Zy2 - T,,
The results of the above calculations are usually presented as in Tables V

and VI.
How did we obtain the expected mean squares shown in Tables V and
VI? These are a consequence of the assumptions made concerning the
nature of our observations. What are these assumptions? It is customary
t o assume that any observation will consist of certain parts which when
added together produce the value recorded. In our example there are only
two parts: one part due to the treatment employed and a second part due
to the experimental unit (and all extraneous sources of variation). We
TABLEV
Analysis of Variance for Among and Within Treatments (Unequal Numbers of
Observations for Each Treatment)
Degrees of Sum of Expected nican

Source of variation freedom squares Mean square square
Among treatments t-1 T#" y',,/(t - 1) u2 t A i=l

niTi2
Among experimental
2 2
1
units treated alike
(within treatments) (ni - 1) E,, Evu/ (ni - 1) 0 2
i=1 i=
1
Total
i
i= 1
ni -1
c y2
TABLEVI
Analysis of Variance for Among and Within Treatments (Equal Numbers of
Observations for Each Treatment)
Degrees of Expected mean

Source o€ variation freedom Sum of squares Mean square square
Among treatments - I
Within treatments t ( n - 1) E"t< E,,/t(n - 1) 0 2
Total tn - 1 ZY2
usually refer to these two parts as the treatment e$ect and the error efect,
respectively. Since one customarily deals with deviations from the mean
when considering various effects, the observation in our example would
most frequently be thought of as containing three parts: a mean effect
(average of all the original treatment effects); a treatment effect which is
now a deviation from the average of all the original treatment effects; and
an error effect. Symbolically this appears as:
where
Yil = observation recorded from the jth experimental unit
subjected to the ithtreatment
f i = average or mean effect
7%= effect of the ithtreatment measured as a deviation
from the average of all original treatment effects
error effect due to: (1) t h e y b experimental unit sub-

e,j =
jected to the ithtreatment, and (2) all extraneous
effects
+
For example, Y S 5= p f 7 3 €36 and represents the observation re-
corded for the 5"' experimental unit treated with
treatment No. 3.
It is clear that p and the ~i are constants such that i

i=l
n ~=; 0. The
cij(the error effects) are assumed to be independently and normally dis-

tributed with mean zero and common variance u2. Other mathematical
models might be used but not only does the one chosen lend itself to
simple mathematical manipulation but it is an excellent approximation
to the real world.
Now the usual hypothesis to be tested using such an analysis as
presented above is that there are no differences among the true effects
of all the treatments being investigated. Mathematically this may be
expressed as H: T, = 0. (i = 1, . . . , t ) . The test procedure,is to calcu-
late F where:
mean square for among treatments
F =
mean square for within t,reatments
and compare this with the tabulated values of F (Snedecor, 1948)

t
for the
appropriate degrees of freedom, namely, n1= t - 1 andn2 = 2
i-1
(ni - 1)
or t(n - 1). If the calculated value of F exceeds the tabulated value,

Fa(,,,,,,,
the hypothesis is rejected. There are also other tests which might
be performed (e.g., tests of selected treatment comparisons), but these
will be discussed subsequently.
TABLEVII
Fat Absorbed by Batter of 24 Doughnuts'
Fat no.
1 2 3 4 5 6 7 8
Doughnut (grams)
1 164 173 177 178 163 175 178 155
2 172 161 183 191 165 193 146 166
3 168 190 191 197 144 178 141 149
4 177 180 169 182 177 171 150 164
5 156 197 179 185 165 163 169 170
6 195 167 187 177 176 176 182 168
Totals 1032 1068 1092 1110 990 1056 966 972
Means 172 178 182 185 165 176 161 362
I Lowe (1936).
An example of the above type of design is provided by Lowe (1935),

who investigated the amount of fat absorbed in the cooking of doughnuts
to see if this is affected by the type of fat used. Eight different fats were
employed in the experiment. The data, which have been discussed in
some detail by Snedecor (1948), are presented in Table VII.
The appropriate subdivision of the total sum of squares is into two
parts: the sum of squares associated with the variation among treatment
(fat) means, and the sum of squares associated with the variation among
batters treated alike (ie., among batters within fats). In this example,
t = 8 and n = 6 and thus the required sums of squares are found as
follows :
Zy2 = (164)' +
* * * +
(168)2 - (8,286)'/48 = 9,193
Tuu= ((1,032)' +
* * * +
(972)')/6 - (8,286)'/48 = 3,527
Eu,= 9,193 - 3,527 = 5,666
These results are presented in Table VIII. To test the hypothesis that
TABLE VIII
Analysis of Variance for Fat Absorption by Doughnuts1
Degrees of Expected mean

Source of variation freedom Sum of squares Mean square square
+ (95) 1
8
Among treatments 7 3527 503.9 u2 ~ i *
i= 1
Among experimental
units treated alike
(within treatments) 40 5666 141.6 U2
Total 47 9193
1Lowe (1935).
there are no differences among the true average amounts of fat absorbed
by doughnuts while cooking for the eight different fats, compute:
mean square for among treatments
F =
mean square for within treatments
= 503.9/141.6 = 3.56
which has degrees of freedom nl = 7 and n2 = 40. Since this exceeds the
1% point of F for the specified degrees of freedom (since F = 3.56 >
= 3.12), the hypothesis is rejected as stated. It is concluded
Po.01(,,40)
that the eight fats are not all alike with respect t o their absorption by the
doughnuts during cooking.
Frequently there exists an interest in other comparisons (or contrasts)
among the treatments than that specified by the general hypothesis that
the true effects of all the treatments being investigated are the same. As
an illustration, suppose that the fifth, seventh, and eighth fats in Lowe’s
experiment were animal fats and the remainder were vegetable fats. It
would be of some interest to make a comparison of “animal” us. ‘(vege-
table” fats as well as comparisons among the three “animal” fats and
among the five “vegetable” fats. These comparisons may be made by
computing the following sums of squares:
(( Animal us. vegetable l 1 fats:
(990 + 966 + 972)2 + (1032 + 1068 + 1092 + 1110 + 1056)2
(6) (3) (6)( 5 )
Among “animal fats 1 1 :

+ +
(990)2 (966)2 (972)? - (990 + 966 + 972)2 = 52
6 (6)(3)
Among ((vegetablefats 1 1 :
(1032)2 + (1068)2 + (1092)2+ (1110)2+ (1056)*
6
- (1032 + 1068 + 1092 + 1110 + 1056)2 = 619
(6) (5)
By putting the results in the form of an analysis of variance table (Table
IX), it is noted that what has been accomplished is a subdivision of the
TABLE
IX
Analysis of Variance for the Doughnut Experiment Showing Hypothetical Treatment
Comparisons1
Degrees
of
Source of variation freedom Sum of squares Mean square
Among treatments 7 3527 503.9

“Animal” vs. “vegetable” fats 1 2856 2856
Among “animal fats” 2 52 26
Among “vegetable fats” 4 619 154.8
Among experimental units treated
alike (within treatments) 40 5666 141.6
Total 47 9193
1 Adapted from Lowe (1935).
“among treatments l 1 sum of squares so that the selected comparisons

may be interpreted.
180 B. .OSTLE AND ROBERT 0. TISCHER
If the appropriate P tests are made to provide information concerning

the selected comparisons, namely:
mean square for the selected treatment comparison
F =
mean square for within treatments
it is apparent that almost all of the variation among fats may be asso-
ciated with the difference between “animal” fats on the one hand and
“vegetable” fats on the other. Neither the differences among the “ani-
mal” fats nor the differences among the “vegetable” fats are statistically
significant.
Overman and Li (1948) compared, by means of a tastiiig panel, the
flavors resulting from the use of five fats in pastries. The members oflthe
panel were selected primarily on the basis of availability and no effort was
made t o determine the judges’ abilities in advance. This procedure for set-
ting up panels sometimes may be necessary. It is not, however, an ideal
method for insuring a high degree of precision in the results obtained.
Five samples of pastry, representing the five fats, were tested each
day for ten days. Ten replicate lots of pastry, prepared under controlled
conditions, were presented for judging, a different order of presentation
being used each day. A scale from one through five was represented
by descriptive terms of, not edible, score = 1; poor, score = 2; fair,
score = 3; good, score = 4; and excellent, score = 5, with the numbers
being assigned after judging was completed.
A simple test of consistency of judges was made using the range of
each judge:
Judge A 33 C D 1C F G H J K
Averagerange 2.2 2.4 1 1 3.0 1.6 2.6 2.4 1.2 1.8 2.0
The judge exhibiting the smallest range was considered most consistent.
Another simple device of use in determining consistency of judges is
the mean score and sum of deviations from the mean. (Table X.) The
differences among the four judges are readily apparent from the total
deviations over all fats.
The simple analysis of variance (Table XI) indicates that large
differences in flavor were evident among the five fats tested as well as
large differences in the judges’ scores within fats. These facts are evident
from a consideration of the large F values, both of which are highly
significant at P = 0.01.
Comparisons of the discriminating ability and consistency of the
judges were made by an additional use of the analysis of variance. An
individual analysis was made for each judge among fats and within fats
exemplified by the analysis for Judge A.
STATISTICAL METHODS I N FOOD R E S E A R C H 181
Judge A
Mean
Source of variation Degrees of freedom square
Among n fats (n - 1 ) 11.02
Within fats for m replicates
(error variance) ( n - l ) ( m - 1) = 9 X 5 = 45 0.6244
11.02
Variance ratio = F ~ , u ,= -= 17.65.
0.6244
The data for the ten judges are presented in Table XII. Since all of the
F values are above P = 0.01, all judges had some discriminating ability
but it apparently varied considerably from judge to judge. Whether the
TABLEX
Mean Scores of Each Judge and Sum of Absolute Values of Deviations from His Own
Mean Scores for Flavor of Pastry Made with Five Different Fats112
All fats
Fat no. total
Judge and score identification 1 2 3 4 5 deviation
Judge C
Mean score 3.7 3.7 3.9 3.6 2.8 ....
Sum of deviations from mean 4.2 4.2 1.8 5.6 6.4 22.2
Judge D
Mean score 3.1 2.8 3.9 1.8 1.9 ....
Judge E
Mean score 4.3 4.3 4.6 3.6 2.9 ....
Judge H
Mean score 4.5 3.8 3.1 2.1 1.8 ....
* From Overman and Li (1948).
2 Ten replicate lots of pastry were scored for each fat.
TABLEXI
Analysis of Variance-Combined Flavor Scores for Pastry Made with Five Different
Fats'
Degrees of
Source of variation freedom Variance F
Fat 4 60.9530 81.93

Judges 9 9.1736 12.33
Error 486 0.7440 ._...
Total 499
I From Overman and Li (1948).
182 B. OSTLE AND ROBERT 0 . TISCHER
variance ratios may be compared quantitatively to say judge H was best

and judge D worst in discriminating ability is open to some question.
An appropriate test of consistency is Bartlett’s (1937) (see also
Snedecor, 1948) of homogeneity of variances which, in this case, would
be applied t o the error variances in Table XII. Bartlett’s test is made by
comparing an adjusted x2 value with the x2 value for an appropriate
number of degrees of freedom where:
where
2.3026 = log, 10
n = number of columns,
k = number of rows in a two way classification
n
1log 52 = c
n
1
(log 2)
The adjusted x2 value in this test was 51.72 with nine degrees o
freedom. The tabular value for x2 is 21.666 for P = 0.01, indicating thab
the panel used was not homogeneous.
b. Simple Treatments: Subsampling: Occasionally several determina-
tions may be made on each experimental unit. When such an event takes
place, the analysis of variance table is extended to provide for the varia-
tion “among determinations on each experimental unit ” (i.e,, one further
subdivision of the total sum of squares is made). As an example of this
type, consider the data collected by Peterson, Tucker et at?. (1951) pre-
sented in Table XIII. The appropriate analyses of variance are presented
in Table XIV. However, we are unable to verify the sums of squares for
the variation among leaves since data on individual leaves were not given;
only the mean values were published (see Table XIII).
This analysis of variance provides another example for testing a null
hypothesis and also gives further insight into the concept of components
of variance. The expected mean square for “between leaves” on the same
plant contains only one component of variance, since the only factor
which affects (or produces) this variation is “leaves.” The expected mean
square for “among plants in same group” contains two components of
variance, since this source of variation reflects the variation among the
mi2
2
d
P
F
TABLEXI1
The Error Variance and Variance-Ratio of Each Judge's Scores for Pastry Made with Five Fats' i
c3
Judge B
A B C D E F G H J K g
Error variance 0.6244 0.7533 0.3356 1.3667 0.3978 0.8333 0.8689 0.2111 0.5178 0.5200 ?!.
Variance-ratio 17.65 5.22 5.45 4.37 11.89 14.12 5.14 60.78 11.20 32.06 q
0
1 From Overman and Li (1948). 0
U
TABLE XI11
Mean Ascorbic Acid and Moisture Contents of Individual Turnip Green Leaves
Plant 1 Plant 2 Plant 3 Plant 4 Plant 5 Mean
Moisture, %a
Group 1 72.96 73.10 66.10 78.46 77.26 73.58
Group 2 81.20 73.10 78.48 78.91 77.24 77.78
Group 3 74.08 76.37 73.48 76.18 77.20 75.46
Group 4 74.03 76.91 76 90 74.28 73.74 75.00
Group 6 77.59 76.78 75.79 76.44 78.03 76.93
Group 6 78.30 78.98 77.82 76.41 75.05 77.32
Ascorbic acid (dry basis; mg. per 100 grams)
Group 1 823 816 906 1167 858 894
Group 2 1354 993 1182 888 983 1080
Group 3 813 754 606 606 750 706
Group 4 674 889 766 1012 938 856
Group 5 1164 1018 814 1023 1119 1028
Group 6 1146 1186 1124 1384 858 1204
L.S.D. for “plants in groups” = 339 (1% level of significance)
L.S.D. for “groups of plants” = 290 (1% level of significance)
1 From Peterson, Tuoker el al. (1951).
f These moisture values are somewhat lower than those usually found, but it should be emphasized
that the planta were quite mature and grown in the greenhouse.
TABLEXIV
Analyses of Variance for Moisture and Ascorbic Acid Contents of Individual Turnip
Leaves’
Mean square
Degrees of Ascorbic acid Expected mean
Source of variation freedom Moisture, % (dry basis) square
Groups of plants 5 25.8817 316,664** UI* + 2up2 + lOu,*
Plants in groups 24 13.3952 53,624** ci* + 2up2
Leaves 30 9.5128 15,227 uz*
Variance components:
Moisture Ascorbic acid (dry basis)
~i’: 9.5128 15,227
sp*: 1.9412 19,148
so*: 1.2486 26,304
1 From Peterson, Tuoker el al. (1961).
** Signifioant a t 1 % level.
means of the observations on each plant and these means will vary not
only because of the variation from plant to plant but also because of the
two leaves from each plant. Similarly, the mean square for “among
groups” reflects the variati,on among the means of all the observations
for each group. These means will vary because of three contributing fac-
tors: (1) variation among groups, (2) variation from plant to plant within
the 8ame group, and (3) variation between the two leaves on each plant.
Thus the expected mean square involves three components of variance.
The coefficient of up2 = 2, because there were two leaves from each plant.
The coefficient of ug2= 10, because there were ten observations, two on
each of five plants in each group. Another way of looking a t the coeffi-
cients described above is to observe that each plant mean is the average
of two observations, whereas each group mean is the average of ten
observations.
The following model was assumed in the analyses of Table XIV;
Yz,k = p -k g L -k p,, -k l,jk i = 1, . . . ,6

j = l , .. .,5
k = 1,2
where the g,, pi3, and li3k are independently and normally distributed with
expected values zero and variances up2, u p 2 , and u?, respectively. It is
clear, certainly, that Y z ] k represents the kth observation (leaf) from the jth
plant in the ithgroup, where g, is the true effect of the itbgroup, pi, is the
effect attached to the jth experimental unit plant in the P group, andli3k
is the effect associated with the kth leaf from the jthplant in the ithgroup.
Now, in this experiment, the null hypotheses t o be tested are H1:ug2= 0,
and H2:up2 = 0; that is, that there are no differences among the true
effects of the groupings or of plants on the ascorbic acid and moisture
contents of turnip green leaves. If Hl is true, then, for example,
mean square for among groups
F =
mean square for plants
= 316,664/53,624 = 5.91
with degrees of freedom nl = 5 and n2 = 24. Looking up values in the

F-tablefordegreesoffreedomnl = 5andnz = 24, thevaluesFo.06(6,24) = 2.62
and F0.01(6,24) = 3.90 are found. Since the calculated value of F exceeds
HI is rejected and the groupings appear to have an appreciable
Fo.olc~,ac),
effect on the ascorbic acid content of turnip green leaves.
The method of analysis has been described which is appropriate to a
completely randomized design in which we sampled from each experi-
mental unit. Consider now the computation of the various sums of
squares in such a design. As an example consider a process for the con-
version of sugar to lactic acid in which there is a choice of using one of
two microorganisms known to carry out the fermentation. An appropriate
substrate is prepared and divided into two unequal portions. These por-
tions are again divided into a number of 100-ml. samples. The number of
samples in each of the two portions is as follows: No. 1, 4 samples; No.
2, 3 samples. Each of the samples is inoculated with one or the other of
the two microorganisms (4 samples being inoculated with one micro-
186 B. OSTLE AND ROBERT 0 . TISCHER
organism and 3 with the other). The fermentation is allowed t o proceed

for 24 hours and then each sample is examined for the amount of residual
sugar (expressed as mg./5 ml.) to determine the amount of change pro-
duced by each microorganism (the converted sugar having been pre-
viously shown to occur as lactic acid), Varying numbers of determinations
are made on each sample. The null hypothesis is that there is no differ-
ence between the true effects of the two microorganisms as reflected by
the amount of unconverted sugar. Assume that the data presented in
Table XV are obtained:
TABLEXV
Amount of Unconverted Sugar in the Substrate Following a Twenty-four Hour
Fermentation Due to Two Different Microorganisms1
Microorganism no. 1 Microorganism no. 2
Sample no. Sample no.
Run no. 1 2 3 4 1 2 3
1 5.6 5.0 5.4 5.3 7.6 7.4 7.5
2 5.7 5.0 5.4 5.5 7.6 7.0 7.6
3 - 5.1 5.4 - 7.8 7.2 7.5
4 - - 5.5 - - - 7.4
5 - - 5.4 - - - -
Sums 11.3 15.1 27.1 10.8 23.0 21.6 30.0
nif 2 3 6 2 3 3 4
1 Hypothetioal data: coded for easy oaloulation.
In order t o simplify (as much as possible) the presentation of the

method for computing the various sums of squares needed for our analysis
of variance, the following notation will be used:
kth determination on jth sample from itbtreatment
number of determinations on jth sample from the ithtreatment,
that is, k = 1, . . . , nij
number of samples from the ithtreatment, that is, j = 1, . . . ni
2
i=l j-1
nij = total number of determinations in the whole experi-
ment, where t equals the number of treatments
total of the ni, determinations on the j t b sample from the ithtreat-
ment = C
nij
k=l
yi3k
ni
total of all the (2 nij) determinations on the ithtreatment
c
j-1
ni
- 2
j=ljk=l
Yijk=
j=l
Sii
STATISTICAL METHODS I N FOOD RESEARCH 187
T = total of all the ( N ) determinations in the whole experiment

1 ni ni; t ni t
In our example, these quantities are:
nll = 2 Sll = 11.3 nl = 4

n12 = 3 Slz= 15.1 nz = 3
n13 = 5 Sla= 27.1 N = 22
n14 = 2 S14= 10.8 t = 2
n21 = 3 Szl = 23.0 T 1 = 64.3
n22 = 3 Szz= 21.6 T z = 74.6
n23 = 4 s 2 3 = 30.0 T = 138.9
The various sums of squares (this term always refers to corrected sums of
squares; that is, it is the sums of squares of deviations about some average
value) may be calculated most easily in the following order:
T,, = treatment sum of squares =

i=1
t ni
S,, = sum of squares among samples treated alike

1 ni
=
1
11
ni
i - 1 j=l
S,,2/nij -
i=l
Ti2/ 2
j=l
n,,
D,, = sum of squares among determinations on the same sample

= 222
<;I j=,l knTjl
( Yijk- Pij)2
= 222
n'
i = l j=1 k - 1
yijk2 - 5
i - 1 j-1
s,,2/nz,
188 B. OBTLE AND ROBERT G. TISCHER
where Pij = average of the n.j determinations on the jthsample from the
ithtreatment
= S&j/n,j
Pi = average of all the determinations on the ithtreatment
ment = T / N
For the hypothetical example, the sums of squares are:
(5.6)' + * * * +
(7.4)' - (138.9)'/22
902.07 - 876.9641 = 25.1059
(64.3)'/12 +
(74.6)'/10 - (138.9)'/22 901.0568 - 876.8641
=
= 24.0927
[(11.3)'/2 +
(15.1)'/3 +
(27.1)2/5 -I- (10.8)2/2 (23.0)'/3 +
+
(21.6)'/3 +
(30.0)'/4] - [(64.3)'/12 (74.6)'/10] +
901.9036 - 901.0568
0.8468
25.1059 - 24.0927 - 0.8468
0.1664
Careful attention should be given to the preceding calculations; a
thorough understanding of the principles and mechanics of this example
will be an immense help in later illustrations. If each ni = n and each
n,, = d, the above calculations may be materially simplified, since j=l nij 2
TABLEXVI
Analysis of Variance-Sugar Fermentation Data1
Degrees
of Sum of Mean
Source of variation freedom squares square Expected mean square
Between microorganisms 1 24.0927 24.0927 UD* 4-CZU* f '= jt=:
Among samples treated

alike 5 0.8468 0.1694 UD' + clop
Among determinations
on the same samples 15 0.1664 0.0111 un'
Total 21 25.1059
* Hypothetical data (see Table XV).
would reduce to nd and N would equal tnd. The results of the calculations
are presented in Table XVI. We have assumed the model,
YJk = /4 + + + 7% €tj (Sijk i = 1, .. 9 t
j=l,...,ni
k = 1, . . . ,nij
where t = 2, T~ is the true effect of the ithtreatment (and the 7%are fixed
22
t n;
constants so that = O ) , the E z j are N I D ( 0 , d),and the dijk are

i = 1 j-1
N I D ( 0 , ao2). It may be shown that:
N - 22
f ni
i=lj-1
n,j2/N
c1 = t
and
On examining the expected mean squares in Table XVI, it is apparent

that an exact test of the null hypothesis H: T~ = 7 2 = 0 is impossible.
This unfortunate circumstance exists because c1 # c2 and this arises from
the unequal number of determinations per sample and the unequal num-
ber of samples per treatment. This result brings home with great force the
need for equal numbers of observations in the various subclasses. No difi-
culty was encountered when dealing with the data on ascorbic acid in
turnip greens; this was because there were equal frequencies in the various
subclasses.
Often it is desirable to estimate the magnitude of the various com-
ponents of variance. Referring to Table XIV, we see that the components
are estimated by s? = 9.5128, sp2 = 1.9412, and so2 = 1.2486, where
st2, sp2,and so2 estimate ul2, up2,and ug2,respectively. These estimates aid
in determining which factor is contributing (subject to sampling varia-
tion, of course) the most to the observed variation as measured by the
variance concept. Such information may then be used as a guide in
improving the experimental technique and for seeking a more suitable
experimental design in order t o make future experiments more efficient.
This suggests that we consider the efficiency of the plan used relative
to other possibilities (different numbers of samples, etc.) within the same
190 B . OSTLE A N D ROBERT 0. TISCHER
basic design. For example, is the design with five plants per group and
two leaves per plant more or less efficient than a similar design utilizing
four plants per group and three leaves per plant? Before this question
may be answered, a criterion for measuring efficiency must be adopted.
Assume that a design which provides a smaller estimated variance of a
treatment (group) mean than does some other design is the more efficient.
This brings up the question: What is the estimated variance of a treat-
ment mean? Recalling earlier discussions of the estimated variance of a
mean, observe that it is defined as the estimated variance of the indi-
vidual items contributing to the mean divided by the number of items
(observations) in the mean, that is, st2 = sv2/n. Once you know what
factors cause a single treatment mean to vary and, naturally, the number
of observations (plants, leaves, etc.) which make up the particular mean
under consideration, you may easily compute the sample variance and
standard error of the mean. In the example presented by Peterson,
Tucker et al. (1951) two factors cause a treatment (group) mean to vary,
namely, plants in the same group and leaves from the same plant. Another
way of saying this is that the treatment mean would vary from one repe-
tition of the experiment to another because of variation among plants and
variation between leaves from the same plant. Similarly, in the fermenta-
tion example, a treatment mean might vary from one repetition of the
experiment to another because of variation among samples and variation
among determinations. A treatment mean, in the turnip greens example,
is an average of ten determinations (two from each of five plants). I n the
fermentation example, the answer is not so clear cut-the numbers of
samples and the numbers of determinations per sample are different for
the two treatments and thus a definite answer to the question “HOW
many observations? ” can only be given if a specific treatment is designated.
All of the statistical details indicated above are necessary parts of the
analysis. Consider the data on the turnip greens: Using a slightly different
notation, the estimated variance of a treatment mean may be represented
by (for ascorbic acid)
In more general form, letting d equal the number of leaves per plant and
n equal the number of plants per treatment (d = 2 and n = 5 in our
example) :
A comparison may now be made to determine whether the present

design, with five plants per group and two leaves per plant, is more or less
efficient than a similar design employing four plants per group and three
leaves per plant. To compute the efficiency of the design used relative to
the proposed design, it is necessary to make one assumption. This assump-
tion is as follows: even though the number of plants (n)is changed and/or
the number of leaves per plant ( d ) is changed, the assumption is made
that the estimates of at2 and uP2 would remain unchanged. With this
assumption granted, P(Pi) may be calculated under the proposed experi-
mental plan. If we denote this new estimated variance of a treatment
mean by p’(Pi),it is found t o be:
and since this value is greater than the corresponding value under the
original plan (6,055.9 > 5,362.4), the decision is made that the original
design is the more efficient of the two alternatives. In terms of relative
efficiency, the efficiency of the original design relative to the proposed
plan is (6,055.9/5,362.4)100 = 112.9%.
When discussing the analysis of the fermentation data it was noted
that no exact test of the null hypothesis H: 7 1 = 7 2 = 0 was available
owing to the unequal frequencies a t the various stages of the subsampling.
Such a difficulty may be circumvented when it occurs by using our esti-
mates of the components of variance. By these means we shall synthesize
one (or two) mean squares which will have the same expected value if the
null hypothesis is true. These synthesized mean squares will then be used
to form a ratio which is approximately distributed as F . This method,
therefore, provides only an approximate solution. If the expression
cz
- (0.1694) -
c1 [E:
-- 1
1 (0.0111)
+
is obtained this would be an unbiased estimate of uD2 c g 2 , and this is
the same as the expected value of the mean square for treatments if H
is true. Denoting the mean squares, and the constants by which they
were multiplied, used in obtaining our “synthetic mean square ” by
MSI, MS2, . . . and al, az, . . . respectively, Satterthwaite (1946)
gives the following formula for approximating the degrees of freedom
to be associated with the synthetic mean square:
e
W
Is
TABLEXVII
General Form for Analysis of Variance for Two-Factor Factorial Experiment in a Completely Randomized Design: Equal Number of
Observations for Each Treatment Combination
Source of variation Degrees of freedom Sum of squares Mean square Expected mean square E:
a
A a - 1
b
B b - 1
where dfi (i = 1, 2, . . .) are the degrees of freedom attached to M S ,

(i = 1, 2,. . .). In the example: a1 = c2/c1, a2 = -(cZ/cl - l ) , dfl = 5,
and df2 = 15. The approximate F-ratio
24.0927
F =
+
ai(0.1694) a2(0.0111)
is then calculated and compared with the tabulated values of F with
degrees of freedom nl = 1 and n2 = a. It should be clear that if the deci-
sion had been made to synthesize the numerator mean square rather than
the denominator mean square (sometimes both are synthesized) a
different F value (with different degrees of freedom) might easily occur.
This emphasizes still more the approximate nature of the proposed test
procedure. Reference may be made to Cochran (1947) for a further
discussion of this problem.
c. Factorial Treatments: N o Subsampliny. Most experimental work is
concerned with the joint effects of several factors upon the characteristic
being measured. For example, in agronomic research the joint (simul-
taneous) effect of variety and fertilizer upon the yield of various crops is
often studied. Research in food preservation is commonly concerned
with the effects of varying storage temperatures and lengths of storage
time upon the quality of food products. Other examples are readily
available in the literature and some of these will be referred to.
The additional complications which arise when the treatments are
combinations of two or more levels of each of the factors whose influence
is being investigated are few. Thus it does not seem desirable to spend
much time outlining computational techniques. These are adequately
explained in such texts as Goulden (1939), Ostle (1954), and Snedecor
(1948) where the necessary details may be found.
Suppose the number of treatments involved in the experiment, t , can
be thought of as all combinations of a levels (or classifications) of one
factor with b levels of a second factor. That is, t = ab. Then the treat-
ment sum of squares:
T,, =
TI2 + T2' + . * . l't2 T2
-_
n nt
can be subdivided into three parts
A,,, =
A12+. . * + A n 2- _
T2
nb nab
sum of squares due to the first factor
=
Bull =
BIZ +
* * * Bh2
--T2
nu nab
= sum of squares due to the second factor
194 B. OSTLE AND ROBERT Q. TISCHER
(AB),, = T,, - A,, - B,,

= m m of squares due to the interaction between the two factors
Ti = total of all observations associated with the ith treatment
Aj = total of all observations associated with the jth level of the
first factor
Bk = total of all observations associated with the kth level of the
second factor
It has been assumed that the same number of observations (experimental
units), TI, were included for each treatment combination. The results just
obtained are usually presented as in Table XVII. The appropriate F test
for examining the average effect of any one of the two main effects or the
interaction effect is performed by forming the ratio of the mean square
for the effect in question divided by the mean square for experimental
error.
Cook (1941) has made a study of Canadian Wiltshire bacon using a
completely randomized design to investigate in duplicate the effect of
storage temperatures, permeability of wraps, one period of storage and
several methods of thawing, heating, and smoking on the color and color
stability of rib-in pork backs.
The storage temperatures used were 6.6', -12.2', -23.4', and
-29.0' C. The wraps were (1) glassine, (unsealed), (2) double glassine,
(unsealed) and (3) sulfite paper laminated aluminum foil (sealed). The
details of the thawing and curing procedures are shown in Table XVIII.
The analysis of variance of color after 48 weeks storage is shown in
Table XIX.
All of the four main effects were highly significant at the 0.1% level
of probability. Over half of the two-factor interactions (differential
TABLEXVIIZ
Wiltshire Bacon Study-Thawing and Curing Procedures1
Tempera- Time, Additional exposure
Stage Method ture, "C. hr. after sampling
Thawing I n air (wrapped) 10.0 25 6 hr. in air a t 7.5" C.
In brine 4.5 5 2 hr. in air a t 7.5" C.
In pickle 4.5 434 2 hr. in air at 7.5" C.
In water 38.0 156 1f.i hr. in air at 7.5" C.
Cure Wiltshire pickle 3.6 88 Sampled on removal

Maturation I n air (covered) 3.2 168 Sampled on removal
Subsequent Smoked in commercial 55 754 Sampled for color on fol-
treatment smoke house lowing day, and for fat
Heated in laboratory 55 73i analyses five days later.
oven Temp. 10' C
I From Cook (1941).
TABLEXIX
Analysis of Variance-Color Data in Wiltshire Bacon Study’
Degrees Mean square
of Total
Variance due to freedom Blue Green Red (brightness)
Temperature 4 13.9*** 34.4*** 74.8*** 328.0***
Wrapping 2 23.2*** 57.8*** 122.0*** 539.0***
Thawing 3 12.3*** 43.9*** 140.0*** 476.0***
Stage of conversion 2 101.0*** 367.0** * 387.0*** 2376.0***
Temperature X wrapping 8 13.6*** 30.9*** 53.4*** 273.0***
Temperature X thawing 12 6 . 4 * * * 17.1*** 27.2*** 140.0***
Temperature X stage 8 1.4 1.3 0.8 9.0
Wrapping X thawing 6 5.1* 12.0* 21.3** 96.8*
Wrapping X stage 4 0.8 1.3 1.2 9.1
Thawing X stage 6 13.3*** 43.3*** 149 .O**’ 489.0***
Error 124 2.0 4.2 6.7 35.2

IFrom Cook (1941).
* Indicatea 6 % level of significance compared with error.
** Indicatea 1 % level of significance compared with error.
*** Indicates 0.1 % level of significance compared with error.
effects) were significant and of no greater magnitude, in general, than the

main effects. On this basis the decision was made to use the error term
in the test of significance. The same conclusion might have been reached,
with improved theoretical justification, on the assumption that the levels
of the main treatment effects were not randomly chosen. In this event it is
necessary to make tests of significance using the generalized error in
preference to the appropriate interaction or combination of interaction
terms.
Examination of the error term indicates that it contains four three-
factor interactions and one four-factor interaction:
Interadion Degrees of freedom
TXWXTh 24
T X W X S 16
WXThXS 12
TXThXS 24
TXThXS XW 48
-
Total (error) 124
As reported, this analysis and the accompanying analysis for stability of
color in the bacon contains no replication. As a result, there exists no
appropriate term for evaluating the three- and four-factor interactions.
This difficulty could have been remedied by replicating the experiment
twice. This arrangement would add 180 subclasses to the analysis, each
possessing 1 degree of freedom. The “within subclasses’’ mean square
with 180 degrees of freedom would estimate u2. This mean square could
then be used to test the significance of the four main effects and all of the
eleven interactions.
Use of this procedure is advisable where an estimation of all inter-
actions is important for any reason and especially important where
further investigations are to be made of the magnitudes of individual and
combined sources of variation.
d. Factorial Treatments: Subsampling. As might be expected, data are
often encountered where some degree of subsampling has been carried
out on the experimental units. The calculation procedures for such cases
have been examined in a previous section of this paper. No further com-
plications are introduced because of the factorial treatment combinations.
This will be apparent upon examination of the following examples.
Sharpe and Gammon (1951) studied the sources of error in analyses
of leaves of pecan trees using two locations, two varieties, four trees of
each variety in each location, and four determinations (two samples from
each tree and two ashings per sample) of calcium, potassium, magnesium,
and phosphorus (Table XX).
The analysis of variance (Table XXI) indicated that varieties differed
only with respect to calcium level. The two orchards yielded different
responses in all four analyses for minerals. In addition, the behavior of
varieties in the two orchards (VxO)was different.
There were four trees in each of four groups resulting in
ab(n - 1) = (2)(2)(4 - 1) = 12 degrees of freedom for trees
The mean square indicated that considerable differences existed among

trees in an experimental unit.
Two determinations of each mineral were made for each of the
minerals for each tree resulting in (2 - 1)16 = 16 degrees of freedom for
sample duplicates. Each duplicate was ashed twice to yield thirty-two
pairs of ash determinations with a total of 32 degrees of freedom.
The estimated variance components were obtained by equating com-
ponents with mean squares and solving for a single component. The cal-
culation for vt for calcium would be:
0.03902 = 9, + 2v, + 49,

-0.00560 = -9, - 29,
0.03342 = 49,
0.03342
9, = - 4
= 0.00836
With the use of more generally accepted notation this calculation would
be :
MSTia - MSsd = 8a2 268' + +
481' - &a2 - 28a'
where Tib = trees in block; sd = sample duplicates.

The variance components shown in Table XXII indicate that the
largest variation stemmed from vt,
the trees in a block. Variance com-
ponents for ashing were also rather high and an additional experiment
was run (Table XXII) to discover whether the ashing procedure or the
determinations performed on the ashed sample needed improvement.
The variance components in Table X X I I are all larger for determinations
TABLEXX
Mineral Composition of Foliage from Pecan Trees'
Orchard and Tree Foliage composition (as percentage of dry wt.)l
variety no. Ca K Mg P
Orchard l--(;"urtis 1 1.57 0.56 0.364 0.174
2 1.29 1.18 0.313 0.166
3 1.32 1.27 0.424 0.177
4 1.20 0.97 0.407 0.162
- __ -
Total 5.38 3.98 1.508 0.679
Orchard 1-Schley 1 1.36 0.55 0.442 0.171

2 1.28 0.80 0.325 0.151
3 1.24 0.95 0.322 0.163
4 1.37 0.77 0.346 0.157
__
Total 5.25 3.07 1.435 0.642
Orchard 2-Curtis 1 0.95 1.31 0.179 0.182

2 0.84 1.45 0.111 0.159
3 0.80 1.27 0.154 0.161
4 0.96 1.09 0.141 0.166
-
Total 3.55 5.12 0.584 0.668
Orchard 2-Schley 1 1.42 1.74 0.253 0.191

2 1.16 1.48 0.196 0.194
3 1.28 1.94 0.278 0.201
4 1.25 1.63 0.197 0.186
- --
Total 5.11 6.79 0.924 0.772
1 From Sharpe and Gammon (1961).
Average of four determinations per tree.
TABLEXXI
Analysis of Variance-Mineral Content of Pecan Foliage' m
Degrees of Mean square Expected mean
Source of variation freedom Ca K Mg P square 8
Total 63
Variety 1 0.43070** 0.05210 0.017523 0.001106
Orchard 1 1.03280** 5.89280** 0.514627** 0.003570**
Variety X orchard 1 0.61420** 1.78350** 0.042282** 0.004865**
Trees per experimental unit 12 0.03902** 0.18707** 0.008148** O.O00274**
Sample duplicates 16 0.00560 0.00565 0.000359 0.000082 r l
Ash duplicates 32 0.00379 0.00514 0,000437 0.000069
Estimates of Variance Components
Trees in block (Vt) 0.00836 0.04535 0.001947 0.000048
Samples (V.) o.oO09o O.OOO25 0.000000z 0.000006
Ashing (V,) 0 00379 0.00514 0.000437 0.oooO69
* From Shsrpe and G a m m o n (1951).
2 Computed value wa8 negative.
** Significant at 0.01 level.
than for ash duplicates, indicating that the analyses (determinations)

performed on each ash duplicate were the source of the largest error.
The occurrence of negative estimates of variance components is
worthy of comment. Under the usual assumptions made in analysis of
variance, the existence of negative (population) components of variance
is not possible. Thus the occurrence of negative estimates of essentially
positive quantities is somewhat frustrating to the research worker. How-
ever, such estimates do arise occasionally, as the example has just shown.
Since negative estimates are annoying, it is customary to estimate the
unknown population parameter by zero when such a situation occurs.
T A B LXXII
~
Analysis of Variance-Mineral Content Data on Duplicate Determinations of 12
Samples’
Degrees Expected
of Mean square mean
Source of variation freedom Ca K Mg P square
Trees and samples 11 0.03887 0.06482 0.029687 0.000814
Ash duplicates 12 0.00737 0.00363 0.000706 0.000109 V s 4-2V.
Determinations 24 0.00421 0.00261 0.000438 0.000163 V s
Estimates of Variance Components
Ash duplicates (V.) 0.00158 0.00051 0.000134 0 .OOOOOOa
Determinations (Vd) 0.00421 0.00261 0.000438 0.000163
1 From Sharpe and Gammon (1951).
* Computed value waa negative.
This, of course, is a biased procedure, but it does lead to aesthetically

satisfying results.
2. Randomized Complete Block, Latin. Square, Split Plot, and Other Designs
In practice, research is not always conducted to produce data corre-
sponding to a completely randomized design. Often it will be possible t o
predict varying responses among groups of experimental units even when
they are subjected to the same treatment. Under such conditions it is
clear that if different treatments are t o be investigated they should not
be imposed on groups of experimental units which differ markedly, for
then any comparison among treatments would be confounded with a
comparison among groups.
If the experimental units are placed in groups (the units in a group
are assumed t o be fairly homogeneous) which reflect the researcher’s
judgment as to differential responses not attributable to the treatments
under investigation and then the various treatments are assigned at
random to the experimental units within each group the “treatment-
groups” confounding may be avoided. If the number of experimental
200 B. OSTLE AND ROBERT c). TISCHER
units in each group, or block, is equal to the number of treatments being

investigated, a randomized complete block design results.
The calculations for a randomized complete block design are not
difficult, proceeding as follows:
Z Y 2 = total sum of squares
22
b 1
Yij2- T2/bt
i=l j=1
Buu = block sum of squares
2
i-1
Bi2/t - T2/bt
Tu, = treatment sum of squares

Tj2/b - T2/bt
j-1
experimental error sum of squares
1Y2 - B,u - Tuu
where B, = total of all observations in the ithblock
Tj = total of all observations associated with the jth treatment
T = total of all observations
If the experiment is planned in more detail than a randomized com-
plete block design implies, it may be helpful t o consider such designs as
Latin squares, split plots, and others. These designs are not difficult to
analyze and may, in certain instances, prove profitable.
The computational techniques for all the designs that are available
may be found in texts on statistical methods or experimental design
(Cochran and Cox, 1950; Kempthorne, 1952; Anderson and Bancroft,
1952). Some of the examples to be presented later will be of the types
referred to, whereas others will be of more complex nature. It is hoped
that the examples, together with some reading in appropriate references,
will provide information concerning the uses (and limitations) of analysis
of variance techniques.
Doxtator (1937) investigated the quality of canned sweet corn, more
specifically the quality to be canned; data on puncture tests a t various
stages of maturity were obtained. Data were obtained in two different
years for a number of crosses; these were analyzed using analysis of vari-
ance methods based on a randomized complete block design. The results
were similar in both years, as may be seen in Table XXIII. The authors
noted that replicates were not statistically significant and remarked
that apparently soil variation among replicates was not such as to affect
STATISTICAL METHODS IN FOOD RESEARCH 20 1
resistance to puncture. The differences among crosses were clearly

significant.
Further experimentation (with 11 selected cultures: 10 crosses and
normal Golden Bantam) was carried out in 1931, 1932, and 1933, and
apparently these data were combined with the appropriate crosses for
1929 and 1930 to give the analysis of Table XXIV.
This procedure is ill-advised since no method exists for matching a
particular replicate number for one year with the same replicate number
in another year. This difficulty could have been avoided by extracting a
mean square for “replicates within years ” and using the interaction of
this term with cultures, as error (Table XXV).
TABLE XXIII
Analysis of Variance-Puncture Values on Golden Bantam Sweet Corn Crosses’
1929 1930
Degrees Degrees
of Mean of Mean
Source of variation freedom square freedom square
Replicates 2 191.4 2 511.4
Crosses 52 2947.9** 60 1436.6**
Exp. error 104 302.1 120 260.1
Total 158 182
1 From Doxtator (1937).
** Significant at 1 % level.
TABLE XXIV
Analysis of Variance-Puncture Tests of 11 Selected Sweet Corn Varieties and Crosses1
Degrees of Sum of Mean Standard
Variation due to freedom squares squares F error
Replicates 2 118 59 .26 -
Replicates X years 8 2680 335 1.45 -
Years 4 12072 3018 13.06** -
Cultures 10 43778 4378 i8.95** -
Cultures X years 40 49813 1245 5.39** -
Error 100 23059 23 1 - 3.9
Total 164 131520
1 From Doxtator (1937).
** Exceeds the 1 % level.
TABLEXXV
Alternative Analysis of Variance for Sweet Corn Puncture Test Data
Degrees of
Source of variation freedom
Years 4 (Years - 1)
Replicates within years 10 (Replicates - 1) X years
Cultures 10 (Cultures - 1)
Cultures X years 40 (4 X 10)
Replicates within years X culture 100 (10 x 10)
Total 164 (n - 1)
rn
0
TABLEXXVI N
Schematic Diagram of Split-Plot Design for Puncture Data on a Single Variety of Golden Bantam Corn'
Harvest date
1 2 3 4 5 6 7 8
1
Ears 1 2 3 411 2 3 411 2 3 4
-
Punctures
1
-
2
3
-
4
5
-
6
7
-
8
-
9
-
10
1 Derived from data of Doxtator (1937).

Incidentally, the use of one asterisk to indicate significance a t the 1%

point (as was originally done by this author) is misleading, since it is not
the accepted form. It is preferable to use * = significant at P = 0.05 and
** = significant a t P = 0.01, in agreement with common usage.
TABLEXXVII
Elements of Analysis of Variance for Corn Puncture Data'
Source of variation Degrees of freedom
Replicates ( R ) ( R - 1) = 2
Dates ( D ) ( D - 1) = 7
Ears ( E ) (E-1) = 3
Replicates X dates ( R - 1)(D - 1) = 14
Ears X dates ( E - l)(D - 1) = 21
Error (a) +
( R - l ) ( E - 1) ( R - - 1) = 48
Total between ears 95
Error (b) within ears ( P - l)P.E.D.= 864
A further study of the characteristics of the puncture test data was

directed toward a determination of the effect of sampling methods.
Table XXVI shows a schematic diagram of the type of design used.
The elements of the analysis of variance are shown in Table XXVII.
The analysis of variance for this split plot design is of special interest
because more than one error term is involved.
The analysis is presented in a rather peculiar order, since it is only a
standard split-plot design with subsampling of the split-plot experimental
units. A more reasonable presentation (at least from the point of view that
what is customary is usually more reasonable, if not for the more rigorous
reason of better pointing out the proper test procedures) would be the
following (Table XXVIII) :
TABLEXXVIII
Suggested Alternative Analysis of Variance for Sweet Corn Puncture Data'
Degrees
of Mean
Source of variation freedom square Expected mean square
Replicates 2 60.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Dates 7 1269.7upz + +
lOuiZ 4&2'+ ('29$)Zdj'
Error (a) 14 55.6upz 1&1* + + 400~'
Ears 3 1840.1upz + +
10~1' (2495)Zebz
Dates X ears 21 76.4 upz + 1 0 +~( 3 ~
% 1~
)ZZ(de)j~~
Error (b) 48 76.7 upz + 10~1'
Among punctures within ears 864 3 . 8 up¶
Total 959
Another example of the use of analysis of variance is provided by

Gough and Lantz (1950), who investigated the effect of variety and grow-
ing location on the niacin, thiamine, and riboflavin content of dried
beans. Owing to difficulties in obtaining samples at harvest time, separate
analyses were run to examine the effects of: (1) localities and varieties
TABLEXXIX
Mean Square Vitamin Contents for Eight Varieties of Dried Beans Grown at Three
Different Localities in New Mexico in 1949'
Degrees
of Mean squares
Source of variation freedom Niacin Thiamin Riboflavin
Total 23 ....... ...... .......
Localities 2 248.0** 31.4** 0.28**
Varieties 7 21 .o* 2.9* 0.039**
Interaction 14 2.2 0.79 0.005
I From Gough and Lantz (1960).
* Differenoes significant at the 6 % level.
** Differences significant at the 1 % level.
within each of three years, and (2) years and varieties for the same
locality. For example, Table XXIX summarizes the variety and location
differences for the year 1949, and Table XXX summarizes the year and
variety differences for one particular location. Location and variety
differences were significant (statistically) and the effect of years cannot
be altogether discounted. It is unfortunate that samples were not avail-
able for all, and the same, varieties and locations for each of the three
TABLEXXX
Mean Square Vitamin Values for Dried Beans Grown at Eming, New Mexico, in 1947,
1948, and 19491
Degrees
Of Mean squares
Source of variation freedom Niacin Thiamine Riboflavin
Total 23 ...... ..... ......
Years 2 28.4** 1.8** 0.005
Varieties 7 32.1** 1.6** 0.05**
Interaction 14 0.9 0.093 0.005
1 From Gough and Lantz (1960).
** Differences aignifioant at the 1 % level.
years (this is inferred from the authors' remarks), for a combined analysis
of variance might have yielded much useful information. For example,
the joint effect of the various factors could have been examined by con-
sidering the proper interaction mean squares.
Some interesting points concerning analyses of variance arise in an
article by Cone and Ashworth (1949), who examined several methods for
counting bacteria in reconstituted ,milk using the plate count. The vari-
ables investigated were the effect: (1) of four temperatures used in recon-
stituting, (2) of the time of holding the reconstituted sample prior t o
plating, (3) of prolonged mechanical shaking (2, 5, and 15 minutes) with
beads and without beads, (4) of four alkaline solutions and water at room
temperature compared with water a t 50” C. for reconstituting samples,
and ( 5 ) of five alkaline solutions, water, and Ringer’s solution a t 50” C.
in reconstituting samples. The data were presented as logarithms of
actual counts-a method which adds measurably to the readability of
this type of data.
The analysis of variance applied to each of these sets of treatments is
exemplified by the results from the first comparison (Table XXXI).
TABLEXXXI
Analysis of Variancc-Effect of Reconstitution Temperature of Milk Powder on Log10
Bacterial Plate Count1
Degrees of
Total 119 48.9668 .. . .... .
Powders 14 47.3932 . . . .... .
Duplicates 60 0.2388 0.0040
Treatments (temperatures) 3 0.6583 0.2194**
Powders X treatments 42 0.6765 0.0161
1 From Cone and Ashworth (1949).
** Significant a t P = 0.01.
The four treatments were the temperatures of reconstitution, and

these were found t o differ significantly a t the level P = 0.01. The authors
stated that this difference was “significant beyond the one percent con-
fidence level ”; the use of the word confidence ” in this manner should be
((
avoided. The F test used for determining significance is

treatments mean square - 0.2194
--=
13.6
F =
powders X treatments mean square 0.0161
The F value required for significance a t the 1% level with 3 and 42
degrees of freedom is F o . ~ ~ =( ~4.29.
, ~ ~Since
) 13.6 exceeds 4.29, the mean
square for treatments is significant a t the 1 % level.
Other characteristics of these results which could have been examined
are: (1) differences between duplicate counts a t each temperature and
within temperatures, and (2) the powders x counts interaction.
The analysis of variance would have been more easily understood if
the mean square for “Duplicates” were identified as the “among counts
within subclasses” mean square or as the “among counts on experimental
units treated alike” mean square. This is an “individuals treated alike’’
mean square derived from the sixty pairs of duplicate readings. Each pair
of readings constitutes one subclass with (n - 1) = 1 degree of freedom.
The results might better have been presented as in Table XXXII.
TABLE XXXII
Alternative Analysis of Variance for Bacterial Plate Count Data1
Degrees of
Powders 14 47.3932
Treatments (temps.) 3 0.6583 0.2194**
Powders X treatments 42 0.6765 0.0161
Among count8 on experimental units
treated alike 60 0.2388 0.0040
Total 119 48.9668
1 Derived from data of Cone and Ashworth (1940).
** Signifioant at P = 0.01.
The approach to the problem encountered apparently was a step by

step attempt designed to find the best procedure in each group, more or
less independently. This procedure has the distinct advantage of sim-
plicity and economy, as only a relatively small number of the total com-
binations in the reconstitution and counting procedure are actually tested.
It has the disadvantage that the results of one set of tests may well have
an important bearing on those of another set-a fact which would not
always be disclosed in a step by step procedure.
Another method of attacking this problem would have been t o in-
corporate all the sets of conditions in a single design. For example, if
fifteen powders were used as replicates, the design might include:
1. 15 powders.
2. 4 temperatures of reconstitution.
3. 4 times of holding the reconstituted sample.
4. 2 methods of shaking.
5 . 3 times of shaking.
6. 7 reconstitution liquids.
All combinations of these variables would be
(15 X 4 X 4 X 2 X 3 X 7) = 10,080 determinations
as compared with 321 determinations (ignoring duplicate determinations)
completed in the work reported. It is usually impractical to make as
many as 10,080 determinations; this is probably the main reason why
complete factorial designs are not often applied to an experimental
situation of this kind.
An alternative procedure would be the use of two levels of each
variate in a 26 design involving only 64 determinations used as a trial run
t o define the importance of each variable as well as the degree of inter-
action among the variables.
AS the number of levels of each variable is increased, the amount of
information provided about the factors is increased but unfortunately, so
also is the complexity of the design. If three levels of the six variables
were employed (a 3" factorial) the result would be 729 combinations
(determinations) containing information similar to the 26 design but of a
more extensive nature. If a 46 design were desired, it would probably be
wise t o use fractional replication t o reduce the labor and expense neces-
sary to complete the work. Fractionally replicated designs have been
discussed by Finney (1945), Fisher (1947), Kempthorne (1947, 1952), and
Cochran and Cox (1950).
Fractionally replicated designs find use in the examination of a rela-
tively large number of factors where it is known (or assumed) that
high-order interactions are nonexistent or negligible. Under these condi-
tions it may be shown that an appropriate fraction of the total number
of determinations indicated in the complete replicate may be used t o
yield information about the main treatment effects and about the two
factor interactions. This procedure allows for a survey of the effects of all
factors in a single experiment which usually involves less effort than any
complete factorial design involving the same treatments.
Tischer and Kempthorne (1951) used a one-ninth replicate of a 3'
factorial in an investigation of the influence of variations in technique and
environment on the determination of consistency of canned sweet corn.
Through use of this design only 243 of the total of 2187 determinations in
a single replication were carried out. The statistical procedure by which
the experimental results were evaluated was that of the analysis of vari-
ance. The resulting analysis is shown in Table XXXIII except that the
sums of squares for nonsignificant two-factor interactions are not pre-
sented separately. The existence of effects or interactions is judged by the
criterion of statistical significance.
Table XXXIII indicates that temperature of corn, degree of con-
sistency, time of spreading, angle of cone, and surface of cone have highly
significant effects on consistency readings. Interaction between tem-
perature and consistency and between angle and surface of cone were also
shown t o be significant. In the present case the experiment achieved a
very high degree of sensitivity because of the large number of observations.
Verification of the validity of the experimental design used was
achieved by obtaining an estimate of the error of repeated observations
on the same lots of corn. This was found t o be 1.46, which is very little
different from the error variance of the experiment of 1.57. The similarity
of these two error variances demonstrates that high-order interactions are
208 B. OSTLE AND ROBERT c). TISCHER
trivial; this was the basis of the fractional replication design. There was
some evidence of a higher variance for the readings between cans on
samples with thin consistency than for those with thick consistency. It
was felt, however, that this could be ignored in applying the analysis of
variance.
TABLEXXXIII
Analysis of Variance-Effects of Seven Variables on Consistency of Canned Sweet
Corn'
Degrees of Sum of Mean
Source of variation freedom squares square F
Block 8 38.98 ....... .........
Error 136 213.46 1.57 .........
Treatments 98 7547.89 ....... .........
A. Rate of lifting 2 6.33 ....... .........
B. Temperature of corn 2 577.93 288.96 184.05**
C. Degrees of consistency 2 6688.31 3344.16 2130.04**
D. Time of spreading 2 98.40 49.20 31.34**
E. Methods of cleaning 2 1.05 ....... .........
F. Angle of cone 2 17.28 8.64 5.50**
G. Surface of cone 2 16.79 8.39 5.34**
Two-factor interactions 84 141.80 1.69 1.10
BXC 4 20.23 5.06 3.22*
E X G 4 13.38 3.34 2.13
F X G 4 18.96 4.74 3.02*
Total 242 7800.33 ....... .........
1 From Tischer and Kempthorne (1961).
* Denotes significance at the 6 % level.
** Denotes significance at the 1%level.
One should remember that the statistical significance of an effect is
not related t o its practical importance. The practical importance of the
effects of the factors tested may be observed in Table XXXIV in which
the mean values for each of the factors is given for each level. For exam-
ple, the mean reading with 10 seconds allowed for spreading was 7.2, with
20 seconds 7.4, and with 30 seconds 7.6. There was a consistent increase
in readings with increase in time of spreading, so that in routine use the
time of spreading should be specified. The use of 10 seconds rather than
30 seconds will result in a lowering of the reading by approximately 0.4
inch. Variation of approximately 0.2 inch due to rate of lifting, method of
cleaning, angle of cone, and surface of cone over the ranges investigated
indicates that changes in these factors probably will produce little
variation in the final readings.
The effect of the temperature of the corn sample was very pronounced,
in accordance with general knowledge. Account of this fact should be
taken when one set of observations is compared with another. If readings
a t 70" F. (21' C.) are designated as the standard of comparison, readings

taken at 40" F. (4"C.) should have 0.3 inch added, and readings taken a t
100" F. (37" C.) should have 0.6 inch subtracted in order t o standardize
them.
The estimate of interaction of temperature and consistency is shown
in Table XXXV. Examination of the means does not indicate a strong
trend and the interaction, although statistically significant, is of little
practical importance. The interaction arises because a change of tem-
TABLE XXXIV
Estimates of Effects of the Seven Variables on Consistency of Canned Corn'
Variable Level of variable Mean reading, inches
Rate of lifting, sec. 0.34 7.4
1.0 7.5
3.0 7.4
Temperature of corn, "F. 40 7.0
70 7.3
100 7.9
Consistency Thin 9.2
Medium 7.0
Thick 6.1
Time of spreading, scc. 10 7.2
20 7.4
30 7.6
Method of cleaning Cold water 7.4
Hot water 7.4
Hot water plus detergent 7.4
Angle of cone 2'43" 7.3
5"43" 7.4
8'43" 7.5
Surface of cone, sq. cm. 200 7.5
250 7.4
300 7.3
TABLE XXXV
Interaction of Temperature and Degree of Consistency in the Determination of
Consistency in Canned Corn'
Differences in readings
between pairs of test
Temperature ("F.) temperatures ( O F . )
Consistency 40 70 100 Mean 70-40 100-70 100-40
Thin 8.7 9.2 9.7 9.2 0.5 0.5 1.0
Medium 6.6 6.9 7.4 7.0 0.3 0.5 0.8
Thick 5.7 5.8 6.7 6.1 0.1 0.9 1.0
Mean 7.0 7.3 7.9
perature from 40" to 70" F. has little effect on thick corn and a compara-
tively large effect on thin corn.
V. REGRESSION
TECHNIQUES
Regression is a term used by statisticians t o describe the relationship
between two or more variables. It may therefore be considered as a
description of the form of the interrelation among the variables, or how
one variable changes in value as the others are permitted to vary. In
simple cases, the relationship among the variables may become evident
without the aid of formal analytical techniques. Much more common,
however, is the problem of deciding with some certainty if, and to what
degree, the variables are related.
1. Linear Regression
a. One Independent Variable. It is convenient to begin with linear
(straight line) regressions and, as the simplest case, those involving only
Y
FIQ.4. Example of simple linear regression where the two variables are perfectly
related (from hypothetical data).
one independent variable. To make the presentation of the ideas under-
lying regression easier to assimilate, consider the diagram of Fig. 4. If
all points lie on the line (as shown), then Y (the dependent variable) is
perfectly related to X (the independent variable). Further, for each unit
change in X , there is a change of fl in the value of Y . The relationship
between X and Y is given by the equation:
y=ar+px
where
d
,6 = tan 6 = - ( Y )
ax
is known as the regression coefficient in the population and LY is the
population Y-intercept.
I n ordinary experimental work, however, the line relating X and Y
would be derived from data in which values of Y were observed repeatedly
for $zed values of X . In this case, the ( X , Y ) points representing the
observations might appear as shown in Fig. 5, commonly termed a scatter

graph or scatter diagram. This diagram shows the Y values obtained for
each value of X . Further, it gives some indication of the distribution of
Y values corresponding to each value of X , and a straight line drawn
approximately through the mean Y values (one for each.X value) pro-
vides a linear estimate of the relationship between the two variables.
A different presentation of this illustration is shown in Fig. 6, where
the values of Y are presented in the form of normal distributions for three
sets of observations (corresponding to XI, X p ,and X , ) . Now if a set of Y
values were observed for each of 20 X values, the data might be repre-
sented as they are in Fig. 7, where each dot represents not a single Y
Y
FIQ.5. Scatter diagram of a linear regression showing distributions of Y values for

fixed X values (from hypothetical data).
observation at each value of X but instead is the mean of all Y values for
each value of X .
It is evident that a line may be fitted visually to the points in Fig. 7
so that it passes among the points leaving approximately half the points
on one side and half on the other. However, it is equally evident that
the degree of precision with which the line is fitted depends a t least upon
the acuity of the operator, the number of points, and their placement.
In the statistical approach, this difficulty is avoided through use of the
method of least squares. By this means a straight line may be fitted t o any
set of points relating two (or more) variables so that the sum of the
squared deviations of Y (measured from the fitted line) will be a mini-
mum. This is accomplished by equating to zero the partial derivative
with respect to each parameter. Solution of the resulting equations yields
expressions for the slope and the intercept in terms of the observed X’s,
Y’s, and n, the number of observations. Derivations may be found in
FIG.6. Diagram of a linear regression showing Y distribution curves for fixed x

values (from hypothetical data).
I
I
I
II
I
I
I
0 . I I
I
I
I
I
I
I
I
I
I
I
I
I
FIQ.7. Diagram of a linear regression where each dot represents the mean of all Y
values associated with a given X value (from hypothetical data).
Hoe1 (1947), Ostle (1954), Kendall (1946), Mood (1950), and Cram&
(1946).
The estimated slope is usually calculated as:
fi = b = Zxy/Zx2
- ZXY - (ZX)(ZY)/n
-
2x2 - ( Z X ) 2 / n
and the estimated intercept is
&=a=F-bx
If piis the actual value of the P observation and pi is the estimated

value, the sum of the squares of the deviations about the estimated
regression line is
i=l
2 ( Y 1- Pi)*
where
and an estimate of the variance of the deviations about the regression

line is given by
n
The hypotheses of most interest which may be tested when simple

linear regression has been employed are the following:
1. H:P = Po
that is, the true slope of the regression line equals some specified value,
00.Common values of Po are zero and unity.
that is, the value of Y when X is zero is equal to some specified value, (YO.
A common value of (YO is zero.

These two hypotheses may be tested by computing, respectively:
with (n - 2) d.f., and
2. t = (a - (Yo)/S,
with (n - 2) d.f. where
and
4
i=l
We might note that, if Po = 0, this hypothesis may be tested using the

methods of analysis of variance (see Table XXXVI) and computing:
mean square due to regression
F =
mean square of deviations about regression
TABLE XXXVI
Simple Form of Analysis of Variance for Linear Regression
Degrees of
Due to regression 1 r2Zyl r2.Zya
Deviation about regression n-2 (1 - r2)2y2 (1 - r2)Zy*/(n - 2)
Total n-1
It should be clear, of course, that
and
where
In statistical terminology, r is referred to as the sample coefficient of

correlation between X and Y , and it yields a measure of the degree of
linear association existing (in the sample) between the two variables.
An example of the application of this technique is afforded in a simple
set of values produced in research on a titration method for determination
of moisture in sweet corn (Tischer, 1952).
Twenty-one moisture determinations were made on aliquots of a
single sample of corn a t temperatures ranging from 61' to 90' F. (Table
XXXVII). It was necessary to discover whether the titration values
presented were significantly related to temperature.
TABLE XXXVII
Titration ( Y ) and Temperature ( X ) Values in a Determination of Moisture in Sweet
Corn1
Y X
4.18 61
4.21 62
4.12 61
4.35 66
4.25 64
4.36 66
4.44 70
4.54 70
4.44 71
4.74 76
4.61 74
4.64 75
4.78 78
4.88 78
4.82 79
5.38 86
5.09 86
5.02 85
5.37 90
5.29 88
5.38 88
7 = 4.709 8 = 74.95
Zy2 = 3.5893 Zza = 1850.9524
Zzy = 78.9990
B= 0.0427
6 = 1.5100
1 From Tisoher (1952).
The analysis of variance is presented in Table XXXVIII where:

Sum of squares due to regression =
-
-
Mean square -
Sum of squares about regression =
-
- 1850.9524 (3.5893) - (78.9990) -
-
1850.9524
-
--0.2176
Mean square
n - 2
--- =
0*2176 0.0114
19
3.3716
Fcl,le, = -= 295.75
0.0114
The tabular value of F = 8.18 for P = 0.01 and the regression of
titration value on temperature is highly significant. From this it may be
concluded that values obtained in this titration must be corrected for
temperature. The regression equation P = 1.5100 0.0427X yields +
the amount of correction to be applied in each case.
TABLEXXXVIII
Analysis o€ Variance Due to and about Regression of Temperature on Moisture
Titration Value for Sweet Corn*
Degrees of
Due to regression 1 3.3716 3.3716
Deviation about regression 19 0.2176 0.0114
Total 20 3.5893
1 From Tischer (1962).
Another example involving both analysis of variance and linear

regression is provided by Geise et al. (1951), who used three methods for
measuring the moisture content of sweet corn for canning. These were the
Steinlite moisture tester, the Brown-Duvel moisture tester, and the
vacuum oven. Five varieties of corn were tested a t moisture levels ranging
from 62 to 79%.
The mean moisture contents with the three methods on 51 samples
ranging from 62% to 79% moisture and taken from five varieties of sweet
corn are given in Table XXXIX. Each sample mean is the average of
three determinations. The average percentage moisture of all samples
70.4 with the vacuum oven, 70.0 with the Brown-Duvel tester, and 70.5
with the Steinlite tester.
For the purposes of comparing the three methods, the vacuum oven
was used as the standard. Analyses of variance of the differences in
moisture determinations between the vacuum oven and the Brown-
Duvel tester and between the vacuum oven and the Steinlite tester are
given in Table XL. A test for the equivalence of two scales in measuring
differences among treatments given by Cochran (1943) is the ratio of the
pooled mean square for varieties and mean and the mean square for error.
This test indicates that the Brown-Duvel tester and the vacuum oven
are equivalent in measuring the differences in moisture content among the
TABLE XXXIX
Means of Triplicate Determinations of Moisture Content of Sweet Corn Using Five
Varieties and Three Methods'
Moisture Content, %
Variety Vacuum oven Brown-Duvel Steinlite
Golden Cross 72.8 73.2 72.5
72.5 67.8 71.4
74.0 76.4 73.5
70.5 69.8 70.8
70.9 69.5 70.7
67.9 68.1 68.4
68.9 67.6 69.0
67.9 66.2 67.5
66.4 63.7 67.8
66.7 65.8 68.3
62.8 66.8 65.2
Golden Glory 71.0 71.2 72.4
70.4 68.7 69.8
69.2 68.3 69.6
69.2 68.6 69.7
67.6 67.5 68.7
66.1 69.4 67.3
65.8 66.2 66.2
65.9 64.2 67.5
65.1 65.0 66.8
63.3 60.5 65.4
Tendermost 78.1 75.0 75.1
78.5 77.5 77.6
75.1 78.3 74.8
74.6 75.2 74.2
74.8 75.2 73.8
72.5 71.0 71.0
72.0 73.7 70.8
72.5 71.9 71.9
69.4 69.3 70.6
69.9 69.6 68.9
68.4 67.2 68.6
67.4 67.0 68.7
Victory Golden 76.9 75.1 75.5
77.3 77.6 75.7
73.2 74.3 73.1
71.7 68.6 71.5
71.8 71.6 71.2
68.7 69.6 69.2
68.3 69.8 69.3
66.2 65.4 67.9
Iochief 78.2 77.5 75.9
77.3 75.4 75.1
72.8 70.3 71.6
218 B . OSTLE AND ROBERT G. TISCHER
TABLEXXXIX (Continued)
Moisture Content, %
Variety Vacuum oven Brown-Duvel Steinlite
Iochief (Continued) 72.7 70.8 72.4
71.2 72.3 71.0
68.4 68.4 69.2
68.7 67.0 69.3
67.2 67.2 67.1
66.4 65.5 67.8
66.1 66.4 67.4
Instrument mean 70.4 70.0 70.5
1 From Geise et al. (1951).
varieties. The mean difference between the two methods was not quite
significant at the 5 % level as measured by the ratio of the mean square
for mean and the mean square for error. Additional evidence based on the
linear regression of the 51 mean determinations by the Brown-Duvel
70
Z
Y I
Y, =3.3+0.95 x
60 62 64 66 60 70 72 74 16 70 0
VACUUM OVEN MOISTURE (PER CENT)
FIG.8. Regression of Brown-Duvel on vacuum oven moisture values (from Geise

et al., 1951).
tester on those by the vacuum oven indicates that the two are equivalent
in measuring the moisture percentage in sweet corn that has been frozen.
+
This regression is Y 1= 3.3 0.95X (see Fig. 8). If the two are equivalent
the Y-intercept should equal zero, the regression coefficient should equal
TABLEXL
Analyses of Variance-Moisture Values between Methods and among Varieties for
Sweet Corn'
Vacuum oven-Brown-Duvel, Vacuum oven-Steinlite
Source of variation mean square mean square
Mean 1 10.645 0.201
Varieties 4 0.724 3.930
Error 46 2.975 1.215
1 From Geise et al. (1951).
one, and neither of the observed values in the above equation differs
significantly from the expected.
The analysis of variance of the differences between the determinations
made with the vacuum oven and the Steinlite indicates th a t the two
methods are not equivalent and that they do not differ b y a constant,
70
76 -
:
W
74-
0
Y, = 18.37 t 0.74 X
260
&6 60 62 64 66 68 70 12 74 76 70 0(
FIG. 9. Regression of Steinlite on vacuum oven moisture determinations (from

Geise et al., 1951).
even though the means of all determinations by the two methods are
almost identical. Th at the determinations are not the same on the indi-
vidual samples for the two methods is also shown by the regression
+
Y t = 18.37 0.74X (see Fig. 9). The value 18.37 differs significantly
(at the 1% level) from 0, the regression coefficient differs significantly
from 1, and there is sufficient evidence to conclude th a t the regression is
linear. The differences between the vacuum oven and Steinlite and be-
tween the vacuum oven and the Brown-Duvel are plotted against the
determinations by the vacuum oven in Figs. 10 and 11.
0 vs;s.~~i~i-f--
.&&
- “drn/*. v.o.-v.o.
,r-6-*
*:* . 3
0
0
W I I I I I I I I
0
62 64 66 60 70 72 74 76 78 8
Fro. 10. Regression of (vacuum oven-Steinlite) on vacuum oven readings showing

how the difference between the two methods varies over the range 62-78% moisture
(from Geise et al., 1951).
FIQ. 11. Regression of (vacuum oven-Brown-Duvel) on vacuum oven showing

wide variations between the two methods over the range from 62-78% moisture (from
Geise et al., 1951).
The scatter diagram shown in Fig. 10 is a comparison of the results
obtained with the vacuum oven with deviations of the Steinlite results
from those of the vacuum oven. It should be apparent from this figure
that use of the vacuum oven as a standard of comparison yields results
which are not identical with those derived through use of the Steinlite
instrument. Within the range of 68 to 73% moisture as determined by
the vacuum oven, deviations are generally within 1 %. When the range is
extended to include 62 to 7975, the deviations increase to approximately
3 %.
Observation of Fig. 11 indicates that the deviations of the Brown-
Duvel determinations from those of the vacuum oven are consistently
large over the entire moisture range tested.
Actual corrections for either the Steinlite or the Brown-Duvel could
be made from the regression lines (Figs, 8 and 9) or perhaps more con-
veniently from a chart containing the same information. That is, what
would be desired is an estimate of what reading would have been obtained
from a particular sample had the vacuum oven been employed rather
than either the Steinlite or Brown-Duvel instruments. Assuming one of
the last two instruments had been used (owing to economy of time and/or
cost and a reading obtained), what is the equivalent reading that would
have been realized had the vacuum oven been utilized? This could be
read from the graph, or from tables as mentioned, but it is evident that
at best this estimate must be of an interval type. Reference may be made
to Eisenhart (1939) and to Ostle (1954) for a discussion of appropriate
formulas when estimating X from Y having calculated a regression of
Y on X .
Cook el al. (1934) also made a comparison of three methods for
measuring the moisture content of grain. It was found that the air oven
(130" C.), the Brown-Duvel apparatus, and the vacuum oven yielded
different results, the magnitude of the difference depending upon the level
of moisture content. In the comparison of the two methods used for
grinding samples of grain and in subsequent comparisons the results were
plotted in a single scatter diagram (Fig. 12). A diagonal line extending
from the origin to 25% moisture on each coordinate was included t o
provide a picture of the ideal situation ( k where
, each method gives the
same result). Finally linear regressions were calculated for each subgroup
corresponding to the two methods of grinding to demonstrate its influ-
ence. These were tabulated (Table XLI) with the standard error of dupli-
cates and the standard error of prediction. Similar data were tabulated to
illustrate comparisons among several types of grain. Inferences drawn
from these comparisons were largely based on differences in standard
errors.
The linear regressions do not appear to have been tested to indicate
whether significant differences exist owing to the influence of methods of
grinding or t o the types of grain compared. That is, the hypotheses
H 1 : a = 0 and H2:P = 1 should have been tested as was done by Geise
et al. (1951). Had these comparisons been made it might have been
possible to: (1) demonstrate more clearly the influence of method of
grinding and type of grain or (2) in the event that significant differences
did not exist among regressions within a subgroup, to pool the estimates
ua 200
ci
- 180
2 160
W
a
3 140
c
u,
p 120
a
Q
I 00
100 120 140 160 180 200 220 240
% MOISTURE BY VACUUM OVEN
FIG.12. Moisture content of hard red spring wheat by 130°C. air oven and vacuum
oven methods (from Cook et al., 1934).
to obtain an improved comparison of the methods used in measuring

moisture content.
TABLEXLI
Standard Error of Duplicates, Linear Correction Equations and Observed and Net
Standard Errors of Prediction of Vacuum Oven Percentage Moisture Content of
Hard Red Spring Wheat by the 130" C. Air Oven Using Two Grinding Methods1
Standard Standard error
Grinder error of of prediction
employed duplicates Correction equations Observed Net
Hobart 0.069 Mv = -0.644 +
1.066M- 0.31 0.30
Wiley 0.050 Mv = -0,926 f 1.095Ma 0.25 0.24
1 From Cook et al. (1934).
M , = moisture content as determined by vacuum oven.
M . = moisture content a8 determined by air oven.
Standard errors apply to mean of duplicates.
Porter et al. (1947) investigated the apparent increase in concentration

of carotene in vegetables after cooking and found that the fresh and
cooked values were correlated. As a result, regressions were calculated
using the formula:
b = ZXY/ZX2
where Y = mg. of carotene/100 g. dry weight of fresh spinach and

X = mg. of carotene/100 g. dry weight of cooked spinach from the same
lot. The regression shown in Fig. 13a has a slope of 0.761 and a standard
error of estimate of 5.2 mg. It was found th at similar comparisons made
10 -
5 60:
$50'
\ -
p0 4 0 --
30-
2Q I I I l~i l l l l l ' ' ' ' ' '
2b 30 40 50 60 70 80 90 100
mgms/100g -COOKED
FIG.13a. Regression of carotene concentration in milligrams per 100 g. dry weight

of raw and cooked spinach. ( X represents actual values obtained.) (From Porter
et al., 1947.)
2" 1'8 ' 1 " 1 ' " '

2 3 4 5 6 7 8
mgmr /IOOg-COOKED
FIG. 13b. Regression of carotene conccntration in milligrams per 100 g. moist

weight of raw and cooked spinach. ( X represents actual values obtained.) (From
Porter et al., 1947.)
on a moist weight basis (Fig. 13b) yielded a slope of 0.897 and a stand-
ard error of estimate of 0.706 mg. It should be pointed out th a t both of
these estimates were made by calculating:
b = slope = Z X Y / Z X 2
that is, the authors have postulated a regression line passing through the
origin.
In the equation for a straight line where:

P=a+bX
the intercept may have any value and this value must be calculated to
obtain the value of Y corresponding to X . In the event that intuition or
experience indicates that a decrease of one variable to zero should be
accompanied by a corresponding decrease in the other, then it may be
logical to use a simple slope function to calculate the regression, assuming
that the intercept is actually zero. However, in view of the small amount
of additional labor necessary to release this restriction and the advan-
tages obtained, it seems advisable to estimate both the slope and the
intercept, especially in the range of X under consideration.
In Table XLII are shown regression coefficients for raw vs. cooked
carotene contents for four vegetables, spinach, green beans, Fordhook
chard, and rhubarb chard. Details of the experiment are contained in the
table. Two regression coefficients were calculated for each set of replica-
tions. Calculation of the regression for fresh value on cooked is based on
the assumption that the cooked values were measured without error, and
the regression of cooked values on fresh is based on the assumption that
the fresh values are measured without error. One or the other of these
assumptions is often convenient to make, even though neither is entirely
correct.
If neither variable may be considered without error, some information
may be gained by computing the correlation coefficient. With the use of
the data in Table XLII this is easily accomplished since:
Correlation coefficient = T
or in other words, the correlation coefficient is the square root of the

product of the slope of Y on X and the slope of X on Y . For example, for
spinach, using the first two regression coefficients in the table,
biz = 0.761
bz,
~~
= 1.30 and
T = d 0 . 7 6 1 X 1.30
=m
= 0.995
and the change in carotene content in cooked spinach is highly correlated

with a similar change in the fresh product.
TAEILE XLII
Factors for Adjusting Carotene Values of Processed Vegetables'
Carotene Carotene
before after Regression
processing processing coefficient
No. of per 1OOg. per 100 g. Fresh Cooked
repli- dry wt. dry wt. value on value on
Vegetable cationsa (mg.) Type of processing (mg.) cooked fresh
Spinach (local market) 22 53.2 400 g. cooked in twice its weight of boiling water 69.6 0.761 1.30
3!3
Green beans (Lad- 4 2.94 Steam blanched, 174" F. for 2.5 min. 3.23 0.905 1.10
c)
F
*
rettes stringless green 4 2.93 Steam blanched, 208" F. for 3.75 min. 3.26 0.907 1.09 z
pod) 8 3.41 Canned-boiled 10 min. 3.68 0.928 1.07
8 2.90 Frozen-boiled 12 min. 3.25 0.891 1.12 8
-__ U
u)
Fordhook chard 4 62.9 (1) 6 g. NaCl added, then cooked in twice its weight of 66.0 0.942 1.05
boiling water z
4 62.9 (2) 6 g. NaCl added, then cooked in water clinging to 61.2 1.03 0.973 2
leaves after last rinsing 0
4 62.1 (3) 8 lb. cooked in steam-jacketed kettle (9 g. NaCl 69.0 0.893 1.11
added after cooking), held on steam table 10 min. z
M
4 62.1 (4) Same as (3),held on steam table for 1hour 62.1 0.983 1.00 m
-____-__ P
Rhubarb chard 4 57.1 Same as Fordhook (1) 70.S3 0.802 1.24
4 57.1 Same as Fordhook (2) 60.6 0.938 1.06
4 59.6 Same as Fordhook (3) 74.13 0.784 1.26
4 58.6 Same as Fordhook (4) 72.1' 0.811 1.23
1 From Porter et al. (1947).
1 Eaah replication is the average of triplicate determinations except the spinach which is in duplicate.
a The probability that this is a chance variation from the fresh value is 1/100. N
4 The probability that this is a chance variation from the fresh value is 1/20. N
cn
Hlynka et al. (1949) made a study of ten electrical meters designed for
determination of the moisture content of wheat. Table X L I I I shows the
regression equations of moisture estimates obtained with each meter on
the vacuum oven estimate and the standard errors of estimate. It is
apparent that considerable difference exists in the regression character-
istics of the meters. The slopes range from 0.106 t o 2.47 and the intercepts
TABLEXLIII
Calibration Equations and Standard Errors of Estimate for Electrical Moisture
Meters'
Standard
error of
estimate
Meter Unit of measurement Calibration equation* % moisture
Tag-Heppenstall 10-log of resistance, ohms Y = 2.47X + 4.54 0.23
Universal 10-log of resistance, ohms Y = 2.24X + 6.37 0.28
Tag-Dielectric Meter reading Y = 0.109X + 3.62 0.35
G.R.L. Meter reading Y = 0.132X + 10.07 0.36
Patterson Meter reading Y = 0.186X + 7.29 0.41
Steinlite Meter reading Y = 0.106X + 11.09 0.44
Mullard Meter reading Y = 0.876X + 10.88 0.48
N.P.L.
Switch Pos. 1 Meter reading Y = 0.116X + 14.57 0.50
N.P.L.
Switch Pos. 3 Meter reading Y 0.131X + 9.25
= 0.49
1 From
2 Y
X
-
=
Hlynka et al. (1949).
vacuum oven moisture.
unit of measurement.
from 3.62 to 14.57. The standard errors of estimate vary from 0.23 t o
0.50. The two meters exhibiting the lowest standard error were con-
sidered to be the most desirable. This comparison might have been
improved by applying a test of linearity to the regressions, since any
marked deviation from linearity would affect the precision of the results
so that the errors of estimation would change from one moisture level t o
another.
Toepfer et al. (1946) have made use of regression in relating the
sterilizing value of a process with its length. Process value (Fo)
was related
t o process time (minutes) in a linear regression. A minimum safe process
line was constructed a t a distance 2.6 times the standard error below the
regression line which defines the probability (P = 0.005) that any indi-
vidual container might yield an Fo value below this minimum.
More properly, the minimum safe process limit of this regression
might have been derived from:
STATISTICAL METHODS I N FOOD R E S EA RCH 227
where a is the probability th at an individual container might yield a n F o

value below L.
The results of this computation are dependent on X,; as Xo changes,
the value for L will describe a curve similar t o that shown in Fig. 14. The
deviation of L from P (the estimated regression value) increases as the
0
0 10 2
PROCESS TIME (MINUTES)
FIG. 14. Regression of process value on process time for 240°F. processes of snap
beans in pint jars; parallel constructed a t 2.6 times the standard error of estimate
below the regression line (calculated from data of Toepfer et al., 1946).
X Ovalues diverge from the mean (a,8 ) . This method is useful for esti-
mating the precision with which an individual value may be estimated.
Use of a standard error estimate yields instead the average lower limit
associated with all values for a given probability. I n this case only the
negative half of the confidence band was used because interest is confined
t o calculating the probability (or risk) that a process value ( F o ) might be
low enough t o allow the formation of toxins in the food product being
processed.
An interesting example of the use of the regression technique was
made by Paul et al. (1952) in work on the relationship of sulfur content
to sweet corn maturity. Five varieties of corn were harvested on four
dates, differing with each variety, and determinations were made of

sulfur as methionine and cystine.
Regressions for sulfur content and harvest date were calculated
(Table XLIV) for each variety. The results show that the slopes of the
regression lines are nearly zero, indicating only a slight effect for date of
harvest on sulfur content. By interpolation of these regression lines it was
possible to derive data equally spaced with respect to harvest date-an
accomplishment which was not practical in the original determinations.
These derived data were used in an analysis of variance which showed
that sulfur content (on a dry basis) did not change with maturity. It is
well known that the per cent of solids in sweet corn increases with
maturity, and when sulfur content was expressed on a wet basis, it was
found to increase with increase in total solids.
Regressions of per cent total solids on date of harvest were calculated
t o yield the best estimate of total solids for each variety for each (original)
harvest date. These values are approximately equivalent to the com-
plement of the per cent moisture. They differ from the true complement
because they are derived data and are all located exactly on the regression
line.
Since no significant differences were found in sulfur content within a
variety due to harvest date, these estimates were pooled for each variety
and used to calculate the amount of sulfur on a wet basis. These values
were then used to’calculate regressions for methionine, cystine, and total
sulfur for each variety (Table XIJV). In contrast with the regressions
calculated on a dry basis, all of these regressions have positive slopes,
indicating that a definite trend exists between sulfur expressed on a wet
basis and maturity of sweet corn.
A subsequent analysis of variance on the “wet basis” data indicated
that differences among maturity dates were significant at P = 0.05
except for the Golden Glory variety. The failure of this variety to exhibit
significant differences may be attributed to excessive variation in the
individual determinations.
Gangstad and Snell (1943) have made detailed use of regression tech-
niques in a study of the texture of sweet corn. Three standard varieties,
one experimental variety, and the parent inbreds were tested with a
puncture meter to measure resistance (in grams) in relation to dry weight
and maturity (days after pollination). Dry weight is a reliable index of
maturity for sweet corn, and on this basis it was postulated that puncture
resistance should be a valid measure of some function of dry weight and
days after pollination (maturity).
Plots of puncture resistance against the product of dry weight and
days after pollination for ’each variety suggested that linear regressions
%+
*
5
8
TABLEXLIV
Regressions of Per Cent Sulfur on Date of Harvest for Sweet Corn'
L
( X = date of harvest; Y = per cent sulfur calculated on a wet basis) E
Variety Methionine Cystine Total :
0
GG Y = 0.0125 + 0.000457X +
Y = 0.0031 0.000054X +
Y = 0.0155 0.000514X
GX Y = 0.0147 + 0.000179X +
Y = 0.0006 0.000500X +
Y = 0.0153 0.000679X 5
VG Y = 0.0072 + 0.000611X +
Y = 0,0001 0.000325X +
Y = 0.0072 0.000926X
T Y = 0.0068 + 0.000520X +
Y = 0.0021 0.000331X +
Y = 0.0090 0.000849X 2
I Y = 0.0073 + 0.000410X +
Y = 0.0237 0.000291X +
Y = 0.0108 0.000701X $
I From Paul et d.(1952). !z
#m
*
might be useful in describing these relationships. The regressions, are

shown in Table XLV; they were tested for significance of differences
between regression coefficients and for significance in residual differences
between varieties. They were found to be significantly different a t the 1%
level in each case.
Significant differences among the regression coefficients indicated that
these coefficients were probably not all drawn from the same population.
This technique does not serve t o distinguish the degree of difference of one
TABLE XLV
Regression Equations of Maturity on Puncture Values for Sweet Corn Inbreds and
Hybrids'
Inbreds
C6 +
Y = 0 . 8 3 0.918X
C13 +
Y = 3.60 0.580X
P39 +
Y = 1.01 0.401X
143Y +
Y = 1.28 0.780X
P51B Y = 2 . 1 0 +0.281X
Hybrids
C13 X C6 +
Y = 2.24 0.828X
+
Y = 2.20 0.745Xt
P39 X C13 Y = 2.40 + 0.481X

Y = 2.31+ 0.491Xt
P39 x P43Y Y = 1.00 + 0.574X
Y = 1.14+ 0.590Xt
P39 X P51B Y = 1.73 + 0.389X
Y = 1.55 + 0.341Xt
1From Gangstad and Snell (1943).
The method of least squares was used to obtain the linear regression equation:
where is the mean of puncture resistance, Y = (? - b%) + b X .
b is the regression coefficient.
x is the mean of maturity taken as the product of the days after pollination and % dry weight.
Not corrected for class interval.
t Theoretical equation obtained by simultaneous solution of inbred equations.
or more regression coefficients compared with others-a result which

might be obtained by means of orthogonal comparisons. These are com-
parisons between which the correlation is zero (Fisher, 1947).
The second equation listed for each of the four hybrids (Table XLV)
is the result of the simultaneous solution of the equations for the com-
ponent varieties. This combination was made to test the hypothesis that
the resulting equation would approximate the relationship between mean
puncture resistance and the product of days after pollination and dry
weight, thereby offering a method for predicting the results of the cross
in these terms before the cross is made. Comparison of the pairs of regres-
STATISTICAL METHODS I N FOOD RESEARCH 23 1
sion equations for the four hybrids indicates marked similarity in the
pairs, suggesting that the method may be a useful one.
Peterson, Tucker, and Comstock (1951) investigated the loss of
moisture in turnip green leaves a t room temperature t o determine the
rate of loss by whole leaves, quarters, and sixteenths of leaves. The
TABLE XLVI
Average Percentage Retention in Weight of Turnip Green Leaves Held at Room
Temperature'J
Initial
dry
matter, Time, miriirtrs
%J 0 30 60 90 12J 180 240 300
Greenhouse plants:
Whole leaves 20.5 100 96.98 95.06 92.83 91.43 88.81 84.63 81.48
Cut in quarters 21.8 100 96.72 94.37 91.82 89.18 84.36 79.72 75.54
Cut in sixteenths 18.5 100 96.48 93.41 90.53 87.53 81.63 76.33 70.86
Field plants:
Whole leaves 12.2 100 89.58 80.68 73.90 66.78 54.15 42.80 32.11
Cut in quarters 14.8 100 93.85 89.03 84.34 79.93 70.88 62.67 53.46
Cut in sixteenths 13.2 100 94.56 88.42 84.81 80.44 71.81 63.51 55.32
1 From Peterson, Tucker, and Comstock (1951).
* Mean temperature, 26' C.; mean humidity, 43 %.
moisture losses were measured at intervals for 300 minutes (Table XLVI).
Samples were obtained both from greenhouse and field plants. The raw
material in each day's sample was chosen a t random.
Linear regressions of percentage weight loss on time in hours were
calculated for each sample, for each day, and for each of the two sources
(Table XLVII). The linear regression was found to account for a large
TABLE XLVII
Percentage Water Loss per Minute for Turnip Leaves, Whole, Quarters, and
Sixteenths'
Type of Grrenhouse Field
sample I day 11 day 111 day Avg. I day I1 day I11 day Avg.
Wholeleaves -3.99 -3.41 -3.44 -3.61 -13.26 -14.40 -12.26 -13.91
Quarters -5.33 -5.97 -3.36 -4.89 - 9.45 -99.29 - 8.75 - 9.16
Sixteenths -6.67 -5.68 -5.04 -5.80 - 6.40 -11.04 - 8.86 - 8.77
Average -4.77 -10.41
1 From Peterson, Tucker, and Comstock (1951).
portion of the variance and it was concluded that the regression coeffi-
cients could be considered a satisfactory measure of moisture loss.
Analysis of variance of the regression coefficients (Table XLVIII)
indicated that the variances for origin (field vs. greenhouse) and the inter-
action of size x origin were significant at the 1% level. Analysis of co-
TABLE XLVIII
Analysis of Variance for the Regression Coefficients of Turnip Leaf Weight on Drying
Time1
Degrees of Mean
Source of variation freedom square
Origin (field us. greenhouse) 1 143.48**
Reps in origin 4 2.34
Size 2 3.51
Size X origin 2 19.06
Size x reps. in origin 8 1.16
1From Peterson, Tucker, and Comstock (1951).
**Significant at 1 % point.
variance, however, showed that after adjustment for differences in

original dry matter, the origin mean square was not significant.
Hurwicz and Tischer (1952a) derived a mathematical function to
express the temperature a t any point within a cylindrical container of
beef during heat processing. I n the form used in computation this equa-
tion is:
Log (RT - CT) = log Al[(RT - IT)Jo(plr) sin Xl(z Z)] +
- [k(m2 X12)(log10e)lt +
in which the variables are:
1. The can temperature = log (RT - CT) = y;
2. The time after the beginning of the process a t which temperature
is taken = t.
In terms of a linearized function, the slope would be
-k(p12 + X12) loglo e = b
Since this function is linear in semilogarithmic co-ordinates and
log [A1(RT - IT)Jo(pIr)sin Xl(z Z)] = a +
a constant, it may be expressed as y = a +
bt, a simple linear function.
In this form the expression was susceptible to the ordinary method of
linear regression and this technique was employed t o compare the theo-
retical with the regression estimate of can temperature (Hurwicz and
Tischer, 195213).
This situation illustrates. a not too obvious fact, that the constants
in a linear regression may be either simple or complex in nature. It is
obvious, of course, that an expression containing any number of constants
may be expressed in simple linear form if the relationship it represents is
linear or if a linear transform can be found.
b. Two or More Independent Variables. Multiple regression is, as the
name suggests, just an extension of the simple regression situation with
one independent variable to one involving two or more concomitant
variables. For simplicity our discussion will be restricted to multiple
linear regression. The estimating equation in the case of multiple linear
regression is:
+
P = bo blXl+ bzXz * * + +
bkxk
where bo, bl, . . . , b k are determined by use of the least squares principle.
XI, XZ,. . . , xk represent, of course, the k independent variables.
The computation procedures associated with multiple linear regression
are not difficult. They are, however, time-consuming. Rather than utilize
valuable time and space by outlining these methods here, we refer
the reader to Ostle (1954), Snedecor (1948), and Dwyer (1951).
Adams et al. (1952) used a 4 X 4 X 3 X 3 factorial design to test the
effect of several levels of subtilin, spores of a putrefactive anaerobe, and
time and temperature of processing on the spoilage of comminuted beef.
Using the data for per cent of spoiled tubes after 15 days incubation the
variables were related in a multiple linear regression:
P = 157.1658 + 9.1944X1 - 0.4500X2 - 2.3497Xa - 0.0598X4
in which :
P = per cent of spoilage after 15 days incubation
X1 = log of the spore concentration
Xz = processing temperature ( O F . )
X3 = processing time (min.)
Xq = subtilin concentration (p.p.m.)
The correlation values associated with the regression were :
rz,l, = 0.3675**
r,,, = 0.2700*
rZav= -0.2117
rz,a,= -0.659**
where * = significant at 5% level, and ** = significant a t 1 % level. This

indicates that the most significant effects were due to spore and subtilin
concentrations. This expression facilitates prediction of the per cent of
spoilage after 15 days of incubation if the values of X1 through Xq are
known.
A more general expression of increased usefulness could have been
written if the data were available for a series of incubation times. I n this
case, the regression could be recalculated to include the term b6X6 where
x6 = incubation time (days) ; b6 = the estimate of the slope value asso-
ciated with incubation time.
The use of a complete, factorial design in conjunction with a multiple

regression of this form provides concerted results which i t is often im-
possible to obtain by any other means.
2. Nonlinear Regression
a. One Independent Variable. As might be expected, the sample data
will not always suggest a straight-line relationship. In fact in many experi-
mental situations a straight line is the exception rather than the rule.
Growth curves (such as for bacterial
multiplication) are, perhaps, the foremost
examples in support of this remark.
If the relationship between the variates
is such that it may be represented by a
I
polynomial regression (of degree of curva-
SECONDDEGREECURVE ture k 2 2), the researcher will find the
appropriate methods of solution easy to
apply, that is, if the data lead t o curves of
the types illustrated in Fig. 15.
Estimating equations may be found,
such as:
P = al + blx + c1X2
P = a2 + b2X + czX2+ d 2 X 3
THIRD DEGREE CURVE
P = a3 + bsX + c3x2 + d3X3+ d3x4
using the method of least squares. The sub-
scripts have been changed from equation
to equation so that the reader will not gain
the impression that, for example, the a in
the first equation is the same as the a in
the second equation.
These solutions are usually found by
FOURTH DEGREE CURVE
the same methods as those applied to multi-
FIG.15. Illustration of forms ple linear regression problems. Occasionally,
of second-, third-, and fourth- however, if the various values of X occur
degree curves (from hypotheti-
cal data). a t equal intervals and if the same number of
observations are made on Y for each X ,
the calculations may be made easier through the use of orthogonal
polynomial techniques. For details the reader is referred to Anderson and
Houseman (1942) or Ostle (1954).
In the treatment of data which exhibit curvilinear relationships such
as those of Bard (1951) the curves may often be fitted through use of
orthogonal polynomials. Where the values of the independent variable are
equally spaced, this technique makes it possible to fit a curve, calculating

each degree of curvature separately, without recalculating previous terms
when a new one is added. In the Bard study, the nature of the
regression curves of moisture content on processing time for canned beef,
as determined by the use of orthogonal polynomials, indicated that a t a
processing temperature of 255’ F. the highest order regression which gave
a significant reduction in variance was th at of a third order. At a process-
ing temperature of 240’ F. a second-order regression curve was the last t o
reduce the variation by a significant amount. For the 225” F. processing
temperature the regression of moisture content on processing time was
significant a t the level of a third-order regression. These data are sum-
marized in Table XLIX.
The treatment mean moisture content values (based on 16 determina-
tions per treatment) and their corresponding estimated values for linear,
second, and third-order regressions are tabulated in Table L. From these
data i t was possible to plot the curves shown in Fig. 16.
I n spite of the fact that the third-order regression for the 240’ F.
processes was not “significant,” for purposes of comparison with the other
t w o processing temperatures (255 and 225” F.) all three regression curves
were plotted a t the third order. (The reduction in variance gained b y
plotting the third-order regression was substantial; i t is reasonable t o
assume from inspection of the mean data in Table L th a t some error due
t o sampling had prevented this process from following the same trends as
those on either side of it.)
Inspection of the curves in Fig. 16 reveals that the decrease in moisture
content occurs very rapidly during the first stages of processing a t each
temperature. A minimum point was reached in all three processes which
apparently was an inverse function of time and processing temperature.
After the minimum point was reached only slight changes in the moisture
content were observed with extension of the time of processing.
If the relationship is nonlinear and further, if it is not of polynomial
form, i t may sometimes be transformed into a linear function and treated
by the standard methods for linear regressions. A common example of this
type of function is
Y = apx
T o transform this to a linear function we take logarithms as follows
log Y = log a + (log p)x
which is of the usual linear form. The only difference is that, when obtain-
ing our least squares solution, we now consider X and log Y as our
variables, and estimate log a! and log 0. Once the solution has been ob-
TABLEXLIX
Analysis of Variance for Order of Regression for Moisture Content from
Short-Process Canned Beef Investigation1
Degrees of Sum of Mean
Source of variation freedom squares square
266' F. Processing Temperature
Total Ya 9 38450.1454
Correction for mean 1 38357.2224
Deviation from mean 8 92.9230
Linear regression 1 37.4934 37.4934*

Deviation from linear regression 7 55.4296 7.9185
Second term 1 40.3898 40.3898**

Deviation from second term 6 15.0398 2.5066
Third term 1 7.5932 7.5932*

Deviation from third term 5 7.4466 1.4853
$40" F . Processing Temperature
Total Yl 9 38615.4933
Linear regression 1 57.1916 57.1916**

Second term 1 35.9707 35.9707 *

Deviation from second term 6 31..4430 3.5738
Third term 1 7.0499 7.0499

.326'F. Processing Temperature
Total Y9 9 38456.8738
Linear regression I 40.0984 40.0984*

Second term 1 32.0466 32.6466**

Deviation from second term 6 15.8074 2.6345
Third term 1 9.3555 9.3555*

1 From Bard (1951).
* Significant a t 6 % point.
** Significant a t 1 % point.
STATISTICAL METHODS I N FOOD R E S E A R C H 237
T A B L.L~
Treatment Mean Moisture Contents and Corresponding Estimates for Regression
Curves from Short-Process Canned Beef Investigation'
Process time Mean moisture Y for linear Y second-order Y third-order
in minutes X content Y regression regression regression
966'F. Processing Temperature
0 73.48 68.44 71.80 73.02
5 67.49 67.65 68.49 67.88
10 63.65 66.86 67.72 65.69
15 64.16 66.07 64.03 63.24
20 63.59 65.28 62.88 62.88
25 63.33 64.49 62.45 63.20
30 63.16 63.70 62.74 63.87
35 64.11 62.91 63.85 64.46
40 64.66 62.12 65.48 64.26
Regression coefficient -0.7905 0.1201 -0.0876

2.40" F . Processing Temperature
0 74.01 69.36 72.55 73.73
10 67.94 68.38 69.18 68.59
20 65.56 67.40 66.49 65.40
30 63.75 66.42 64.48 63.72
40 63.14 65.44 63.16 63.16
50 63.98 64.46 62.52 63.28
60 63.13 63.48 62.57 63.66
70 63.65 62.50 63.30 63.89
80 63.79 61.52 64.71 63.53
Regression coefficient -0.9763 0.1139 -0,0844

286' F . Processing Temperature
0 73.49 68.56 71.59 72.95
15 67.14 67.74 68.50 67.82
30 64.06 66.93 66.06 64.80
45 64.02 66.11 64.27 63,40
60 63.88 65.29 63.12 63.12
75 63.69 64.47 62.63 63.50
90 63.64 63.66 62.79 64.05
105 63.66 62.84 63.60 64.28
120 64.18 62.02 65.05 63.69
Rearession coefficient -0.8175 0.1805 -0,0972

1 From Bard (1051).
The third order regression estimated moisture content values (240' F. processes) were included for
purposes of comparison with those for the 255" F. and 225" F. processes.
tained, however, it is customary t o “transform back again” and present

the results in the original form, namely:
P = abX
where log a estimates log a and log b estimates log 0.
Other nonlinear equations involving one independent variable may
also be converted t o linear form but, in most cases, the required trans-
formation will be different. Sometimes, of course, no simple transforma-
75t
255 O F . PROCESS
-.-.-
---- 2 4 0 OF. PROCESS
2 2 5 OF. PROCESS
72
0 10 20 30 40 60 80 7 0 60 90 100 I0 I2
TIME OF PROCESSING IN MINUTES
FIG.16. Regressions of moisture content on time of processing canned beef (Bard,

1951).
tion can be found and then other approaches to the problem must be
considered.
b. Two or More Independent Variables. If the nonlinear relationship
involves several independent variables, the derivation of an estimating
equation or function that will satisfactorily account for the variation
observed among observations on the dependent variable may prove t o be
a difficult task. That is, if the postulated function relating the variables is
of a highly complex nature, the estimation of the unknown parameters
will not be easy. Occasionally, however, difficulties may not really be as
great as they appear a t first glance. An example of such a situation is
afforded by some work done by Geise (1951).
This worker made a n investigation of the influence of moisture con-

tent, alcohol-insoluble solids, pericarp, and kernel size on the differences
between pairs of canned sweet corn samples. The four criteria were first
determined objectively. Identical samples were then submitted to judges
in triangular tests to determine the magnitude of the just-detectable differ-
ences. Variation among judges was consistently high enough to make it
difficult t o determine the form of the relationship among the variables.
A multiple linear regression of the form
Y = Po + PIX, + PZX, + PYX3 + P4X4

was estimated in which
Y = number of correct judgments of difference out of fifteen

XI = moisture difference
X2 = alcohol-insoluble solids differences
X3 = pericarp difference
X4 = kernel size difference
The first set of determinations yielded the regression equation :
F = 7.64 + 0.647X1 + 6.038x2 + 0.027X3 f O.622X4
in which none of the slopes (bi) were significant and R 2 = 0.227.

On a second trial an eight-factor regression yielded:
P = 7.45 + 1.278x1 - 4.924Xs + 0.103Xa + 1.327x4 + o.327xi2

+ 82.637X2' - 0.044Xa2- O.452Xd2
in an effort t o determine whether or not a curvilinear function involving
the squares of the variables would account for a greater amount of
the variation than was accounted for by the simple linear regression.
No improvement was noted in terms of significance of the slopes and
R2 = 0.263 is about the same as in the first equation.
I n a third trial a ten-factor regression was run in which the four vari-
ables and their six two-factor cross products were included:
Again the b, were not significant and R2 = 0.279, indicating that the
first and simplest form apparently provided as good a n estimate of Y as
either of the more complicated equations.
240 B. OBTLE AND ROBERT 0. TISCHER
This regression was based on the assumption that the decisions of the
judges may have involved more than one attribute a t a time. This
assumption of nonlinearity is only one of many similar and equally
logical assumptions which might have been adopted. Another assumption
might have been that the individual variables and their two-, three-, and
four-factor cross products are likely to contribute information. This would
have led to another fifteen-factor regression of the form:
An unique solution of this type of problem apparently does not exist.

At least there is no way to prove that many variables, other than those
included, do not exercise an influence on the results. In addition] no
method exists for deciding] logically, what are the “best” relationships
among the variables.
3. Biological Assays
Biological assays are usually used to determine the quantity of a sub-
stance in a preparation in terms of some specific effect. The toxicity of a
poison and the biological activity of a vitamin are two examples. Such
assays may be carried through to indicate, as directly as possible, their
specific effect; however, it may be necessary to use an indirect procedure.
Determination of the biological activity of a vitamin using laboratory
animals to indicate the response expected upon administration of the
same vitamin to human subjects may be thought of as a fairly direct
approach. Determination of the toxicity to laboratory animals of a drug
intended for use as a stimulant for human beings would be considered
indirect.
Bliss (1950) has discussed the design of biological assays indicating
that the treatment of the dosage-response curves which result from the
assay is of primary importance. These curves are often assymetrically
sigmoid in shape and the first task of the statistician is to select a trans-
formation or “metameter” which will convert this relationship to a
straight line over the range being considered.
Knudson (1950) has outlined some of the characteristics of micro-
biological assays for the determination of potency of a preparation for
several types of response including expressions appropriate for lineariza-
tion of the dosage-response curves, calculation of the slope, .and also the
errors.
Miller (1950) has discussed methods of linearization which have been
applied to biological assays exhibiting quanta1 responses.
The work of Bliss (1951) on the use of statistical methods in vitamin
research is a comprehensive publication well worth the detailed attention
of food technologists interested in this type of analysis. Many designs are
discussed in detail, accompanied by computation directions and worked
examples of the methods.
VI. CORRELATION
The term correlation is used to express the degree of association be-
tween two variables. This association may be positive, negative, or non-
existent. Correlation does not necessarily connote a cause and effect
relationship but is rather a measure of the capability of one variate to
predict the level of the other.
0 .
* *
.
0..
. 0
0 .
.*
0 .
A B c
FIQ.17. Illustration of three types of correlation between two variables (hypo-
thetical data).
Simple correlation is usually illustrated by scatter diagrams such as

Fig. 17, A , B , and C.
In Figure 17, A , the seven observations are located in such a fashion
that there appears to be no unique linear relationship of any importance
between the variables.
In Figure 17, B, the association is considerably closer and a straight
line may be fitted among the points which indicates that as Y increases,
X also increases and the two variables are associated. Figure 17, C, yields
the same information for an inverse relationship between X and Y .
The correlation coefficient of the population p is usually estimated
by the sample value T and may be calculated as:
xi - 8 ) ( Y , - F)I2
-
- W)Z 2 (Yi
n
i= 1
- Y)2
or in the usual form for computing:
r =
- (zXd2 - covariance of xy
Bx2Zy2 2/ variance z . variance y
Now it is evident that
/) = -zxy and b ZXY
-7
ux 2x2 zy
and therefore
r,, = db,, * b,,
the geometric mean of the two possible regression coefficients.

+
The estimator r varies from 1 to - 1. Values approaching f 1 denote
a relationship which is very useful in relating Y t o X and values approach-
ing zero indicate a decreasing degree of association. A value of zero for T
indicates that Y is not linearly related to X .
Harris et al. (1952) approached one portion of the problem of insect
and rodent contamination in wheat and wheat flour through use of
correlation. Five test procedures were applied to uncleaned and cleaned
wheat and the results of each test were compared with insect fragment
determinations on the flour. Correlation coefficients were calculated to
show the degree of association in each pair of tests (Table LI). The results
indicate that the cracking and staining tests on wheat yielded insect frag-
ment counts which were highly correlated with those performed on the
flour prepared from the wheat. Both tests were significant when applied
to uncleaned and cleaned wheat.
This application of correlation is one of the most common. It is
usually used as a sorting method, resulting in the elimination of com-
parisons which do not yield sufficiently reliable information for the pur-
pose at hand.
Beyond defining the degree of association, correlation is of little use.
If the form of the relationship is required, regression methods or other
forms of analysis must be applied.
TABLEL I
Correlation of Wheat Tests for Insects t o Fragment Counts in the Flour'
No. of Correlation
samples coefficient
On uncleaned wheat
Flour us. cracking 266 0.90
Flour us. stain 266 0.83
Flour us. surface flotation whole insects 266 0.25
Flour us. surface flotation insect fragments 266 -0.08
Flour us. pickout 266 0.13
On cleaned wheat
Flour us. cracking 266 0.92
Flour us. stain 18ga 0.80
Flour us. surface flotation whole insects 1822 0.23
Flour us. surface flotation insect fragments 1832 0.00
Flour us. pickout 2112 -0.01
* From Harris el al., (1952).
2 Cleaning in some mills involved so much wheat breakage that this test could not be run.
TABLEL I I
Correlation Coefficients between Quality Factors of Beef'
Correlation Probable Probability
Relationships coefficient error value
Percentage press fluid to quantity of juice -0.01 k0.07 . . . ....
Percentage press fluid to percentage fat therein -0.25 k0.06 <0.02
Percentage press fluid to carcass grade -0.04 k0.07 >0.7
Percentage press fluid to percentage fat in edible
portion of rib -0.08 k0.07 >0.4
Percentage press fluid to percentage f a t in rib eye -0.11 -1-0.07 >0.3
Percentage fat in press fluid to quantity of juice +0.22 k0.07 <0.05
Percentage fat in press fluid to quality of juice + O . 36 -1-0.06 <0.01
Percentage fat in press fluid to carcass grade +0.54 k0.05 <0.001
Percentage fat in press fluid to percentage fat in
edible portion of rib + O . 61 +0.05 <0.001
Percentage fat in press fluid to percentage f a t in
rib eye +O. 71 k0.04 <0.001
Quantity of juice t o quality of juice +O. 45 -1-0.06 <0.001
Quantity of juice t o percentage fat in rib eye +O. 30 +0.07 <0.01
Quality of juice to percentage fat in rib eye +O. 39 *0.06 <o. 001
1 From Gaddis et al. (1950).
Gaddis et al. (1950) investigated the relationships between several

quality factors of beef and found a high degree of correlation among those
tested (Table LII).
Graphical interpretation of the relationship between scores for
quantity of juice and per cent of fat in the press fluid of beef (Fig. 18)
indicated that the relationship is curved. A separation of the data was
made, dividing the determinations into two groups, one for fat between
0 and 2% and another for fat levels beyond 2%. By this means it was
TABLE LIII
Correlations of Objective and Subjective Measures of Quality for 13 Varieties of Raw,
Frozen, and Canned Yellow Sweet Corn Hybrids’
Correlations
Condition Characteristics correlated (“r” values)
Raw Growth degree days vs. % cut off +0.684* *
Raw Growth degree days vs. Succulometer -0.646**
Raw Growth degree days vs. % AIS +0.916**
Raw Growth degree days us. % soluble solids + O . 882**
Raw Growing days vs. % cut off + O . 687**
Raw Growing days us. Succulometer -0.617**
Raw Growing days vs. % AIS +o. 911**
Raw Growing days us. % soluble solids + O . 868**
Raw % AIS vs. % soluble solids +0.896**
Raw % AIS vs. succulometer -0.711**
Raw-canned Succulometer vs. succulometer +0.594**
Raw-frozen Succulometer us. succulometer +O .835**
Raw-canned % AIS us. % AIS + O . 936**
Raw-frozen % AIS vs. % AIS +0.899**
Raw-canned % Soluble solids us. % soluble solids +0.740**
Raw-frozen % Soluble solids vs. % soluble solids +0.476**
Canned Tenderness us. % pericarp -0.892**
Canned % AIS vs. tenderness -0.776**
Canned % AIS vs. % soluble solids +O .725**
Canned % AIS vs. Succulometer -0.828* *
Canned % AIS us. % pericarp +O .798**
Canned % AIS us. flavor -0.145
Canned Succulence us. % moisture content + O . 861
Canned % AIS vs. % moisture content -0.915**
Frozen Tenderness us. % pericarp - 0.537**
Frozen % AIS vs. tenderness -0.702**
Frozen % AIS us. % soluble solids f O . 360*
Frozen % A18 us. Succulometer + O . 745**
Frozen % AIS us. % pericarp +o. 579**
Frozen % AIS us. flavor -0.109
Frozen Succulence vs. % moisture content +0.765**
Frozen % AIS us. % moisture content -0.787**
1 From Oould et al. (lQ51).
* Indioates significance at the 5 % level.
** Shows signifioance at the 1 % level.
found that the correlation was T = +0.41 k 0.07, (P = 0.01) for the
first group and T = -0.13 f 0.13 (P = 0.5) for the second group. If
these values are compared with the correlation over all of the data
(T = -0.25 f 0.06, P = 0.02), the effect of the curvature on the correla-
tion value is immediately apparent, indicating that only in the first part
of the relationship is there any appreciable association between the
variables.
In the event that the remaining comparisons were curved relation-
ships, the restriction of linearity in the correlation would again result in

an underestimate or an incorrect estimate of the degree and type of
association.
3! 4.00
u) I I I I I
0 1.00 e.00 3,OO 4.00 5.00 6.0(
PERCENT FAT IN PRESS FLUID
FIG.18. Relationship between score for quantity of juice and per cent fat in press
fluid of beef (from Gaddis et al., 1950).
S
S 10 IS 20 t?S 30
PERCENT ALCOHOL INSOLUIILE CONTENT (RAW CORN)
FIQ.19. Relationship of per cent alcohol-insoluble solids content of fresh sweet

corn to per cent alcohol insoluble solids of frozen sweet corn (from Gould et al., 1951).
In any event, this difficulty may be resolved by determining the form
of the relationship, if it is other than linear, and subsequently taking
account of this fact in the analysis.
Gould et al. (1951) applied correlation methods to thirty-two com-
parisons of quality factors in raw, canned, and frozen sweet corn (Table
LIII) . With the exception of two comparisons of alcohol-insoluble solids

with flavor on canned and frozen sweet corn, all of the correlation coeffi-
cients were significant. An illustration of degree of association may be
found in Figs. 19 and 20. Figure 19 shows the regression of alcohol-
insoluble solids of frozen corn on ctlcohol-insoluble solids in raw corn.
The correlation coefficient is r = 0.904, indicating a high degree of asso-
ciation-a fact which is apparent from the close grouping of points about
the regression line.
I-
z
c
w
2'8
2.4
9 e.0-
a
t
a
0
16-
c
L
W
a
0 r-0.58
-
*
g 1.2 -ACTUAL VALUES
0- REGRESSION LINE
0 8 10 IS 20 25 30
PERCENT n;lwma INSOLUBCE SOLIDS CONTENT
FIG.20. Relationship of alcohol-insoluble solids content to per cent pericarp con-

tent for frozen whole kernel yellow sweet corn (from Gould et al., 1951).
I n Figure 20, relating pericarp content to alcohol-insoluble solids, the

points are fairly widely dispersed and the resulting correlation coefficient
is r = 0.58.
VII. ANALYSIS
OF COVARIANCE
Another technique which can often be used to advantage by food

research workers is that known as analysis of covariance. Essentially, this
technique is an extension of analysis of variance methods to include cases
where measurements are obtained on two or more characteristics from
each experimental unit. A moment's thought should convince us that this
is, then, only a scheme for combining regression and analysis of variance
into a single technique. Clearly, this device, where applicable, should
permit a more discriminating analysis of the sample data.
A covariance analysis may be performed on appropriate data for any
of the standard designs; that is, completely randomized, randomized
complete block, Latin square, split plot, and so on. The technique is the
TABLE LIV
Abbreviated Form for Analysis of Covariance
Degrees Deviations about Degrees
Source of of Slims of squares and products regression of
248 B. OSTLE AND ROBERT G. TISCH ER
same, however, in every case and the essential steps are summarized in
Table LIV. To test the hypothesis of no differences among the true effects
of the treatments after adjusting for the concomitant variable, the F
ratio :
F = (ST+, - sE)/f1
SB/(.f2 - 1)
is calculated and compared with the appropriate values in the F table.
An example of such an analysis is provided by Peterson, Tucker et al.
(1951), who studied moisture content of turnip greens with a view to
examining the variation due to leaf-to-leaf differences, plant-to-plant
differences, and time held prior to making the moisture determinations.
A Latin square design was used. The data on moisture are shown in
Tables LV and LVI. The data on leaf weights are not given, except as
TABLE LV
Moisture Content of Turnip Greens, as Affected by Drying and Size'
Increasing leaf size
Plants A B C D E Mean
1 86.67 87.15 88.29 88.95 89.62 88.14
2 85.40 84.77 85.40 87.54 86.93 86.01
3 87.32 88.53 88.50 89.99 89.68 88 * 80
4 84.92 85.00 87.29 87.85 87.08 86.43
5 84.88 86.16 87.83 85.38 88.51 86.55
Mean 85.84 86.32 87.46 87.93 88.36 87.18
f From Peterson. Tuoker st d. (1861).
TABLE LVI
Moisture Content of Turnip Greens as Mected by Holding Time and Leaf Size'
Increasing leaf size
Time A B C D E Mean
10:12 84.92 86.16 88.29' 87.54 89.68 87.32
10:31 85.40 88.53 87.29 85.38 89.62 87.24
10:54 87.32 85.00 87.83 88.95 86.93 87.21
11:09 84,88 87.16 85.40 89.99 87.08 86.90
11:27 86.67 84.77 88.50 87.85 88.51 87.26
Mean 85.84 86.32 87.46 87.93 88.36 87.18
From Peterson, Tucker et al. (1851).
averages for each plant and for each leaf size. However, the resulting
covariance analysis may be summarized as in Table LVII.
We see that, even after adjusting for differences among leaf weights,
the variation in moisture content among plants is significant. However,
the variation in moisture content among different leaf sizes was not
significant after adjusting for leaf weight. This is a decided change from
TABLELVII
Covariance Analysis-Moisture Content of Turnip Greens Adjusted for Leaf Weight'
Deviations about regression
De- De-
grees grees
of Sums of squares and of
Source of free- products free-
variation dom Zz* ZXY Zy9 Zys - ( Z X ~ ) ~ / Zdom
X ~ F
Time 4 3.7512 0.9110 0.5423
Plants 4 31.9057 8.0621 29.4231
Size 4 41.0502 30.0063 22.9950
Error 12 7.1497 3.9015 9.7788 7.6498 11 0.6954
+
Plant error 16 39.0554 11.9636 39.2019 35.5372 15
Difference for testing among adjusted plant means 27.8874 4 6.9718**
+
Size error 16 48.1999 33.9078 32.7738 8.9202 15
Difference for testing among adjusted leaf size means 1.2704 4 0.3176
1 Abridged
-
from Peterson. Tucker et al. (1961).
** Significant at P 0.01.
the significant result that would arise from an ordinary analysis of vari-
ance. It would seem, though, that leaf size and leaf weight must be closely
related and this no doubt explains the effect of carrying out the covari-
ance (regression) adjustment. Also, the error variance is reduced from
s y 2 = 9.7788/12 = 0.8149 to 8.2 = 0.6954, showing a considerable in-
crease in precision from utilizing the covariance technique.
VIII. OTHERSTATISTICAL
TECHNIQUES
There are many other statistical techniques in addition t o those pre-
sented in the preceding sections of this paper which can be used to advan-
tage by the food research worker. Many of these are concerned with
enumeration data (ie., data which arise by counting), and others are
recently developed methods for dealing with measurement data. Exam-
ples of the latter are control chart techniques, sequential analysis, pro-
cedures involving the sample range in place of the sample standard
deviation, and nonparametric and distribution-free techniques. Since
these methods have as yet received little attention by food research
workers, published examples are difficult if not impossible t o find.
However, we have mentioned these methods so that interested persons
may consult appropriate references (Ostle, 1954; Snedecor, 1948; Goulden,
1939; and Dixon and Massey, 1951) for the details of operation of par-
ticular techniques.
Especial attention should be paid, though, to statistical procedures
designed for handling enumeration data, since these techniques have
been widely used by food research workers for some time. Specifically,
practically all sensory difference tests are of such a nature that only
enumeration data result. Since this area of food research has been very
popular in recent years, the accompanying statistical procedures have
been widely used. Reference is made to such techniques as triangle tests
and similar types. Boggs and Hanson (1949) have presented a compre-
hensive review of the use of such techniques in food research. We shall,
however, take the time to illustrate the mechanics of the triangle test,
since it is so important to many food research workers.
In taste discrimination experiments, the triangle test proceeds as
follows: A person is given three samples, two of which are alike and one
which is different, and is asked to select the one which is different. It is
obvious that even if the person does not taste the samples, he has a
probability of one third of selecting the one sample which is different
due to chance alone. The question which the triangle test is designed to
answer is this: “ I s the person actually capable of discriminating between
the two different samples?” The decision will be based on the following
line of reasoning: if he chooses the odd sample correctly a sufficient num-
ber of times in repeated trials, his ability to discriminate between the two
“treatments” will be acknowledged. How many correct choices shall be
required before it will be admitted that the person can discriminate be-
tween the two treatments? This will, of course, vary with n (the number
of trials) and the preassigned probability of type I error. In such experi-
ments, an error of the first kind (or type I error) is the accepting of a
person as a qualified judge who is not able to discriminate between the
two treatments, that is, the null hypothesis is that a person cannot dis-
criminate between the two treatments.
Suppose, then, that it is desired to decide whether a particular candi-
date should be accepted as a qualified judge. He is presented with the
three samples properly randomized and asked to choose the one which is
different. This operation is repeated several, say, n times and a record is
kept of the number of times r that the odd sample is identified. If we
assume the willingness t o run a 5 % risk of accepting as a qualified judge
a person who cannot really discriminate between the treatments, and
if the probability of his making r or more correct decisions in n trials just
due to chance is less than or equal to 0.05, he is accepted. If the proba-
bility of his making r or more correct decisions in n trials just due to
chance is greater than 0.05, he is rejected. That is, if he is accepted when
he gets T correct out of n and the probability of his getting r or more
correct out of n just by chance was less than or equal to 0.05, a risk of
5 % or less is run that he is not a qualified judge.
To illustrate the method of evaluation of “due to chance alone” in n
independent trials consider the binomial theorem expansion:
(a + b)" = a" + (;)a"-% + (;) an--2b2 +. * . + (n - 1 ) ab7-1

+ b" (1)
If a = (1 - rl). and b = TI,
Now, if ~1 is interpreted as the probability of a success in any one trial,

we see that the first term on the right-hand side of Eq. (6)represents the
probability of zero successes and n failures in n independent trials.
Similarly, the second term represents the probability of one success and
n - 1 failures in n independent trials, and so on, until the last term,
which represents the probability of n successes and zero failures in n inde-
pendent trials. Thus the probability of r or more successes in n inde-
pendent trials is :
P(r or more successes in n independent trials)
= P(exact1y r successes) +
P(exact1y r 1 successes)+ ... +
+
P(exact1y n successes)
r.(1
= (").' - 7r1)n-r + ;4 l ) T 1 r + y l - T1)"+1 + * . . + TI"
= c
W=r
(;JT1Yl - ?rl)n--w (4)
In our triangle test TI = % and thus Eq. (4) becomes
To evaluate this probability is often a tedious job, especially if there are

many terms to be summed. I n this regard, it is good to know a quick,
convenient method of writing down the values of the coefficient
W = 0, 1, . . . , n in the binomial expansion. A device known as Pascal's
(;);
triangle is the favorite method for doing this (Table LVIII). The rule for
forming the values of the coefficients should be clear to the reader.
Thus, if n = 7 and it is desired t o evaluate the probability of five or

more successes, Eq. (6)together with the Table LVIII gives us
7
= 0.0448
However, even with the aid of Pascal's triangle, such computation can be
excessive. Fortunately, tables of the binomial probability distribution
have been prepared by the National Bureau of Standards (1950) which
give sums such as we require for values of ?rl going from 0.01 to 0.50 (in
steps of 0.01) and for values of n from 2 to 49 inclusive.
Having outlined the general procedure in the problem of applying
triangular tests as they are more often encountered, given a specified
TABLELVIII
Pascal's Triangle Giving Binomial Coefficients
n Values of );(
1 1 1
2 1 2 1
3 1 3 3 1
4 1 4 6 4 1
5 1 5 10 10 5 1
6 1 6 15 20 15 6 1
7 1 7 21 35 35 21 7 1
8 1 8 2 8 56 70 56 28 8 1
TABLELIX
Number of Successes Needed in Triangular Tests for Significance at Certain
Probabilities of Type I Error'
n = Number of independent trials
Probability
of type I
error 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
0.10 - 3 4 4 4 5 5 6 6 7 7 7 8 8 9 9101010
0.06 - 3 4 4 5 5 6 6 7 7 8 8 9 9 910101111
0.01 - - - 5 6 6 7 7 8 8 9 910101111121213
0.001 - - - - - 7 8 8 9 10 10 11 11 12 12 13 13 14 14
1 08th (1864).
Note: All values in the body of the table are "on the safe side." The probability of type I error will
always be less than that specified in the left-hand column for the stated numbers of 8ucc~s8e8in
6he table; in some canes, considerably lass.
probability of type I error, how many correct selections (successes) must

a candidate make before he will be accepted as a qualified judge? This
will, of course, depend on n, the number of independent trials. For values
of n from 2 to 20, a table (Table LIX) of values of T (number of successes)
necessary to allow acceptance of the candidate as a qualified judge are
presented. These are given for 0.10, 0.05, 0.01, and 0.001 probabilities of
type I error.
IX. NEEDSFOR FUTURERESEARCH
The review of the literature necessary to prepare the foregoing report
leads one to the conclusion that considerably more intensive use could be
made of statistics in food research than has been made up to the present
time. This body of research material indicates that in the majority of
cases no account has been taken of any formal statistical design or
analysis. A smaller percentage of the reports contained nominal use of
simple statistical tools and a relatively small percentage are complete
with statistical design and analysis. Considering the economies which are
apparent in the application of statistical methods to food research it
would seem to be profitable for food researchers to apply these methods
in as many cases as possible. It is not necessary nor is it recommended
that all become statisticians; on the contrary, it appears to be necessary
only t o have a clear idea of the fundamental principles of statistics and
their application to add to the existing body of knowledge already
held by the researcher. In situations where no statistical help of any kind
is available, this may entail some difficulty. However, even in these cases,
the researcher who really is interested in making an application of
statistics will find that there are a number of excellent texts which may be
used to learn the principles necessary.
One application of statistics which is used only infrequently in the
food industry is statistical quality control of production processes. This
type of control is not at all new and has been applied with great success
in other industries. At the present time it is beginning to be applied in the
food industry.
Perhaps the most significant application of statistical quality control
to food processing might be its application to grading of both the raw and
the manufactured product. Grades are in existence for most large-scale
food operations either emanating from the Production and Marketing
Administration or in the form of industry grades. However, these grades
ordinarily do not take advantage of the tools of statistical quality con-
trol. Application of these tools to procurement, manufacturing, handling,
storage, and transportation should yield information which would lead to
greater efficiency in the processes and ultimately to a greater profit.
254 B. OSTLE AND ROBERT Q. TIBCHER
In the formulation of new processes and new methods of analysis and

in the comparison of different processes and methods of analysis the use
of statistics is especially important. Research on methods often involves
comparison of two or more methods in different locations using different
operators who perform the analysis. In this case, statistical tools are
available which make it possible to compare these methods and processes
in terms of statistical differences and to conclude finally that there
actually is, or is not, a real advantage in one or another procedure when
it is carried on under carefully stated conditions. If statistical methods are
not used in making these comparisons, reliance must be placed usually in
experience and in intuitive processes, in which case the amount of con-
fidence which may be placed in the results is usually in considerable doubt.
A similar situation is found in the introduction of new varieties and
hybrids of fruits and vegetables and of new breeds of livestock and in
other endeavors where in many cases years of research go into consider-
ation of the qualities of a raw product from the point of view of its growing
characteristics and its appearance, size, color, and other attributes
judged in the raw state. In the past, many of these research efforts have
virtually ended a t this point without extension of research t o a consider-
ation of the manufactured product. In these cases statistical methods
may be used for determining the characteristics of differences among these
raw products which might make them especially acceptable as a finished
product. If this kind of research is followed up by similar considerations
of the manufacturing process and of the acceptability of the final product,
the whole procedure takes on a different form and appears to be a con-
siderably more useful evaluation.
After a product or a process has been appropriately evaluated in terms
of its usefulness and adequacy in the manufacturing operation the evalu-
ation of the resulting product in terms of its acceptability to the consumer
remains to be done. This also should be done with the aid of statistical
methods, since there is at present in many cases no substitute for the
subjective evaluation of a quality attribute of a food product. Where
they exist, objective methods are usually used in conjunction with sub-
jective methods in an effort to correlate them, with a view to the ultimate
use of the objective method as a simple and inexpensive index of the
quality of the product in question. In terms of research, however, in many
cases this may not be done because no objective methods of evaluation
exist. A good example concerns the evaluation of the flavor of a food
product where, with practically no exceptions, subjective evaluation
must be used.
In the use of subjective methods of evaluation much needs t o be
known concerning the statistical characteristics of the procedures used.
Questions which may be asked are: Which quality attributes are most
important? How many judges should be used? What kind of judges
should be used? What is the influence of age, smoking, time of day, tem-
perature, humidity, etc., on the judgments made? What inferences may
be drawn from the resulting data? These and many other questions have
been answered only partially with respect to subjective evaluation,
primarily because the body of orderly knowledge on this subject is still
rather small. Another consideration is the fact that differences in pro-
cedure for different commodities are still important and in many cases
their magnitude is unknown.
If a food product is produced in the field and manufactured in the
plant under careful statistical quality control and if the product is
evaluated by an appropriate panel of judges, the task still remains of
discovering whether the product in question is acceptable to the con-
suming public. This is a type of subjective evaluation and analysis which
is related to, but quite different from, ordinary panel judging. It should
be recognized that this type of evaluation ordinarily involves a relatively
large consuming population, the characteristics of which in many cases
are not known. What procedure should be used in surveying such a
population, what its distribution may be, and how reliable are the results
obtained are questions which we have only begun to answer. However,
only diligent search for methods and procedures for consumer acceptance
surveys will lead to sound results, and these methods will yield best
results if they are statistically sound.
With regard to the expression of the results of research in the food
field, the literature indicates that in most cases the objective appears to
be a description of differences. Although the description of differences
between products and processes is certainly a legitimate aim, there exists
another aim which is in many cases more productive; this aim is to relate
the results obtained either to fundamental physical and chemical laws or
by means of statistical or empirical procedures to establish the trends
which may be evident among the variables being considered. A corollary
to this is a consideration of processes in which it is desired to find a maxi-
mum or minimum value. There appears to have been very little applica-
tion of these techniques in food technology. As an example one might
consider the formulation of a processed food product containing several
ingredients in which it may be desired to determine whether a new
formulation is better than the existing one. In this case a simple difference
test would be sufficient; however, there would be in many cases consider-
able advantage in testing the individual and combined effects of the
ingredients in terms of their acceptability to determine, if possible, the
trends which occur upon addition of more or less of each ingredient. In
this consideration then, one would gain information which should make
it possible to decide which combination of ingredients would yield the
best product. More often than not in the past, this kind of result has been
obtained by guess work rather than by orderly procedures of any kind.
Although there exist many more sound statistical methods than are in
constant use in the food field, the serious food researcher will usually find
that some of his research problems lead into areas where appropriate
statistical methods for solution of the problem a t hand do not exist. In
the event that research leads to curved relationships among the variables
being considered it is often found that no statistical method is available
for converting these data to an easily usable form. I n many cases trans-
formations are used on the data to affect a degree of rationalization and
simplify the problem of analysis. I n some cases, however, even this treat-
ment does not yield the required result. If an attempt is made to apply
the fundamental laws of physics and chemistry to the data derived from
food research, it is often found that no statistical treatment of these laws
has been made and as a result no least squares or other solutions may be
in existence. In these cases it would be especially helpful to the food
researcher to have available designs which have as their aim the solution
of these more complicated curved relationship designs which take
account of the fundamental laws which might be involved and which lead
to a consideration of the function of the variables involved in addition
to or instead of merely the differences between or among sets of values.
One other type would be desirable, and that is a design which would yield
information concerning maximum or minimum values in a system of
variables.
It would be of considerable help to the food technologists to have
available improved practical texts dealing with statistical methods and
especially written for those with little training in statistics. If such a text
would contain information of direct interest in the food field it would
probably serve as an encouragement to those who do not have training in
statistical methods.
Of a special importance in food research are consideration of operator
comparisons, methods for discovering the best process in a group, in-
formation on appropriate designs for obtaining survey information, and
further information on the use of analysis of variance in food research
operations. If this material could be presented largely in the form of
worked examples accompanied by the limitations of the method, it would
be of great use in food research.
Those food industry workers who already make use of statistical
methods in research should consider a more systematic and illustrative
treatment in food journals of the material presented. The aim in these
writings should be to present in as complete form as possible the statistical

processes and procedures used so that in addition to being a report of
research information the article might also be used by others as a source
of information on the application of statistical methods.
REFERENCES
.4dams, A. T., Ayres, J. C., and Tischer, R. G. 1952. Effect of subtilin on spoilage of
thermal processed beef. Food Technol. 6 (11), 421.
Anderson, R. L., and Bancroft, T. A. 1952. “Statistical Theory in Research.” Mc-
Graw-Hill Book Co., New York.
Anderson, R. L., and Houseman, E. E. 1942. Tables of orthogonal polynomial values
extended to N = 104. Zowa Agr. Expt. Sta. Research Bull. No. 297, 595.
Bard, J. C. 1951. Objective measurement of changes in beef during heat processing.
Unpublished M. S. Thesis, Iowa State College Library, Ames, Iowa.
Bartlett, M. S. 1937. Some examples of statistical methods of research in agriculture
and applied biology. J. Roy. Stat. Soc., Suppl. 4, 137.
Bliss, C. I. 1950. The design of biological assays. Ann. N . Y . Acad. Sci. 62, 877.
Bliss, C. I. 1951. Statistical methods in vitamin research. In “Vitamin Methods”
(P. Gyorgy, ed.). Academic Press, New York, Vol. 2, 445.
Boggs, M. M., and Hanson, H. L. 1949. Analysis of foods by sensory difference tests.
Advances in Food Research 2, 219.
Cochran, W. G. 1943. Comparison of different scales of measurement for experimental
results. Ann. Math. Stat. 14, 205.
Cochran, W. G. 1947. Some consequences when the assumptions for the analysis of
variance are not satisfied. Biometrics 3, 22.
Cochran, W. G., and Cox, G . M. 1950. “Experimental Designs.” John Wiley & Sons,
New York.
Cone, J. F., and Ashworth, U. S. 1949. Comparison of methods reconstituting milk
powder for the plate count, with an analysis of variance. Food Research 14 (2), 165.
Cook, W. H. 1941. Canadian Wiltshire Bacon. XVI. Colour and colour stability of
pork after frozen storage and conversion to bacon. Can. J. Research 19 (Sect. D),
85.
Cook, W. H., Hopkins, J. W., and Geddes, W. F. 1934. Rapid determination of mois-
ture in grain. I. Comparison of 130” C. air oven and Brown-Duvel method. Can.
J. Research 11, 264.
Cramer, H. 1946. “Mathematical Methods of Statistics.” Princeton University
Press, Princeton, New Jersey.
Deatherage, F. E., and Reiman, W. 1946. Measurement of beef tenderness and
tenderization of beef by tendency process. Food Research 11, 525.
Dixon, W. J., and Massey, F. J., Jr. 1951. Introduction to Statistical Analysis.
McGraw-Hill Book Co., New York.
Doxtator, C. W. 1937. Studies of quality in canning corn. J. Am. SOC.Agron. 29, 735.
Dwyer, P. S. 1951. “Linear Computations.” John Wiley & Sons, New York.
Eisenhart, C. 1939. The interpretation of certain regression methods and their use in
biological and industrial research. Ann. Math. Stat. 10, 162.
Finney, D. J. 1945. The fractional replication of factorial arrangements. Ann. Eugen.
12, 291.
Fisher, R. A. 1947. “Design of Experiments.” Oliver and Boyd, Edinburgh, 4th Ed.
Caddis, A. M., Hankins, 0. G., and Hiner, R. L. 1950. Relationships between the
amount and composition of press fluid, palatibility, and other factors of meat.
Food Technol. 4 (12), 498.
Gangstad, E. O., and Snell, R. S. 1943. A practical method for quantitative evaluation
of texture in sweet corn inbreds and hybrids. J. Am. SOC.&Ton. 40 (lo), 885.
Geise, C. E. 1951. Influence of objective quality factors on subjective evaluation in
sweet corn. Food Technol. 7 ( l ) , 15.
Geise, C. E., Homeyer, P. G., and Tischer, R. G. 1951. Comparison of three methods
for determination of moisture in sweet corn. Food Technol. 6 (6), 250.
Gough, H. W., and Lantz, E. M. 1950. Relation of variety and locality to niacin,
thiamine and riboflavin content of dried beans grown in three years. Food Re-
search 16 (4), 308.
Could, W. A., Krantz, F. A., and Mavis, J. 1951. Quality evaluation of fresh, frozen
and canned yellow sweet corn. Food Technol. 6, 175.
Goulden, C. H. 1939. ‘‘Methods of Statistical Analysis.” John Wiley & Sons, New
York.
Harris, K. L., Nicholson, J. F., Randolph, L. K., and Trawich, J. L. 1952. An investi-
gation of insect and rodent contamination of wheat and wheat flour. J . Assoc.
O$k. Agr. Chemists 86 (I), 1.
Hlynka, I., Martens, V., and Anderson, J. A. 1949. A comparative study of ten elec-
trical meters for determining moisture content of wheat. Can. J . Research 27F,
382.
Hoel, P. G. 1947. “Introduction to Mathematical Statistics.” John Wiley & Sons,
New York.
Hurwicz, H., and Tischer, R. G. 1952a. Heat Processing of Beef 1. A theoretical con-
sideration of the distribution of temperature with time and in spare during
processing. Food Research 17 (4), 380.
Hurwicz, H., and Tischer, R. G. 1952b. Heat Processing of Beef 11. Development of
isothermal and isochronal distributions during heat processing of beef. Food
Research 17 (0), 518.
Kempthorne, 0. 1947. A simple approach to confounding and fractional replication
in factorial experiments. Biometrics 34 (Parts 3 and 4).
Kempthorne, 0. 1952. “The Design and Analysis of Experiments.” John Wiley &
Sons, New York.
Kendall, M. G. 1946. “The Advanced Theory of Statistics.” Griffin, London, Vols. I
and 11.
Knudson, L. 1950. Statistics in microbiological assay. Ann. N. Y. Acad. Sci. 62, 889.
Lowe, B. 1935. Data from the Iowa Agricultural Experiment Station.
Miller, L. C. 1950. Biological assays involving quanta1 response. Ann. N . Y. Acad.
Sci. 62, 903.
Mood, A. M. 1950. “Introduction to the Theory of Statistics.” McGraw-Hill Book
Co., New York.
National Bureau of Standards. Tables of the Binomial Probability Distribution.
Applied Mathematics Series 6. U. 5. Gov’t Printing Office. Washington, D.C.
1949.
Ostle, B. 1954. “Statistics in research.” The Iowa State College Press, Ames, Iowa.
Overman, A., and Li, J. C. 1948. Dependability of food judges as indicated by an
analysis of scores of a food-tasting panel. Food Research 13 (6), 441.
Paul, D. L., Ostle, B., and Tischer, R. G. 1952. The relation of sulfur content of sweet
corn to its stage of maturity. PTOC.Am. Soc. Hort. sci. 60, 393.
Peterson, W. J., Tucker, H. P., and Comstock, R. E. 1951.Loss of moisture in turnip
green leaves a t room temperature. Southern Coop. Ser. Bull. No. 10,35.
Peterson, W. J., Tucker, H. P., Wakeley, J. T., Comstock, R. E., and Cochran, F. D.
1951. Variation in moisture and ascorbic acid content from leaf to leaf and plant
to plant in turnip greens. Southern Coop. Ser. Bull. No. 10, 13.
Porter, T., Wharton, M. A., and Bennett, B. B. 1947. Evaluation of carotene content
of fresh and cooked spinach. Food Research 12 (2), 133.
Sane, R. H., Powers, J. J., Morse, R. E., and Mills, W. C. 1950. The p H and total
acidity of raw and canned pimientos. Food Technol. 4,279.
Satterthwaite, F.E.1946. An approximate distribution of estimates of variance com-
ponents. Biometrics Bull. 2, 110.
Sharpe, R. H., and Gammon, N., Jr. 1951. Sources of error in foliar analysis of pecans.
Proc. Am. SOC.Hort. Sci. 68, 120.
Skok, J. 1951. Quality of artificially ripened tomatoes for canning purposes. Food
Research 16 (2), 147.
Snedecor, G. W. 1948. “Statistical Methods.” Iowa State College Press, Ames, Iowa.
Tischer, R. G., 1952.Unpublished data.
Tischer, R. G., and Kempthorne, 0. 1951. Influence of variations in technique and
environment on the determination of consistency of canned sweet corn. Food
Technol. 6, 200.
Toepfer, E. W., Reynolds, H., Gilpin, G. L. and Tauhe, K. 1946. Home canning
processes for low-acid foods. U.S. Dept. Agr. Tech. Bull., p. 930.
Flavonoid Compounds in Foods
BY E. C. BATE-SMITH
Low Temperature Research Station, University of Cambridge, and Department of
Scientific and Industrial Research, Cambridge, England
CONTENTS
Page
...................... 267
1. Properties Depending upon Their General Phenolic Character.. . . . . . . 268

a. Destruction of Ascorbic Acid.. . . . . . . . . . . . . . . . . . .
b. Reactions with Metals.. . . . . . . . . . . . . . . . . . . . . . . . . . .
c. Antioxidant Action. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
a. Color . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
c. Pharmacological Action. . . . . . . . . . . . . . . . . . . . . . . 275

.............................. 277
111. The Genetic Situation.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
1. Genetic Situation in Fruits and Vegetables.. . . . . . . . . . . . . . . . . . . . . . . 280
a. Apples. . . . . . . . . . . . . . . . . . . . . . . . . . . .
b. Peaches.. . . . . . . . . . . . . . . . . . . . . . . . . .
c. Grapes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
d. Potatoes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
e. Onions.. . . . . . . . . . . . . . . . . . . . . . . . . . ........ 283
IV. Systematic Distributiori . . . . . . . . . . . . . . . . . . . . . . ........ 283
1. Anthocyanins . . . . . . . . . . . . . . . . . . . . . . . . . . ................ 283
2. Anthoxanthins . . . . . . . . . . . . . . . . . . . . . . . . .
a. Zsoflavones. . . . . . . . . ............................... 286
3. Catechins and Leuco-anthocyanins. . . . . . . . . . . . .
2. Tea.. . . . . .
c. Fermentation. . . . . . . . . . . . . . .................... 293
..........................
261
262 E. C. BATE-SMITH
I. INTRODUCTION
The task of reviewing the subject of flavonoid compounds in foods
has been made very much easier by the appearance recently of several
reviews dealing with particular aspects of the subject. One of these,
“Theories of the Biogenesis of Flavonoid Compounds” (Geissman and
Hinreiner, 1952), is especially valuable in providing a complete survey of
these compounds, in tabular form, and a discussion of their chemical and
biological relationships. The definition of flavonoid compounds em-
ployed by these authors is as follows: “The flavonoid compounds are
characterised by their possession of a Cs-C8-Ce carbon skeleton con-
sisting of two aromatic rings linked by an aliphatic three-carbon chain.
Chiefly on the basis of the oxidation state of this aliphatic fragment the
very large number of compounds included in the flavonoid classification
is subdivided into such well-known types as anthocyanins, flavones,
chalcones, etc.”
Flavone, the type substance of the whole class, which occurs as a dust
on the leaves and stems of certain species of Primula, has the structure
(I). The other naturally occurring compounds of this class have a number
of phenolic hydroxyl or methoxyl groups substituted in each of the two
aromatic rings, usually in certain characteristic patterns (Table I)
e.g., apigenin (11). Flavones substituted with hydroxyl specifically in the
6 Co
I
111 IV
0
V VI
FLAVONOID COMPOUNDS IN FOODS 263
3 position are known as flavonols, e.g., quercetin (111).In the flavanones,

e.g., naringenin (IV), and flavanonols, e.g., taxifolin (V), the double bond
between carbon atoms 2 and 3 is reduced; together with the isoflavones
(VI), these four types are sometimes collectively known as anthoxanthins.
The catechins, and probably the leuco-anthocyanins (whose structure
is not completely known) are derived from a reduced flavone, i.e., a
flavan molecule; catechins, e.g., catechin (VII), being flavan-3-01s.
Anthocyanins, in the colored anionic form in which they usually occur,
are flavylium salts (e.g., cyanidin, VIII). The anthocyanins may, how-
ever, occur in a colorless form, known as a pseudo base, possibly (IX),
+
VII VIII
and the leuco-anthocyanins probably have some structure such as (X),
related both t o this pseudo-base form of the anthocyanins and to the
catechins.
IX X
Closely related to the flavones and anthocyanins are the coumarins
(XI) and chlorogenic acid (XII), which may be regarded as derivatives
O;:,; Ho(j
of cinnamic acid.
CHOH
CH-COOCk ‘CHOH
\ / \ I I
CH ‘c< CH, CH2
\Ci/OH)COOH
XI XI1
These, although not, properly speaking, flavonoid in structure, have
hydroxylation patterns similar to that of the B ring in flavones (Table I)
and must therefore be kept in view in any survey of flavonoid compounds.
TABLE I
The Principal Naturally Occurring Flavonoid, and Some Related, Compounds
H O G C H , O G
Cinnamic acids
CH=CH. COOH CH=CH.COOH CH=CH .COOH CH= CH .COOH
p-Coumaric Caffeic Ferulic Sinapic
Coumarm
Scopoletin R = H Fraxetin
R = CH, Isofraxidin
Anthocyanidin:
HO
\
+
ooH
&*OH OR'
OR
OH CH
Pelargonidin Cyanidin Peonidin R = R' = H Delphinidin
R'-H, R-CHI Petunidin
R' = R = CHI h5alvidin
Catechins
I Catechins Gallocatechins
Flavonols
Kaempfeml
I R = H Fisetin
R - OH Quercetin
Isorhamnetin
-
K = H Robinetin
R OH hlvricetin
Flavones
Apigenin Luteolin I Chrysoeriol I Tricin
H O P ,CHOH
O o H
Flavanonols
OH co
Katsurenin
(Aromadendrin)
Taxifolin
I I Ampelopsin
Flavanones
-
R H Liquiretigenin
R = OH Xaringenin
R = H Butin
R = OH Ericdictyol I Homoeriodictyol
I
Chalcones
HO' ' ? H a o H
U c 6 C H
Dahlia chalcone Butein I
- - -
Isoflavones
R = R' H Daidzin Orobin
R OH, R' H Genistein
R = H , R'-CHa Formononetin
R - OH, R E CHI Biochanin-A
The only other types of flavonoid substances which have to be con.

sidered are benzalcoumaranones (aurones, cf., Bate-Smith and Geissman,
1951), e.g., sulphuretin (XIII, R = H) and aureusidin (XIII, R = OH),
chalcones, e.g., butein (XIV), and dihydrochalcones, e.g., phloretin
(XV). The relationships between these types are admirably illustrated
by Geissman and Hinreiner (1952). Table I1 is a somewhat modified
TABLE^ I1
Equivalent Level of Oxidation of 3-Carbon Fragments
Catechins -CHz.CHOH.CHOH- Dihydrochalcones -CO.CHn.CH2
Leuco-antho- Structure uncertain Chalcones
cyanins }
Flavtmones
--CO.CHz.CHOH .
Anthocyanins -CHz.CO.CO Flavones -CO.CHz.CO-
or
-CHOH.CHOH.CO-
Benaalcoumaranones -C0.C0.CH2-
Flavanonols -CO.CHOH.CHOH-
Flavonols -CO.CO.CHOH-
version of their chart. It shows in a purely formal way the level of oxida-
tion of each of the three carbon atoms of the central ring or chain. It is
interesting to note that in the jlavanone series no higher state of oxidation
of the Cafragment than that of the flavonols is possible; and similarly no
higher state of oxidation in the JEavan series is possible than that of the
anthocyanins.
XIII XIV
From the examples given in Table I it will be noted that variation

within a type is due to hydroxylation and methoxylation of the two
benzene nuclei. As pointed out above, an important point to note is that
the pattern of hydroxylation and methoxylation is common to a great
many types. In some families of plants, and especially in the Rutaceae
and Malvaceae, still more highly hydroxylated and methoxylated
flavonoids occur. Some of these are dealt with in connection with citrus
fruits (p. 289).
1. Glycosidation
With the exception of the catechins and possibly the leuco-antho-
cyanins, the flavonoid compounds occur in the plant as glycosides in which
certain of the phenolic hydroxyl groups are combined with sugar residues.
The sugar-free molecules shown in Table I are termed aglycones. If an
extract of plant material is chromatographed on paper, numerous spots
will be found which give the typical reactions for flavonoid compounds.
Usually these are clearly visible by their fluorescence in ultraviolet light,
and many change both their visible and fluorescent colours on fuming with
ammonia. If the extract is hydrolyzed with mineral acid, a chromatogram
of the extract now shows only a few spots-often only one-reacting as
described. The numerous spots on the original chromatogram are, in fact,
numerous glycosyl forms of just a few parent substances. The variety
of glycosyl forms arises both from the number and variety of sugars that
can combine with any one phenolic hydroxyl group and from the numerous
hydroxyl groups capable of glycosylation present on the flavonoid mole-
cule. Sugars which commonly occur in glycosyl combination with fla-
vonoid substances include galactose, arabinose, xylose, and, especially,
glucose and rhamnose. These can, and often do, occur not only attached
as single sugar residues to particular hydroxyl groups, but as di- or tri-
saccharides, and several positions on the same molecule may be so
glycosylated. Thus it is quite possible for three or four different glycosides
of each parent phenolic compound to be clearly visible on a chromato-
gram. I n such circumstances the particularization of any one of these as a
constituent of the foodstuff in question (unless it far outweighs all other
flavonoid constituents in quantity or is known to have some property of
especial importance in regard to the behavior of the food) would be more
misleading than instructive. Furthermore, the practice of giving specific
glycosides-or even specific aglycones-names derived, as has usually
been the case, from the botanical species from which they were first
isolated, cannot be extended to the hosts of new compounds which, we can
anticipate, will be isolated and characterized with the help of chromato-
graphic methods. Finally, it remains to be seen how much of the work
on the characterization or identification of flavonoid compounds recorded
in the literature is reliable.
So far as is known, the catechins are never glycosylated, and the
leuco-anthocyanins seem also to occur, as a rule, uncombined with
sugars. Johnson et al. (1951) report, however, that the main component
of peach tannin (which from the description they give resembles in every
respect a leuco-anthocyanin) is associated with carbohydrate, since
glucose appears on hydrolysis with HC1.
The catechins in tea occur mainly in the form of 3-galloyl esters such
as catechin-3-gallate (XVI).
XVI
Many anthocyanins are “ acylated ” with aliphatic or aromatic hydroxy-
acids, attached as a rule to the glycosyl sugar residues.
11. PROPERTIES OF FLAVONOID COMPOUNDS SIGNIFICANT IN FOODS
1 . Properties Depending upon Their General Phenolic Character

I n spite of their frequently brilliant color, it is not so much this
property which is important in foods as their tendency to undergo dis-
coloration. This is due to their general phenolic character which allows
them to serve as effective substrates for oxidase action. They are, in fact,
of all classes of phenolic substances, those most universally present in the
plant kingdom. Seldom does the analysis of a plant extract fail to reveal
one or more substances of flavonoid character, and if these are absent,
chlorogenic acid or one of the closely related coumarins is almost certain
to be present. Thus the flavonoid compounds and coumarins are the most
commonly available substrates, actual or potential, for polyphenol-
oxidase or peroxidase activity.
Little work has been done with the specific aim of showing that
flavonoid substances are competent to act as substrates for phenolase,
but what has been done allows no room for doubt that they are. The most
extensive work in this connection, that of Roberts and Wood (1951a, b,
1953) with tea oxidases and polyphenols, will be discussed in detail later
on (p. 292). Having established that the mixed polyphenolic substances
were acted upon by the purified phenolase from tea, they proceeded to
show that individual catechins, flavonols, chalcones, etc., were similarly
attacked. They showed, moreover, that certain glycosides of flavonols
were highly resistant to attack by the enzyme, although the correspond-
ing aglycones were themselves readily acted upon. In the author’s own
laboratory (Baruah and Swain, 1952) it has, similarly, been shown that
the glycosides quercitrin and rutin are not acted upon by potato poly-
phenolase, whereas the corresponding aglycone, quercetin, is rapidly
oxidized.
It is evidently of great importance whether the phenolic compounds
are present in the free state, or as glycosides, in the tissues of the plant.
FLAVONOID COMPOUNDS I N FOODS 269
Where both active polyphenoloxidases and competent phenolic sub-

strates are present, the result of the action of the one upon the other is to
produce a discoloration of the tissues, usually brown and therefore known
as “enzymic browning.” A review of this subject has appeared so recently
(Joslyn and Ponting, 1951) th at there is n o need to embark on any
detailed description of it here. Browning ensues when tissues are damaged
by cutting or bruising, by physiological injury such as storing fruits and
vegetables in inappropriate atmospheres, and by freezing and thawing.
It is, in fact, a n indication of post-mortem change, and is, in almost all
circumstances connected with food, undesirable. What interests us here
is whether, by attention to the particular nature of the flavonoid sub-
strates, any means of control of enzymic browning suggests itself.
Occasionally, but only very occasionally, a proper development of
browning is a desired step in the preparation of a food product. This is so,
for instance, in the manufacture of tea and cider. It is necessary in these
few instances t o ensure that the right phenolic substrates are present in
the raw material and that the conditions are right for the enzymes to
act upon them.
a. Destruction of Ascorbic Acid. An indirect outcome of the un-
restrained action of polyphenoloxidase in plant tissues is the total and
rapid destruction of the ascorbic acid in the tissue. This might, in certain cir-
cumstances, be a more serious disadvantage from the viewpoint of the use of
the material as food (for instance in citrus products, and vitamin concen-
trates) than the discoloration itself. Enzymic browning does not, in fact,
begin until all, or almost all, of the ascorbic acid has been destroyed (Reid
(1952) in apple juice; Miller and Heilmann (1952) in pineapple).
The reason for this is that the polyphenol, in the course of oxidation,
can reversibly transfer oxygen to ascorbic acid, being itself reduced t o its
original state. The sequence in such a cycle of changes might be indicated
as follows:
0
OH It 0 OH
where R is a substituent group such as
OH CO
acid, and DHA dehydroascorbic acid (cf. Bate-Smith and Morris, 1952).
b. Reactions with Metals. I n common with other phenols, the fla-

vonoid compounds give color reactions with a number of heavy metals.
I n many instances insoluble complexes are formed, and this enables those
compounds which are usefully colored to be employed as dyes. It was, in
fact, on account of their value as dyes that the chemistry of these sub-
stances was first studied, and especially so by A. G. Perkin. The colors
produced by treatment with ferric salts are often characteristic of par-
ticular hydroxyl groupings ; thus the catechol grouping gives a greenish
reaction, the pyrogallol grouping a blue or black reaction, and the resor-
cinol grouping a reddish reaction. With the flavones and related com-
pounds having a ketonic group a t carbon atom-4 and a Bhydroxyl group,
ferric chloride gives a deep brown color. Usually, with their multiplicity
of hydroxyl groups on two benzene nuclei, the flavonoid compounds give
rather undifferentiated reactions with ferric chloride, in which green or
greenish-brown shades are commonest because of the greatest frequency
of occurrence of the catechol grouping.
An interesting reaction is that of the flavones with aluminum ions.
The very faintly yellow-colored flavones can be made to dye tissues a
bright yellow. An instance occurred within the author’s experience, when
a tripe-dresser complained that when boiled with onions in a n aluminum
kettle his tripe became bright yellow in color. This was due to the querce-
tin glycosides (rutin, etc.) present in the yellow-skinned onions he had
happened to use. Had he chosen purple-skinned or white-skinned varieties
this difficulty would probably not have been encountered, since these
contain much smaller quantities of flavone (cf. p. 283).
The chelation of metal ions has been discussed by Clark and Geissman
(1949). They suggest that several sites might be concerned in chelation.
In quercetin, for instance, sites A and B can both be considered as
competent to form chelate complexes (XVII). Site C is perhaps even
more likely to be involved.
w-
C H
XVII
Reference to Table I will show what a large number of flavonoid com-

pounds have either or both of these active groupings. “ T h a t chelation
FLAVONOID COMPOUNDS IN FOODS 27 1
with copper actually takes place with these substances can be shown
qualitatively by the formation of coloured substances (in solution or as
insoluble precipitates) when copper sulphate is added to their alcoholic
solution ” (Clark and Geissman, 1949).
This ability to chelate heavy metal ions has been brought forward to
account for some pharmacological activities shown by flavonoid com-
pounds. Clark and Geissman showed, however, that the potentiation of
effects of epinephrine by flavonoid compounds, tested upon an isolated
smooth muscle preparation and attributable to the metal-chelating
properties of the compounds, did not correspond with their vitamin-P
activities as reported in the literature (cf. p. 275).
c. Antioxidant Action. In common with other phenolic compounds,
flavonoid compounds, especially the more highly hydroxylated members,
might be expected to have antioxidant properties. If we consider, for
instance, the constitution of such well-known and effective antioxidants
as the alkyl gallates (XVIII) and nordihydroguaiaretic acid (XIX), it is
easy to understand how the catechins, their gallate esters, and perhaps
XVIII XIX
even such commonly occurring representatives as cyanin and quercetin

might be quite powerful antioxidants. Lard and beef fat can be consider-
ably stabilized by stirring the melted fat for a few minutes with commer-
cial tannic acid or other crude tannin, followed by filtration (Spannuth
et al., 1946). Quercitrin has been tested as an antioxidant for walnuts
(Cruess and Armstrong, 1947). Although it afforded some protection, it
was not so effective as nordihydroguaiaretic acid. Some unpublished
results (Banks, 1943) show that extracts of flower petals can exert a
strong protective action against the haem-catalyzed oxidation of un-
saturated fatty acids. Especially effective were extracts of wallflowers,
peonies, azaleas, and roses, whereas those of pale-colored lupins, white
clover, and calendula were least effective. The antioxidant action seems
to have been shared by, but was not confined to, the anthocyanin pigment.
It seems highly probable that it was due to the phenolic constituents in
general, of which the flavonoid compounds form the major fraction.
2. Properties of Certain Classes of Flavonoid Compounds
a. Color. Several classes of these compounds possess brilliant coloring
properties. Pre-eminent among these are the anthocyanins, both by the
depth and quality of their color, and by the frequency of their occurrence.
Although there is a general tendency for the color to become bluer in hue
from left to right of Table I ( i e . , with increasing hydroxylation of the
flavylium radical), circumstances within the cell may profoundly modify
the color produced by the same anthocyanin. Thus the blue cornflower
and the red rose are both pigmented with cyanin (cyanidin-3,5-digluco-
side), but in the former petal the cyanin is “co-pigmented” with apigenin,
itself colorless, which has the, a t present unexplained, property of bluing
the color of cyanin. Acid conditions will cause a reddening of the hue of
anthocyanins, alkaline conditions, a bluing, so that the pH of the cell-sap,
or, in the case of processed foods, of the aqueous phase, will be highly
important in determining the color of a food containing anthocyanin
pigment.
The chalcones are yellow in color, tending towards orange as hydroxyl-
ation increases. Although they do not appear to have been reported in
foods, they may well be present, since butein occurs in flowers of the
Papilionaceae (Butea frondosa) and several of the tribe Heliantheae of the
Compositae, of which the artichoke (Helianthus tuberosus) is a member.
They may be met with as added coloring matters; for instance safflower
contains a chalcone, carthamin. They may, moreover, be produced from
flavanones during the processing of foods containing these flavonoids,
since the flavanone ring readily opens on heating, the isomeric forms
tending towards an equilibrium :-
The Zeuco-anthocyanins are themselves colorless, but give rise to red-

brown colored products when heated in aqueous solution, especially if the
solution is strongly acid. The color of stewed pears may well be due to
leuco-anthocyanins present in the fresh tissues, but this has, apparently,
never been investigated. Several classes of flavonoid compounds, them-
selves colorless, are unstable to alkaline or oxidizing conditions, producing
brown discolorations (quite independently of enzymic browning). The
calechins and Jlavanonols are examples of such types of substances. The
latter, especially, can be detected by the formation of an orange-brown
stain when fumed for some time with ammonia vapor.
The Jlavones themselves are almost colorless. Yellow colors in plant
tissues are usually due to carotenoid pigments, especially the xantho-
phylls and their epoxides, a group of pigments beyond the scope of the
present review. Even in flowers, only in rare instances such as Butea

frondosa, a few Scrophulariaceae, Papaveraceae, and Heliantheae have
yellow pigments been shown to be other than carotenoid. The absolute
limit of coloration to be expected of flavones is probably indicated b y the
pale yellow color of cotton flowers, which contain glycosides of quercetin
(111) and gossypetin (XX).
OH
XX
b. Taste. Except for naringin, the 7-rhamnoglucoside of naringenin, a

flavanone, the flavonoid compounds are not conspicuously endowed with
taste. Naringin is extremely bitter, and it is to this substance that the
bitterness of grapefruit (cf. Kesterson and Hendrickson, 1953) and sour
oranges' is due. If, however, we distinguish (as we must) astringency from
bitterness, the flavonoid compounds are by no means of negligible
importance; on the contrary, some of them, for instance the catechins,
are among the most important astringent substances in nature. Astrin-
gency is, strictly, concerned with the sense of touch, and is due to the
coagulation of the proteins of the saliva and mucous epithelium by com-
bination with the astringent body (cf. E. G. McDonough, 1935). The
action is akin t o that which occurs during tanning, and these substances
are, therefore, t o be regarded as tannins. Furthermore, we have reason
to suppose (Bate-Smith and Swain, 1953a) th a t the leuco-anthocyanins
also belong to this class of materials, and they are even more widespread
than the catechins. Such of them as have been tested are, in fact, strongly
astringent. This property is, in general, undesirable in foods. Where a
food is apt t o contain astringent substances, varieties are bred and
selected for cultivation in which astringency is reduced to a tolerable
level. Examples of this are apples and peaches. To proceed in the process
of elimination of astringency beyond a certain point often, however,
results in flatness and insipidity, so that the possession of factors for
astringency is, in many food plants, from the genetic point of view a
desirable character. Ciders, for instance, are blended to a degree of
astringency adjudged optimal for the consumer market concerned.
Wines, especially red wines, are also required to have a proper degree of
* Bitterness in navel orange juice
(cf. Marsh, 1953) is due to the presence of limonin,
which is not a flavonoid compound.
astringency. An important part of the art of chocolate manufacture seems

to lie in the selection of varieties which, after an intricate series of proc-
esses, will yield a finished product of exactly the desired degree of astrin-
gency. In each of the instances mentioned the astringent constitutents
are flavonoid in nature. Incidentally the bitterness of chocolate is quite
unconnected with its astringency, and is largely due to the presence of the
purines theobromine and caffeine. These questions have been fully dis-
cussed in a recent article (Bate-Smith, 1954a).
The realization that the leuco-anthocyanins are probably the com-
monest representatives in Nature of the substances known generally, but
indefinitely, as tannins enables a great deal of existing information to be
brought within the framework of this review. “Tannin” is a heading
commonly included in tables of the composition of foods, but it is well
recognized that the analytical figures recorded under this heading are no
more than an indication of the approximate amounts of materials which
react with the reagent employed (usually permanganate in the cold)
under standard conditions. As Charley and Harrison (1939) remark, with
regard to the application of a modification of the permanganate method
to the analysis of fruit juices: “This estimation is particularly useful
when bitter fruit is available and when it is desirable to blend the astrin-
gency equally throughout other bulks of juice . . . This method does
not give a true tannin figure, but the data obtained do bear a definite
positive relationship to the astringency of the juice.” It might be asked
what is a “true tannin figure,” short of a detailed analysis of the concen-
tration of every specific substance possessing tanning properties in the
system!
Potassium permanganate reacts in the cold, in weakly acid solution,
with many other substances than tannins. An indication of the kinds of
substances contributing to the permanganate titer of a plant extract is
given by the data in Table 111, representing results (unpublished) ob-
tained a t Long Ashton Research Station,2 Bristol, England. ( C j . also
Williams, 1953 and Kaiser, Pollard and Williams, 1953.)
Clearly, from these data, any ortho- or paradihydroxybenzene deriva-
tive will react with permanganate, with the absorption of 4-5 equivalents
of oxygen, but neither monohydroxy nor metadihydroxy derivatives are
reactive unless the molecule is otherwise open to attack (as in the side
chain of cinnamic acid and its derivatives). The same workers have also
shown that the Folin-Denis reagent is even more general in its reactivity
-monohydroxy as well as dihydroxyphenols are included in the range
of substances with which it reacts. Clearly, a good deal more discrimina-
tion is needed in the methods employed before chemical data recorded
2 By Miss M. E. Kieser, Dr. A. Pollard, and Mr. A. H. Williams.
under the heading “tannins” can be related to any particular class of

compound.
As it happens, the catechins and leuco-anthocyanins can be diff eren-
tiated from other known tannin-like substances in that they give a red
color-reaction with vanillin and concentrated hydrochloric acid ( c f .
Bate-Smith and Swain, 1953a). Although at present employed only as a
qualitative test for their detection, this reaction might well serve as a
basis for their quantitative estimation, since it depends on the 5,7-dihy-
droxy- substitution of the A ring, which is common to all the known
TABLEI11
Titer of Various Compounds Reacted with Cold, Weakly Acid Permanganate
Atoms oxy- Atoms oxy-
gen equiva- gen equiva-
Mol. lent to Mol. lent to
Substance wt. K M n 0 4titer Substance wt. KMnOl titer
Catechol 110 4.6 Tannic acid 322 7.9
Resorcinol 110 - Cinnamic acid 148 3.9
Hydroquinone 110 5.2 Caffeic acid 180 6.7
p-Hydroxybenzoic acid 138 - Chlorogenic acid 354 6.7
Protocstechuic acid 154 4.7 Quinic acid 192 -
2 : 4 Dihydroxybenzoic Tyrosine 181 -
acid 154 - Dihydroxy-
2: 5 Dihydroxybenzoic phenylalanine 197 5.5
acid 154 5.1 Catechin 290 4.3
Gallic acid 170 5.8 Phlorizin 436 -
catechins and leuco-anthocyanins. (A carbonyl group a t position 4

interferes with the reaction so that flavonoid compounds containing this
group, such as flavones and flavanones, do not react.)
Catechins and leuco-anthocyanins readily form condensation prod-
ucts, and it seems to be more particularly the condensation products of
intermediate size which possess tanning properties, i.e., the property of
cross-linking and precipitating proteins. These products, either pre-
existing or formed in the expressed juices on standing, contribute to the
turbidity, “body,” and other properties, desirable or undesirable, of
beverages such as wine, beer, and tea.
Finally, it has been observed that in the presence of “tannins” the
fading of anthocyanins is considerably retarded. This may be not the
least important, in food technology, of the properties of this very impor-
tant group of flavonoid substances.
c. Pharmacological Action. In recent years the flavonoid compounds
have come into prominenceit might almost be said have achieved
notoriety-in connection with their so-called vitamin-P activity. The
concept was first employed by Armentano el al. (1936) in explanation of
276 E. C . BATE-SMITH
the improvement in capillary resistance after administration of the juicc

of Hungarian red peppers or of lemons. The substance responsible was
not ascorbic acid. An active preparation from lemon juice (2 g. from
200 kg. lemons) believed to be a mixture of flavones, was called “citrin.”
The main component, hesperidin, was tested by Scarborough and his
collaborators (1938, 1940, 1945) and found to produce certain of the
effects described by the earlier workers. Lavollay et al. (1943), of a
number of flavonoid compounds studied, found d-epicatechin (prepared
by epimerization of d-catechin) to be the most active. Activity is not,
however, restricted to one or two specific substances. It seems to be shared
in greater or lesser degree by most flavonoid compounds and is, if only
on these grounds, considered to depart from proper vitamin character.
The whole question has been reviewed quite recently by Scarborough
and Bacharach (1949).
Hesperidin derivatives have been shown to act as inhibitors of
hyaluronidase (Beiler and Martin 1947, 1948) and, as a rather remarkable
application of this property, hesperidin solubilized by phosphorylation
has been shown to suppress fertility (Sieve, 1952). It is suggested that
the phosphorylated hesperidin, which can be administered orally, pro-
duces its effect by inhibiting the liquefaction of a hyaluronic acid gel
which is necessary before the spermatozoon can penetrate the ovum.
Since the results vary with different preparations, Martin (1953) suggests
that the effect is due to one particular component of a complex mixture
of phosphorylated derivatives. Insofar as both the aglycone hesperetin
and its 7-rhamnoglucoside hesperidin, unless phosphorylated, are ex-
tremely insoluble, it is unlikely that any such action can be expected of
the flavanones themselves when present in foodstuffs.
The isoflavone genistein (cf. Table I), which is present in subterranean
clover (Trifolium subterraneum) and probably in other Trifolieae, pos-
sesses oestrogenic properties (Bradbury and White, 1951, 1953). It seems
likely that genistein is the causal agent in the sheep infertility problem
first encountered in Western Australia in 1941, since this has been shown
t o be due to excessive ingestion by the sheep of oestrogenic substance
present in subterranean clover (Bennetts and Underwood, 1951).a
A direct action of flavones on smooth muscle (distinct from the poten-
tiation of epinephrine action mentioned earlier) has been reported. Most
conspicuous is the action on rabbit intestine reported by Ferguson
(1948) and Ferguson et a2. (1949, 1950) and offered as a possible explana-
tion of bloat in ruminants following overconsumption of herbage plants
8 Information kindly supplied by Dr. G . S. Pope, National Institute for Research in
Dairying, Shinfield, Reading, England.
such as lucerne (alfalfa). Lucerne was shown by these authors, a t certain

stages of growth, to have an exceptionally high content of tricin (see
Table I), but it could not be demonstrated that the oral administration
of flavones produced the condition. This does not rule out the possibility
that paresis of the rumen is a contributory cause of bloat, since the
etiology of the condition is extremely complicated.
The pharmacological action of coumarins and related unsaturated
lactones has been reviewed recently by Haynes (1948).
d. Fate in the Body. The little that is known about the flavonoids
after ingestion or injection (cf. Scarborough and Bacharach, 1949) sug-
gests that they are excreted more or less unchanged in the urine. I n the
case of the bright purple-red pigment betanin of the beetroot, which is a
nitrogenous anthocyanin (cf. p. 285), certain individuals possess the
idiosyncracy of excreting the pigment in the urine. Czimmer (1937) made
the interesting observation that after administration of pelargonin the
urine is a t first colored red, but later is colorless, becoming red when
acidified. This suggests that the anthocyanin is excreted as the pseudo-
base. Czimmer (1936) also reported that, after administration of flavonol
(probably quercetin) glucoside extracted from Forsythia sp., only the
urine of carnivora contained the pigment, that of herbivora being free
from flavonol. The likeliest explanation would seem to be that the pig-
ment is destroyed in the rumen or caecum of the herbivorous animal.
An effect of administration of the drug uva ursae is to cause darkening
of the urine. Although leuco-anthocyanins are present in the plant, the
effect is probably due to the glucoside of hydroquinone, arbutin, that the
plant, in common with many Ericaceae, also contains.
There is perhaps one misconception which might be corrected here,
and that is that the catechins are intestinal astringents. I n actual fact, in
the case of the tea catechins at least, the reverse is the case; they have a
cathartic effect.
e. Toxicity. All workers are agreed that, phlorizin excepted, flavonoid
compounds, administered orally, are virtually nontoxic. If, in fact, this
were not the case, the very considerable amounts which are consumed in
one form or another in fruits and vegetables would long ago have made
their presence unpleasantly felt. Some invertebrates accumulate large
quantities of flavone in their organs, conspicuously moths of the family
Sphingidae, whose wing color is due to flavone pigmentation (Thompson,
1926). Flavones are also accumulated in the epidermal tissues of the snail
Heliz pomatia (Kubista, 1950). There is, however, evidence that flavones
and chalcones are toxic to certain bacteria (Schrauffstiitter, 1948;
Schrauffstatter and Bernt, 1949).
278 E . C. BATE-SMITH
111. THEGENETICSITUATION
The kinds of foods that are grown are determined by their desirability
for direct consumption or by their suitability for consumption after
manufacturing operations. Each particular kind of food exists in many
varieties, some of superior, some of inferior, quality, and new, superior
varieties are continually being produced by the breeder. It is an advan-
tage to the breeder to have as much information as possible about the
separate variable properties which are concerned in the over-all “quality ”
of a product. In the previous section we have discussed some of these
properties, and the various ways in which flavonoid compounds may
affect them.
Some of the genetic variations in which flavonoid compounds are con-
cerned can be discerned by mere visual inspection: such variations, for
instance, as the presence or absence of red or yellow color in the skin or
flesh of fruits; others, such as astringency, can be discerned by simple
organololeptic test. But when, as is frequently the case, the genetic
situation is complicated, it may be important to the breeder to know, for
instance, exactly what pigments are present in a particular strain, or how
to detect variants possessing undeveloped characters which may be
elicited in further crossings.
The need for deeper exploration of such genetic situations, and the
importance, in certain manufacturing processes, attaching to the presence
, ~ it likely that from these direc-
or absence of particular s ~ b s t a n c e smake
tions of plant breeding and food technology detailed knowledge of the
nature and distribution of the flavonoid compounds in foods is most likely
to come. So far as the genetic situation is concerned, more information is,
however, at present being obtained from studies not immediately con-
nected with the use of plants as foods. From these studies it is already
apparent that in the plant different classes of flavonoid compounds are
biosynthetically related. (By way of contrast, it is equally apparent that
there is no biosynthetic connection between the flavonoids and the
carotenoids.) This means, for instance, that where a red anthocyanin
pigment is suppressed, a yellow chalcone or a virtually colorless flavone
may appear in its place, or that colored varieties may appear from the
crossing of two colorless ones. Chemical studies related to investigations
of pigment inheritance have also shown the impossibility of inferring,
from appearance alone, the precise genetic makeup of a particular
phenotype. A realization of the complexity of the genetic situation is
4 In this connection, a paper by Marsh (1953), although it concerns the occurrence of
a substance, limonin, which is not a flavonoid compound, in navel orange juice, is
especially pertinent.
especially desirable when we come to consider the records in the literature

of the systematic distribution of this or that flavonoid compound in
edible plants.
The most detailed investigations to date in this field have been carried
out with flower petal pigments. The cases of Dahlia variabilis and Antir-
rhinum majus are particularly instructive. In the dahlia, the genetic
situation has been worked out by Lawrence (1931), and chemical studies
by Lawrence and Scott-Moncrieff (1935), Price (1939), Bate-Smith and
Swain (1935b), and Nordstrom and Swain (1953a) have shown that the
following flavonoid compounds in various combinations, and numerous
glycosyl forms, may be present:
Genes
A, B ilnthocyanidins: Pelargonidin, cyanidin
Flavones: Apigenin, luteolin
Flavanones: Naringenin (eriodictyol?)
Elb Chalcones: 2’,4,4’Trihydroxychalcone,butein
Their occurrence is determined by the genes A , B, I , and Y , as indicated.
A and B differ in the depth of anthocyanin coloration they determine.
In an “abnormal white’’ form, still another “pigment,” which gives a
deep mahogany color when fumed with ammonia and which is possibly a
flavanonol, is present. The genetic status of this form has not yet been
ascertained, but it may perhaps represent the recessive condition, a b y i.
A particularly important aspect of Lawrence and Scott-Moncrieff’s
work is that they showed the following interactions of genes:
1. Y suppresses I .
2. Y in presence of A or B suppresses cyanidin formation.
3. I in presence of A or B suppresses pelargonidin formation.
The position in D. variabilis is complicated by the fact that it is a
polyploid, and each gene can therefore be multiplex. Some of the genes
are additive in their effects, others fully effectivein the simplex condition.
Similar or even greater complexity can be expected in large numbers of
cultivated plants, so many of which are hybrid and polyploid in their
genetic makeup.
I n still finer detail, an even greater complexity in the representation
of specific glycosides is revealed, Thus in a blue-mauve variety of dahlia
(constitution probably A1--4b12--4y), Nordstrom and Swain (1953a) found
the following specific glycosides: cyanidin (3.5?) arabino glucoside, and
traces of a pelargonidin diglucoside; apigenin-7- and 4’-glucoside and
7-rhamnoglucoside (rhoifolin) ; luteolin-5-glucoside and 7-diglucoside as
well as a naringenin monoglucoside and an unknown -diglucoside.
6 Nordstrom and Swain (1953b), more recently still, have shown the aurone 8~1-
phuretin (cf. p. 266, S I I I ) to be present in some yellow dahlias.
Similarly in parsley, which in the literature is quoted as containing

the glycoside apiin 7-(apiosyl-glucosyl)-apigenin,Nordstrom et al. (1953)
have found, in addition to apiin, 7-(apiosyl-glucosyl)-luteolin, an apigenin
glucoside, and a naringenin derivative.
In garden forms of Antirrhinum majus yellow flowers contain the
aurone glucoside aureusin (XXI, R = glucose), luteolin and apigenin gly-
OH
XXI
cosides, and a naringenin glycoside. Here the formation of luteolin is sepa-

rately controlled by a specific gene, and its presence or absence cannot be
detected by the phenotypic appearance of the flowers, which are equally
yellow whether it be present or not (Seikel and Geissman, 1950).
The moral to be drawn from these and other studies with similar out-
come is that the reporting of the occurrence, in a particular species, of a
particular member of this class of substances does not mean that it is
present in every variety of that species, nor that other flavonoid com-
pounds may not also be present. Only occasionally, however, will it be of
practical significance whether any particular glycoside of a particular
flavonoid compound, is present; it will, as a rule only be important to
know whether members of a particular group, such as flavones, flavonols,
catechins, leuco-anthocyanins, etc., are present, since it is the properties
of these substances as a group which will usually matter.
One reservation must, however, be made, that is, as to whether the
particular member is competent to act as a substrate for the polyphenol-
oxidases present. We have already (p. 268) seen that the 3-glycosides of
quercetin and myricetin are not attacked by the tea polyphenolase
system, whereas the aglycones themselves are readily attacked. It may
prove t o be important, in any tissue, whether the specific glycosides
present are attacked by the enzyme present, or introduced from other
sources, as in products of mixed origin. Such considerations as these un-
doubtedly operate in the manufacture of such products as cider and
Perry.
1. Genetic Situation in Fruits and Vegetables
a. Apples. Owing to the degree of polyploidy of cultivated forms
(Darlington and Moffett, 1930), variation in characters is almost con-
tinuous. The inheritance and distribution of anthocyanins are controlled
by a number of genes, and red color (due to cyanidin glycosides) appears

to be dominant over absence of red (Wellington, 1924). White-fleshed
varieties are of two kinds, one of which (heterozygous for yellow) carries
a suppressed factor for yellow flesh color. Although yellow-skinned apples
contain glycosides of quercetin (Sando, 1937), it is not clear to what
extent yellow skin color or yellow flesh color is due to these compounds
or to carotenoids6-a case in point where more precise information would
assist in the interpretation of breeding results. The apple certainly con-
tains catechins, leuco-anthocyanins, and many other flavonoid com-
pounds, as well as chlorogenic acid (Bradfield et al., 1952).
b. Peaches. An excellent example of the linking together of breeding
experiments and critical examination for edible quality is provided by the
work, summarized by Bailey and French (1949), that has been done on
peaches. It is, in the first place, appreciated that “edible quality” is
rather an elusive and complex thing, and has itself to be analyzed so that
its components can be separated for purposes of scoring in the families
raised. Even so, many of the component characteristics have to be
evaluated subjectively, so that assessment is a matter of personal taste.
“Hence in this particular area of fruit genetics it is a question of whether
genetic analysis measures the taste of the peach or the taste of the
taster.” The characteristics recorded were (1) flavor characteristics :
acids, sugars, tannins, and essential oils; (2) feel in the mouth: firmness,
mealiness, stringiness. Characteristics in the fruit other than those con-
cerned with edibility were skin color, color around the pit, and flesh color.
In the determination of “catechol tannin” we have at least some sort of
an objective measure of astringency, that is, the component in edible
quality with which flavonoid compounds are associated.
Both red skin color and tannin are transmitted from parent to off-
spring, high color and high tannin giving correspondingly high scores
in the seedlingsproduced, and vice versa (Blake, 1939-40 ;Weinberger, 1944).
A white-fleshed variety, “Sunbeam,” which does not become brown on
wounding, was found by Kertesz (1933) to contain no flavonoid com-
pounds other than tannins and anthocyanins-a suggestion supported by
the work of Johnson et al. (1951). Their description of the main compo-
nent of the peach tannin preparation corresponds exactly with that of a
leuco-anthocyanin (cf. p. 267). d-Catechin and chlorogenic acid are also
present.
Yellow flesh color is due to carotenoid pigment (McKinney, 1937), so
that the genetic situation as between white and yellow flesh does not
concern us here.
6 The yellow color of the flesh of the variety Tydeman’s Late Orange (Cox’s Orange
x Laxton’s Superb) is due to carotenoid pigment (Bate-Smith, unpublished).
c. Grapes. The anthocyanin pigment of the skins of European black

grapes (Vitis vinifera) was shown by Willstatter and Zollinger (1915,
1917) t o be the 3-glucoside of malvidin (oenidin). American black grapes
are crosses from V . labrusca, V . riparia, and V . aestivalis with V . vinifera.
Anderson and his collaborators (1923, 1926, 1928) and Shiner and
Anderson (1929) found that the pigment from these grapes seldom gave
the correct theoretical methoxyl content, and also gave a n intense
reaction with ferric chloride, which malvidin does not give. From the
variety “ Ives,” regarded by Hedrick as “practically pure V . labrusca,”
both malvidin and monomethylated delphinidin (petunidin) were
isolated. Crosses known to contain V . vinifera blood always contained
malvidin, and i t was concluded that the factor for malvidin formation
carried by V . vinifera was dominant in these crosses.
I n crosses of colored with white grapes, white skin color is always
recessive t o colored (Hedrick and Anthony, 1915).
White grapes contain quercetin and its 3-glucoside, isoquercitrin
(Williams and Wender, 1952a). Here the flavone and the anthocyanin
have quite different hydroxyl substitution patterns. Vitis vinifera and
V . heterophylla humifolia were reported by Robinson and Robinson
(1933, 1934) t o contain leuco-anthocyanin yielding cyanidin on digestion
with hydrochloric acid, and Durmishidze (1951) and Durmishidze and
Bukin (1951) report 1-gallocatechin amounting to 45-54% of the total
tannin present. This fruit provides therefore a n interesting example of
the varied states of oxidation and methylation in which the different
classes of flavonoid compounds exist in identical or nearly related species.
d . Potatoes. As an example of similar genetic work on vegetables, that
on the potato is outstanding. The skin color of the tubers may vary from
“white” t o (a) red or (b) blue or purple, and in intensity from very faint
to deep. Various authors differ as to the genetic interpretation of the
results of crossing, but most postulate the presence of a basic gene, acting
together with a red-producing ( R ) or a purple-producing gene ( P ) .As in
the garden dahlia and in apples, polyploidy is likely to complicate the
genetic interpretation. As regards flesh color, purple flesh in the variety
Congo Black was dominant to white (Krantz, 1922). (It seems, in fact,
to be generally true that presence of anthocyanin is dominant to absence.)
The pigment in yellow-fleshed varieties is mainly carotenoid, but other
phenolic compounds may be present in the flesh including chlorogenic
acid, coumarins (scopoletin?), and the substance of at present unknown
constitution responsible for the blackening of some varieties during
cooking. The amount of this substance, if not in fact its presence or
absence, is almost certainly an inheritable varietal factor.
Work currently in progress a t the Agricultural Research Council’s
Potato Genetics Station, Cambridge, England,’ on Solanum rybinii,

suggests a somewhat different genetic interpretation for this diploid
cultigen. The shoots of all seedlings are colored with anthocyanin, so that
all plants have the faculty for pigment production. The tubers may have
blue, red, or colorless skins; for colorless skins the presence in a homo-
zygous condition of a recessive inhibitory factor seems to be necessary
and this is epistatic to color-producing factors. Red pigmentation is
recessive to blue. If P is the factor for blue pigment, p will represent the
red-skinned condition; if i is the factorial condition for suppression of
color, I will represent the necessary condition for color development.
Other genes (some of which are linked with those for tuber skin-color)
control flower color, flecked patterning of the flower, and “spectacle”
patterning around the eyes of the tuber. The blue pigment appears to
be due to a cyanidin glycoside, and the red to a pelargonidin glycoside,
but this is probably too simple a view because in some genotypes the two
pigments may occur together.
e. Onions. In this vegetable, genetically controlled factors can be
manipulated to the advantage both of the cultivator and the food proc-
essor. Forms may occur having red, yellow, or white bulbs, the first
containing in the outer scales a glycoside of cyanidin, the second gly-
cosides of quercetin, and the last neither pigment. When pigment is
present it is accompanied by protocatechuic acid, which effectively
protects the bulb against invasion of the pathogenic fungus Colletotrichum
circinans. The presence of quercetin can be undesirable in some food-
processing operations; an instance has already been described (p. 270).
According to Rieman (1931) allelomorphic genes, W , Wv, and w govern
red pigment, yellow pigment, and white condition, respectively, W being
dominant to Wu and Wu to w. Clarke et al. (1944) postulate a basic gene,
c, for any pigment, a gene R for red, the recessive T producing yellow.
Either of these interpretations would imply that the formation of cyanidin
is alternative to that of quercetin, and involves an additional step in the
synthetic process. Such a relationship is understandable when the antho-
cyanin and flavone are as similar in structure as these two substances are.
IV. SYSTEMATIC
DISTRIBUTION
1 . Anthocyanins
Most of what we know about the systematic distribution of the antho-
cyanins, and in fact a good deal of their chemistry also, is due t o Sir
Robert Robinson and his collaborators. The results of their studies up t o
1939 were summarized and discussed in a paper in the Philosophical
7 By K. S. Dodds and D. H. Long.
Transactions of the Royal Society of London (Lawrence et al., 1939). I n

this paper the occurrence of pelargonidin, cyanidin, peonidin, malvidin,
petunidin, and delphinidin in the flowers of more than 400 species of
plants was recorded, with their glycosidic condition. The arrangement of
the data in systematic order of families allowed any regularity in the
distribution, if present, to be discerned, but no striking regularity,
relatable to systematic affinities, was apparent. Overall, however, the
data clearly showed the predominance of cyanidin (including peonidin)
over pelargonidin and delphinidin (including malvidin and petunidin) in
flowering plants. It was further observed that the occurrence of cyanidin
was also correlated in some degree with a woody habit in the plant.
Particularly relevant from our present point of view are the data for the
distribution of anthocyanins in fruits, also recorded in this paper. The
data for those of the fruits which are edible are produced in Table IV.
TABLEIV
Anthocyanins Present in Edible Fruits
Common name Botanical name Anthocyanin
Banana, red Musa coccinea Pelargonidin-3-monoside
Bilberry Vaccinium myrtillus Malvidin-Bmonoside
Cowberry Vacciniumvitis-idaea Cyanidin-3-monoside (galactose)
Cranberry, American Vaccinium (Ozycoccus) Peonidin-3-monoside (glucose)
macrocarpus
Eggplant Solanum melongena Delphinidin-3-pentoseglucoside
Eggplant, common Solanum melongena Delphinidin-3-bioside
var. esculentum
Elderberry Sambucus nigra Cyanidin-3-pentoseglucosideand
monoside
Grape Vitis vinifera Malvidin-3-monoside
Fig Ficus can'ca Cyanidin-3-monoside
Gean Prunus avium var. Cyanidin-3-monoside
Mulberry Morus nigra Cyanidin-3-monoside
Plum Prunus communis var. Cyanidin-3-monoside
Pomegranate Punica granatum Delphinidin diglycoside
Raspberry Rubus idaeus Cyanidin-3-bioside
Sloe Prunus spinosa Cyanidin-3-pentoseglucoside
Strawberry, alpine Fragaria vesca
Pelargonidin-3-monoside
Strawberry, Virginian Fragaria virginiana
Apart from these, very few identifications of anthocyanins in foods seem

to have been made, though some of those listed have since been con-
firmed. Thus Sondheimer and Kertesz (1948) have confirmed that the
pigment of strawberries is pelargonidin monoglucoside. In Jonathan and
Stayman Winesap apples the pigment has been identified as cyanidin-3-
galactoside (Sando, 1937).
I n vegetables, betanin, the pigment of red beets, calls for special note.
This is a nitrogenous anthocyanin, and its constitution has not yet been
established8 (cf. Pucher et al., 1938; Chmielewska, 1938). Apart from the
Chenopodiales in which they are repeatedly found the nitrogenous
anthocyanins are limited in their distribution to four other orders, but
only a few of the species mentioned are used as food: in Cactales, Opuntia
spp., and in Caryophyllales, Mesembryanthemum spp. Chmielewska
el al. (1936, 1938) have found that the anthocyanin of red cabbage is a
bioside of cyanidin acylated with sinapic acid, and that of a purple-
skinned variety of potato (Chmielewska, 1935) a bioside of malvidin '
acylated with p-coumaric acid; otherwise no identifications of the antho-

cyanins in vegetables appear to have been attempted.
2. Anthozanthins
No survey of the systematic distribution of flavones, flavonols, and
flavanones has been made comparable with that carried out on the antho-
cyanins. This is due in the first place to the absence, until recent years,
of any tests for their identification comparable in simplicity with those
available for the anthocyanins; and in the second place, to the wider range
of variety of these compounds and their glycosides found in nature.
The valuable summary in Klein's Handbuch der Pjlanzenanalyse, Vol.
I11 (2), 1933, enables a general view to be made of the identifications
recorded up to that date. The warning previously given must however
be repeated here, that conclusions from such identifications must be
drawn with caution because of the variability of these compounds within
a species from one genotype to another, and also because the reported
presence of one of these compounds in a plant does not necessarily mean
that others are absent. Indeed, chromatography shows a t once how many
and varied are the flavonoid compounds present in most plant extracts
(see for instance Fig. 1, p. 293). Usually, however, when hydrolyzates of
the extracts are examined, these are found to derive from no more than
one or two aglycones. It is therefore more important as a rule to know
what aglycones are present than to try to determine in detail and with
great labor the constitution of a few of the glycosidic combinations in
which they happen to occur in the particular specimens examined.
Bearing these considerations in mind, there seems nevertheless to be
no doubt that the analog of cyanidin, quercetin, is by far the most widely
distributed of the flavonols. I n Klein's Handbuch it is reported as occur-
ring in 87 species, representing 47 genera, the next most frequent,
kaempferol, and myricetin, being reported in 14 and 17 species, respec-
tively. Of the flavones, those derived from catechol, i.e., luteolin and its
8 Although Chemical Abstracts, Index, 1949 gives the structure for betanin chloride a8
4'-amino-5, 7-dihydroxyflavylium chloride, no authority for this can be traced.
methyl ethers, are again the commonest. In the flavanones, hesperitin, the
4’-methylated analog of luteolin, has been most frequently reported.
There is, however, some confusion in nomenclature regarding this sub-
stance and its glycoside, hesperidin, and the records of its occurrence
cannot be completely trusted (cf. Klein’s Handbuch, Zoc. cit., p. 849).
There are fairly strong indications, from these recorded data, that
the flavones (and flavanones) occur for the most part in herbaceous
plants, whereas the flavonols occur more particularly in woody ones.
The flavones are reported as occurring in 28 families of the dicotyledons,
20 of which are placed by Hutchinson (1946) in Herbaceae, and only 8 in
Lignosae; whereas the flavonols are reported in 47 families of Lignosae
and in only 15 of Herbaceae. The same tendency has been noted by the
author in chromatographic studies of the anthoxanthins in flower petals.
These studies have also revealed the universality of these compounds in
flowering plants. It is exceptional to find a plant in which flavones or
flavonols are not present, and frequently two, three, or four aglycones
occur together. Only rarely is it possible to identify these by chromato-
graphic evidence alone, and it must be long before a really adequate pic-
ture can be formed of their systematic distribution.
a. Isojiavones. These appear to be very restricted in their distribution,
having been reported so far only in Leguminosae, in Iris, and in one
species of Prunus. The isoflavones in Soja have received most attention.
It is now fairly certain that the structures of isogenistein and methyl
isogenistein reported by Okano and Beppu (1939) are incorrect (Baker
et al., 1952) and that the Soja isoflavones are daidzein and genistein on1
(Baker, personal communication). The yellow pigments of Osage orange
(Maclura pomifera, Moraceae) are complex isoflavones, having a second
fused pyran ring in the 7, 8 position. They bear some resemblance in
structure to the hop resins, humulone and lupolone, so important in beer
manufacture. Two other flavonoid compounds, also occurring only in
Moraceae, should be mentioned here, viz., morin (XXII) and cyano-
maclurin (XXIII). The latter is probably ring-closed between the
2’( = 6’) and 3 positions. It occurs in the jackfruit, Artocarpus integrifolia.
XXII XXIII
b. Aurones. Besides aureusidin (XXI, R = H), isolated from Antirrhi-
num majus, and leptosidin (XXIV), isolated from Coreopsis and Leptosyne
species by Geissman and co-workers, Shimokoriyama and Hattori (1953)
have recently isolated sulphuretin (XXV) from Cosmos sulphureus. These
substances are, therefore, of more frequent occurrence than, u p t o
recently, has been recognized, and if present in foods would, because of
their brilliant golden-yellow color, be of considerable importance as
pigments.
SXIV
0
ssv
c . Chalcones. These occur only infrequently and have not in fact been
reported t o occur in any common food-plant ; but since they may readily
be formed by opening of the pyrone ring from flavanones, they may be
found as secondary products from foodstuffs containing flavanones. This
seems t o be especially likely in the case of hesperidin, and it has been sug-
gested that the resulting chalcone imparts a bitter taste t o the fruit
(Kotodi, 1950). They are much more deeply colored, yellow to orange,
than the corresponding flavanone, and any such formation of chalcone
may therefore have a marked effect on color.
Phlorizin is a dihydrochalcone. Apart from its common occurrence in
rosaceous trees, and especially in the root bark, it is otherwise reported
as occurring only in Micromelum teprocarpum, a relative of Citrus.
Although present in the seeds, shoots, and leaves of apples, it appears to
be entirely absent from the flesh-a fortunate fact in view of its poisonous
character.
3. Catechins and Leuco-anthocyanins
As a first approximation, the systematic distribution of the catechin-
leuco-anthocyanin group of substances can be taken to coincide with
that of “tannins,” as recorded in the botanical literature, since the con-
densed, or catechin, tannins, which this group comprises, outnumber by
far the hydrolyzable (tannic acid) group (cf. Rottsieper, 1946; Russell,
1935). The tanniniferous families of the dicotyledons are listed by
Metcalfe and Chalk (1950). These include almost all the woody plants
from which edible products are obtained, but herbaceous families are
notable for their scanty representation. The expectation is, therefore, that
catechins and leuco-anthocyanins will generally be absent from the
tissues of herbaceous plants, and this expectation is fully realized, at
least in so far as the dicotyledons are concerned (Bate-Smith and Lerner,
1954).
In Klein’s Handbuch (Zoc. cit., p. 410), very few identifications of
catechins are recorded. Apart from tea and cacao, recent work has shown
them to be present in: (1) apples (epicatechin, commonly, catechin in
some cider varieties (Bradfield, 1952, Williams, 1952)); (2) pears (epi-
catechin only in perry varieties in small amounts (Williams, 1952)) ; (3)
peaches (d-catechin? (Johnson et al., 1951)).
Leuco-anthocyanins were reported by Robinson and Robinson (1933,
1934) to occur in almonds, grapes, apples, walnuts, chestnuts, and brazil
nuts: and in the seed coats of groundnuts, runner beans, and loquats. An
analysis of the anthocyanidins produced from all the specimens examined
by these authors (most of them heartwoods) showed a great preponder-
ance of cyanidin (84%), followed by delphinidin. Bate-Smith (1954b)
reports a similar preponderance of cyanidin-producing sources, the
remainder producing delphinidin. (It is interesting to note that the
natural catechins, derived from catechin and gallocatechin, are re-
stricted to the same pattern of hydroxyl substitution as, it seems, are the
leuco-anthocyanins.)
A wider systematic survey of the occurrence of leuco-anthocyanins
in leaves (Bate-Smith and Lerner, 1954) has shown that in dicotyledons
these substances are, with few exceptions, confined to woody families
(those included by Hutchinson in his division Lignosae). Outside this
division, so far the only species giving a positive reaction are in the
families Polygonaceae, Plumbaginaceae (which contain many woody
species), Primulaceae Crassulaceae, Ficoidaceae (Aizoaceae), and Oxali-
daceae. All of these, except Plumbaginaceae and Primulaceae, provide
edible products, mainly leafy vegetables or salad plants.
I n the monocotyledons, the presence of leuco-anthocyanins, in some
part of the plant at least, appears to be the rule, even among the herba-
ceous members. Iris pseudacorus, for instance, recognized as a tan-
niniferous plant, gives an intensely strong leuco-anthocyanin reaction in
leaves and rhizome. Sugar cane, although negative in the aerial growth,
gives a positive reaction in the rootstock. Cereals such as sorghum and
barley contain tannins; the amount varies widely in different varieties
of the former (Menaul, 1923). The tannin in the latter is of importance
in relation to the control of turbidity in beer. Luers and Stauber (1931)
state that barley tannin forms a red phlobaphene compound on heating,
and appears to belong to the same group as quebracho tannin (Le., the
condensed tannins). In asparagus, the tannin which is sometimes present
may give rise to complaints of excessive astringency (Culpepper and
Moon, 1935). All too little information is given, however, in these and
similar cases, for a decision t o be made as t o the precise nature of the
tannins present.
4. General Tendencies in Distribution
Following pp indications that can be most clearly discerned in the
case of the leuco-anthocyanins, there is a tendency for the organs of
woody plants t o be distinguished from those of herbaceous plants by the
types of flavonoid compounds that are found in them. The data relating
to anthocyanins can be reviewed in this light, and what was there re-
marked about the preponderance of cyanidin in woody plants will be
found to hold for the Lignosae and Herbaceae of Hutchinson’s classifi-
cation. If the argument were taken to its extreme limit, the situation
would be represented as follows:
Lignosae Herbaceae
Cyanidin types predominate. Pelargonidin and delphinidin types pre-
dominate.
Flavonols predominate. Flavones and flavanones predominate.
Leuco-anthocyanins and/or catechins uau- Leuco-anthocyanins and catechins usually
ally present. absent.
Generally speaking, the expectation will be for the lignose situation

to be found in fruits and nuts, and the herbaceous in leafy vegetables.
These expectations are being realized in the most recent work of Williams,
Ice and Wender (1952) and Williams and Wender (1952a, b, 1953) on the
flavonols of fruits. Studies such as these, employing chromatographic
methods for the isolation and unambiguous identification of flavonoid
compounds, are most valuable, and their extension t o leafy vegetables
and seeds would add immensely to our knowledge of the occurrence and
significance of these substances in foods as a whole.
V. SOMESPECIALCASES
1 . Citrus Fruits0
In these fruits occur the following highly methoxylated flavones and
flavanones which have not been reported as occurring in any other
natural source :
8 Cf.also footnotes on pp. 273 and 278.
Auranetin Tangeretin
Nobiletin Ponkanetin
as well as naringenin, isosakuranetin, eriodictyol, and hesperitin (cf.

Table I).
As is so often the case with cultivated plants, the systematic botany
of citrus fruits is exceedingly confused, especially since the recognized
species seem t o be almost completely interfertile. Swingle, in Webber and
Batchelor’s The Citrus Industry (1938), recommends the terminology
given in the second column of Table V. Since other authors, whose work
will be quoted, have employed other names, synonyms are given in the
third column.
TABLEV
Botanical name
Common name (Swingle) Synonyms
Citron Citrus medica
Lemon Citrus limon
Lime Citrus aurantifolia
Sweet orange Citrus sinensis Citrus aurantium vars.
Sour (bitter, or Seville) orange Citrus aurantium
Grapefruit Citrus paradisi }Citrus decumana
Pummel0 (shaddock) Citrus grandis
Tangerine (mandarin orange) Citrus reticulata Citrus nobilis
Citrus deliciosa
Citrus nobilis var. deliciosa
Ponkan tangerine Citrus reticulata Citrua poonensis
var. austera
Trifoliate orange Poncirus trifoliata Pseudaegle trifoliata
Citrus trifoliata, etc.
Swingle (loc. cit., p. 392) gives the following distributions of flavones
and flavanones:
Hesperidin in C. limon, C . aurantium, C . sinensis, C. medica.
Naringin in C. aurantium, C . paradisi, C. grandis.
Eriodictyol in C . limon, C. sinensis.
Tangeretin in C . reticulata.
“ Aurantamarin,” presumably naringin, is reported (according to
C. Tanret, 1886) as responsible for most of the bitter taste of the sour
orange (C. aurantium).
“ Ponciridin,” presumably isosakuranetin, in Poncirus trifoliata.
Tseng (1938) found nobiletin in the drug chen-pi prepared from the
skins of C. nobilis.
Patnayak et al. (1942) found the following distributions:
Hesperidin in C . aurantium vars. Batavias and Kamatas; C . medica
vars. Naranja and Dabba.
Naringin in C . decumana vars. Shaddock and Mathrepala.
Auraneting in C . aurantium var. Kamatas.
The bitterness of C . decumana vars. was attributed by these authors
to naringin.
Ichikawa and Yamashita (1941) isolated ponkanetin from the Ponkat
tangerine, C. poonensis.
Yamamoto and Oshima (1931) reported the isolation from the skins
of C . limon f . ponderosa a new glycoside, citronetin, to which they ascribed
the constitution of a 7-rhamnoglucoside of 5,7-hydroxy-2‘-methoxy-
flavanone. In the same year, Shinoda and Sato, referring to the synthesis
of citronetin, ascribe the constitution 5,7-hydroxy-3’methoxyflavanone.
Neither of these constitutions would appear reasonable by analogy with
the other co-occurring flavonoid compounds in citrus fruit. It should
be considered whether citronetin is not, in fact, identical with iso-
sakuranetin (5,7-hydroxy-4’-methoxyflavanone), which bears the same
relation to naringenin as does hesperitin to eriodictyol.
It would be expected (cf. Section 111) that the particular flavonoid
compounds found in hybrids would depend, in the first place, on whether
their formation was a dominant or recessive character. In the Sampson
tangelo (C. paradisi X C . reticulata) Nelson (1934) found no nobiletin.
Krewson and Couch (1948) report the occurrence of rutin (quercetin-3-
rhamnoglucoside) in the Satsumelo ( C . paradisi x C . reticulata var. of
Satsuma group), which presumably indicates its presence also in a t least
one of the parent forms.
The chances of using the specific distribution of flavonoid compounds
either for the identification of the origin of a citrus product, or for clari-
fication of taxonomy (as Swingle suggests), will depend on the develop-
ment of easy methods of identification. With such completely meth-
oxylated compounds as nobiletin and tangeretin this is not an easy
matter, because of the absence of any phenolic hydroxyl function. For
9 This name substituted by V. V. S. Murti, S. Rangaswami, and T. R.. Seshadri, 1948
Proc. Ind. Acad. Sci. 28A, 19, for earlicr aurantin.
this reason also, these substances cannot, of course, be glycosylated, and

they occur not in the cell sap but in the oil glands of the fruits. They do
not even make deeper-colored salts with bases, because of the absence
of hydroxyl groups, and are therefore difficult to locate on chromatograms.
The best hope would seem to lie in demethylation and identification of
the resulting nor-compounds.
In addition to these flavones and flavanones, many varieties of citrus
fruits contain anthocyanin pigment. This does not appear t o have been
identified, but that in blood oranges has been studied by Matlack (1931),
Carrante (1941), and Patan& (1948). The color of pink grapefruit and
Indian pummel0 is due t o lycopene, a carotenoid (Matlack, 1934, 1935).
Citrus fruits also contain, in greater or lesser degree, " phlobaphene
tannin." Hardy and Warneford (1925) reported that the coloring matter
of lime juice (variety not specified) was a sap-soluble phlobatannin. The
author has been unable to confirm this observation in limes of West
Indian origin; nor has Mrs. N. H. Lerner (personal communication) in
West African limes. Faint phlobaphene reactions were, however, obtained
with the albedo of commercial sweet oranges and tangerines, varieties
unknown. These also gave fairly strong, but rather anomalous, vanillin
reactions.
Technical uses of naringin as a by-product of the citrus industry are
discussed by Kesterson and Hendrickson (1953).
2. Tea
The polyphenols of tea comprise catechins, flavones, and simpler
molecules such as gallic acid, the whole system presenting a picture of the
greatest complexity even in the green leaf before fermentation (cf. Fig.
1). The changes during fermentation increase the complexity still more.
a. Catechins. Roberts and Wood (1951a, b, 1953), from their own
work and that of Tsujimura and Bradfield and his co-workers (for refer-
ences see Roberts, 1952) conclude that the naturally occurring catechins
are Z-epicatechin, d-catechin, 1-epigallocatechin, d-gallocatechin, and the
3-galloyl esters of Z-epicatechin and Z-epigallocatechin, the last pre-
dominating markedly over all other catechins. During the manufacture
of green tea, the steaming or immersion in boiling water partially epimer-
izes the catechins, so that in green tea, as Roberts and Wood (1951b)
showed, the unnatural optical isomers may be found. These are separated
from the natural ones when extracts are chromatographed on paper with
water as the mobile phase.
b. Flawones. The large number of individual anthoxanthins revealed
on a chromatogram of an extract of unfermented tea (Roberts and Wood,
1951b) are probably glycosyl forms of a much smaller number of agly-
cones. Osima and Ka (1936) reported the presence of kaempferol and

Tsujimura (1941) that of quercetin and kaempferol in green tea. I n a
chromatogram of the aglycones from fresh leaves, the author has ob-
served myricetin, quercetin, and kaempferol, the first in greatest amount.
It is likely that it is from glycosides of these three flavonols that all the
spots giving flavone reactions derive. Other “ anthoxanthin” spots are
due to chlorogenic acid and similar substances (Roberts and Wood, 1953).
c. Fermentation. I n this complicated process, oxidation and con-
densation of the polyphenolic substances play a major part. The rolling
RF
0 0.2 0.4 0.6 0.
r
I+ I I I
FIG.1. Map showing location of the chief anthoxanthins found in tea-leaf juice.
Crosses indicate positions of identified polyphenols (I - 10). Color reagent, NHs vapor.
Solvents left to right, phenol; downwards n-butanol-acetic acid (Roberts and Wood,
1951a).
process is, in effect, a deliberate bruising treatment which allows rapid
enzymic oxidation, as described earlier (p. 268), to take place. As might
be expected, the catechol and pyrogallol residues of the catechins are
strongly attacked by the polyphenoloxidase system, with the formation
of dark-colored oxidation products. The anthoxanthins, possibly because
they are present almost wholly in glycosidic combination, show little
change during fermentation.
In fermenting juice, the gallocatechins are oxidized before the cate-
chins and epicatechins are attacked. I n the earlier stages of fermentation
gallic acid shows a marked increase, presumably by liberation from ester
combination. Similar changes are observed in fermenting tea leaf, in
which, as the catechins disappear, more complex substances are formed
which are probably condensation products (Harrison and Roberts, 1939).
d. Bearing on Quality in Tea. The term “quality11in tea has a rather
specialized significance, about which there is not at present any general
agreement. Various factors contribute to over-all quality, of which
“quality,” in this specialized sense, is one. A particular taster might, for

example, recognize “strength,” “briskness,” body, flavor, etc., and also,
in addition, “quality.” These factors, it will be observed, are not clear-cut
in their attribution to specific physical properties, and it is therefore
unlikely that they can be accounted for in terms of specific chemical
constituents. “Strength,’] relating especially to taste, will be referable in
part to the bitterness imparted by caffeine,in part to the astringency and
taste (if any) which the “tannins” and other polyphenols may contribute.
“Briskness ” is probably related to astringency specifically, but acidity
may also play a part. “Body” and L‘smoothnessJ’are likely to be con-
cerned with the colloidal constituents, that is, to the degree of condensa-
tion of the polyphenolic constituents. Flavor is concerned mainly with the
volatile constituents, i.e., the essential oils and esters. “Quality,’] it has
been suggested, may be defined as an especial attractiveness that a par-
ticular tea may possess, in respect of any of these factors or a combination
of them, but especially, perhaps, in relation to flavor. Finally, the color
of the infusion is a factor to be taken account of in quality. This, above
all, is related to the flavonoid constituents of the leaf and the changes
they undergo during fermentation. It is towards an understanding of the
processes which take place during fermentation, and how they might be
more closely controlled to the advantage of the finished product, that
research in tea chemistry is now being most actively pursued.
3. Cacao
Traditionally, there are two main types of cacao, Forastero (“for-
eign”) in which the beans are purple in section, and Criollo (“native”)
in which they are cream-colored. The purple pigment of the former was
shown by Robinson (cf. Knapp, 1937) to be a cyanidin-3-glycoside. The
latter were shown by Knapp and Hearne (1939) to contain a leuco-
anthocyanin and Forastero beans were found by Hallaslo (1949) to con-
tain a leuco-anthocyanin in addition to anthocyanin. Both types contain
Z-epicatechin (Freudenberg el al., 1932).
With the aid of chromatography, Forsyth (1948, 1949, 1952a, b) has,
during the last few years, immeasurably advanced our knowledge of the
polyphenolic substances in cacao and their changes during fermentation.
There are three anthocyanins in Forastero beans, the two main pigments
being a cyanidin monoglycoside, probably of glucose, and an acylated
cyanidin bioside, probably of arabinose and glucose; and the third, in
minor amounts, a diglycoside yielding glucose and arabinose in the pro-
portion about 3 :1.
10 Unpublished work of the British Food Manufacturing Industries Research Asso-
ciation.
The leuco-anthocyanins consist of two main fractions one, sugar-free,

RE about 0.4 in butanol-acetic acid, and one remaining on the origin when
chromatographed and yielding sugar on hydrolysis. Both yield cyanidin
when digested with hydrochloric acid.
There appear to be at least six catechin-like constituents, of which
I-epicatechin is the principal one, with d-catechin, I-epigallocatechin, and
d-gallocatechin also present.
For isolation of these various constituents in bulk, fractionation with
ethyl acetate and chromatography on cellulose pulp were employed.
a. Fermentation of Cacao. In practice, the fermentation of cacao is
carried out in sweat boxes, in which the pods are heaped. The pulp sur-
rounding the beans ferments, producing heat and acetic acid, and killing
the beans. Forsyth has shown that the fermentation is anaerobic, and
that the browning of the beans, which was previously thought to be due
to the action of polyphenoloxidase, is nonoxidative. It is mainly accom-
panied by a destruction of the anthocyanins, and their condensation to
products resembling the nonmotile leuco-anthocyanin fraction. Oxidation
only takes place during the subsequent sun-drying of the beans. The
changes during fermentation are vitally important in determining the
flavor and aroma of the final product.
This review was prepared as part of the program of work of the Food Investigation
Organization of the Department of Scientific and Industrial Research. The author
wishes t o acknowledge the help he has received from Professor T. Wallace and the
staff of the Long Ashton Horticultural Research Station, and Dr. E. A. H. Roberts,
Dr. E. B. Hughes, Dr. K. S. Dodds, and Dr . T. Swain.
REFERENCES
Anderson,'R. J. 1923. Anthocyanins in Norton and Concord grapes. N.Y. State Agr.
Ezpt. Sta. Tech. Bull. No. 96.
Anderson, R. J. 1928. Anthocyanins in Seibel grapes. N . Y . State Agr. Expt. Sta. Tech.
Bull. No. 146.
Anderson, R. J., and Nabenhauer, F. B. 1926. Anthocyanins in Isabella grapes. N . Y .
State Agr. Expt. Sta. Tech. Bull. No. 123.
Anderson, R. J., and Nabenhauer, F. B. 1928. Anthocyanins in Clinton grapes. N . Y .
State Agr. Ezpt. Sta. Tech. Bull. No. 146.
Armentano, L., Bentsdth, A., Bbres, T., Rusznydk, I., and Szent-Gyorgyi, A. 1936.
The effect of flavins on capillary permeability. Vitamin P. Deut. med. Wochschr.
62, 1325.
Bailey, J. S., and French, A. P. 1949. Inheritance of certain fruit and foliage charac-
ters in peach. Mass. Agr. Expt. Sta. Bull. No. 452.
Baker, W., Harborne, J. B., and Ollis, W. D. 1952. The synthesis of 5:7:2'-tri-
hydroxyisoflavone. Chemistry & Industry 1962, 1058.
Banks, A. 1943. Report on the effect of antioxidants on the oxidation of aqueous
suspensions of linoleic acid. Record Memo. (unpublished) No. 24 of the Torry
Research Station.
Baruah, P., and Swain, T. 1952. The effect of I,-ascorbic acid on the in vitro activity
of polyphenoloxidase from potato. Biochem. J . 66, 392.
Bate-Smith, E. C. 1954a. Astringency in foods. Food 23, 124.

Bate-Smith, E.C. 1954b.Detection and identification of anthocyanidins formed from
leuco-anthocyanins in plant tissues. Biochem. J. (in press).
Bate-Smith, E. C., and Geissman, T. A. 1951.Benralcoumaranones. Nature 167, 688.
Bate-Smith, E. C., and Lerner, N. H. 1954. Systematic distribution of leucoantho-
cysnins in leaves. Biochem. J . (in press).
Bate-Smith, E. C., and Morris, T. N. 1952. The nature of enzymic browning. I n
“Food Science.” The University Press, Cambridge, p. 55.
Bate-Smith, E. C., and Swain, T. 1953a. Identification of leuco-anthocyanins as
“tannins” in foods. Chemistry & Industry 1963, 377.
Bate-Smith, E. C., and Swain, T. 1953b. The isolation of 2’.4.4’-trihydroxychalkone
from yellow varieties of Dahlia variabilis. J . Chem. SOC.1969,2185.
Beiler, J. M., and Martin, G. J. 1947. Inhibitory action of vitamin P compounds on
hyaluronidase. J. Biol. Chem. 171, 507.
Beiler. J. M., and Martin, G. J. 1948.Inhibition of hyaluronidase action by derivatives
of hesperidin. J. Biol. Chem. 174, 31.
Bennetts, H. W., and Underwood, E. J. 1951. The oestrogenic effects of subterranean
clover. Australian J . Expt. Biol. Med. Sci. 29, 249.
Blake, M. A. 1939-40. Some results of crosses of early ripening varieties of peaches.
Proc. Am. Soc. Hort. Sci. 37, 232.
Bradbury, R. B. and White, D. E. 1951. Chemistry of subterranean clover, I. J.
Chem, SOC.1961, 3447.
Bradbury, R. B., and White, D. E. 1953. Chemistry of subterranean clover, 11. J .
Chem. Soe. 1963, 871.
Bradfield, A. E. 1952.Personal communication.
Bradfield, A. E., Flood, A. E., Hulme, A. C., and Williams, A. H. 1952.Chlorogenic
acids in fruit trees. Nature 170, 168.
CarrantB, V. 1941. Pigmentation of Italian oranges. Ann. regia staz. sper. frutticolt. e
agrumicolt., Acireale 16, 193.
Charley, V. L. S., and Harrison, T. H. J. 1939. Fruit juices and related products.
Imp. Bur. Hort. Tech. Commun. No. 11, 88.
Chmielewska, I. 1935. Pigments of violet potatoes. Roczniki Chem. 16, 491.
Chmielewska, I. 1936.The coloring matter in red cabbage. 11. Roczniki Chem. 16,384.
Chmielewska, I. 1938.Investigations on the coloring matter of red beets, Beta vulgaris
L. Roczniki Chem. 18, 1.
Chmielewska, I., Smardrewska, I., and Kulesza, J. 1938. Red coloring matter of
cabbage (Brassica oleraceae). 111. Roczniki Chem. 18, 176.
Clark, W. G., and Geissman, T. A. 1949. Potentiation of effects of adrenaline by
flavonoid (“vitamin PI’-like) compounds. J. Pharmacol. Exptl. Therap. 96, 363.
Clarke, A. E.,Jones, H. A,, and Little, T. M. 1944. Inheritance of bulb color in the
onion. Genetics 29, 569.
Cruess, W.V.,and Armstrong, M. 1947. Experiments with antioxidants for walnuts.
Fruit Products J . 26, 327, 344.
Culpepper, C. S., and Moon, H. H. 1935. Composition of the developing asparagus
shoot in relation to it8 use as a food product and a material for canning. U.S.
Dept. Agr. Tech. Bull. No.462.
Czimmer, A. G. 1936.The pharmacological action of flavonol glucoside of the species
Foraythia. Arch. exptl. Pathol. Pharmakol. 183,587.
Czirnmer, A. G. 1937. Action of pelargonin. Arch. exptl. Pathol. Pharmakol. 187,565.
Darlington, C . D., and Moffett, A. A. 1930.Primary and secondary balance in Pyrus.
J . Genet. 22, 1929.
Durmishidze, S. V. 1951. 1-Gallocatechin in the composition of grapevine tannin.
Doklady Akad. Nauk S.S.S.R. 77, 859.
Durmishidee, S. V., and Bukin, V. N. 1951. Physiological properties of the tannins and
colouring matters of grapes. Doklady Akad. Nauk S.S.S.R. 76, 703.
Ferguson, W. S. 1948. A possible relation between the action of clover juice on isolated
muscle and bloat in cattle. Nature 161,816.
Ferguson, W. S., Ashworth, R. de B., and Terry, R. A. 1949. Identity of a muscle-
inhibiting flavone in lucerne. Nature 168, 606.
Ferguson, W. S., Ashworth, R. de B., and Terry, R. A. 1950. Nature of a muscle-
inhibiting compound in lucerne and its possible connexion with bloat in cattle.
Nature 166, 116.
Forsyth, W. G. 1948. Colour reagents for the paper chromatography of sugars.
Nature 161, 239.
Forsyth, W. G. 1949. A method for studying the chemistry of cacao fermentation.
Nature 164, 25.
Forsyth, W.G. 1952a. Cacao polyphenolic substances. 1. Fractionation of the fresh
bean. Biochem. J . 61, 511.
Forsyth, W.G. 1952b. Cacao polyphenolic substances. 2. Changes during fermenta-
tion. Biochem. J . 61,516.
Freudenberg, K., Cox, R. F. B., and Braun, E. 1932. The catechin of the cacao bean.
J . Am. Chem. SOC.64,1913.
Geissman, T. A., and Hinreiner, E. 1952. Theories of the biogenesis of flavonoid com-
pounds. Botan. Rev. 18, 77.
Hardy, F., and Warneford, F. H. S. 1925. Coloring matter of lime juice. Znd. Eng.
Chem. 17, 48.
Harrison, C. J., and Roberts, E. A. H. 1939. The fermentation process in tea manu-
facture. IV. Tea tannin and its fermentation products. Biochem. J . SS, 1408.
Haynes, L. J. 1948. Physiologically active unsaturated lactones. Quart. Revs. (London)
2, 46.
Hedrick, V. P., and Anthony, R. 0. 1915. Inheritance of certain characters of grapes.
N . Y . Expt. Sta. Bull. No. 45.
Hutchinson, J. 1946. “A Botanist in South Africa.” Gawthorn Ltd., London.
Ichikawa, N., and Yamashita, T. 1941. The constitution of “ponkanetin,” a new
flavanone derivative from the peel of Citrus poonensis, Hort. J . Chem. SOC.Japan
62, 1006.
Johnson, G., Mayer, M. M., and Johnson, D. K. 1951. Isolation and characterization
of peach tannins. Food Research 16, 169.
Joslyn, M. A., and Ponting, J. D. 1951. Enzyme-catalyzed oxidative browning of fruit
products. Advances in Food Research 3, 1.
Kertesz, Z. I. 1933. The oxidase system of a non-browning yellow peach. N . Y . State
Agr. Expt. Sta. Tech. Bull. No. 219.
Kesterson, J. W., and Hendrickson, R. 1953. Naringin, a bitter principle of grapefruit.
Univ. Florida Agr. Expt. Sta. Bull. No. 511.
Kieser, M. E., Pollard, A., and Williams, A. H. 1953. The occurrence of leuco-antho-
cyanins in perry pears. Chemistry and Industry 1969, 1260.
Knapp, A. W. 1937. “Cacao Fermentation.” Bale, Sons and Curnow, Ltd., London.
Knapp, A. W., and Hearne, J. F. 1939. The presence of leuco-anthocyanins in Criollo
cacao. Analyst 64,475.
Krantz, F. A. 1922. The application of genetic principles to potato breeding. Proc. Am.
Hort. Sci. 19, 124.
Krewson, C. F. and Couch, J. F. 1948.Isolation of rutin from a citrus hybrid. J . Am.

Chem. SOC.70, 257.
Kotodi, E. P. 1950. The chemical nature of the substances causing bitterness in
products obtained by the processing of mandarins. Doklady Akad. Nauk S.S.S.R.
73, 763.
Kubkta, V. 1950. Flavones in Helix pomatia L. Ezperientia 6, 100.
Lavollay, J., Parrot, J. L. and Sevestre, J. 1943. On the nature of vitamin P. Epi-
catechin, the substance up to the present having the most action on capillary
resistance. Compt. rend. 217, 540.
Lawrence, W. J. C. 1931. The genetics and cytology of Dahlia oariabilis. J . Genet. 24,
257.
Lawrence, W.J. C., Price, J. R., Robinson, G. M., and Robinson, R. 1939. The dis-
tribution of anthocyanins in flowers, fruits and leaves. Phil. Trans. 230B, 149.
Lawrence, W. J. C. and Scott-Moncrieff, R. 1935. The genetics and chemistry of
flower colour in Dahlia: A new theory of specific pigmentation. J . Genet. 30, 155.
Liiers, H., and Stauber, J. 1931. The tannin of barley husk. Wochschr. Brau. 48. 93.
103, 117.
McDonough, E.G.,1935. Astringency. Drug & Cosmetic Znd. 37, 733.
McKinney, G.1937. Carotenoids of the peach. Plant Physiol. 12, 216.
Marsh, G.L. 1953. Bitterness in navel orange juice. Food Technol. 7 , 145.
Martin, G. J. 1953. Phosphorylated hesperidin. Science 117, 363.
Matlack, M.B. 1931.Observations on the red color of the blood orange. Plant Physiol.
6,729.
Matlack, M. B. 1934. Pigment of the Indirt red pummel0 (Citrusgrandis Osbeck).
J . Wash. Acad. Sci. 24, 385.
Matlack, M. B. 1935. Pigments of pink grapefruits, Citrus grandis (L., Osbeck). J .
Biol. Chem. 110, 249.
Menaul, P. 1923. A method for the quantitative estimation of tannin in plant tissue.
J . Agr. Research 26, 257.
Metcalfe, C. R., and Chalk, L. 1950. “Anatomy of the Dicotyledons.” The Clarendon
Press, Oxford, Vol. 2.
Miller, C. V., and Heilmann, A. S. 1952. Ascorbic acid and physiological breakdown
in the fruits of the pineapple (Ananas wmosus L. Merr.). Science 116, 505.
Nelson, E. K. 1934. The occurrence of a pentamethyl flavonol in tangerine peel. J.
Am. Chem. Soc. 66, 1392.
Nordstrom, C. G.,and Swain, T. 1953a.The flavonoid glycosides ofrDahlia variabilis.
J. Chem. SOC.1963, 2764.
Nordstram, C. G.,and Swain, T. 1953b. Isolation of a beneylidenecoumaranone
(aurone) from a yellow dahlia. Chemistry & Industry 1963, 823.
Nordstrom, C. G., Swain, T., and Hamblin, A. J. 1953. The flavone glycosides of
parsley. I. The isolation of 7-(apiosyl-glucosyl)-luteolin. Chemistry & Industry
1963, 85.
Okano, K., and Beppu, I. 1939.Coloring matters in soybean. I. Isolation of four kinds
of isoflavone from soybean. J. Agr. Chem. SOC.Japan 15, 645.
Osima, Y., and Ka, H. 1936. Kaempferol from tea leaves. J . Agr. Chem. SOC.Japan
12, 975.
P a t a d , G. 1948. Anthocyanins extraction from oranges. Ann. regia staz. sper. frutti-
colt. e agrumicolt., Acireale 17, 1.
Patnayak, K. C., Rangaawami, S., and Seshadri, T. R. 1942. Chemical investigation
‘of ‘India fruits. 111. Characteristic crystalline components of certain citrus fruits
(oranges of the Circars). Proc. Zndian Acad. Sci. M A , 10.
Price, J. 1939. The yellow colouring matter of Dahlia variabilis. J . Chem. SOC.1939,
1017.
Pucher, G. W., Curtis, L. C., and Vickery, H. B. 1938. The red pigment of the root
of the beet (Beta vulgaris). I. Preparation of betanin. 11. A method to determine
betanin. J . Biol. Chem. 123, 61, 71.
Reid, H. H. 1952. Personal communication.
Rieman, G. H. 1931. Genetic factors for pigmentation in the onion and their relation
to disease resistance. J . Agr. Research 43, 251.
Roberts, E. A. H. 1952. The Chemistry of tea fermentation. J . Sci. Food Agr. 3, 193.
Roberts, E. A. H., and Wood, D. J. 1951a. A study of the polyphenols in tea leaf by
paper chromatography. Biochem. J . 49, 414.
Roberts, E. A. H., and Wood, D. J. 1951b. The separation of polyphenols in tea leaf
by paper chromatography with water as a mobile solvent. Biochem. J . 49, xxxiii.
Roberts, E. A. H., and Wood, D. J. 1953. Separation of tea polyphenols on paper
chromatograms. Biochem. J . 63, 332.
Robinson, G. M. and Robinson, R. 1933. A survey of anthocyanins. 111. Notes on the
distribution of leuco-anthocyanins. Biochem. J . 27, 206.
Robinson, G. M., and Robinson, R. 1934. A survey of anthocyanins. IV. Biochem. J .
28, 1712.
Rottsieper, E. H. W. 1946. Vegetable tannins. Forestal Land, Timber and Railways
Co. Ltd., Harpenden. (Privately printed.)
Russell, A. 1935. The natural tannins. Chem. Revs. 17, 155.
Sando, C. E. 1937. Coloring matters of Grimes Golden, Jonathan and Stayman
Wmesap apples. J . Biol. Chem. 117, 45.
Scarborough, H. 1940. Deficiency of vitamin C and vitamin P in man. Lancet 11, 644.
Scarborough, H. 1945. Observations on the nature of vitamin P and the vitamin P
potency of certain foodstuffs. Biochem. J . 39,271.
Scarborough, H., and Bacharach, A. L. 1949. Vitamin P. Vitamins and Hormones 7, 1 .
Scarborough, H., and Stewart, C. P. 1938. Effect of hesperidin (vitamin P) in capillary
fragility. Lancet 11, 610.
Schrauffstatter, E. 1948. The bacteriostatic action of chalcone, flavanone, flavone and
flavonol. Experientia 4, 484.
Schrauffstatter, E.,and Bernt, H. 1949. Antibacterial action of curcumin and related
compounds. Nature 164, 456.
Seikel, M.K., and Geissman, T. A. 1950. The pigments of yellow Antirrhinum majus.
J . Am. Chem. SOC.73, 5725.
Shimokoriyama, M., and Hattori, S.1953. Anthochlor pigments of Cosmos sulphureus,
Coreopsis lanceolata and C.saxicola. J . Am. Chem. SOC.76, 1900.
Shiner, R. L., and Anderson, R. J. 1929. Anthocyanins in Ives grapes. N . Y . State Agr.
Expt. Sta. Tech. Bull. No. 152.
Shinoda, J., and Sata, S. 1931. Synthesis of polyhydroxychalkones and polyhydroxy-
flavanones. IX. Synthesis of citronetin and a few flavanones. J . Pharm. SOC.
Japan 61, 576.
Sieve, B. F. 1952. A new antifertility factor. Science 116,373.
Sondheimer, E.,and Kertesz, Z. I. 1948. Anthocyanin of strawberries. J . Am. Chem.
SOC.70, 3476.
Spannuth, H. T., McGuine, T. H., and Crapple, G. A. 1946. Improvement of keeping
quality of edible fats by some commercially available tannins. Oil &: Soap 23, 110.
Swingle, W. T. 1938. In “The Citrus Industry.” (Webber and Batchelor, eds.)
University of California Press, Berkeley and Los Angeles, Vol. 1. 129.
Tanret, C. 1886. On some constituents of the peel of the bitter orange. Acud. Sci.
Paris 102,518.
Thompson, D. L. 1926.The pigments of butterflies' wings. Biochem. J . 20, 73.

Tseng, K-F. 1938. Nobiletin. J . Chem. Soc. 1958, 1003.
Tsujimura, M. 1941. Flavones in green tea. Sn'. Papers Znst. Phys. Chem. Research,
Tokyo. 38, 490.
Weinberger, J. H. 1944. Characteristics of the progeny of certain peach varieties.
PTOC.Am. Soc. Hort. Sci. 48, 233.
Wellington, R. 1924.An experiment in breeding apples, 11. N.Y. State Agr. Expt. Sta.
Tech. Bull. No. 106.
Williams, A. H. 1952. Personal communication.
Williams, A. H. 1953.The tannin apple juice and cider. Chemistry and Industry 1965,
540.
Williams, B. L.,Ice, C. H., and Wender, S. H. 1952. The isolation and identification
of quercetin and isoquercitrin from black currants (Ribes nigrum). J . Am. Chem.
SOC.74, 4566.
Williams, B. L., and Wender, S. H. 1952a.The isolation and identification of quercetin
and isoquercitrin from grapes (Vitis vinijera). J. Am. Chem. Soe. 74, 4372.
Williams, B. L.,and Wender, S. H. 1952a.The isolation and identification of kaemp-
ferol and quercetin from strawberries (Fragaria chiloensis). J. A m . Chem. Soc. 74,
5919.
Williams, B. L., and Wender, S. H. 1953.Isolation and identification of quercetin and
isoquercitrin from apricots (Prunus armeniaca). Arch. Biochem. and Biophys. 45,
319.
Willstiitter, R., and Zollinger, E. H. 1915. Pigment of grapes and bilberries. Ann.
408, 83.
WillstSitter, R., and Zollinger, E. H. 1917. Pigment of grapes and bilberries. Part 11.
Ann. 412, 195.
Yamamoto, R., and Oshima, Y. 1931. A new glucoside, citronin, from the peel of
lemon ponderosa (Citrus limon Burm. f. ponderosa Hort.). J . Agr. Chem. SOC.
Japan 7 , 312.
The Color Problem in Foods
BY G. MACKINNEY AND C. 0. CHICHESTER
Department of Food Technology,
University of California, Berkeley
CONTENTS
Page
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11. Color in Selected Commodities.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Sugar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
...........................
4. Tomatoes and Tomato Products.. . . . . . . . . . . . . . . . . .

5. Peas, Spinach, and Similar Chlorophyll-Containing V
6 . Strawberry Preserves
111. Some Theoretical Considerations

1. Discrimination.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
2. The Uniform Chromaticity Scale (U.C.S.), . . . . . . . . . . . . . . . . . . . . . . . . 332
3. The N.B.S. Unit.. . . . . . . . . . . . . . . . . . . . . . . . . .
4. MacAdam Ellipsoids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5 . Alternative Spaces.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IV. General Considerations. . . . . . . . . . . . . . . . . . . . . . . . . .
V. Instrumentation .......................
2. Subtractive Colorimeters ......................... 340
4. Spectrophotometers
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
FOREWORD
“Most color measurements are made for the purpose of specifying a
particular color, frequently to ensure a color match. Unlike a dyestuff,
manufactured under controlled conditions, a foodstuff is subject to the
vagaries of the weather and other variables which affect the natural color-
ing matters. Consequently the color of a food becomes an important
criterion of quality. Is the color characteristic of what we expect from
good quality raw material, ie., is the color natural? Has it undergone
changes suggestive of deterioration in the foodstuff itself? How much may
the color vary before we feel the product to be of inferior quality? Answers
301
302 G. MACKINNEY A N D C. 0. CHICHESTER
to these questions of course depend on how strongly we are conditioned

to show color preferences. This in turn depends on the product. Butter
must be yellow, although considerable latitude in the degree of yellowness
is tolerated. A greenish discoloration sometimes observed in eggs, on the
other hand, is never acceptable.”’
I. INTRODUCTION
The science of color is hampered in its development by the fact that
we are all color-conscious, scientists, artists, and laymen alike. We resist
attempts to restrict the use of everyday terms in describing our experi-
ences, even though our description would thereby gain exactness. We are
equally unwilling to reject familiar terminology in favor of an entirely
new one, defined so as to be free from ambiguity.
As estimated by the Committee on Colorimetry of the Optical Society
of America (1953, p. 129), the “volume of the discrimination solid for
reflected’color is about 7,500,000 threshold elements.” (For a discussion
of discrimination steps, thresholds or limens, see Section 111, “Some
Theoretical Considerations.”)
We must therefore accept the view that a trained eye can differentiate
between a vast number of different shades of color. This can be achieved
only by direct comparison of series of colored samples, and even then the
number of distinguishable shades is limited by the illuminants used and
the background conditions under which the comparisons are made. Also,
the level of illumination is of critical importance, since different sensitive
elements in the retina, the rods and cones, are responsible for vision a t
low and high levels, and the type of response, i.e., the color perception, is
different in the two cases.
The overriding color problem is to obtain an objective, precise, and
reproducible procedure for measuring quantitatively the differences which
can be perceived. To do this, we must impose certain restrictions on our
concept of color and its attributes. “Color consists of t’he characteristics
of light other than spatial or temporal inhomogeneities,” light itself then
being defined as the aspect of radiant energy responsible for human
vision. The Committee on Colorimetry (1953) can thus concentrate on
what Troland (1929) has called “opaque surface character,” which can
be determined by reflecting objects with highly diffusing (matt) surfaces.
These surfaces, according to Troland, “carry most of the typical color
phenomena of everyday experience.” As we depart from the diffuse
opaque surface, we become increasingly concerned with gloss (specular
1 From a talk by G . Mackinney on “Color Measurement of Different Commodities,”
Quartermaster Food and Container Institute Symposium, November 3-4, 1953,
Chicago, Illinois.
THE COLOR PROBLEM IN FOODS 303
or mirrorlike reflection), transparency, and turbidity. The specular com-

ponent of reflection is eliminated instrumentally in most cases, and cor-
rections can be made for turbidity. When we come to the transparent
colored body, further restrictions are necessary if we are to specify its
color successfully. If we have two portions of a given transparent object
and view them under identical conditions, except th a t the light paths
differ in length in the two cases, we shall find two colors differing not only
in the brightness factor but in chromaticity2 a s well. The color of a trans-
parent bulk commodity is thus uniquely defined only when the length of
the light path is included in our conditions of measurement.
I n our presentation, we omit discussion of many charts and dic-
tionaries of color ( e g., Ridgway, Maerz and Paui) which may be familiar
t o many. We justify the omission on the ground that more precise or more
convenient systems exist for measurement of color in foods. This is not
to argue that these charts have outlived their usefulness or th a t they will
necessarily be abandoned. We shall emphasize, in our discussion, the
usefulness of the C.I.E.3 system (and the 2, y, chromaticity diagram) and
also the Munsell color space. Because one instrument (Hunter Color-
Difference Meter) discussed here measures quantities in a transformed
color space, we cannot safely ignore the problems arising from non-
uniformity of color spaces, and procedures for effecting transformations.
I n this introduction, we shall attempt, for the reader unfamiliar with
these concepts, an elementary discussion of the problems of color space
and specifically of the C.I.E. color space. We shall begin by considering
the colors, R, G , and B of any convenient red, green, and blue light sources
a t the apexes of a Maxwell triangle (Fig. la). We can mix these primaries
in any proportions we wish, and the resulting colors must lie within the
confines of this triangle. I n the center, we shall have white. We cannot,
however, reproduce any color which is redder than our selected red
primary, and the same limitation applies to green and blue colors. T o
represent all possible colors in a plane, resort is made to imaginary or
unreal primaries. This is purely a mathematical device, somewhat analo-
gous t o asserting one’s net worth not merely in terms of a n outstanding
bank balance, but in addition, whatever overdraft the bank might be
induced t o carry, and whatever credit one might secure. The base line for
calculations is simply changed.
Extending the corners of our Maxwell triangle, to locate our new
2 By chromaticity, we refer to attributes equivalent to hue and saturation, inde-
pendent of photometric brightnes!, lightness, or value.
Commission internationale de 1’Eclairage formerly I.C.I., the Internationa1,;Com-
mission on Illumination. Reports of this Commission are to be found in the Journal
of the Optical Society of America.
primaries wherever we decide, we may again make mixtures, and in this

space there will be a line of demarcation (the plot of the pure spectral
colors) between what is real and what is imaginary. The line joining the
extremes of this plot delimits the purples. We thus create a chromaticity
diagram, Fig. lb. This, however, does not completely define the color,
which may be dark or light. Reverting to the original Maxwell triangle
for a moment, we may visualize our central spot varying through an
infinite series of grays from white to black. We complete our definition
4 S p e c t r u m locur
‘pu r p I c s
FIG.la. Maxwell triangles (triangular co-ordinates). The color triangle obtained by

use of primaries red, blue, and green (R, G, and B) chosen for convenience or availa-
bility. They do not have to form an equilateral triangle, nor is white necessarily a t the
precise center. Even if pure spectral colors had been selected, the plot of the spectrum
locus shows that no three real primaries can be selected which will match all possible
colors.
of a color, therefore, by erecting a perpendicular to the chromaticity
plane along which the apparent brightness may be plotted. We have thus
created a volume which can be referred to rectangular co-ordinates, and
which constitutes a “color solid” or “color space,” as in Fig. 2. The
International Commission on Illumination provided in 1931 a standard
observer, standard (though unreal) primaries, and standard illuminants
and the basis for a standard system of co-ordinates.
In matching colors by reference to three primaries, we are dealing in a
trichromatic system. The X Y 2 tristimulus values of C.I.E. space are
created by simple linear transformation of the color space of R G B .
Given any second system, R1, GI, B’, the two may be related thus:4
In Figure la, we locate a color by the proportions of R, G,and B required for a match.
In a transformation, the absolute amounts of R, G, and B must be used.
R' = alR + blG + clB

G' = azR + b& + czB
B1 = arR + b& + c3B
With nine parameters, a1 to c3, and eight degrees of freedom, we have
considerable flexibility and some choice as to the properties of a new space
G'
FIG.lb. The same spectrum locus is traced. The primaries R1, GI, and B 1 are no
longer real but the area enclosed by the broken line contains all possible real colors,
and they are all reproducible by appropriate mixtures of R1, GI,and B1. This is not
possible in all cases where R, G, and B are used, unless we are willing to adopt other
artificial devices, such as negative colors. Thus R might be made equivalent to R 1 by
adding to it negative amounts of primaries B and G, represented by the blueness and
greenness differences inherent in (R1 - R). The enclosed area constitutes a chroma-
ticity diagram.
in any transformation we may propose. The X , Y , and 2 tristimulus

values thus enable us to locate a color in the color space, but they are in-
convenient for direct comparison. Trichromatic coefficients (x,y,z) were
therefore derived, such that
X Y
x=X+Y+Z) y=x+Y+z'
etc., and consequently x y+ + z = 1. Hence a color is fully defined by
z, y, and Y . It is unfortunate that the C.I.E. color space is nonuniform
306 0. MACKINNEY AND C. 0. CHICHESTER
with respect to perceptibility differences. Thus if we examine two colors

separated by a given distance in the red with two separated by the same
distance in the green, the differences in color perceived by the eye will be
smaller for the latter.
As we approach the extremes of black and white, our ability to dis-
tinguish the colors becomes increasingly uncertain. The color solid is
X
FIG.2. The C.I.E.ChromaticityDiagram (rectangular co-ordinates).This diagram
constitutes a transformationof Fig. l b with rectangular co-ordinates in terms of 2, y,
the C.I.E. trichromatic coefficients derived from the original X Y 2 tristimulus
values,
The lines OD, ODI, OD11,are lines of constant dominant wave length.
The inner figure indicates a locus of constant purity.
therefore highly irregular in shape, and this is true of any color space
including the Munsell.
From the foregoing, it should be clear that three measurements are
needed to determine color. It is not essential that we use a trichromatic
system. A color match may be achieved by mixing white with a spectral
color (except in the case of purples, where the complementary green must
be subtracted). The spectral color required for a given match is termed
the dominant wave length. The degree of saturation is given by the per
cent purity (100%white = 0% purity, and 100% spectral color = 100%

purity) , and a measure of photometric brightness completes the specifica-
tion. In the C.I.E. space, we may designate them by 2, y, and Y or,
alternatively, by conversion to the quantities termed dominant wave
length, purity, and lightness. In the Munsell space, the appropriate terms
are hue, chroma, and value.
It is disconcerting that, while the mind can readily grasp these
distinctions and recognize these three attributes of color, they are not so
Purity. per cent
FIQ. 3. Whiteness diagram. For dominant wave length 575 mp. This diagram
was suggested to us by Professor A. C . Hardy. Original data may be found in articles
by Judd, Paper Trade Journal, 1935, 1936.
clearly differentiated by the eye. The whiteness diagram, Fig. 3, illustrates
this point. If we follow the parabola passing through the point-purity
0.4,brightness 7&to the point represented by purity, 2.0, brightness 76,
the same perceived whiteness is observed. In other words, the eye will
accept as equivalent in whiteness, specimens with different purities, pro-
vided the brightness (lightness, in present terminology) is suitably ad-
justed. Similar adjustments are made by the eye with respect to hue-
chroma or dominant wave length-purity changes. This has been observed
in tomatoes, where perceived redness is not solely an attribute of hue. To
some extent, this may be helpful, in forcing us to recognize that all three
308 G. MACKINNEY AND C. 0. CHICHESTER
attributes of color determine the perceived color, that the lightness scale
is not, outside the central gray point, a purely neutral component.
The historically earlier Munsell color space was developed as a result
of an entirely different concept. The C.I.E. system permits a unique
specification of a particular color in a reference framework that, regardless
of certain conveniences, is highly artificial; whereas the Munsell system
is an attempt at systematic sampling of the color solid, by use of perma-
nent reproducible standards. The hues are arranged on a 100-point
circular scale from 1R (red) to 10R, then 1YR (yellow red) to lOYR, and
similarly through Y to YG, G (green), BG, B (blue), BP, P (purple) and
from lORP (red purple) back to 1R. The 10-point value (lightness) scale
normal to the plane of the circle ranges from 0 / , black to value lo/,
white. The chroma or saturation is arranged radially from the central
achromatic gray, chroma / 0 , to that of the fully saturated hue. The
numerical value depends on the saturation attainable for a given hue and
value, As an example, for hue 5R, attainable chromas at values 1/ and 7/
are approximately /11 and /16 respectively. As Judd (1952) observed, a
collection of color chips based on the radial organization favors a closer
sampling near the achromatic gray, and a correspondingly poorer sam-
pling of the saturated colors. Judd’s discussion of systematic sampling of
the color solid is particularly interesting in its suggestion of a collection of
chips organized on a three-layer packing principle.
Because the Munsell collection has now been evaluated in terms of
C.I.E. space, accurate interpolations are possible. The most accurate
Munsell specifications are now based on C.I.E. data, and it may well be
that the greatest value of the Munsell space in the future will be in
enabling us to see how a specified color is perceived under standardized
conditions. The Munsell quantities are uniformly reported in terms of
hue, value, and chroma, in this order. Thus the designation 7.5YR/4.1/5
signifies a hue of 7.5YR, a value of 4.1, and a chroma of 5.
Five important treatises on color have appeared in recent years.
“Handbook of Colorimetry” (Hardy, 1936) is a direct outcome of the
eighth session of the International Commission on Illumination in 1931.
It enables us to compute psychophysical values for brightness (later
termed luminous reflectance and now, lightness) and chromaticity
(dominant wave length and purity) from the raw spectrophotometric
data. In these psychophysical terms, a colored sample may be uniquely
defined with respect to its color, provided that we follow a standard
procedure for determining the spectral reflectance or transmittance of
the sample. The perceived color, however, is not defined, and we enter
here into psychological considerations as distinct from those of psycho-
physics. The C.I.E. quantities z and y (the trichromatic coefficientswhich
define the chromaticity) and Y (the tristimulus quantity which defines

the lightness) are directly computed by aid of a standard observer,
standard illurninants, and standard primaries from the spectrophoto-
metric curve, by use of tables compiled by Hardy and his associates. The
nontechnical reader for whom this presents difficulty is advised to begin
with three Tintometer pamphlets (G. J. Chamberlin, 1951; A. J. Fawcett,
1952; Tintometer Ltd., 1950; or with General Electric Bulletin LD-2 by
Sturrock and Staley, 1950).
“An Introduction t o Color” (Evans, 1948), presented in nonmathe-
matical terms, is especially useful in its discussion of color perception and
of the distinction between color as calculated in psychophysical terms and
color as perceived by the mind. This author points out that a colored
object or scene may yield knowledge, create an illusion, or even involve
hallucination. Such experiences are explained simply by Evans as follows :
when we think we know the color of an object, we do not look to verify
the assumption. Consequently, we can be victims of a genuine hallucina-
tion. The observer does not look at the stimulus but relies on his memory.6
It is important to realize that the definition of color psychophysically is
in essence a definition of what a carefully standardized eye receptor is
presumed to record in response to a given stimulus, under prescribed con-
ditions, and has nothing to do with how it may be interpreted by the
brain.
The Dutch edition of “Physical Aspects of Colour” was completed
by the late P. J. Bouma in 1945. It has been translated into English with
some additions by de Groot (Bouma, 1947). The concept of color space is
extremely well developed, together with the problem of transformation
from one system to another, as, for example, the U.C.S. system (uniform
chromaticity scale), in which ideally the colors would be uniformly dis-
tributed over the color plane, as judged by the eye. The historical develop-
ment of color science is also thoroughly treated.
The National Bureau of Standards has been in the forefront of research
in color measurement and in the development of lightness and chromatic
6 The psychologist is unlikely to agree with the interpretation tha t hallucinations are
involved. As Professor D. Kresch has observed: “The relation between stimulus and
sensation is not simple, but it is systematic.” The example frequently cited is the
result achieved b y the experimenter who uses spinning discs to match the color of a
purple banana. The disc is invariably too yellow, apparently because of the experi-
menter’s knowledge t h at the ripe banana skin ought t o be yellow. There are those
who would analyze problems of color preferences in terms of the Gestalt, i.e., in con-
tezt, and a t the other extreme, in purely subjective terms. Granger (1952)seems to
steer a middle course, concluding t h at “color preferences are objective in the sense
that: (a) they are to a considerable extent independent of personal taste, and (b) they
are dependent to some degree on inherent stimulus properties.”
310 Q. MACKINNEY AND C. 0. CHICHESTER
scales. “Color in Business, Science and Industry” (Judd, 1952), written

by a member of the Bureau of Standards, is therefore especially valuable
for its analysis of the problems; the worker with color measurements to
make will find detailed appraisals of methods and available instruments,
and assistance in the procedures for computation.
The latest of the five treatises discussed here is entitled “The Science
of Color” by the Committee on Colorimetry of the Optical Society of
America (1953). A slight letdown is possibly inevitable, which a rereading
does not entirely banish. The report is designed to include an elementary
discussion to attract and hold the attention of casual readers, with a
gradual transition to more advanced exposition. The food technologist
tackling the psychophysics of color is apt to be impatient when con-
fronted with distinctions between operational and relational definitions.
He learns that “the use of quantitative concepts would specialize the
relational definitions in an undesirable manner” (loc. cit., p. 220). Con-
tinued study makes the reader more aware of the problems facing the
Committee. Is color to be defined as a sensation or in terms of perception?
How many attributes of color are to be admitted?
The five treatises necessarily cover much common ground. “Hand-
book of Colorimetry” will remain a standard so long as color is specified in
terms of the 1931 C.I.E. conventions. The other books cover a wider field,
and collectively they provide a balanced perspective which an individual
viewpoint can rarely achieve, and all five have proved their worth to the
present writers.
To some extent, the food technologist is today at a crossroads, and
decisions which are made in the next few years will govern our whole
approach to the problem of color measurement and color specification.
Shall we, for example, specify color in terms of the C.I.E. quantities, or
shall we proceed by measurement of color differences? How do the Mun-
sell, Lovibond, and other systems fit into such a plan?
In a book devoted to advances in a field, we cannot write a treatise on
color, even if it lay within our competence to do so. We must therefore
presuppose that the reader will acquaint himself with the concepts dis-
cussed in the above-mentioned monographs. Additional background is
furnished by the following National Bureau of Standards publications :
Spectrophotometry (Circular 484), Photoelectric Tristimulus Colorimetry
with Three Filters (Circular C429) , Colorimetry (Circular 478). We shall
refer to these and to research articles in the Journal of the Optical Society
of America, when necessary, and proceed to discuss their application to
foods, emphasizing those problems with which we have had direct, if
limited, experience.
As the first reviewers on color for Advances in Food Research, we feel
justified in making an individual selection. That subsequent reviewers
will supplement our choice and modify our interpretations should be

obvious. " Colour, in Theory and Practice "by Murray (1952) for example,
has just come to our attention via a review in Nature, and numerous other
monographs and papers might well have been cited, but it has not seemed
necessary to our purpose to attempt completeness at this stage.
In the last analysis, we are interested in color in foods because the
consumer of foods has certain color preferences. He will buy and eat prim-
arily according to those preferences. It is surprising how revolting an egg
can appear, particularly when poached, if the hen has been fed a diet
adequate in vitamin A but devoid of carotenoid. The yolk is an unnatural
ivory-white, though perfectly normal from a nutritional standpoint. It has
frequently been stated that color involves problems in physics, psychology,
philosophy, and aesthetics. It is not unfair to caution the reader that in the
opinion of some of our colleagues in psychology and philosophy, only the
physical aspects have been adequately treated.
We have arbitrarily chosen seven commodities around which to build
our discussion. The selection is governed by the fact that work has been
done on these commodities which serves to illustrate the problems in deter-
mining color in different regions of the color plane: whiteness in sugars
and flours; yellowness in oils; greenness in certain vegetables; redness in
tomatoes and tomato products; darkening of strawberry preserves; the
color of grape juices, jellies, and wines. Three of these commodities are
standard articles of international commerce: sugar, flour (more correctly,
wheat), and oils, and the price is determined by the condition of the
respective world markets for sugar, grain, and oils. This cannot be said of
wines, and certainly not of canned fruits or vegetables.
Consequently for the first three, we may expect to find professional
groups keenly aware of the necessity for describing an offering in terms
that can be understood in the world market. Thus in this country there is
established the New York Sugar Trade Laboratory and a United States
National Committee on Sugar Analysis, an affiliate of the International
Commission for Uniform Methods of Sugar Analysis. This is probably the
most organized of the three.
By contrast, we have a series of local markets for canned fruits and
vegetables. Even though standards prevail on a nationwide basis, these
are not always adequately defined. A concerted attempt is being made,
therefore, to improve color specifications. This is especially true for
tomato products.
11. COLORIN SELECTED
COMMODITIES
1. Sugar
The individual sugar crystal is transparent. The whiteness of the mass
is due to reflection of light from surface and subsurface layers. The light
3 12 G. MACKINNEY A N D C. 0. CHICHESTER
from subsurface layers undergoes refraction as it passes through the

crystals, and the quality of the light transmitted is affected by selective
absorption to the extent that colored impurities may be present.
Measurement of color of sugar products has been discussed by Browne
and Zerban (1941). We shall omit consideration of Stammer’s colorimeter
and the Stammer degree scale, even though this has been extensively used
in the past by the European beet-sugar industry. The modern approach
requires transmittancy determinations on colored sugar solutions and on
“liquid sugars,” while reflectance measurements may be made on solid
sugars. I n 1952 neither the Association of Official Agricultural Chemists
nor the International Commission for Uniform Methods of Sugar Analysis
and its affiliated United States National Committee on Sugar Analysis had
adopted standard methods for determining the color of sugars and sugar
products, although there was general agreement that for a standard
method of color determination, precise photoelectric spectrophotometers
would have to be used. There was still a controversy about the proper
method for preparing the solutions the transmittancies of which were to
be measured.
According to Zerban et al. (1951), Peters and Phelps of the National
Bureau of Standards first devised a procedure for determination of the
color of sugar and sugar products based on an “absorbancy index”
( a 5 ~ 0a)t 560 mp. This index is the negative logarithm of the transmittancy
for a solution of concentration of 1 g. of dry substance in 1 ml. of solution
for a thickness of 1 cm. The value a t 560 mp (‘ . . . is equivalent colori-
metrically to the sum total absorption over the visible range.” Zerban
et al. (1951) provided experimental verification by measurements of
transmittancies a t various wave lengths of 60% ’ solutions filtered through
Celite filteraid, of seventy-six different refined sugars. The dominant wave
length varied from extremes of 578.2 to 572.0 mp, and the purity from
4 to 63.1.
A similar study on raw cane sugars by Zerban et al. (1952) leaves no
doubt that the index a660is equally useful as a single-value function t o
describe the color of a sugar solution whether refined or raw.
Two questions arise in connection with these studies: how was it
possible to develop a single value function for expressing the color of a
sugar, and why should a single measurement a t 560 mp be a practical
solution to the problem?
As noted above, Zerban et al. (1952) determined the tristimulus values
by use of ten selected ordinates and Illuminant C (Hardy, 1936), and
dominant wave length, purity, and brightness were recorded. Corre-
lating asBO (or T 6 6 0 ) with these three components determining the color,
they obtained a multiple regression equation in which the influence of
T H E COLOR PROBLEM IN FOODS 313
each on a680 (or Taco) is determined. Separate expressions are needed for
raw and refined solutions, but the conclusions are identical, th a t a t 560
mp, the influence of purity and dominant wave length is small compared
with the effect of the brightness component. The following facts are t o be
noted: first t ha t the dominant wave for the solutions, centering around
575 mp, is not far removed from the peak (555 mp) for the visibility
function, on which the brightness is based. The dominant wave length is
in essence a spectral center of gravity. Small variations at high trans-
mittancies in this region should not have any marked influence. T h e
purity is a measure of the distribution on either side of this spectral center.
Here we observe differences. As one may deduce from the published
transmittancy curves for the refined sugars, on the long wave length side,
ie., above 575, the distribution is essentially uniform, while below 575 mp
divergencies are increasingly apparent with increase in yellowness. T h e
yellowness of the raw sugars differs significantly from that of the refined,
and the transmittaricy ratios at 420 mp relative to 560 mp are substan-
tially higher for the latter. Thus a 4 2 0 bears a less direct relation to the
brightness. “The refined sugars have throughout a higher purity at equal
brightness than the raw.”
It is t o be expected, therefore, for this type of problem, where the
dominant wave length is reasonably close t o 555 mp and where the
photometric brightness is high, that a single-value determination around
555 or 560 mp should be obtained which can be correlated with the visu-
ally observed color. An accurate absorbancy index requires a nonturbid
solution. Provided such solutions have the same purities and the same
dominant wave length, a single measurement a t any arbitrarily chosen
wave length below 575 mp would be satisfactory. Selection of 560 mp has
the advantage of minimal differences in slope, or purity, in the event th a t
the purities are not strictly equivalent in the solutions under examination.
Gillett et al. (1949) approached the whole problem in a different
manner. They did not attempt to obtain a clear solution but corrected for
turbidity by a measurement in the near infrared (using a Wratten No.
88A filter which does not transmit below 720 mp). Turbid matter was
then added t o 50% sugar solutions and straight-line plots were obtained
for a series of filters for concentration of turbid matter as a function of
- log T. They then determined the color by a measurement with a
Corning No. 554 filter, maximum transmittance a t 420 mp. This choice
emphasizes yellowness in the solution. The turbidity is then determined
from the infrared reading, and the 420 mp reading is corrected for this
degree of turbidity as it affects the No. 554 filter. It is not clear to us
that this precise procedure has been used on products varying from
<<
. . . light-colored washed raw sugar liquor . . . to dark-colored
314 G . MACRINNEY AND C. 0. CHICHESTER
affination green sirup . . . ” (Meads and Gillett, 1949), but the crux
of the problem is now clear.
As already stated, a definitive answer is possible in terms of an
absorbancy index (or transmittancy) which can be correlated with the
color in terms of the C.I.E., provided always that we are dealing with
nonturbid solutions. Regardless of the practical difficulties in achieving a
clear solution without loss of color, this tying-in with the C.I.E. system
is an enormous asset to the value of the absorbancy index. With regard
to the procedure of Gillett et al. (1949) its successful application is predi-
cated upon two assumptions: (1) that the turbidities of the tested
products are due t o turbid matter with essentially the same properties
(including the light-scattering power) as the turbid matter used (ben-
tonite) in calibrating; and (2) that the corrected second reading a t 420 mp
is a direct measure of the concentration of coloring matter of reasonably
uniform composition. The latter requirement means that the ratios of
- log T a t any two wave lengths we may care to choose shall be constant
throughout the series of samples under testsEThe method is attractive
because of its speed and simplicity, and it works undoubtedly because the
assumptions are valid the majority of the time. Zerban et al. (1952), how-
ever, have demonstrated exceptions to the procedure. Furthermore it is
tacitly implied that color and turbidity are separable. If it be argued that
the turbid matter cannot be removed without modification of the color
of the resultant clear solution, we are immediately confronted with the
dilemma that the color (corrected for turbidity) describes a system we
cannot hope to see in practice. This is not to minimiee the practical
usefulness of the method in routine control.
Wiklund (1950) in his report to the International Commission for
Uniform Methods of Sugar Analysis analyzed the difficulties yet un-
resolved. He pointed out that a source of error in transmittance measure-
ments lies in polarization differences between standards and unknowns.
He cites Evans (1948, p. 213): “ . . . A system of material standards
having the same energy distribution over the range expected in the sample
may be made to give very high precision results . . . Such systems, of
course, may be correlated satisfactorily with another system such as the
I.C.I. because, in effect these eliminate the observer as a variable.” We
shall revert later to the question of material standards. Whalley (see
Wiklund, 1950) remarked that the infrared and 420 mp measurements
by the method of Gillett et al. (1949) apply only to high-grade refined
sugars, and even then are dependent on particle size of the turbid
matter.
Since the impurities are yellow, we shall stipulate that the choice shall be between
400 and 550 mp.
In March 1953, the United States National Committee adopted the

following procedure: the Tsao measurement is now limited to raw cane
sugars, the solutions being prepared and filtered according to specific
instructions. Because the solutions still contain fine particles, the Taao
reading is converted to an “attenuation index,” the term “specific
absorptive index” being abandoned. In the case of white sugars, the
recommendation is that the color be determined by measurements at
420 and 720 mp. Since the white sugar solutions cannot be easily filtered,
a correction is applied for turbidity to the 420 mp reading. Experimentally
some variation is found in the relationship between turbidities at 420 and
720 mp, respectively, though it is approximately twice the attenuation
index a t 720 mp. The factor 2 is therefore chosen, and the color of the
sample is given by the measured index a t 420 minus twice the index a t
720. It may be further added that Gillett (1953) has reviewed the whole
subject of color measurement in sugars and sugar products.
In the grading of sugar cane sirup and edible sugar cane molasses,
Broeg and Walton (1952) discussed the use of inorganic salt solutions
developed by the American Molasses Company. These are used for
grading products according to the permissive grades of the United States
Department of Agriculture for sirup and molasses from sugar cane. This
of course is recognized as a temporary expedient.
Bruce and Turner (1952) developed polished glass standards, now in
use, for determining color specifications and tolerances for maple syrup
and honey, based on the C.I.E. system. They can be used for both clear
and cloudy liquids.
Reflectance measurements of white sugars have been made by Gillett
and Meads (1952). They observe that the amount of color is extremely
small, that 94% or more of the incident light is reflected by high-quality
white sugars of normal granulated grain size. Even lower quality sugars
will reflect 90% of the incident light. They further remark that “more
finely divided particles appear whiter than coarse particles, both to the
eye and to photoelectric devices. Furthermore differences in luster in-
fluence results, and differences in apparent grayness sometimes confuse
the detection of yellowness in the sugar.”
The single-value reflectance measurement made by Gillett and
Meads (1952) is not immediately convertible to C.I.E. values. A light
blue Corning Daylight type No. 590 filter with a 200-w. projection lamp
simulates daylight from the north. The photogenerative type of cell used
(Gillett and Holven, 1943) is presumably a selenium cell with character-
istics given in Fig. 22 of the bulletin by Sturrock and Staley (1950). This
requires special filters if it is to be corrected to approximate the spectral
sensitivity of the human eye. Otherwise, it is somewhat more sensitive
316 G . M A C K I N N E Y A N D C. 0. CHICHESTER
in the red and substantially more so in the blue. For detecting traces of
yellowness, the unmodified sensitivity would seem more desirable.
2. Flour
Flour is opaque and its whiteness is due to nonselective diffuse surface
reflectance. There has not been such interest, internationally, since the
world market is for the whole grain, rather than flour. Thus the profes-
sional group in the United States (American Association of Cereal
Chemists) has no committee on the measurement of flour color, although
collaborative studies have been made on methods for determining the
yellow pigment content-chiefly carotenoid in nature-of wheat flour.
This lies outside the realm of the present review. Four main factors
determine flour color, according to Kent-Jones (1952) :
1. The grade of the flour (contamination of endosperm with bran).
2. The yellowness (carotenoid pigments not completely removed by
bleaching).
3. The granularity-(the larger the granules, the darker and duller the
flour).
4. Dirt, smut, etc.
The yellowness of flours is caused by selective absorption of blue by
yellow-colored naturally occurring pigments or impurities. The problem is
strictly twofold: measurement of the total reflectance (a brightness or
lightness value) and an independent measurement in the blue. If a flour
is gray owing to inclusion of dark particles, the total reflectance or
luminosity is decreased by nonselective absorption so that only one
measurement, that of diffuse reflectance, is needed. It is interesting to
note that the flour technologists appear to ignore yellowness in reflectance
measurements, presumably because this can be controlled by the bleach.
If not, the yellow is determined indirectly by extraction of the yellow
coloring matter, which is then estimated as such. Thus Irvine and
Anderson (1952) deliberately choose to make reflection measurements a t
600 mp to ensure that the carotenoids will not affect the answer. Kent-
Jones (1952) uses a filter with maximum transmittance at 530 mp where
the pigments still do not have too much effect. Investigations are far less
advanced here than in the sugar industry. While the single-value measure-
ment for color may be justified, results to date consist of photometric
brightnesses a t some particular wave length which may or may not
correlate with visual brightness, depending upon the nature of the im-
purity. The empirical nature of the work does not impair its usefulness,
though the terminology may cause confusion. The usefulness is shown in
relating this measure of “brightness” in the flour with the crumb color
of the loaf. Irvine and Anderson (1952) have set up an arbitrary score to
demonstrate such a relationship.
It may be pertinent here to review briefly the development of the
Flour Color Grader (Kent-Jones and Martin, 1950; Kent-Jones, Amos,
and Martin, 1950). Prior to the development of this instrument, solvent
extraction of the bran pigments was used. This is time-consuming and is
not an accurate measure of the flour grade. An ash test was also much
used. It depended on the low ash content of the endosperm, 0.3%, con-
trasted with 8.0% for the outer coverings. The Color Grader with its
speed and accuracy was therefore certain of favorable reception. We may
properly inquire, however, whether this method should not evolve in the
direction taken by the sugar industry. The Gillett-Meads instrument
with filters to approximate the spectral sensitivity of the eye would yield
a single-value determination essentially equivalent to photometric bright-
ness; the industry is interested primarily in grayness, since measurements
have been made a t 600 and 530 mp with apparently equally satisfactory
results. Finally, the measure taken must be related to some desideratum,
e.g., crumb color. Consequently either the C.I.E. brightness or an ab-
sorbancy index ax (where the most favorable X has still to be determined)
must be correlated with a corresponding quantity of industrial value
(presumably crumb color).
3. Oils
The American Oil Chemists’ Society (1946) has published a loose-leaf
book of methods which are subject to continuous scrutiny and revision.
The following methods for determining oil color are given:
1. Official Method Cc 13a-43, applicable t o animal fats and to all fats
and oils too dark to be read by the Wesson method.
2. Official Method Cc 13b-45 (Wesson method) corrected November,
1947, applicable to all normal fats provided the samples are not turbid.
3. Tentative Method Cc 13c-50, revised October, 1951, applicable to
cottonseed, soybean, and peanut oils.
4. Tentative Method Cc 8d-48 (formerly Cc 8d-47), revised January,
1948, refined and bleached color, applicable to tallows and greases for
soap. This deals with samples requiring special treatment prior to appli-
cation of the Wesson method. Since it is a special case of the latter,
involving no differences in principle in the color determination, it will not
be considered further.
5. Tentative Method Ka 3-47. This method utilizes a series of eighteen
Gardner Standard Colors and is applicable to natural and synthetic
drying oils of the same hue as the Standards. These consist of inorganic
salt solutions, and a simple direct comparison is made, so that the sample
receives a numerical rating from 1 to 18, or “darker than 18,” according

to its position on the Gardner Scale.
The first method employs F.A.C.’ Standard Colors. These consist
of inorganic salt solutions also. Unlike method (5), applicable to oils of
constant hue, this series of solutions makes provision for changes in hue.
The fats are classified in five groups as follows: light-colored, predomi-
nantly yellow, dark (red cast), very dark (predominantly green), and very
dark (predominantly red).
The second method employs Lovibond glasses. These glasses have
been evaluated in terms of C.I.E. quantities; this is of great importance
because it represents the first case in this discussion where such a possi-
bility is embedded in an official method.
For comparison, a standard white block of magnesia is required, and
specifications for a (‘Wesson-type” colorimeter are approved, together
with provisions for the illuminant and for viewing by visual means. The
conditions are very carefully specified. Thus in the viewing booth, the
level of external illumination a t the top of the colorimeter box shall be
not less than 1 nor more than 5 foot-candles. This obviates errors due to
the Purkinje effect involving too low a light intensity (rod vision), yet
extraneous light is held to a relatively low level.
The Lovibond grading of vegetable oils is discussed in some detail
by Judd (1952, pp. 209-212). It was shown by McNicholas (1935) and is
obvious from the classification of oils in method Cc 13a-43 that there are
two groups of pigments, green (chlorophyll-containing) and yellow
(frequently due to considerable quantities of carotenoid, but probably
also a t times to brownish pigments in varying amount). As Judd points
out, these vary independently.
Let us now take a specific example of the procedure. The official
method is as follows: “a. Refined Oil. Use only 1 yellow glass, 35 yellow
for refined cottonseed oil and refined peanut oil, 70 yellow for refined
soybean oil. Use not more than 2 red glasses up to and including 13.0 red,
and not more than 3 red glasses above 13.0 red. ”
To digress for a moment, color matches are of two kinds. Thus an
unknown carotene solution may be matched against a standard carotene
solution or against selected Lovibond glasses. I n the former, the spectral
composition of the transmitted light is identical in the two cases. They
can differ only in brightness, and when a match is achieved, the colors
will appear identical regardless of the illuminant. I n the second, the
spectral composition of the transmitted light is not identical, although
the match may be readily achieved. It is valid only under specified condi-
tions of illumination and viewing. This type of match is said to be
7 Fats Analysis Committee, A.O.C.S. and A.C.S.
metameric. Furthermore, since the match involves judgment on the part

of the observer with respect both to hue and to brightness, it is more
dependent on the observer than the former (non-metameric), where only
one variable, brightness, is involved.
Matches following the official method just cited are clearly metameric.
As Judd correctly points out, no one-dimensional grading system can
yield perfect color matches, because of variability in the two pigment
groups which determine the color of an oil. The official method is one-
dimensional, as it prescribes a single yellow, e.g., 35Y, 70Y, etc., with a
variable red, for this particular category of oils. This has probably been
an inevitable development, since a two-dimensional system would have
led to uncertainties in interpretation. Furthermore, the difficulties are not
too great in practice, or the method would not enjoy the confidence of the
industry.
Method Cc 13c-50, a photometric method, is designed to obviate some
of the difficulties mentioned under use of the Lovibond. The measure-
ments are made a t selected wave lengths, for which empirical constants
are developed to correlate the reading with the Lovibond. This reading,
the so-called “photometric color ” D, a modified optical density, is defined
thus:
+ +
D = 1.29D4~0 69.70560 41.2Dazo - 56.4Dmo.
The method is based on collaborative work by the Color Committee of

the Oil Chemists’ Society. Thomson (1947) related the thinking which
led to the selection of specific wave lengths for the measurement. The
spectral transmittances of three groups of oils are given, light, medium,
and dark. Work at Swift and Company apparently led originally to the
selection of 400, 540, and 670 mp to cover a wide range of oil colors. Dark
tallows do not transmit appreciably below 600 mp. Light oils, on the other
hand, may transmit from 5 to as much as 50% at 420 mp (5-cm. cells).
In the absence of chlorophyll, the corrections at 670 and 620 mp will be
negligible, except possibly for a dark tallow, where the contribution a t
620 mp may be appreciable.
An interesting extension of the photometric color method of the
A.O.C.S. is found in the work of Moster and Prater (1952) on their
measurement of color of Capsicum spices. The common method used at
present (according t o these authors) involves matching an alcohol
extract of the spice against combinations of red and yellow Lovibond
glasses. “Variations (in the yellow) have only a slight effect upon match-
ing the red glasses, and it has become a widespread practice to use a
20-yellow glass and express the Lovibond value as the number of addi-
tional red units needed to obtain a match . . . the widespread experience
of the industry indicates that this color does correlate with the tinctorial
value of these spices.” It appears from further observations that the same
problem arises with Capsicum when the experimental variability is
limited to one dimension. Moster and Prater therefore developed a
spectrophotometric procedure which they call the Gentry method. They
prepare alcohol extracts which transmit insignificant amounts of blue.
Therefore the 2 tristimulus value, and hence z, the trichromatic coeffi-
cient, are both zero. Under these circumstances x defines the chroma-
+ +
ticity, since z y 0 = 1. Thus, from the properties inherent in the
extract, they have a one-dimensional chromaticity scale which can be
compared directly with the one-dimensional red scale of the Lovibond.
Because, as before, two pigment groups are involved, carotenoids and
chlorophyll or chlorophyll derivatives, they required measurements a t
a minimum of two wave lengths. For chlorophyll, 663 mp is a natural
choice, the point of maximum absorption in the red for chlorophyll a
in the solvent used. The transmittances of chili and paprika extracts
below 540 mp are so low, except a t high dilution, that choice of the second
wave length is limited to the yellow and orange regions. Several wave
lengths were considered, and because the color range for the samples was
great, from 11 to 36 red units on the Lovibond, it was necessary to place
the samples in two groups. The wave length 569 mp was selected for low
Lovibond red numbers, and 577.5 mp for the high. The authors showed
that for a substantial overlap, either wave length might be chosen. Two
equations were then formulated:
1. Gentry (color) units = T669 + 0.2(100 - T683).
2. Gentry (color) units = + 0.2(100 -

T6,7.6 T663) - 14.2.
The first term on the right is the principal factor in determining the
size of the Gentry unit. The higher the chlorophyll concentration, the
lower will be the value of T 6 6 3 , and the greater therefore the numerical
value of 0.2(100 - T 6 6 B ) ; a low Gentry number is indicative of a high
red color, and therefore the T 6 6 3 correction factor gives a higher Gentry
number with correspondingly less “redness.”
The Gentry unit was expressly designed to correlate with the x tri-
chromatic coefficient,and there is an almost exact inverse proportionality
between it and x over the range 1: = 0.67 to 1: = 0.59. (Recall that in these
+
samples x y = 1, so that Moster and Prater have developed a one-
dimensional scale on the basis of their two measurements.)
Over this range of Lovibond reds, x (and the Gentry Units neces-
sarily also) shows deviation from linearity with the Lovibond reading
below x = 0.64. Below a Lovibond reading of 22 red, the limitations
imposed by insisting on using the Lovibond one-dimensionally (ie., with
T H E COLOR PROBLEM IN FOODS 32 1
a fixed yellow) become apparent, because a two color-component system

is now involved.
The oil chemists have avoided the more serious of these difficulties
by alternative methods, using F.A.C. Standards, for example, for different
hues. Melvin, et al. (1953) have studied the spectral characteristics of
green soybean oil and have suggested optical density values at selected
wave lengths which would delimit Grade 2 for crude, refined, refined and
bleached oils. Thomson (1953) has further discussed single-number
systems with particular reference to the measurement of color of cotton-
seed and soybean oils.
4. Tomatoes and Tomato Products
The Production and Marketing Administration of the United States
Department of Agriculture promulgated standards of identity, including
color specifications based on the Munsell system for tomato pulp (pur6e) ,
juice, paste, and catsup. We shall consider the specifications for pulp, the
grade of which is determined by a numerical point score, 60 for color and
40 for absence of defects. (The color score for catsup is only 25% and for
juice, 30% of the total score.) The appropriate section of the specification
is quoted below:
“I. COLOR-The score for the factor of color is determined by comparing the
color of the product with that produced by spinning a combination of
the following Munsell color discs:
Disc l-Red (5R 2.6/13)-(Glossy finish)
Disc 2-Yellow (2.5YR 5/12)-(Glossy finish)
Disc 3-Black (N1)-(Glossy finish)
Disc 4-Grey (N4)-(Mat finish)
(A) Canned tomato puree (tomato pulp) that possesses a good red, ripe tomato
color may be given a score of 51 to 60 points. “Good red, ripe tomato
color ” means the typical color of well-ripened tomatoes. This color con-
tains as much or more red than that produced b y spinning the specified
Munsell color discs in the following combinations: 65 percent of the area
Disc 1; 21 percent of the area Disc 2; 14 percent of the area either Disc 3
or Disc 4, or any combination of the two.
(C) If the canned tomato puree (tomato pulp) possesses a fairly good red tomato
color, with red predominating, a score of 42 to 50 points may be given.
Canned tomato puree (tomato pulp) that falls into this classification shall
not be graded above U. S. GRADE C or U. S. STANDARD, regardless
of the total score for the product. “Fairly good red tomato color” means
the typical color of tomatoes which may not be well-ripened. This color
contains as much or more red than that produced b y spinning the specified
Munsell color discs in the following combinations: 53 percent of the area
Disc 1; 28 percent of the area Disc 2; 19 percent of the area either Disc 3
or Disc 4, or any combination of the two.
“ (D) If the color is definitely off-color for any reason or the product fails to meet
the requirements of paragraph (C) above, a score of 0 to 41 points may be
322 0 . MACKINNEY AND C. 0. CHICHESTER
given. Canned tomato puree (tomato pulp) that falls into this classification
shall not be graded above U. S.GRADE D or SUBSTANDARD, regard-
less of the total score for the product.”
We must now consider details of the Munsell system. Two points
require consideration-the nature of Munsell color space and the deter-
mination of a color match. Munsell notation is in terms of hue, chroma,
and value. Hue and chroma are essentially measures of C.I.E. chroma-
ticity-dominant wave length and purity, respectively-and value corre-
sponds with the photometric brightness. However the analogy must not
be pushed too far. A constant hue a t different values has different domi-
nant wave lengths. This is shown in plots of constant hue loci on the C.I.E.
chromaticity diagram in the final report of the Optical Society of America,
Subcommittee on the Spacing of Munsell Colors (1943). This plot, shown
in Fig. 4, is of interest also in illustrating the extent to which Munsell
color is reproducible, Thus, the point (z = 0.24, y = 0.43) is the limit of
the locus for constant hue 5G at value 9, and this is the limit of attainable
chroma for 5G a t this value. It should be clear that we cannot have high
chromas (or purities) at high brightnesses for surface colors. Only with
illuminants is this theoretically possible, and even here the color percepti-
bility will be a factor a t high brightness levels. Thus Judd (1952) describes
the dilemma of the couple who would make their white oleomargarine
yellow by viewing it under a sodium vapor lamp. With a white table
cloth, the unpigmented margarine still appears white.
There are two common methods of matching colors by the Munsell
procedure, by the use of specially prepared chips of known hue, chroma,
and value, and by spinning discs made by allocating wedge-shaped
percentages of the disc area to specified papers supplied by the Munsell
Color Company. This is an important asset to the Munsell system. Not
only can the papers be evaluated in terms of C.I.E. quantities, but where
color matches have been made, the color can then be reproduced and seen
by any observer without reference to the original sample, the color of
which was under examination.
There are, necessarily, limitations to the above. Normally, the
specular component of the reflected light is eliminated, and since the color
matches are metameric, viewing conditions must be specified. Also, the
greatest usefulness lies in the measurement of surface colors. The object
should be opaque and nonfluorescent. The nature of the surface of tomato
puree is far from ideal. Although many difficulties arise in borderline cases,
and the result is dependent on the skill and color vision of the observer,
the method has served a useful purpose. If however we examine the word-
ing of the PMA specification, a sample “shall contain as much or more red
than that produced by spinning the specified Munsell discs . . . ,” it
seems reasonable to interpret the phrase “as much or more red” in line
with the early work of MacGillivray (1937), who stated that the dominant
/x,
wave length was a measure of desirable redness, and that the brightness
lI - l //
,’ ,/ I
I
\
\
‘.
FIG.4. Plots of some Munsell Hues on the C.I.E. chromaticity diagram at two
values.
Plots of the loci of ten hues, 5G, 5BG, etc. to 5YG, on the C.I.E. chromaticity
diagram are shown as broken lines for value 1, superimposed on plots of the loci of the
same hues, solid lines, for value 9. The limits of plots for value 9 are given by the
heavy solid line.
Plots of constant chroma have been omitted. They would appear as ovoids around
the central gray. Detailed plots for 40 hues were prepared by the sub-committee
of the Optical Society of America on the spacing of Munsell colors, J. Opt. SOC.
America 83,385 (1943),and by interpolation, data in the two systems are completely
in terconvertible.
The simplified diagram presented here shows clearly the distinction between hue
and dominant wave length.
and purity were of relatively less significance. We shall revert to this
later, but may note again an attempt to reduce the number of variables.
The volume of tomato products is so great that, a concerted attack
has been made on tomato color measurement, in part to establish more
precisely the determination of the color grade of a sample, and in part, by
324 G . MACKINNEY AND C. 0. CHICHESTER
determining the color of a raw tomato, to predict the color grade of the
product. The bulk of the work to be reported makes use of the Hunter
Color-Difference Meter, and we shall have occasion to discuss the relation
of the aL,bLl and L scales*to the C.I.E. quantities and also in terms of the
chromaticity co-ordinates a, 0; the uniform chromaticity scale, U.C.S. ;
and AE, the color difference in N.B.S. units. Since aL and bL are deter-
mined in relation to a standard red reference tile, they represent differ-
ences, al - a2 and bl - b2, respectively.
Younkin (1950a, b) has plotted Munsell renotations for value 3/ in
terms of Hunter’s aL and bLl including the portion from Chroma /4 to
chroma /8 and from hue 2.5R to 2.5YR, where colors of tomato purees
lie. “Colors that plotted near the 2.5YR hue line were undesirable, while
colors that plotted near the 7.5R hue line were highly desirable. Thus the
hue of a puree was evaluated with respect to its proximity to one of the
constant hue lines. If hue differences were equivalent to 0.3 of a unit on
the aL scale, observers noted differences in color.” Chromas of / 6 were
weaker and less desirable than those of /8, but differences had to be
greater than 0.5 to be readily perceptible. Younkin further noted that as
L diminished, colors became darker and more attractive, provided L was
greater than 23, but that below 23, the puree colors became objectionably
dark. An L of 23 represents a lightness of Y tristimulus value equivalent
to 5.29%. Younkin (1950b) then evaluated tomato puree colors from
fruits selected to represent the upper and lower limits of U.S. No. 1 and
No. 2 in terms of aL and bLl from which it appeared that the fruit was
graded almost exclusively on hue, with the line represented by 0.5YR
separating the two grades. A second test was made with other inspectors
and the hue was again shown to be critical. If in the narrow range repre-
sented, we assume the 0.5YR hue can be plotted as a straight line in terms
of aLand bL then we may obtain for the line of constant hue, 0.5YR1that
aLN 2.5bL - 8.4, a t this Munsell value. If aL exceeds this, the grade is
fancy. Important contributions have also been made by Robinson et al.
(1951) and Robinson et al. (1952). In a comparative study of methods for
measuring color of tomato juice, Robinson et al. (1952) point out that
two juices of the same lycopene content but of different insoluble solids
content present quite different appearances, and consequently a method
is to be preferred that does not depend on a quantitative determination of
pigment. Because computations based on reflectance data by spectro-
photometry are tedious, tristimulus filters have been developed by Hunter
(1942). The characteristics of three, amber, green, and blue, respectively,
have been adjusted so that from three readings, A , G , and B , the tri-
stimulus values X , Y, and 2 may be readily computed. Like aL/bLlthe
8 See Section V, “Instrumentation,” and Section 111.5, “Alternative Spaces.”
THE COLOR PROBLEM I N FOODS 325
ratio ( A - G ) / ( G - B) is directly related to the dominant wave length.

Robinson et al. (1952) compared the Maxwell spinning disc, the Beckman
DU spectrophotometer, the Photovolt Reflection Meter, and the Hunter
Color-Diff erence Meter. The various methods were compared and
evaluated on the basis of correlation with PMA grades. The values ob-
tained for the ratio a L / b Lin the tomato samples were independent of the
brightness (or Munsell value), since the ratio is linearly related to the
dominant wave length. The disadvantages of the spinning discs are set
forth in some detail.
I n a n earlier paper, Robinson et al. (1951) made a detailed study of
tomato grades which were correlated with the Hunter aL/bL ratio. They
remarked that the correlation between this ratio and the PMA color score
fell when the tomato juice samples had widely differing chromaticities.
Their general consideration is summed up as follows: “Colors of the same
dominant wave length are interpreted as slightly different hues when
chroma and value differ. In the tomato color region, according to the
Munsell system, higher chroma is interpreted as yellower hue, while
higher value is interpreted as a redder hue.” The dilemma is then pointed
out that the Munsell system most nearly corresponds t o the interpretation
of human observers, yet individual observers differ in their interpreta-
tions. When samples are similar in value and chroma, then the C.I.E.
dominant wave length or some similar measure, such as a L / b L , is not only
satisfactory but much easier to use.
Friedman el al. (1952) expressed some concern as to these conclusions.
The pressure is great on all workers to find an acceptable industrial solu-
tion. We have on the one hand the data of Younkin (1950b) clearly
showing that as we approach the 7.5R hue, we have desirable redness, and
apparently unequivocally, that the 0.5YR hue delimits standard from
fancy grades. If with Robinson et al. (1952) we accept uL/bL as a measure
of redness, and recognize further that this is a measure of dominant wave
length, this latter quantity is independent of brightness and purity. Now
we are entitled to compare New York and New Jersey samples, solely on
the basis of aL/bL ratios. Friedman et al. (1952) selected seven such New
York samples with the highest aL/bL ratios, 1.59 to 1.62. They then
selected three standard and three fancy New Jersey samples, for which
aL/bL ranged from 1.29 to 2.32. To present a common basis for compari-
son, the Y, z, y C.I.E. quantities were calculated so that they in turn
could be referred back to PMA grades determined by spinning discs with
the appropriate Munsell papers. It makes no difference for this com-
parison that each batch of Munsell papers must be specially calibrated.
An ideal disc 5 R 2.6/13 has precise and ascertainable tristimulus values,
as is the case for the other Munsell discs specified by PMA. We are con-
326 G. MACKINNEY AND C. 0. CHICRESTER
cerned solely with the question: Can we grade these samples, regardless
of variety, differences in processing treatment, storage conditions, etc., in
an objective manner with respect to their degree of desirable redness?
In the simplest terms, an a L / b Lratio of 1.6 for a sample graded fancy
in New York, would not necessarily yield a fancy grade in New Jersey,
according to the data presented by Younkin (1950a, b, c), though it might
if L is 25 or below. Robinson et al. (1952) present convincing evidence that
inspectors must have their own prediction equations for satisfactory
grading. If, as seems t o be established by their work, there is (or may be)
a sufficiently large effect of brightness on hue, and if also samples are going
t o differ substantially in purity (or chroma), then the aL/bLratio must be
used with reserve as a measure of desirable redness in tomato products.
Consequently the correlation between PMA grades and the a and b values
obtained for a series of samples by Kramer (1950, 1951) necessitates
uniform chroma and value. This is perfectly possible but by no means
inevitable. The L values for the seven New York samples with a t / b L ratios
of 1.59 to 1.62 vary within the narrow range 28.0 to 28.8, corresponding
to brightness variations from 7.84 to 8.29%. The New Jersey samples
show much greater variation. A nationwide a L / b h standard would be
most unfortunate a t this juncture without further definition.
Buck and Sparks (1952) have used the Hunter instrument to correlate
the color of heated extracts of whole tomatoes with the color of the finished
catsup. We shall have occasion to consider these contributions in Section
V, “Instrumentation.”
6. Peas, Spinach, and Similar Chlorophyll-Containing Vegetables
There has been relatively little work evaluating the greenness of vege-
tables such as peas or spinach, though there has been considerable effort
expended to measure chemical changes related to color changes, e.g., the
conversion of chlorophyll t o pheophytin (Dutton, et al., 1943; Mackinney
and Weast, 1940) and also the disappearance of chlorophyll with ma-
turity, as, for example, in lemon peel (Eastmond, 1950). Eastmond et ul.
(1951) give a series of reflectance curves for fresh, stored, frozen, and
dehydrofrozen peas (whole, cooked). In addition they consider the effect
+
of two storage temperatures (- 10’ and 10” F.) and storage times up
to one year. In the former series differences were observed in luminous
reflectance and small changes in dominant wave length, ranging from
16.7 t o 21 % in Y and from 566.0 to 569.1 mp in A. These changes resulted
in a yellowing of the product. In the second series, a graying was observed.
A measure of greenness was obtained from the expression ( G / A ) B +
where G , A , and B are the green, amber, and blue filter readings. A similar
index can be obtained from selected reflectances. Eastmond et al. (1951)
chose a greenness index expressed as R 6 4 0 / R 4 9 0 +

RbgO.It will be noted
that in most of their curves, the reflectance at 490 mp is fairly constant,
and even more so a t 470 mp. R4Q0 is therefore relatively fixed, and the
ratio R 6 4 0 / R 4 9 0 depends on the reflectance at 540 mp. The higher the
reflectance at 580 mp, the greater the increase in yellowness and conse-
quently the greater the loss in greenness. Like the Gentry unit in relation
to the Lovibond (see Section 11.3, “Oils”), the green index is inversely
related to greenness.
In view of the limited data on these commodities we deemed it worth-
while t o make a few color measurements on extreme cases. Blanched
frozen peas and spinach were therefore thawed, portions blended, then
centrifuged to remove air bubbles. Other portions were heated in the
presence of dilute oxalic acid, and treated as before. Canned peas and
TABLEI
Color Data on Peas and Spinach*
Frozen blanched Canned t
Thawed, blended With oxalic, blended bhded
Peas $
Y, % 20.3 18.8 25.9
X 0.379 0.438 0.405
Y 0.498 0.454 0.445
A, m P 565.5 576.5 573.4
Spinach
Y, % 5.26 2.81 5.69
X 0.344 0.413 0.413
Y 0.442 0.424 0.429
A, mr 563.5 577 576.2
* Mackinney and Chichester (unpublished data).
t Canned eamples, from the local market, do not have comparable total pigment contents, so that no
significance can be attached to the changes in Y as between canned and fresh.
1: Y is the C.I.E. tristimulus Y,z and II the trichromatic coefficients; method, 10 selected ordinates.
spinach were also blended and centrifuged. Samples were then placed in
special leucite cells and measurements were made using the amber, green,
and blue filters with the Photovolt Reflection Meter, and by the 10-
selected ordinate method, with the Beckman Spectrophotometer. Results
are shown in Table I. The frozen peas thawed and blended exhibit a
dominant wave length almost exactly that reported by Eastmond et al.
(1951). The canned peas show a shift of almost 8 mp, and the spinach
of 12 mp. The difference in Y values between peas and spinach is due
primarily to difference in chlorophyll content. The changes observed are
thoroughly plausible, as the conversion of chlorophyll to pheophytin
causes a marked diminution in the absorption ca 570 mp, and an increase
ca 535 mp in solutions of the extracted pigment. It would be of interest to
determine whether spinach canned after blanching at 160’ F. (the Thomas
patent) has in fact a significantly lower dominant wave length than that
canned after steam-blanching. The Photovolt readings on the pea samples
were in reasonable agreement with those obtained from reflectance
measurements for the selected ordinate method. Those for the spinach
were not, differences of 7-8 mp in dominant wave length being noted.
The reason for this is that at the low brightness level, with a high pigment
content, the very low blue readings on which tristimulus 2 depends could
not be estimated within better than a 50% reading error.
6 . Strawberry Preserves
Strawberry preserves and jams were included for two reasons. They
contain anthocyanin pigment, the pelargonidin 3-glucoside (Sondheimer
and Kertesz, 1947), but in relatively low concentrations, 3 to 10 mg. per
100 g. preserve, and secondly they show a marked tendency to brown a t
unfavorable storage temperatures. Strawberry preserves are customarily
made in the United States from frozen strawberries. These are drawn from
freezing storage as needed, thus obviating unnecessary storage of the final
product. Unlike the jam, where no attempt is made to secure whole fruit,
the strawberry preserve contains a high proportion of whole fruit and
pieces of fruit which have been only partially distintegrated during
handling. The preserve therefore consists of opaque berries and portions
of berries suspended in a jelly highly translucent to red light.
We have, in consequence, an awkward problem in color measurement,
owing to inhomogeneity of the material. If we examine the jelly portion,
freed from all turbid matter, there is virtually no light reflected. This may
be illustrated very simply, by placing the jelly in a container the inner
surfaces of which have been painted black. The jelly appears black. If
however a white background is substituted the jelly will have its cus-
tomary red color, the red light being reflected from the white surface and
transmitted back through the jelly. By using a sample holder of sufficient
depth, with a white background, the jelly will have a negligible 2 tri-
stimulus value, and there is furthermore relatively little fluctuation in
dominant wave length or purity. The situation is in some respects com-
parable with that found in Capsicum. However, whereas chlorophyll may
modify the color of the spice, this is not true in any normal instance for the
strawberry. The dominant wave length, for the jelly, is constant, ca 605
mp,within fairly narrow limits, and a surprising number of samples will
have chromaticities represented by x, 0.65 - 0.66; y, 0.34 - 0.33. As
Eastmond (1950) has shown with raspberries, if it were solely a matter of
securing a measure of anthocyanin, an R S f O or Tala would serve the pur-
pose, the maximum for the anthocyanin absorption appearing a t or near
520 mp. We have, however, the browning to contend with at the same
time. Measurements 011 a number of samples of industrial origin, both
normal and dark, indicate that changes in appearance are governed
almost exclusively by changes in the brightness.
The question arises as to why there should be so little change in
dominant wave length during browning. The answer lies in the nature
of the spectral curves. The brown pigments show a more general absorp-
tion, high in the blue, low in the red, than the natural anthocyanin which
they replace. Until most of the anthocyanin has disappeared, however,
this will not seriously affect the relative proportions of light reflected from
different spectral regions. It will result only in a darker red of essentially
the same hue. There is, on heating, as in the preparation of the preserve
itself, a substantial loss in anthocyanin. When the packer speaks of de-
veloping the color during the cooking process, he is redistributing the
anthocyanin in the berry-syrup mixture. I n spite of the ‘(deepening” of
the color, there is in fact a net loss of pigment. This ‘(deepening” is
physically not unlike the color change in a green leaf, held under water
in an evacuated flask to eliminate air. Substantially less light is reflected
from the surface, and the leaf, originally opaque, becomes partially trans-
lucent. The lightness, or Y tristimulus value, therefore, should be a
useful measure of the changes in a strawberry preserve. Loss in antho-
cyanin without concomitant browning may result in a slight increase in
lightness, though the paling is not likely to be detected visually. Browning
would immediately be reflected in lower tristimulus green, or G readings,
and in Y computed from reflectance data. As shown by Shah and Worth-
ington (1953) for pur6ed frozen strawberries, either the Hunter L value
or the Photovolt G reading provided a measure of color differences in
their samples.
Unfortunately, for similar results with a strawberry preserve, a pro-
cedure must be agreed upon whereby the opaque berries are consistently
and uniformly dispersed. Centrifuging will not eliminate air bubbles
incorporated as a result of blending. It may prove feasible to force the
preserve gently through a coarse mesh screen and secure consistent repro-
ducible results. The values obtained by this procedure differ from those
made on the jelly. The 2 value is no longer negligible, as there is sub-
stantial surface reflection. Increase in turbid matter has not, in our
experience, changed the dominant wave length, but the purity has fallen
appreciably from ca 95-98% t o SO-SO%, depending upon the sample.
The browning, however, is immediately detected by a decrease in Y .
Possibly enough has been said to indicate that color measurement in a
preserve, containing opaque and also highly translucent portions, depends
upon carefully standardized procedures.
7. Wines
As with other commodities, suggestions have been made for color
specifications in terms leading toward a single-value function. Thus
Boutaric et al. (1937) suggest the optical density a t 520 mp (the peak for
anthocyanin absorption) and the ratio of the densities a t 480 and 640 mp.
There is no a priori reason why some such measure should not provide a
uaeful scale for specifying the color of a given wine, though as Winkler and
Amerine (1938) point out, its greatest utility would be for red wines
possessing similar transmission curves.
With wines, however, it would be much more difficult t o correlate the
data with the C.I.E. system. Winkler and Amerine distinguish between
brightness (lightness on a gray scale) and brilliance, which represents
freedom from turbidity. Furthermore the gamut of wine colors from dark
red Burgundies to rose wines and to tawny ports or white wines, is such
that a photometric color index of the kind developed for oils would be
most difficult to interpret. A series of such indices might be set up for
each wine type, but this would not assist in identifying types, particularly
where blends have been made. Winkler and Amerine therefore computed
C.I.E. quantities from transmission data obtained spectrophotometri-
cally. They used light paths from 0.25 cm. in the case of a dark red Petite
Sirah, 0.5 t o 1.0 cm. for California Ports, t o 4.0 for Grenache (light pink),
and 3.0 to 4.0 for various white wines.
Dominant wave lengths varied from the complementary of 504 mp
(a purple) t o 650 mp for the Petite Sirah, 590 to 595 m p for the ports, and
587 to 590 mp for the white wines.
We select in the following table examples of wines from the work of
Winkler and Amerine (1938) and of Amerine and Winkler (1941) t o
indicate the range of variation in color. A cursory inspection of the
table indicates some of the difficulties. The Grignolino is “typical” and
“light orange pink,” the Port “too brown,” and “light amber red,” the
white wines, as is well known, normally light amber. The coefficient z in
the three cases is negligible, 0.027, 0.028, and 0.033, but if all were ad-
justed t o standard depth, 1.0 cm., this would not be true. This is because
at any given wave length and pigment concentration, the optical density
(a logarithmic function of T , the transmittance) is determined by the
depth, d, and
T e-d,
Q:
the C.I.E. values are determined from T itself. We must therefore consider
the wines in one sense to be multicolored or polychroic, insofar as the
color perceived with the transmitted light changes with the depth of wine
being viewed.
THE COLOR PROBLEM I N FOODS 33 1
TABLE11.
Colors of Wines*
Wave length, Pu- Depth,
5 2/ mp rity Yt cm .
Petite Sirah 0.649 0.316 655 73 9.6 0.25
Alicante Bouschet 0.635 0.309 504c 68 2.4 0.5
Muscat 0.615 0.344 616 71 15.4 0.5
Mission 0.603 0.380 599.5 86.2 24.6 1 .o
Port 0.572 0.400 596 83 30.4 1 .o
Grignolino 0.565 0.408 593 78 34.1 4.0
Mixed whites 0.550 0.417 590 72 38.8 3.0
Grenache 0.520 0.410 592 52.4 57.3 4.0
* From Winkler and Amerine (1938) and Amerine and Winkler (1941).
t Y ia the C.I.E. triatimulus Y,z and I/ the trichromatic coefficients.
111. SOMETHEORETICAL
CONSIDERATIONS
1. Discrimination
In the foregoing sections, we have selected for discussion what we
believe to be significant contributions to the color problem in foods. A
certain line of development is followed-the selection of a unidimensional
color scale or index is attempted, according to the color problem to be
evaluated, which wherever possible is related to the C.I.E., possibly via
the Lovibond or Munsell systems. This is in actuality an attempt to
correlate visual estimation of a color with an instrumentally measured
numerical score. Bouma (1947, ch. XII) discusses discrimination by the
eye. If a color match has been made, the question is raised as to how far
one can modify one of the colors before the eye can perceive the differ-
ence. The amount of change to render the difference perceptible is a
“discrimination step,” “threshold,” or “limen.” The discrimination may
be applied to the brightness, the wave length or hue, and to the purity
or chroma. The various thresholds are determined individually, keeping
the other two variables constant. If we can just distinguish the spectral
color X + AX from the spectral color X, then we can plot AX as a function
of X for the visible spectrum, but as Bouma points out, results are in-
fluenced by brightness, size (i.e., areas) of the spots under comparison,
environment, and other factors.
With respect to purity ( p ) discrimination, it appears that p can fall
from p = 1 to p = 0.5 before AX as a function of X changes appreciably.
As p approaches zero (white), the number of distinguishable colors be-
comes much less. In the immediate neighborhood of the white point, hue
discrimination is small. This is true also a t low brightness levels, i.e., as
we approach black ( Y -+ 0).
For a color space to have maximum usefulness, all steps of just dis-
tinguishable differences, whether of brightness, hue, or purity, should be
represented by equal distances along the appropriate axis; i.e., each series
of thresholds or limens should be equally spaced. This procedure involves
the development of a uniform chromaticity scale. This requires construc-
tion of the color space in a Euclidean geometry, possessing no ruffled
planes as in the index of fading as defined by Nickerson (1936). A second
method compensates for the known distortions of C.I.E. space by use of
MacAdam ellipsoids.
2. The Uniform Chromaticity Scale (U.C.S.)
Judd (1952) expresses a “ . . . somewhat dismaying suspicion that
a strictly uniform tridimensional color scale cannot possibly be de-
veloped.” Nevertheless, the success that the Bureau of Standards has
achieved is convincing proof that an extremely useful approximation can
be made, and for maximal usefulness we must understand some of the
limitations involved. The range of brightness over which measurements
can be made extend over nearly five orders of magnitude and Bouma
(1947, p. 226) plots B/AB as a linear function of log B , where AB is the
brightness limen. The ratio is called the sensitivity of the eye to differences
of brightness. This sensitivity possesses the characteristics of a logarithmic
function and is markedly lower in dark surroundings. The Munsell system
has a 10-step series from white to black. I n the simplest form, these steps
V were based on the following relationship with the reflectance:
v = 1022%
Thus the reflectance for step 5 was 0.25; for step 7, 0.49, etc., step 10
(white) having a value of 1.0.
Judd traces the development of modifications needed to take into
account the reflectance of the background as well as of the gray chips
representing the various steps. As shown by Bouma, it is only for a light
background that B/AB is a linear function of log B. Consequently
Munsell value must be determined against such a background. In sum-
mary, he accepts the equation given above as a convenient first approxi-
mation, and states that it can be applied to lightness differences perceived
among chromatic colors as well as to those among grays.
Judd then considers chromaticness and shows how Munsell hue and
chroma scales can be developed, and explains the irregularity of the
Munsell Color Solid. The chroma loci are determined with color chips on
a middle gray-to-white background and this prevents detection of chroma
and hue differences a t low values. The limits in “ideal Munsell space”
were plotted by Nickerson and Newhall (1943).
Judd developed a U.C.S. triangle based on a projective transforma-
tion of the C.I.E. co-ordinates (z,y) to new co-ordinates (r,g), such that
approximately equal distances were obtained for perceptually equal

differences. Results were plotted on a triangular grid in terms of R, B ,
and G. Of the various other transformations, mention should be made of
Hunter’s (a#) co-ordinates (1942). Thkse are defined by the equations:
2.42662 - 1.3631~- p.3214.
+ +
a = 1.0000~ 2 . 2 6 3 3 ~ 1.1054’
a= +
0.5710~ 1.2447y - 0.5708
+ +
1.0000~ 2 . 2 6 3 3 ~ 1.1054’
They may be approximated from readings made with Hunter’s amber,

green, and blue tristimulus filters, in much simpler form:
N ( A -G)/(A + +2G B ) ;
+ +
p N _ 0.4(G - B ) = ( A 2G B ) .
The origin (a = 0, p = 0) represents illuminant C of the C.I.E.
3. The N.B.S. unit

To determine color tolerances, we need ideally an assessment of color
differences on a tridimensional color scale. Judd discusses the various
attempts to measure the color differences. These include Nickerson’s
index of fading I , based on Munsell spacing, the changes in hue, chroma,
and value ( A H , AC, AV) being appropriately weighted for perceptibility
differences, and his own contribution, A E , the N.B.S. unit. This has been
redefined in terms of Hunter’s (a,@ chromatic space. The simplest expres-
sion of this quantity is given by Scofield (1943) :
AE = + +
[(L,- L2)2 (al - a2)2 (bl - b2)2]f4
where L is a function of the square root of Y , and a and b are proportional
to a and ,8, respectively.
In 1945 Bouma summarized limitations of the various U.C.S. systems
as’follows: that in the neighborhood of the white point, and also for
unsaturated colors, they give a fairly correct measure of the number of
steps of just noticeable color differences; that for the spectral locus and
for very saturated colors, agreement is much worse; finally, so far as
thresholds for colorimetric purity are concerned, the U.C.S. system is
useless (Bouma, 1947 p. 241). So far as we know, these views have not
been contradicted. This is not to say, as Bouma points out, that the
U.C.S. system is not serviceable, and furthermore it is wholly proper that
emphasis should be given to the usefulness of the quantity A E to set
color tolerances in various industries.
4. MacAdam Ellipsoids
Hitherto we have discussed measurement of the distance separating two
colors plotted in a visually uniform color space. Instead of transforming the
C.I.E. space into a uniform color space, its known distortions can be
compensated by MacAdam ellipsoids (MacAdam, 1942). Davidson (1951)
Fro. 5. MacAdam ellipsoids. Any point in the color plane may be surrounded by
an ellipse, and in the color solid by an ellipsoidal volume, the boundary of which
represents a finite number of perceptual color differences. For any fixed number of
differences, the size of the ellipsoid depends upon its position in the color space. By
courtesy Dr. D. L. MacAdam by permission of the Optical Society.
calculates the color differences by a simple graphical method, assuming

that any point on the surface of the ellipsoidal volume represents aunit
color difference from a focal point of the ellipsoid. As he points out,
MacAdam ellipsoids are theref ore a means of specifying color tolerances.
In Fig. 5 are shown a series of ellipses within the chromaticity diagram
which illustrate the nonuniformity of C.I.E. color space. The distances
from any marked point to points on the boundary of the ellipse enclosing
it all represent equal chromaticity differences. Since we must take into
account the lightness, we have an ellipsoidal volume. “We may plot two
colors, A and B , of nearly equal chromaticities A , and B, on the conven-
tional 2,y chromaticity diagram, but an ellipsoid surrounding either point
A , or B, will be of little value in determining the number of just per-
ceptible differences between them, unless plots of the two colors on an
2,Y diagram also fall near each other.” Davidson’s solution is graphical:-
A and B represent the loci of two colors differing slightly in z,y and Y .
We plot their chromaticities A,, B,. We then erect plane Y perpendicular
t o the x, y plane and plot A and B such that A A , and BB, are propor-
tional t o Y Aand YBrespectively. We may next visualize A as a point in
color space which we shall enclose by a surface not unlike an egg shell.
The volume contained within this shell is ellipsoidal, and every point
on the shell surface represents an equal perceptual difference from
I)oint A . We join A B which must pass through the shell at some point P .
Thus A P represents a vector along the line joining the two colors. It
constitutes a stated unit perceptual difference along A B . The quantity
,4B / A P then gives us the number of such differences.
6 . Alternative Spaces
Whereas Hunter derived his alpha, beta chromaticity directly from
C.I.E. quantities, Adams (1942, 1943) developed a chromatic value space
and subsequently derived from it a chromatic valence space. He first
modified the tristimulus values so that for illuminant C
x, = Y c ( = Y ) = 2,
where the subscript c indicates the values have been corrected for the
illuminant, the corrections being X , = X/0.9804,2, = 2/1.180. He then
applied the Munsell value function to Y,, deriving V,, calculated from
reflectance ratio, using the Y primary and a magnesium oxide standard
white, from the expression
R,/R,M,o = 1.2219Vg - 0.2311Vv2 + O.2345Vv3- 0.021009V~4
+ 0.0008404V,6.
The other two quantities required t o define the valence space are W,
and 0.4W2, where
W, = V,(X, - Y ) and W, = V,(Z, - Y ) .
In the chromatic value space, Adams (1942) applied the Munsell value
functions by the substitution of X , and 2, in the original value function.
We see that substitution of X , and 2, in the same formula yields two new
quantities, Vz and V, which with V, uniquely define any color. This then
will give slightly better spacing, since we apply the value function to each
tristimulus quantity separately. This is a difficult problem in which

to have a direct reading instrument. The differences (V, - V,) and
(V, - V,) determine the chromatic value. They will be zero for a neutral
gray. If the Munsell hues for a given value are plotted in the Adams
value space, lines of constant chroma describe a concentric series of
ovoids about the neutral gray. An instrument could not be developed to
measure all the quantities in the Adams value space, but the quantities
of a modified valence could be incorporated in a direct reading instru-
ment, the Hunter Color-Difference Meter (Hunter, 1948a, b, Gardner
Laboratory, 1950).
Using the tristimulus filter Y +
G and X E (0.80A 0.20B) we may
substitute and obtain W , = V, (0.80' + 0.2'' - G), o r ~ , ( 1 . 0 2-~ y ) .
0.9804
To approximate V , without recourse to involved circuits, Hunter h w
evolved the function V, g f, = 0.51 (21 4-2oy) which has proved
1+20Y'
quite satisfactory. Multiplying the whole expression by a constant, we
arrive a t Hunter a, = Kl(fg)(1.02X - Y ) , and
br = -[0.4Klf,(Oa847Z - Y ) ] .
These give quite a good visual spacing and together with V , = K2Y(f,)
give a good measure of visual differences, AE.
+
AE = [(AK2Y(fU)2 ( A U , ) ~= (Ab,)']%.
Hunter, however, has not been able to incorporate Y(f,,), i.e.,
= L,
so it is necessary to compute L from R or to accept as a compromise
L = K ( Y ) n . This conversion to L scales unfortunately means that the
rest of the scales are also changed by the same factor, which slightly
decreases their usefulness in estimating color differences. The instrument
can be used with both scales so it merely involves a switching operation
(Gardner Laboratory, 1950).
Saunderson and Milner (1946) developed a zeta space, which like an
omega space of Moon and Spencer (1943) is based upon the Munsell
renotation. The omega space is not as useful a representation and there-
fore will not be discussed. The zeta space is a modification of the Adams
chromatic diagram. Chroma is represented by the expression
Chroma = Ic[(V, - V,)' + 0.16(V. - V , ) 2 ] s
in which Ic represents the proportionality factor between radial distance
on Adams chromatic value co-ordinates (ie., V, - V , and V, - V,) and
THE COLOR PROBLEM I N FOODS 337
the chroma itself. The k is remarkably constant for different Munsell

values and consequently the zeta space transformation is not a function
of Munsell value. This follows from the fact that the radius for constant
chroma is the same for all values. The zeta space is therefore defined by
(ll, {z, lb), where l.2 is proportional to Munsell value and lI and la
define the color plane.
IV. GENERAL
CONSIDERATIONS
I n considering the developments within the food industry, a pattern
may be discerned. Frequently, ad hoc material standards may be set up,
to be replaced by more precise measurements based on one or more
attributes of the C.I.E. color system. A trend toward simplification then
commences, in order to specify the color for grading purposes. The
method has been described as abridged spectophotometry. As Hardy and
Young (1949) point out, the abridged method may serve a dual purpose,
in production control and where necessary, as an estimate of tristimulus
values. Before this method can be successfully applied, three sets of
data are required:
1. The nature of the pigments involved, and reflection or transmission
measurements, determined spectrophotometrically.
2. The basis for a visual color preference, e.g., desired redness in
tomatoes, or whiteness in sugar.
3. Some objective measure which locates the color preferences in color
space. This is best determined by reference to the C.I.E. or related
systems.
Given the foregoing, it becomes possible to determine whether a more
restricted set of data can replace the cumbersome C.I.E. evaluation and
simplify interpretation of the color preference. Where this is practicable,
an instrument can be designed for a given commodity which will assign a
grade on the basis of the color tolerances permitted in the grade specifica-
tion. Where this procedure has not been followed, difficulties have been
encountered.
These difficulties usually involve the assignment of different values
for the color difference between two samples when judged visually and
by the instrument in question. This is best illustrated in the near-white
region, where even the basic C.I.E. system gives incorrect values owing to
incomplete evaluation of the 1931 standard observer. This was first noted
by Jacobsen (1948), though Wald (1945) had already indicated that the
eye was more sensitive t o the blue than is shown by the standard observer.
These findings were corroborated by Judd (1949), who modified the char-
acteristics of the standard observer to account for differences observed
visually in titanium paints but not indicated by reference to the C.I.E.
338 C. MACKINNEY AND C. 0. CHICHESTER
The pattern we have been describing, beginning with simple material

standards, is capable of an additional step, the development of new
material standards duplicating with high fidelity the reflectance curve
of the commodity. We are in no position to offer a critical appraisal of the
possibilities here. We may remark however that the printed color varies
slightly between batches, even with the care exercised by the Munsell
Color Company. Similar difficulties may be encountered with ceramic
tiles. We have been interested therefore to learn of the development of
plastic standards which rather remarkably duplicate the reflectance
curves of tomatoes, obtained in the General Electric Recording Spectro-
photometer. These and similar discs for peaches have been developed by
Monsanto for use with the Agtron instrument of Ma,gnuson Engineers
(Smith and Huggins, 1952).
Matches are thus nonmetameric and are independent of the observer
and the illuminant, since comparisons are confined to the brightness.
Two additional procedures may be mentioned here, to be described
possibly as abridged photometry or abridged colorimetry. Eolkin (1952)
evaluated the “relative color” of purees by black and white photography.
This would obviously be of greatest use, when ordinary film is employed,
in light-colored products subject to browning. Eolkin used the procedure
to measure discoloration in apple sauces. Livingston and Vilece (1953)
substituted a photoelectric method for a similar type of problem.
V. INSTRUMENTATION
With respect to the majority of procedures discussed for the different
commodities, and the instruments used, it is necessary to accumulate
data from widely different sources to evaluate the natural variability of
the commodity, the usefulness of the procedure selected as a criterion for
delimiting a color grade, and finally the instrumental reliability under
operating conditions, whether it be in the laboratory or in the inspection
service in the field.
Work of this type has been initiated by Worthington, Cain, and
Wiegand (1949) on a variety of juices, by Shah and Worthington (1953)
on strawberries, and by Sastry and Tischer (1952) on the anthocyanins of
grapes. Kramer and El-Kattan (1953) have worked out detailed correla-
tions between visual score and Hunter quantities for tomatoes.
1 . Additive Colorimeters
Some of the oldest color-determining instruments are the additive
visual colorimeters. They establish the basis for all color measurements
which fundamentally assign a unique numerical value to a particular
color. When the spinning disc technique is applied to the Munsell color
collection, it is in principle an additive colorimeter. In order to obtain

values for an unknown color, it is necessary to use a minimum of four
segments to secure the necessary three degrees of freedom for a successful
match. Munsell color can be converted to C.I.E. quantities so that data
can be transformed into other standard systems of color specifications.
The Munsell system despite its simplicity and low cost appears less in
favor than it was at one time, although many specifications are still based
upon it. Aside from the labor involved in the measurement and the fact
that a personal judgment is involved, the method has many advantages.
It is the only standard system which will allow a specified color to be
inspected by the eye. Thus one can reproduce the color from a numerical
specification as a color plaque which can be held in hand. Deviations from
these standards may be interpolated with surprising accuracy. The place-
ment of the color chips a t constant hue is uniform with respect to value
and chroma. Munsell colorimetry has some disadvantages. If chips are
not chosen which closely match the color being evaluated, the match is
likely to be metameric, in which case the value assigned is not unam-
biguous. The lighting conditions for making the comparison and the sur-
roundings must be standardized, if the maximum accuracy is to be ob-
tained. The colors to be specified must be of moderate saturation, since
the discs are limited in saturation, owing to the inherent nature of the
dyes and the lack of a wide enough range of dyes. The chips represent
only a narrow range of surfaces, making comparison difficult with unusual
surfaces or other modes of appearance.
The additive colorimeter in which a sample is matched by the addition
or negative addition of three or more primary lights has fallen into disuse.
The number of colors which can be matched with practical primaries
(practical in the sense that they must be a compromise between saturation
and the loss of the major part of the incident light flux owing to high
selectivity) is limited. With moderately saturated primaries the match
becomes metameric, and the match thus made will not stand up under
other illurninants. These instruments are usually constructed with a single
eyepiece so that the field is restricted to two degrees or less. This in turn
reduces drastically the accuracy in matching. Some matches will neces-
sitate the use of negative amounts of one of the primaries further com-
plicating the transformation of the data to other specifications. By adding
more primaries some of these disadvantages may be overcome. Donaldson
(1947)has developed a six-color additive colorimeter in which an initial
approximate match is made with three primaries and the final match is
secured with three more. The instrument is calibrated in terms of C.I.E.
quantities, and since the spectral distribution of the final light flux is
dependent on six filters the matches are essentially nonmetameric, and
the color gamut covered is quite large, owing to the placement of the
six primaries.
2. Subtractive Colorimeters
The visual subtractive colorimeter avoids some of the troubles of the
three-primary additive colorimeters. By using primary filters between
the standard field and the eye, or between a light source and the field, the
emergent beam can be varied with respect to both hue and saturation,
thus assuring a more nearly physical or nonmetameric match. I n the case
of additive mixtures the intensity of impinging flux is varied, which in
turn varies only the intensity of the field, leaving hue and saturation
substantially unchanged. By the use of two primaries and a method of
controlling the illumination, a subtractive instrument becomes easier to
use and interpret. The ability to transform any readings obtained on a
subtractive colorimeter into other specifications is dependent upon the
calibration of the filters in some other system. The permanence of the
filters is vital to success; gelatin filters in particular may not retain their
initial calibration.
The Lovibond Tintometer is probably the most widely used subtrac-
tive colorimeter today. The filters are not continuously variable as in
some of the wedge instruments but are interchanged to vary the color in
discrete steps. The individual filters bear standard specifications allowing
readings obtained to be convertible-at the expense of some labor. The
Schofield-Lovibond modification (Schofield, 1939), using a controllable
light source and two filters at a time, makes the conversion extremely
simple. Since the wedges are made of glass, they are in general permanent.
The instrument, however, suffers the usual disadvantages of a monocular
system in that the field is of necessity restricted to a small angle. I n addi-
tion, the yellow and blue transmit in the far red and this in turn restricts
the producible colors to the lower part of the chromaticity diagram. The
instrument has found wide usage in the specification of oils, which
generally have dominant wave lengths in the red, orange, and yellow
portions of the chromaticity diagram.
The subtractive colorimeter and the additive colorimeters possess one
characteristic which none of the available photoelectric instruments
possess, that is, the ability to give a reproducible specification t o a
fluorescent material.
3. Comparators
There are also instruments used to compare a standard color with the
color of an unknown, i.e., the comparators. These are of little use in
the specification of color as such, since they are restricted essentially t o the
matching of a particular color to that of a standard by adjustment in the
amount of luminous flux transmitted by, or reflected from, the specimen

to be standardized. They may also be used to place a sample within a
color grade. When used in this manner, the answer obtained is unequivo-
cal as to whether the specimen falls within a given tolerance. The dis-
advantage is that this answer does not give any information as to what
causes the mismatch.
4. Spectrophotometers
The basis of much instrumentation and methodology lies in the
requirement of the American Standards Association that “the spectro-
photometer shall be recognized as the basic instrument in the funda-
mental standardization of color.” The Bureau of Standards concurs in this
statement. Basically a spectrophotometric curve, that is, a plot of in-
tensity versus wave length, is certainly the most unambiguous specifica-
tion of color that can be obtained. In one sense this is not color, and data
are open to many interpretations. For instance, we can have two dis-
similar curves which will give the same apparent color sensation under
certain conditions, thus our metameric match. Extremely small variations
in spectrophotometric curves may lead to large variations in individual
appearance. In order to interpret such curves, careful study is required.
If specifications are to be based on the curves by reference to some
standard color system, the conversion must be performed with extreme
care to prevent errors of considerable magnitude. With reference to the
C.I.E. system the eye has the ability to detect a difference in color samples
in some parts of the chromaticity diagram differing by only 0.001 in 2 or
y. This may in many cases be smaller than the uncertainty of the measure-
ment. A study by Nickerson (1935) of various specimens, using the
weighted ordinate method for intervals of 10 mp, gave x and y with an
average uncertainty of 0.004 while the 10 selected ordinate method gave
an average uncertainty in the same samples of 0.0035, which is more than
three times the minimum detectable difference.
The determination of reflectance curves on a large number of samples,
particularly from nonrecording spectrophotometers, is extremely labori-
ous with instruments such as the Beckman, where the specimen holder
has room for only the standard and a single specimen. The subsequent
reduction of the data can be simplified somewhat by the use of automatic
integrating devices on the recording machines and the use of punch card
calculators (Peterson et al., 1944; MacAdam, 1950; Davidson and Imm,
1949; Kaye and Devaney, 1952). Most spectrophotometers are designed
to avoid the inclusion of the specular component in reflection measure-
ments. I n some cases it would be highly desirable to include this but
unfortunately only one of the present spectrophotometers, that is the
General Electric, has any provision for doing this.
342 a. MACKINNEY A N D C. 0. CHICHESTER
There is an unfortunate tendency to regard a spectrophotometric

reading as definitive, and this unavoidably is perpetuated by making the
spectrophotometer the ultimate reference instrument in color work.
Ayres (1949) evaluated the operational accuracy in photometric analysis
and instrumental performance under differing conditions. Color measure-
ments are hampered in that the sample itself may not be altered. This
differs from “ chemical photometry,” where concentrations can be
altered to give maximum reading accuracy. One is limited therefore
in the color work to changes in the reference standard, which may add
appreciably to the labor. Ewing and Parson (1948) show that any
absolute assay undertaken with a number of Beckman spectropho-
tometers is subject to uncertainty greater than that indicated by the
limits of precision of each individual instrument. We have no reason to
doubt that this applies to all spectrophotometers. Caster (1951) studied
variability on a single instrument and found that although duplicates
checked within 0.1 to 0.5%, consistent errors of 3 to 5% were observable,
depending upon the phototube, slit width, lamp intensity, aging, and
similar factors.
Cannon and Butterworth (1953) demonstrated that a linear plot of
Beer’s law is no proof of linearity of spectrophotometer response, and
erroneous absorbancy values may be obtained. They suggest an instru-
mental check by use of independently calibrated neutral filters. Their
findings apply particularly to the barrier-layer photocells, used in such
instruments as the Unicam, the smaller Coleman, the Evelyn, Hunter,
Klett-Summerson, and Photovolt, and are less applicable to the gas-
filled photocells used in the General Electric Recording and Beckman
instruments. Errors in standards for spectrophotometry as well as errors
concerned with the General Electric instrument are discussed by Gibson
(1949).
Goldring et al. (1953) have examined anomalies in spectrophotometric
measurements. These have been classified on the basis of chemical factors,
instrument factors, operational techniques, and mathematical consider-
ations. Of particular concern are errors of which the operator may remain
unaware. The case of mechanically insecure elements in the input elec-
trometer tube is cited. As the photocell compartment shutter is opened
or closed, there can be a substantial shift in the zero of the photometer
system. This is unsuspected when the change in zero is reproducible, the
new position being metastable.
Spectrophotometers do not suffer from the limitations of visual
colorimeters in that the data obtained may be considered objective, but
the time involved in interpretation of the data and the skill needed for
this interpretation prevent their use in routine day-to-day checks on

products except under unusual circumstances.
The uses and limitations of spectrophotometers, such as the General
Electric Recording and the Beckman DU Spectrophotometers, are dis-
cussed in detail by Gibson (1949) and by Judd (1952).
5. Tristimulus Filter Colorimeters
The tristimulus filter colorimeter was apparently developed in order
to retain some of the advantages of the spectrophotometer and a t the
same time to decrease the labor needed to calculate the numerical
specifications of color. They are generally more portable than the spectro-
photometers, cost less, and require much less interpretation of the results.
All of the data of the tristimulus colorimeters can be interconverted to
any of the standard systems of color measurement. Their accuracy
depends upon the fit between the theoretical response of the C.I.E.
primaries and the response achieved by the combination of light source,
filter, and photocell. The fit is in no case perfect, and is usually worst
in the small lobe of the X primary.
Some of the instruments originally developed for use with a set of
tristimulus filters attempted to use a white reference point as their
standard. This usage means that there is a large color difference between
most of the samples measured and the reference point, and it can be
shown that the larger the difference between the standard and the
observed color, the larger the uncertainty in the measurement. With the
Hunter filters and a magnesium oxide white plaque the discrepancy
between the spectrophotometrically determined x or y and the colori-
metrically determined x or y is frequently larger than 0.02, that is, more
than ten times a reasonable chromaticity tolerance for most colorimeter
work. Since these discrepancies are proportional to t,he difference between
the standard and the measured color, this difficulty can be obviated by
the use of specimens of similar spectral composition. Another difficulty
inherent in the measurement by the use of tristimulus filters is that the
more metameric the match between specimens, the larger the discrepancy
in the measurement.
The Hunter tristimulus colorimeter represents a departure from the
usual colorimeters, as it presents its data in a form closely akin in spacing
to the Munsell system. By resistance networks between a series of barrier-
layer photocells which view the reflected light from a specimen through
tristimulus filters and a compensating cell which views the light source,
a reading is obtained which is related t o perceptible differences.
The degree of fit between Munsell spacing and that of Hunter’s a r , br
is discussed by Nickerson (1950). She compared the color space of Hunter

and Adams’s chromatic value space, in terms of Munsell renotation. For
constant Munsell value, the constant chroma lines should fall on con-
centric circles around a neutral gray, if uniform perceptibility spacing is
to be achieved. It was found that the Adams value space gave better
over-all results in terms of uniform chromaticity. The Hunter ovoid is
flattened on the side of the yellow hues, and expanded in the blues and
purples, while the Adams ovoid is flat on the blue side.
The use of barrier-layer photocells in some of these instruments intro-
duces other difficulties. As the radiant energy impinging upon the photo-
cell decreases, the current likewise decreases to such an extent that it is
impossible to obtain accurate readings a t low intensities. In most instru-
ments this occurs at approximately 2 % reflection.
I n order to achieve,higher sensitivity a t low intensities, recourse is
made to the high-impedance outputs afforded by vacuum photocells,
which allow their output to be more easily measured by vacuum tubes.
The Differential Colorimeter of Glasser and Troy (1952) is such an
instrument. In order to achieve direct reading, stability with respect to
slight changes in light intensity during measurement, and balance of dark
currents, two vacuum photocells are used in a balanced bridge. Their
unbalance is measured with a vacuum voltmeter, in a cathode-follower
circuit. I n order to avoid phototube fatigue and dependence upon light
source intensity or gain consistency of the voltmeter, optical null balance
is used, causing the phototubes to operate a t equal response a t balance.
With this arrangement, however, new tristimulus filters had t o be devised
for the source, filter, photocell combination. This instrument can beused
to as low a reflectance as 0.02%.
The Differential Colorimeter reads in terms of per cent reflectance of
the tristimulus filter, which may be converted to X Y 2 tristimulus values,
or directly to Adams’s chromatic value system. As with other tristimulus
colorimeters this instrument performs best if it is used as a difference
meter, i.e., to compare colors which are close together in color space.
All of the current instruments of this type use the recommended angles
of incidences or reflection, either R46-o or RO--45,
and exclude the directional
reflectance. These instruments in general greatly simplify the specification
of nonfluorescent colors, if used within their limitations.
6. Other Instruments
To simplify further the specification of colors there have been de-
veloped a series of instruments dependent essentially on the reduction
of the selected ordinate method as used in spectrophotometry. If one
attempts to measure a series of color specimens whose reflection or trans-
mission curves differ only in extremely small details one can obtain con-
siderable accuracy and still decrease the number of selected ordinates
used. Essentially we enter abridged spectrophotometry. The observer
decides where the characteristic absorption of the average sample occurs
(usually as the result of spectrophotometric measurements) and uses a
few measurements to express the effect of small variations in the curves.
By this method it is intended to reveal very small differences between
like-colored samples rather than to measure the actual color. The reason
for this simplification is obvious-it is capable of differentiating between
an acceptable or nonacceptable product without recourse to any in-
volved or lengthy calculations and may yield a single-valued function
which then may be used to characterize a grade. Naturally any of the
instruments used to specify color may in some cases be used in an abridged
method.
Several instruments have been developed such as the Agtron and the
Purdue Colorimeter that are specifically used for abridged spectro-
photometry of particular products. These instruments can be operated
quite satisfactorily by unskilled operators and the results obtained require
no interpretation. The drawbacks are many. They cannot be used with
any degree of success on colored objects differing materially from those
they were designed for. They must be standardized with nonmetameric
matches. If they are used to compare even moderately metameric
specimens the data obtained may be so far from correct as t o be com-
pletely meaningless. The data obtained even with nonmetameric colors
are generally nonconvertible and therefore valid only for the particular
instrument used, requiring separate specifications to be introduced for
each product and for each instrument.
It would seem appropriate to consider the instrumentation problem
in a specific industrial case, the grading of tomatoes, where two sub-
stantially different approaches have been made, the one developed on the
Atlantic seaboard, the other in California. Two separate problems are
involved: first, the assignment of the correct color grade to a given lot of
tomatoes, determined prior to acceptance by the canner. The second is
the ability to predict the final color grade of the tomato product from
the first set of measurements. This necessitates further data on the
processed material in order to establish a correlation. In the East, the
Hunter Color-Difference Meter has been used for both sets of data. For
both measurements, the tomatoes must be pulped and the pulp deaerated.
The aL/bL ratio is then calculated, and the resulting value checked for the
grade. As noted earlier, Robinson et al. (1952) correlated the ratio with
dominant wave length.
In California, the Agtron has been developed specifically for the pur-
346 a. MACKINNEY AND C. 0. CHICHESTER
pose of assigning a grade to the raw tomato, and is described by Smith

and Huggins (1952). It embodies an abridged spectrophotometric pro-
cedure but differs from those previously discussed in that one measures
directly the ratio of the reflectances at the 546 mp mercury line and at the
neon 640-651 mp lines. Two filters are used (Corning 4-64 and 2-61)
which effectively isolate the two spectral regions from their respective
low-pressure discharge tubes. There are two circuits with different and
adjustable sensitivities for the red and green. Moving the filters switches
the circuits, each of which is adjusted to a “standard level” by means of
plastic discs of controlled color. The 546 and 640 mp regions were selected
partly for convenience and simplicity, and because measurements a t 640
are reasonable sensitive to chlorophyll characteristic of unripe tomatoes
and at 546 they are sensitive to lycopene in the well-ripened fruit.
Measurements are not made on pulped tomatoes, but upon both cut
halves. Field inspection procedure has been outlined by Whipple (1952)
of the Bureau of Fruit and Vegetable Standardization, State of Cali-
fornia. Since both grower and canner representatives may be present, the
cut fruit may be returned to the former to aid picking by relating external
to internal color, or given to the canner to relate the color grades of raw
and finished product.
Since the Agtron as now used makes no measurement of the bright-
ness, results cannot be converted to any of the standard color systems.
It is primarily a field instrument, while the Hunter is primarily a labora-
tory instrument designed for wider applicability. The Agtron type is an
attempt a t reducing the complexities of color aystems to a single meamre-
ment upon which grades of one specific commodity can be determined.
The “Purdue Color Ratio Meter” developed by Desrosier et al. (1952)
has certain features in common with the Agtron and involves essentially
the same considerations in operation, though skin color instead of flesh
color is measured. The Purdue instrument uses two banks of three photo-
cells and a single light source. A ratio is obtained which is indicated as a
single-value function. I n order t o obtain an average reading the latest
machine rotates the tomato. The choice of filters would seem somewhat
less advantageous, the 560 mp region being chosen for the carotenoid and
evaluation of good color development.
Where dependence is placed upon less than three attributes (whether
2, y, and Y , aL, bL, or any equivalent set) the findings of Younkin (1950b)
should be borne in mind. Examining a large number of tomato purdes, he
observed that the majority could be evaluated for appearance from hue
and brightness, but for precise classification of all samples, all three
attributes needed to be considered. Experience in the present authors’
laboratory fully substantiates this conservative view. Genetic strains of
tomatoes light pink in color will be completely devoid of chlorophyll when

fully ripe. It may be surmised that they have the correct dominant wave
length owing to the presence of small amounts of lycopene. The chroma
or purity is so low that no inspector could accept them, yet they might
be acceptable by instrumental check. Instrumentation and current activity
in the color field show that it is in a state of flux, and nothing better than a
guess can be hazarded as to future developments. The extent to which the
Agtron is used depends upon developments both in the general thinking on
the subject and in specific changes that may be projectedin theinstrument
itself. The original blue filter to isolate the 436 mp line (note the article by
Smith and Huggins, 1952) has already been superseded by the Corning green
filter to isolate the 546 mp mercury line. The basicidea, however, has found
acceptance in the State of California, Bureau of Fruit and Vegetable
Standardization, as an acceptable and practical one for easy reference in
field inspection work.
A similar situation exists for the Hunter instrument, where both
circuit changes and filter modifications are in progress. Some thought is
also being given to changing the characteristics of the 1931 standard
observer to conform with Judd’s findings (Hunter, 1952). Matches based
on material standards differing solely in brightness (nonmetameric
matches) are clearly worth exploring. Abridged spectrophotometry may
save an inordinate amount of labor. Color space transformations for
perceived difference evaluation would seem less useful in foods than
elsewhere, since differences must be expected in the perceived colors
within a grade, and it is only the line of demarcation between two grades
that need be determined.
It is of course possible to pick a single standard and to set grades so
that samples within a grade shall not differ by more than a certain number
of N.B.S. units, or by some other measure of AE.
It becomes increasingly apparent that color in foods will be measured
in one of two ways: by abridged spectrophotometry, or by direct color
measurement, preferably in terms of the 1931 C.I.E. conventions in either
the original C.I.E. color space or one of its more useful transformations.
One may also anticipate increased use (when applicable) of carefully
prepared plastic material standards which can be cleaned and polished
as often as necessary, for nonmetameric matches delimiting tolerances
acceptable in a given color grade for a food.
REFERENCES
Adams, E. Q. 1942, X - Z planes in the 1931 I.C.I. system of colorimetry. J . O p t .
SOC.Amer. 83, 168.
Adams, E. Q.1943. Chromatic valence as a correlate of Munsell chroma. J . Opt. SOC.
Amer. 88, 683.
348 0. MACKINNEY A N D C. 0. CHICHESTER
American Oil Chemists’ Society. 1946. Official and tentative methods of 2nd ed.
Chicago, Illinois.
Amerine, M. A., and Winkler, A. J. 1941. Color in California Wines. Food Research 6 , l .
Ayres, G. H. 1949. Evaluation of accuracy in photometric analysis. Anal. Chem. 21,
652.
Bouma, P. J. 1947. “Physical Aspects of Colour,” trans. by de Groot. Elsevier Press,
New York.
Boutaric, A., Ferre, L., and Roy, M. 1937. Recherches spectrophotometriques sur la
couleur des vins. Ann. fals. et fraudes 30, 196.
Broeg, C. B., and Walton, C. F., Jr. 1952. Color standards for sugar cane sirup and
edible sugar cane molasses. Anal. Chem. 24, 832.
Browne, C. A., and Zerban, F. W. 1941. “Handbook of Sugar Analysis,” 3rd ed.
John Wiley & Sons, Inc., New York.
Bruce, B. A., and Turner, A. 1952. Permanent glass color standards for maple sirup
and honey. J . Opt. SOC.Amer. 42, 881.
Buck, R. E., and Sparks, R. A. 1952. Relation of ketchup color to tomato color as
determined by the Hunter instrument. Food Technol. 6, 122.
Cannon, C. G., and Butterworth, I. S. C. 1953. Beer’s and spectrophotometer linearity.
Anal. Chem. 26, 168.
Caster, W. 0. 1951. Variability in the Beckman spectrophotometer. Anal. Chem.
23, 1229.
Chamberlin, G. J. 1951. The C.I.E. colour system explained. Tintometer Ltd., Salis-
bury, England.
Davidson, H. R. 1951. Calculation of color differencesfrom visual sensitivity ellipsoids.
J . Opt. SOC.Amer. 41, 1052.
Davidson, H. R., and Imm, L. W. 1949. A continuous automatic tristimulus integrater
for use with the recording spectrophotometer. J . Opt. SOC.Amer. 39, 942.
Desrosier, N. W., Gaylord, F. C., Kellie, W. F., and Ellis, N. K. 1952. Meter simplifies
color grading of fruits and vegetables. Food Eng. 24(5), 92.
Donaldson, R. 1947. A colorimeter with six matching stimuli. Proc. Phys. SOC.(Lon-
don) 69, 554
Dutton, H. J., Bailey, G. F., and Kohake, E. 1943. Dehydrated spinach: changes in
color and pigments during processing and storage. Ind. Eng. Chem. 86, 1173.
Eastmond, E. J. 1950. Measurement of color changes in foods. Advances in Chem. Ser.
h e r . Chem. SOC.number 3, pp. 3-12.
Eastmond, E. J., Peterson, J. E., and Stumpf, R. R. 1951. Observation of color
changes in some processed and stored foods. Food Technol. 6 , 121.
Eolkin, D. 1952. Relative color evaluation of purees using black and white photog-
raphy. Food Technol. 6, 214-217.
Evans, R. M. 1948. “An Introduction to Color.” John Wiley & Sons, Inc., New York.
Ewing, G. W., and Parson, T., Jr., 1948. Intercomparison of Beckman spectro-
photometers. Anal. Chem. 20, 423.
Fawcett, A. J. 1951. Colour and the visual measurement of colour. Tintometer Ltd.,
Salisbury, England.
Friedman, M. E., Marsh, G. L., and Mackinney, G. 1952. On color in tomato prod-
ucts. Food Technol. 6, 395.
Gardner Laboratory, Inc. 1950. Description and instructions for Hunter color and
color-difference meter (Brochure). Henry A. Gardner Lab., Inc., Bethesda,
Maryland.
Gibson, K. S. 1949. Spectrophotometry. Nat’l Bur. Standards. Circ. 484.
Gillett, T. R. 1953. “ Chemical-Physical Principles of Sugar Technology ”. (Pieter
Honig, ed.). Elsevier Press, New York.
Gillett, T. R., and Holven, A. L. 1943. Grading of soft sugars. Znd. Eng. Chem. 36,
210.
Gillett, T.R.,and Meads, P. F. 1952. Measurement of reflectance of white sugar.
Anal. Chem. 24, 829.
Gillett, T.R.,Meads, P. F., and Holven, A. L. 1949. Measuring color and turbidity of
white sugar solutions. Anal. Chem. 21, 1228.
Glasser, L. G., and Troy, D. J. 1952. A new high sensitivity differential colorimeter.
J . Opt. SOC.Amer. 42, 652.
Goldring, L. S., Hawes, R. C., Hare, G. H., Beckman, A. O., and Stickney, M. E.
1953.Anomalies in extinction coefficient measurements. Anal. Chem. 26,869-878.
Granger, G. W.1952. Objectivity of colour preference. Nature 179,778.
Hardy, A. C. 1936. “Handbook of Colorimetry.” Technology Press, Cambridge,
Massachusetts.
Hardy, A. C., and Young, F. M. 1949. Colorimetry by abridged spectrophotometry.
Hunter, R. S. 1952. Improvement of the Color-Difference Meter. J . Opt. SOC.Amer.
43,289.
Hunter, R. S. 1942.Photoelectric trietimulus colorimetry with three filters. Nat’l Bur.
Standards Circ. C429.
Hunter, R. S. 1948a.Photoelectric Color-Difference Meter. J . Opt. SOC.Amer. 38,661.
Hunter, R. S. 1948b. Accuracy precision and stability of a new photoelectric Color-
Difference Meter. J . Opt. Soc. Amer. 38, 1094.
Irvine, G.N., and Anderson, J. A. 1952.A note on the determination of brightness in
flour. Trans. Am. Assoc. Cereal Chemists. 10,59.
Jacobsen, A. E.1948.Non-adaptability of the ICI system to some near whites which
show absorption in the far blue region of the spectrum. J . Opt. SOC.Amer. 38, 442.
Judd, D. B. 1949.A comparison of direct colorimetry of titanium pigments with their
indirect colorimetry based upon spectrophotometry and a standard observer.
Judd, D. B. 1952. “Color in Business, Science, and Industry.” John Wiley & Sons,
Inc., New York.
Judd, D. B. 1950. Colorimetry. Nat’l Bur. Standards Circ. 478.
Kaye, W.,and Devaney, R.G. 1952.An automatic relative transmission attachment
for the Beckman Model DV Spectrophotometer. J . Opt. SOC.Amer. 42, 567.
Kent-Jones, D.W.1952.Measurement of flour color as affected by grade. Trans. A m .
Assoc. Cereal Chemists 10, 61.
Kent-Jones, D. W., Amos, A. J., and Martin, W. 1950. Experiments in the photo-
electric recording of flour grade by measurement of reflecting power. Analyst 76,
133.
Kent-Jones, D. W., and Martin, W. 1950. A photoelectric method of determining the
colour of flour as affectedby grade, by measurements of .reflecting power. Analyst
76, 127.
Kramer, A. 1950. This meter gives better color evaluations. Food Znds. 22, 1987.
Kramer, A. 1951. Objective testing of vegetable quality. Food Technol. 6, 265.
Kramer, A., and El-Kattan, A. A. 1953.Effect of application of heat on tomato juice
color. Food Technol. 7 , 400-404.
Livingston, G. E., and Vilece, R. J. 1953. Measuring discoloration of food purees in
glass tubes. Food Technol. 7 , 272-275.
350 0. MACKINNEY A N D C. 0. CHICHESTER
MacAdam, D. L. 1942. Visual sensitivities to color differences in daylight. J. Opt.

SOC.A ~ T32,. 247.
MacAdam, D. L. 1950, Method of colorimetric integration using punch-card account-
ing machines. J. Opt. SOC.Amer. 40, 138.
MacGillivray, J. H. 1937. Spectrophotometric and colorimetric analyses of tomato
pulp. Proc. Am. SOC.Hort. Sci. 36, 630.
Mackinney, G., and Weaat, C. A. 1940. Color changes in green vegetables. Znd. Eng.
Chem. 32,392.
McNicholas, H. J. 1935. The color and spectral transmittance of vegetable oils. J.
Research Nat’l Bur. Standards 16, 99.
Meads, P. F., and Gillett, T. R. 1949. Preparation of sugar liquors sirups for color
determination. Anal. Chem. 21, 1495.
Melvin, E. H., MacMillan, D., and Senti, F. R. 1953. Suggested spectrophotometric
methods for grading green soybean oil. J. Am. Oil Chemists SOC.30, 255-259.
Moon, P., and Spencer, D. E. 1943. A metric for color space. J. Opt. Soc. Amer. 33,
260, 270.
Moster, J. B., and Prater, A. N. 1952. Color of capsicum spices. I. Measurement of
extractable color. Food Technol. 6, 459.
Murray, H. D. 1952. “Color in Theory and Practice.” Chapman & Hall, Ltd., London.
Nickerson, D. J. 1935. Disk colorimetry including a comparison of methods for com-
puting tristimulus values for certain disks. J. Opt. Soc. Amer. 26, 253.
Nickerson, D. 1936. The specification of color tolerances. Teztile Research 6 , 509.
Nickerson, D. 1950. Munsell renotation used to study the color space of Hunter and
Adams. J. Opt. Soe. Amer. 40, 85.
Nickerson, D., and Newhall, S. M. 1943. A psychological color solid. J. Opt. SOC.
Amer. 33, 419.
Optical Society of America, 1943. Sub-committee on the Spacing of the Munsell
Color. J. Opt. SOC.Amer. 33, 385.
Optical Society of America Committee on Colorimetry. 1953. “Science of Color.”
Thomas Y. Crowell Company, New York.
Peterson, E. W., Grossman, H. H., Saunderson, J. L., and Caldecourt, V. J. 1944. An
electrical tristimulus integrater. J. Opt. SOC.Amer. 39, 1055.
Production and Marketing Administration. 1945. U. S. Standards for Grades of
Canned Tomato Puree. Jan. 1, 1945. United States Department of Agriculture,
Waahington, D.C.
Robinson, W. B., Ransford, J. R., and Hand, D. B. 1951. Measurement and control
of color in the canning of tomato juice. Food Technol. 6, 314.
Robinson, W. B., Wishnetsky, T., Ransford, J. R., Clark, W. L., and Hand, D. B.
1952. A study of methods for the measurement of tomato juice color. Food
Technol. 6 , 269.
Sastry, L. V. L., and Tischer, R. G. 1952. Stability of the anthocyanin grapes in con-
cord grape juice. Food Technol. 6, 264.
Saunderson, J. L., and Milner, B. I. 1946. Modified chromatic value color space.
J . Opt. Soc. Amer. 36, 36.
Schofield, R. K. 1939. The Lovibond Tintometer adapted by means of the Rotham-
sted device to measure colours on the C.I.E. system. J. Sci. Instr. l6,74.
Scofield, F. A method for determination of color differences. Nat’l Paint, Varnish and
Lacquer Assoc., Sci. Section Circ. 664 (1943).
Shah, J. N., and Worthington, 0. J. 1953. Comparison of several methods for
specifying the color of frozen strawberries. Paper 28, 13th, Annual Meeting,
I.F.T., Boston, Massachusetts, p. 14 of Technical Program and Abstracts.
Smith, T. J., and Huggings, R. A. 1952. Tomato classification by spectrophotometry.
Elect Tonics.
Sondheiner, E., and Kertesz, Z. L. 1947. The anthocyanin of strawberries. J. A m .
Chem. SOC.70, 3476.
Sturrock, W., and Staley, K. A. 1950. Fundamentals of light and lighting. Gen. Elect.
Bull. LD-2,Cleveland, Ohio.
Thomson, P. 1953. Single number systems for the color of transparent objects. J. A m .
Oil Chemists Soc. 30, 259-262.
Thomson, P. 1947. Methods of reading color without Lovibond. J. Am. Oil Chemists
SOC.24, 387.
Tintometer Ltd. 1950. Colorimetry, Part I. Tintometer Ltd. Salisbury, Eng. Lovibond
System. Part 11. Lovibond-Schofield System.
Troland, L. T. 1929. “Psychophysiology.” D. Van Nostrand Company, Inc., New
York.
Wald, G. 1945. Human vision and the spectrum. Science 101, 653.
Whipple, S. R. 1951. Problems in grading canning tomatoes to color standards by
visual methods: and an electronic means of objectively grading color to these
standards. Department of Agriculture Bureau of Fruit and Vegetable Standard-
ization, Sacramento, California.
Wiklund, 0. 1950. Colorimetry, nephelometry, reflectometry and fluorescence in the
sugar industry. Socker 6, 225.
Winkler, A. J., and Amerine, M. A. 1938. Color in California Wines. Food Research 3,
429.
Worthington, 0. J., Cain, R. F., and Wiegand, E. H. 1949. Determination of color
of unclarified juices by reflectometer. Food Technol. 3, 274.
Younkin, S. G. 1950a. Application of Hunter Color-Difference Meter to a tomato color
measurement problem. J. Opt. Soc. Amer. 40, 265.
Younkin, S. G. 1950b. Measurement of small color differences in tomato purees. J.
Opt. SOC.Amer. 40, 596.
Younkin, 5.G. 1950c. Color measurement of tomato purees. Food Technol. 4, 350.
Zerban, &’.W., Martin, J., Erb, C., and Sattler, L. 1952. Spectrophotometric studies
on raw cane sugar in solution. Anal. Chem. 24, 168.
Zerban, F. W., Sattler, L., and Martin, J. 1951. Spectrophotometric studies on refined
sugar in solution. Anal. Chem. 23,308.
Composition of Wines.
Compositionof I. Organic
Wines. I. Organic Constituents
Constituents
BY MAYNARD
BY MAYNARD A. AMERINE
A. AMERINE
Department of
Department of Viticulture
Viticultureand
and Enology,
Enology, College
College ofof Agriculture,
Agriculture, University
University of California
of California
CONTENTS
CONTENTS
Page
Page
I. Introduction..... . . . . . . . . . . . . . ... .,.,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
I. Introduction.. 354
354
11. General
11. General Methods
Metho OP Analysis.. s . ....... .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
356
111. Alcohols and Related Compounds. . 358
Alcohol.....
Ethyl Alcohol.
1. Ethyl
1. ... . . . . . . . . . . 359
2. Methyl Alcohol. . . . 366
3. Higher Alcohols.. . 368
4. Glycerol .............. 372
5. 2,3-Butylene Glycol, Acetylmethylcarbinol, and Diacetyl 377
IV. Aldehydes and Related Compounds.. . . 382
1. Acetaldehyde.. , , . . . . . . . . . . . . . . . 382
2. Acetal . . . . . . . . . . . . . . . . ......... 385
3. Acetone and Benzaldehy ......... 385
4. Hydroxymethylfurfural 385
5. Acrolein 386
V. Acids.. . . . . .... . . 386
1. Titratable Acidity. . . 390
2. Tartaric Acid.. 391
3. Malio Acid. . . . . 395
4. Citric Acid., . . . 397
5. Succinic Acid. . . 402
6. Lactic Acid.. . . . . . . . 403
7.7. Other
Other Fixed
Fixed Acids..Acids......... . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
407
Volatile Acidity
8.8. Volatile Acidity (Acetic (Acetic Acid). Acid).... .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
407
9.9. Other
Other Volatile
Volatile Acids.. Acids. ................ .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
412
10. p H . . . . . . . . . . . . . . . . . . . . . . . . . . . ......... 413
413
VI. Carbohydrates
VI. Carbohydrates.. . . ... .. .. . . . . . . ._. .. .. . . .. ......... .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
418
.
1. Hexoses.. . . . . . . . . . . . . . . . . . . . . . 419
419
2. Pentoses and Related Compounds 6. . . . . . . . . . . . . . . . 423
423
3. Sucrose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
424
. 424
, 426
6. Extract . . . . . . . . . . . . . 428
430
I 435
1. Tannins.. . . . . . . . . 436
2. Color-Red Wines. 441
3. White Wines.. .... ......... ........ 447
353
353
354 MAYNARD A . A M E R I N E
Page
IX. Nitrogenous Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
......................................... 448
................................................ 448
3. In Fermentation and Aging.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
4. In Clouding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
5. Amounts ..................................................... 452
X. Enzymes, Vitamins, and Aromatic Constituents.. . . . . . . . . . . . . . . . . . . . . . 454
1. Vitamins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
2. Aromatic Constituents ....................... 461
XI. Summary. . . . . . . . . . . . . . . . . . . . . 464
Acknowledgments . . . . . . . . . . . . . 466
References.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
I. INTRODUCTION
Whereas Pasteur, Berthelot, and other chemists of the last century
studied wine' from a biochemical point of view, the emphasis in enology
since 1900 has been primarily on determining the percentage of the
various constituents normally present in the wines of a given region or
type, on studies to detect sophistication, or on the influence of processes
on the composition of wines. More recently, however, enologists have
again studied wine from the biochemical point of view. The nature,
source, and fate of the constituents of wine have thus been investigated
and the results interpreted biochemically.
The present review is limited t o the organic components of wines and
to publications in the period 1930 to 1953. Both phases of enology will be
reviewed, but it is obvious that the more significant studies have been
those of a physical-chemical nature. The organic components of wines are
discussed in broad general groups-acids, carbohydrates, alcohols, etc.
For each component, information on the methods of determination is
given first, followed by a review of work on the occurrence or significance
of the compound.
During this period many text books and reviews have been pub-
lished. Among these are Seifert (1938), Garoglio (1941-1942), Vogt
(1945), Ventre (1946-1947), Cruess (1947), Genevois and RibBreau-
Gayon (1947), Ribdreau-Gayon (1947), Sannino (1948), Amerine and
1 In this review wine is the fermented beverage of the fruit of one of the several species
of Vitis. Other terms which may be unfamiliar include: table wine, a wine of 14% or
less of alcohol; dessert wine, a wine of over 14% alcohol; must, unfermented grape
juice with or without the skins; free-run, the juice which drains from the crushed
grapes without pressing; fortified wine, a wine to which alcohol distilled from wine
(fortifying brandy) has been added during or after fermentation; racking, the transfer
of wine (less its sediment) from one container to another; fining, the use of organic
or inorganic materials to assist clarification; and lees, the sediment which is deposited
after fermentation.
COMPOSITION OF WINES 355
Joslyn (1951), Benvegnin et al. (1951)) Frolov-Bagreev and Agabal’gnG

(1951). Amerine (1951) included an annotated list of 171 of the most
important books now available. The problems of reviewing the enological
literature have been studied by Amerine and Wheeler (1951) and Kiel-
hofer (1952), both of whom noted the more important reference sources.
Reviews of work on the chemistry of wine from about 1920 to 1937 were
given by Reichard in Bleyer (1938) and Hewitt (1939). A review of the
period 1927 to 1936 was given by Kielhofer (1937). Practical wine-
making advances are particularly well-covered. A selected list of recent
publications on fermentation physiology, wine-handling, and wine
chemistry have been reviewed by Hennig (1951-1953). German, Russian,
and American articles were included, with lesser attention to French,
Spanish, and Italian work.
Reviews of the process of alcoholic fermentation have been made by
Joslyn (1940a, 1949), Cruess (1950), Genevois (1949b, 1950), Meyerhof
(1952), and Antoniani (1951a, b). Many of the mutual chemical inter-
relationships of microorganisms and musts and wines are reviewed by
Schanderl (1950). Although only a limited amount of chemical data is
included, Schmitthenner (1937) has reviewed 50 years of the micro-
biological aspects of enology.
Germany. The technical enological problems of the Moselle region
were summarized by Petri (1932). Use of ultraviolet light, production of
low-alcohol and normal sparkling wines, and differentiation of wines made
from normal and sugared musts were considered the primary problems.
Kramer (1942) summarized the latest problems in winery practice as
removal of arsenic, influence of yeast strains, malo-lactic fermentation,
aging of wines, wine diseases, and wine analysis. A brief review of the
status of German viticulture and enology on the eve of World War I1 was
made by Husfeld et al. (1939). Among the most important of these from
the chemical point of view were reports on the danger of using pure
aluminum in wineries; recovery of oil, oenin, tartaric acid, and other
products from the pomace; and methods of analysis. Further advances
in German enology were reported by Eckert (1950). Among the subjects
discussed were the errors of the chlorbenzaldehyde procedure for sorbite,
the use of citric acid, and the earthy taste resulting from bentonite. An
excellent account of the analytical procedures used in Germany was given
by Reichard in Bleyer (1938).
Hungary. An excellent summary of the enological work being pursued
in Hungary was given by Dicenty et al. (1935), including a discussion of
such fundamental problems as methods of determining the organic acids
as benzoic esters, detection of concentrate in wines, studies on the use of
potassium ferrocyanide for removal of iron and copper, changes in alcohol
356 MAYNARD A. AMERINE
in wines under various conditions of humidity, and studies on the

fructoselglucose ratio. So65 et al. (1948) also reviewed Hungarian re-
search on the aromatic principles in grapes and with strains of yeasts and
molds and their influence on the process of fermentation.
Other Countries. A review of the applied aspects of viticulture and
enological research in France was made by SBmichon (1948). An informal
review of selected research in enology (mainly American) was presented
by Cruess (1942). Numerous articles on various phases of the biochemistry
of wine appear in the publication of the Institut Biokhimii of the Akad-
emifa Nauk (1947-1950) of Russia. Sherry, brandy, and sparkling wines
are favored subjects for study. K a t a r ’ h (1951) reviewed the enological
work of the Magarach Institute: the malo-lactic fermentation, production
of sherry, use of bentonite, and studies on the redox potential.
11. GENERALMETHODSOF ANALYSIS
Reviews of methods of analysis of wines have been published by
Anonymous (1933, 1934, 1937), Fabre (1936), Dujardin et al. (1938),
Azevedo (1942), Garoglio (1941-1942), Campllonch-Romeu (1945),
RibBreau-Gayon and Peynaud (1947a), Florentin (1948), Sudario (1949) ,
Association of Official Agricultural Chemists (1950), Hennig (1950) ,
Joslyn (1950), Jaulmes (1951), and Amerine (1952). The books by
RibBreau-Gayon and Peynaud and Jaulmes are probably the best discus-
sions of the analysis of wines now available. They include original
research and comments on procedure. General studies and criticisms of
the methods of analysis of wines have been presented by Malvezin (1931),
von der Heide (1934), Marcilla (1934) , Anonymous (1935) , RibBreau-
Gayon and Peynaud (1946a), Valaer (1947) , Garino-Canina (1948,
1950), Godet (1949), and RibBreau-Gayon (1949).
A special congress on wine analysis was held in Rome in 1935. Reports
concerning this congress and comments on the results are given by
Filaudeau (1933) , Benvegnin (1934), Anonymous (1935) , and Marcille
(1937). Godet (1949) noted that too many empirical procedures and
methods of expressing the results were used. He recommended milli-
equivalents per liter. The primary problems in this usage arise in the
analyses where mixtures of substances are determined : total acidity,
extract, etc. He recommended that all methods should be clearly defined
as to objective, principles, apparatus, analytical steps, method of expres-
sion, and sources of error. The 6th International Congress of Viticulture
and Enology in 1950 also studied the general problem of analysis of wine.
The results were summarized by Garino-Canina (1950) and Anonymous
(1952). These reports are valuable in indicating the differences in pro-
cedures and definitions among various countries and the general dissatis-
COMPOSITION O F WINES 357
faction with the lack of preciseness in certain methods: extract, alkalinity

of the ash, tannin, etc. The methods followed in the National Department
of Public Health in Brazil were reviewed by Bicalho (1933) and those
used in Venezuela by Noguero (1942). The Italian (and some other) pro-
cedures for preventing the sophistication of wines and for protecting
named types of wines were reviewed by Garino-Canina (1951b). He noted
the variation in regulations between various countries and the need for
more attention to control of antiseptics.
Reviews of less general interest were given in many reports. Comments
on methods of wine analysis were made by Marcille (1933) with respect
to indicators, standard solutions, and volatile and total acid determina-
tions. fjumuleanu and Ghimicescu (1935) reviewed their micro pro-
cedures. Comparative results of various methods for alcohol, extract, and
total acidity were given by Joslyn and Marsh (1935). A brief summary of
the methods of wine analysis in use in California was given by McCharles
and Pitman (1936). Various analytical methods were discussed by
Bremond (1937~)and Lagneau (1945). Special procedures for deter-
mining the less common fermentation by-products were given in Neish
(1950). SBmichon and Flanzy (1932a) proposed dichromate oxidation of
various acids. HeyrovskJi el al. (1933) applied his polarographic procedure
to the determination of fructose and malic acid in wines. Various im-
provements in the electrochemical methods for wine analysis were
reported by Mestre and Garcia (1947). They used barium hydroxide
titration for determination of ash and sulfates, silver nitrate for chlorides,
and lanthanum nitrate for strong and weak acids. The procedures used in
their studies on sherry wines were detailed by Bobadilla and Navarro
(1949). The detection of grape wine in blackberry wine was studied by
spectrophotometric means by Beyer (1945). Various analytical uses of the
photoelectric colorimeter in wineries were outlined by Barini-Banchi
(1949). Jaulmes (1950) described an elaborate steam-distillation appa-
ratus suitable for the determination of ammonia, alcohol, volatile acidity,
and other constituents in wines.
Chromatography. Ramos and Oliveira (1949) made chromatograms of
port wines. A native clay and aluminum oxide were used. The presence of
elderberry wine in ports could be detected. Use of chromatography as a
means of detecting sophistication was also proposed by Valaer (1949).
Hozumi and Sat6 (1950) used paper-partition chromatography to
demonstrate d-maltose as one of the sugars present in a Japanese white
wine. Sulser (1951) described a procedure to detect tartaric acid, sorbitol,
fructose, glucose, and glycerol in wines and these, citric acid, and hy-
droxymethylfurfural in vinegars. Marsh and Kean (1951) also applied
the technique t o detect grape in berry wine. Identification of foreign
colors which may occur in wines was successfully accomplished by paper

chromatography by Ruf (195213). The general advantages of the pro-
cedure were summarized by Ruf (1952a). Data on the behavior of various
fruits including white and red grapes in a chromatography column of
aluminum oxide were given by Jakovliv (1952).
Statistical Analysis. Application of statistical methods to analytical
data on wines has not been common. Oliveira (1941), however, made a
detailed statistical study of the analyses of various constituents in 600
white and 1071 red port wines. The frequency curves are somewhat
asymmetric, particularly for the volatile acidity; this is, perhaps, to be
expected in dessert wines-where a small constant amount is formed
during fermentation and where the quantity present increases slowly
during aging. The averages were:
Type of No. of Range of 98%
port samples Mean of samples
Volatile acidity (% acetic) Red 1068 0.066 0.035-0.126
White 400 0.058 0.30 -0.099
Total acidity (% tartaric) Red 1070 0.41 0.30 -0.55
White 600 0.37 0 . 2 8 -0.50
Alcohol (% by vol.) Red 1068 19.4 -
White 600 19.3 -
Hickinbotham and Ryan (1948) made a useful application of analyses of

variance to data obtained on the factors influencing the formation of
glycerol during fermentation. Heitz et al. (1951) also applied several
interesting statistical procedures to data obtained in connection with the
influence of aeration and various constituents of the wine on the changes
taking place during baking of sherry. Bohringer (1943) used statistical
procedures in determining the probable error of refractometer readings.
111. ALCOHOLS AND RELATED COMPOUNDS
The organoleptic and physiological properties of alcoholic beverages
are largely conditioned by the ethyl alcohol they contain. Wines are
classified, taxed, and bought and sold according t o alcohol content.
Therefore, great attention has been given to its accurate and rapid
determination. Less attention is paid to methyl alcohol as a component
of beverage wines, but its toxicity has led to many reports on its presence
in brandies and in the wines from which they were distilled. Higher
alcohols are apparently of some importance in the odor of some wines,
and of course they play an important role in brandies. Glycerol, 2,3-
butylene glycol, and acetylmethylcarbinol are related products of lesser
importance. Mixtures of methyl, ethyl, isopropyl, n-butyl, and isoamyl
alcohols can be resolved by partition chromatography according to
Neish (1951), and 2,3-butylene glycol does not interfere. Antoniani
(1951b) has reviewed the theories for the production of glycerol, acetyl-
methylcarbinol, 2,a-butylene glycol, higher alcohols, succinic, citric,
formic, acetic, and lactic acids, methyl alcohol, and isobutylene glycol.
Need of more work on higher alcohols, citric acid, isobutylene glycol, and
formic acid, particularly, was noted.
1. Ethyl Alcohol
Wines may contain from 8% to 24% alcohol2 depending on the sugar
content of the grapes, the amount added during or after fermentation, if
any, and the losses that occur during fermentation or aging. A number of
general studies on accurate alcohol determination have been made, among
them Fabre and BrBmond (1935), Joslyn et al. (1937), and Mica1 (1949).
Fairly complete discussions will also be found in RibBreau-Gayon and
Peynaud (1947a), Joslyn (1950), Jaulmes (1951), and Amerine (1952).
Methods-Ebulliometric. Enology has made more use of ebulliometry than any
other industry. A dozen different instruments are in use throughout the world for the
rapid determination of the per cent alcohol. An historical review of ebulliometry in
enology and data on the proper construction of ebullioscope scales from water :alcohol
boiling point data were given by De Astis (1933). He constructed an ebullioscope on
the basis of these studies. Unfortunately wine is more than a water-alcohol mixture,
and the boiling point is influenced by the other constituents. The amount of change in
the boiling point depends on the concentration of electrolytes and nonelectrolytes, the
tendency of the liquid to separate into two phases, and the volume occupied by the
dissolved materials. Most of the studies have been directed toward reducing these
errors. Joslyn et al. (1937), RibBreau-Gayon and Peynaud (1947a), and Jaulmes (1951)
have discussed the historical and theoretical aspects of the determination. On the
basis of collaborative analyses, Joslyn et al. (1937) concluded that results accurate to
f0.2% may be expected. Prior to using the ebullioscope Korotkevich (1949) distilled
25 ml. into a 50-ml. volumetric flask. Correction factors to apply to the usual empirical
ebullioscope scale were given. His results also agreed with the pycnometer to f0.2%.
Errors in the ebulliometric determination of alcohol in wines of -0.2% to f 1 . 5 %
were found by Lherme (1933), who discouraged their use in white wines and did not
find that a table of correction based on sugar content was acceptable. His results sug-
gest that studies based on an analysis of variance might prove valuable in determining
the influence of the various factors on the apparent boiling point. Bertin (1933)
reported that below 1% sugar there was little influence of the sugar on the boiling
point.
Jaulmes (1951) has emphasized that many ebullioscope scales are incorrectly
constructed as to the relation between boiling point and per cent alcohol. He considers
De Astis’ (1933) data the best for correcting such scales. De Astis (1941) has also given
useful information on the relation of barometric pressure and boiling point. Not only
is this important in the construction of ebullioscope scales, but abrupt changes in
barometric pressure occasionally occur and cause errors. Balestrazzi (1952) presented
a simple method for correcting the ebullioscope reading for changes due to barometric
pressure.
Various Italian ebullioscopes were tested by Cusmano (1949). In the sweeter, more
alcoholic wines the De Astis ebullioscope gave better results than the Malligand. An
* The alcohol content of wines is customarily expressed as per cent by volume.
example of the variation between ebullioscopes is SBmichon and Flanzy's (1934)

report of 0.05% to 0.4% difference in alcohol with three instruments. The Dujardin-
Salleron ebullioscope has been discussed by Churchward and Johns (1940)and Church-
ward (1953),who found that the thermometers supplied with the 1901 model of
Dujardin-Salleron Ebulliometer (official in France since 1907) did not agree-errors
in excess of 0.1"were not uncommon. They also found that for precise work the slide-
rule disc was unsatisfactory owing to irregularities in the graduations. They published
a set of tables for use with the Dujardin-Salleron ebullioscope in which the British
Proof Spirit value is given for each ebulliometer degree-the depression of boiling
point measured in "C. This system has been officially adopted by the Australian
Excise Authorities. The ebulliometer reading on dry wines gives a direct indication
of spirit content. For sweet wine the reading was corrected by the following formula:
True P.S. = Apparent P.S.(l - "B6. X 0.015).
Some errors of the method and of the instruments in the ebulliometric determina-
tion of alcohol were pointed out by Niehaus (1934).Dilution of sweet wines was
recommended. A table for correction of ebulliometric determinations of sweet wines
was given by Churchward and Johns (1940).As Amerine (1952)has indicated: "The
presence of sugars in dessert wines is a complicating factor in the determination of
per cent alcohol based on the boiling point. According to Raoult's law dissolved sub-
stances, such as sugars, raise the boiling point. However, this increase in the boiling
point of sweet wines is usbally somewhat less than that which would be indicated by
the per cent alcohol. This lower boiling point of sugar-alcohol solutions is usually
explained by considering that sugar is insoluble in alcohol and that the sugar thus
occupies a part of the volume that would otherwise contain the water-alcohol mixture.
The per cent alcohol determined by the ebullioscope is thus generally slightly higher
than that determined by distillation procedures. The presence of aldehydes and esters
likewise is a complicating factor, they lower the boiling point and thus indicate too
high a percentage of alcohol. In an attempt to avoid the error due to sugars it is com-
mon practice to dilute sweet wines two or three times. Whether this avoids the error
or not is doubtful as the result must be multiplied by the appropriate factor which
likewise multiplies the relative error."
Bertin (1933),for example, showed that addition of increasing amounts of sugar
(and water to maintain a constant volume) gave an increasing per cent alcohol by the
ebullioscope. He corrected the ebullioscope reading by multiplying the per cent sugar
by 0.05 and subtracting the value obtained. Emiliani (1938)calculated a correction
+
equal to 1.001A 0.0262 - O.OOSAZ, where A is the per cent alcohol determined
and Z the per cent sugar in the wine. Procopio (1939) calculated a correction of
M z - (MzZ o*62),where M1 is the per cent alcohol found with the Malligand
1on
---
ebullioscope and 0.62is a factor which represents the volume occupied by 1 g. of sugar.
Paronetto (1950)has reviewed the various proposals (Bertin, 1933;Emiliani, 1938;
Procopio, 1939) as well as his own which have been made for correcting the alcohol
determination by the ebullioscope according to the amount of sugar present. The
deviation from the true value by the various formulas (in per cent alcohol) were:
-0.21 -0.11 0.00 +0.01 +0.11 +0.21
Outside to to to to to to Outside
Procedure -0.31 -0.31 -0.20 -0.10 +O.lO +0.20 +0.30 +0.31 Total
Bertin 6 2 5 11 12 7 9 20 72
Emiliani 1 1 6 19 29 11 3 2 72
Paronetto - 5 5 27 11 2 - 2 52
Procopio 5 1 6 23 18 16 2 1 72
Venezia (1949) criticized Procopio’s formula based on analysis of 54 wines, and
Procopio (1950~)replied that the anomalous results are inconsequential. Analysis of
86 sweet wines of his own showed the ebullioscope value corrected by Procopio’s
formula to vary from -0.38 to +0.42, but 71 of the values were within f O . l . He
gives further data in a later publication (1950d), using density and a table for calcu-
lating the correction.
Further difficulties in determining the alcohol content of sweet wines using the
ebullioscope were found b y Tartaglia (1950). He also noted that the esters lowered
the boiling point, and some of the difference between ebulliometric and distillation
procedures could be traced to this. He doubted if a correction formula could be used.
Jaulmes (1951) reported very variable results on the same wines according to the
instrument used.
A slightly different type of ebullioscope was described by Fantoni (1949). I n this
a lower flask, containing water, is first heated and the boiling point determined. Then
two-thirds of the wine in an upper flask are distilled into the lower flask, the volume is
brought to the same as that of the wine, and the boiling point is again determined.
Empirical tables were given for calculating the alcohol content. This procedure is
probably better than determining the boiling point directly, but it takes more time.
Double ebullioscopes are on sale in this country, and Errichelli (1952) reported a
new one in Italy. The advantage of these is that the boiling points of the water and
the wine are determined at the same time.
Methods-Hydrometric. The limitations of the hydrometer for determining the
alcohol content were indicated by Luckow (1935). He discussed the problems of
making to volume, clean dry hydrometers, time required for equilibrium, free floating
of the hydrometer, clean cylinders, and other frequently neglected factors influencing
accuracy. The method for the exact determination of alcohol by distillation as used in
Peru was reviewed by Pozzi-Escot (1949). No new points are involved, though he
stresses careful distillation to prevent losses of alcohol (adequate condenser capacity,
dilution of sample, delivery tube in water in the receiver, etc.). Jaulmes (1951) gave a
complete theoretical discussion on the separation of alcohol from wine.
A spiral still, with three spirals in series, vertical axis, and eccentric movement,
was proposed by Piazza and Rouzaut (1939) for the distillation of alcohol from wines.
Results higher by 0.1% to 0.2% compared to those with the ebullioscope were ob-
tained. Using the apparatus Rouzaut (1939) was able to obtain all of the alcohol in
7 minutes. Hanak (1932) proposed an all-glass apparatus which Colombier and Clair
(1936) have adopted for the distillation of wines. The latter investigators claim results
from 0.1% to 0.2% higher than with the usual apparatus.
Marcel and Bastisse (1942) recommended neutralization of the wine prior to
distillation. They suggested use of calcium hydroxide instead of potassium carbonate
for neutralization of the wine before distillation in order to remove foam-producing
materials.
Wines high in sulfur dioxide should be treated for its removal before distillation.
Got (1947) showed that wines of 15% to 18% alcohol and 200 to 400 p.p.m. of sulfur
dioxide yielded results 0.2% to 0.5% low. Use of concentrated alkali in a bubble cap
in the condenser prevents sulfurous acid from distilling over and gives a better distil-
late for the determination of alcohol, according to Rocques (1950).
Zheltkevich (1951) suggested manufacture of smaller alcohol hydrometers in
order to reduce the volume of distillate required. The necessity of checking the scale
of each hydrometer was emphasized b y Luckow (1931). Errors in alcohol hydrometers
of 0.4% were found by SBmichon and Flanzy (1934). Joslyn et al. (1937) on the basis
of collaborative analyses reported that results accurate to +0.1% should be obtained
with alcohol hydrometers. Jaulmes (1951) noted differences in "certified" French

hydrometers of -0.23 to +0.06 of the true value. Later (1953) he found errors of
about f.0.15 between French alcohol hydrometers. These, however, are not graduated
as closely as American alcohol hydrometers. However, he notes that there are errors
due to surface tension and inherent in the tables of temperature correction. He, there-
fore, calculated a new table. The corrections at 10% and 20% alcohol in his table and
in the United States table are as follows:
Temperature, 10 % 20 %
"C. Jaulmes U. S. Jaulmes U. S.
20 (68°F.) 0.80 0.71 1.41 1.33
25 (77" F.) 1.75 1.65 2.87 2.81
There thus appears to be an error of about 0.1 % in the American tables. Jaulmes
and Marignan (1953) have calculated the corrections to be applied when a hydrometer
calibrated a t 15" C. (59" F.) is used a t 20" C. (68" F.).
Methods-Pymmetric. Details for the accurate use of the pycnometer were out-
lined by Twight (1951). He stressed the necessity of an accurate calibration of the
pycnometer and control of temperature. Rankine (1952) obtained results to +0.05%
when proper attention was given to temperature control. He recommended use of a
water bath set slightly above room temperature and pycnometers calibrated to the
same temperature. Botelho (1939) preferred the pycnometer for the determination of
alcohol.
Gomes (1941) has made a thorough study of the factors affecting the distillation
of alcohol from port wines. Among these were: volume distilled, per cent of amyl
alcohol, ethyl acetate, acetic acid, or acetaldehyde present, amount of rectification,
and distillation a t reduced pressure. Using the pycnometer he obtained slightly lower
results with a high acetic acid percentage and slightly higher results when amyl
alcohol, ethyl acetate, or acetaldehyde were present.
Bordas and Touplain (1930) have defined the fundamental terms used in alco-
holometry and recommended the use of the apparent density in commerce rather than
the absolute density. Bordas and Roelens (1930) also prepared some useful tables on
the temperature corrections a t below-zero temperatures. The influence of acetic acid
on the determination of the specific gravity of the alcoholic distillate has been calcu-
1OOOd - 0.42a where is
lated by Errichelli (1950). The true specific gravity equals looo - 0.3981a,
the density as measured and a is grams of acetic acid per liter. This assumes that 42%
of the acetic acid is found in the distillate, which is only an average value. He calculates
that the volatile acidity must exceed 0.146% to have an influence on the density of
the alcoholic distillate in wines of 16% to 18% alcohol, over 0.150% in wines of 13%
to 15% alcohol, and over 0.155% in wines of 10% to 12% alcohol. To simplify the
calculation Errichelli gives a table of 0.42a and 1000-0.3981a for values of a from 0.130
to 0.50. Palieri (1950) has criticized these values as being too high and calculates
+ HAc
approximately that IOOOz HAc - -5 = 10000, where z is the corrected
1.4
density of the alcoholic distillate, HAc the weight of the acetic acid in grams per liter
in the distillate, and D the density of the distillate in the presence of the volatile
acidity.
Methods-Refractometric. A rapid method for determination of alcohol in wines
by refractive index was developed by Newton and Munro (1933). Use of their formula
for the immersion refractometer determination of alcohol was found inapplicable to
Argentinian wines by Pr6lat and Mendivelzda (1934),probably because the formula
assumes that the extract is only sucrose. Use of the immersion refractometer was also
recommended by Sampietro and Invernizzi (1940).Tables and procedures for the cor-
rections due to acidity were given. The practicability and accuracy of the refracto-
metric method for determining alcohol and extract of wines was demonstrated by
Jilke (1950,1951),who obtained very close agreement between the refractometer and
hydrometer values for twenty-three Rhine wines. Use of a direct-reading immersion
refractometer was proposed by Fischl (1942).Formulas for various ranges of alcohol
and sugar were devised and an error of 0.5% indicated.
The theory of the use of the refractometer as applied to alcohol determination in
wine was reviewed by Bohringer (1951a).He obtained good correlation between the
refractive index of the wine and that calculated from the refractive indices of the
alcohol and the extract. He then showed how to calculate the alcohol content of dry
table wines when the refractive index and extract content or the refractive index and
specific gravity of the wine are known. Formulas and tables are given to facilitate the
calculations. Either dipping or sugar refractometers can be used for the determination
of alcohol and extract in wines, according to Vetscher (1947).Equations for both are
given.
Methods-Chemical. Fabre and Br6mond (1935)compared pycnometric, hydro-
metric, ebulliometric, and chemical methods for determination of alcohol and for the
most accurate results preferred the chemical over the pycnometric. The former is
hased on oxidation of the alcohol with dichromate. Prior to chemical determination of
alcohol in sweet wines and vermouth, Paronetto (1938)used a double distillation, the
second one with alkali. A special adaptation of the chemical method for the determina-
tion of alcohol was proposed by Schulek and R6zsa (1939),in which the distilled
alcohol was purified by adsorption on charcoal.
Fessler (1941)gave a practical procedure for the dichromate method. Later modi-
fications of the Fessler procedure were outlined by Zimmermann (1949).Dubrowskaja
(1946)diluted the wine 1 to 60 and used only 8 ml. of the diluted wine plus-5 or 6 ml. of
water. This is made slightly alkaline, and 8 ml. are then distilled into 10 ml. of stand-
ardized potassium dichromate (plus 5 ml. of concentrated sulfuric acid). Then 10 to
15 ml. of 20% potassium iodide are added; after 5 minutes 200 to 250 ml. of water are
added; and the iodine released is titrated with thiosulfate to a starch end point. A
modification of the chemical oxidation procedure to eliminate calculations was de-
veloped by Barini-Banchi (1948).
Ferrari (1949)studied the Hoppler method for the chemical determination of
alcohol. This is based on oxidation with dichromate in the presence of sulfuric acid,
adding iodide, and titrating the iodine with thiosulfate. He studied temperature, time
of reaction, order of adding the solutions, volume, and concentration of alcohol and
the reagent. Results comparable to those of the pycnometer method were obtained.
Cordebard (1939) and Jaulmes (1951)reported better results using a nitric acid-
dichromate mixture.
A microchemical procedure was devised by Prange (1941).Only 1 ml. is required
and the dichromate-oxidation procedure was employed. If the wine was made alkaline
prior to distilling, high results were obtained. A similar micro procedure was used b y
Tomaghelli (1942).An error of only 0.1% was claimed. A colorimetric modification
of the dichromate oxidation method was developed b y Williams and Reese (1950).
They used dilute solutions and measured the violet color of the residual dichromate
ion with s-diphenylcarbazide at 540 m l . Chromic ion does not interfere, and Beer’s
law is followed. Other alcohols interfere but no more so than in other similar pro-
364 MAYNARD A. A M E R I N E
cedures. The economy of time and material and the precision of the results are the
main advantages of the chemical method.
Other Methods. Remy (1932)calculated the alcohol content from the formula
73 - W / n ,where W is the absolute surface tension of the wine and n the ratio of the
absolute surface tensions of water and alcohol less 0.2.The error was said to be f0.13
g. per 100 ml.
Another physical method is that of Etienne (1950,1952)and Etienne and Breyer
(1951). This is based on the principle of extracting the alcohol with an immiscible
solvent and reading the per cent alcohol from the position of the meniscus. They used
70 ml. of pentasol (synthetic amyl alcohol), 28 ml. of toluene, and 1.8 ml. of 10%
hydrochloric acid in a specially calibrated tube. The determination requires about 5
minutes, and an accuracy of f0.5% from 5% to 22% alcohol is claimed. Wines which
tend to emulsify should be pretreated with charcoal. Etienne (1952)has reported on a
collaborative study of about 100 determinations with his tube. In 43% of the cases
the values fell within rt0.15% of those determined by the pycnometer, 60% of the
values were within f0.3, and 94% fell within f0.5. To reduce error he found that the
temperature should be controlled a t 25.5" C. (78"F.);in any case, the test should be
conducted between 21.1" and 29.4" C. (70"and 85" F.)
Ettinger (1947)used a capillary tube, sealed a t one end, containing a drop of wine
to determine the alcohol content of wine. The tubes were slowly heated in a water-
sulfuric acid bath and the temperature of the bath noted when the wine rose in the
capillary. The per cent alcohol is determined from special tables. Reid and Truelove
(1952)have developed a colorimetric procedure using ceric ammonium nitrate. This
appears useful for small amounts of alcohol and may find winery applications for
analyses of still slops, pomace washes, etc.
Alcohol Yield. Ever since Pasteur's studies the alcohol yield in fer-
mentation has interested enologists as well as law enforcement agencies.
Use of sugar by yeasts, losses of alcohol by entrainment, and difficulties
in density determination make the calculation difficult. Furthermore, a
little alcohol may result from hydrolysis of glucosides. Savary (1940)
suggested that the Pasteur balance needed to be revised in certain cases
because he obtained higher than predicted yields. Perard (1939, 1940)
reaffirmed his belief in the Pasteur balance. Rustia (1949) determined the
alcohol yield during fermentation. Using a factor of 0.6 to convert un-
fermented sugar to alcohol (by volume), he then calculated the theo-
retical yield at various stages of the fermentation. He found an increase
in the ratio alcoho1:sugar up t o 10% sugar fermented and a decrease
thereafter. The over-all efficiency varied from 0.52 t o 0.58 compared t o
Pasteur's 0.611 and the theoretical of 0.64. Calculation of the original
sugar content of the must based on the alcohol content of the finished
wines was studied by Benz (1931). He used a factor of 0.45 for the sugar-
to-alcohol conversion (by weight). Niehaus (1937), however, obtained
yields of 0.476 and 0.477. He reported that only very small amounts of
alcohol were lost by evaporation during fermentation even from open
fermenters. The fermentations were conducted on musts with and with-
out the skins.
The alcohol yield from a given amount of sugar is of great interest to

winery operators. Although it should be obvious that there can be no
constant factor of sugar-to-alcohol (because of variation in environ-
mental conditions and microflora), many enologists continue to search
for one. Bassler and Trauth (1934), for example, found the alcohol yield
(by volume) to vary from 57.0% to 61.8% (average 58.7%) of the actual
sugar content. They found that at the higher sugar contents there was
an apparently greater yield of alcohol per degree of sugar fermented.
Muth (1934) obtained similar results. The disagreement on the exact
relation between the sugar content of the must and the resulting alcohol
content of the wine was reviewed by Trauth and Bassler (1936). They
did not find the nonsugar content to vary with maturity, and the alcohol
yield was remarkably constant. However, the yield was larger than
predicted from the sugar constant of the must (which earlier workers
reported as 45% based on a sugar-to-alcohol conversion) and amounted
to 46.4% to 47.1% (by weight). Various formulas for predicting the yield
from density only were investigated. The yield of alcohol (in per cent by
volume of the sugar content) for different types of musts, based on the
actual sugar content of the musts, according t o Miconi (1952a) is given
by the factor: white, 0.60; pink (1 to 2 days on skins), 0.58; red (de-
stemmed, 4 to 6 days on the skins), 0.56; red (not destemmed, 4 to 6 days
on skins), 0.54. The relation of alcohol yield to the specific gravity of the
must as measured by several formulas was studied by Balavoine (1939).
No very definite conclusions were reached as to the preferred procedure.
The problem is that the Oechsle and other hydrometers measure not
sugar but total soluble solids.
In contrast to other investigators Stradelli (1951) calculates that the
amount of alcohol lost by entrainment by carbon dioxide is very small.
Direct evaporation from a hot cap and losses during transfer are more
important. Bolcato et al. (1941) reported losses by entrainment of 0.7%
to 0.8% of the alcohol produced, the loss varying with the height of the
must in the fermenter. Flanzy and Boudet (1949) found a loss of about
0.1% alcohol for a fermentation temperature of about 30" C. (86" F.) and
a negligible loss a t 5" C. (41" F.). The type of fermentation is the other
variable to be considered.
Vogt (1934) showed that musts of low acid and medium-to-high
sugar contents produce more alcohol than would be anticipated from the
specific gravity, and conversely that high-acid low-sugar musts yield less.
The relationship is not exact, since the alcohol is related t o the sugar
content and not to the specific gravity of the must. Fischler (1937) also
indicated that unexpectedly high alcohol yields are obtained in very warm
years, such as 1921 and 1929, probably owing to some of the high-sugar
shriveled berries not being crushed sufficiently to provide their propor-

tional share of the sample. The low-sugar and very high nonsugar solids
of the 1936 Moselle musts were found by Kielhofer and Gunther (1937)
to yield from 55.6% to 62.63% alcohol (by volume), average 59.75%.
In very cold (low-sugar) seasons Seiler (1938) has shown that the alcohol
yield, based on total soluble solids, is notably low-owing, no doubt, to
the greater percentage of acids in the total soluble solids content. I n very
warm (high-sugar) years Fischler (1938) has again demonstrated the
opposite result. Rather large deviations in the alcohol yield from year
to year were noted by Seiler (1936a) in Moselle wines, but no special
corrections for sugaring were found necessary in 26 years experience.
Another factor is the strain of yeast employed. On the same must
Beckwith (1935) obtained 12.2% to 12.8% alcohol, using eight yeasts.
The lower values, however, were wines with residual sugar. Ventre (1936)
reported stable strains of Saccharomyces ellipsoideus which had different
alcohol-producing properties.
2. Methyl Alcohol
Methods. The difficulty of determining methyl alcohol in the presence of ethyl
alcohol was stressed by Ionescu and Popesco (1930). They investigated various pro-
cedures and recommended reactions with salts of mercury either on the pure alcohols
or on products of their oxidation. A review of the methods of determining methyl
alcohol in wines was given by SBmichon and Flanzy (1931b). Their oxidation pro-
cedure to formaldehyde and its determination in the distillate yielded almost quan-
titative results. Flanzy (1934a, b ; 1935a, b) reviewed the previous methods and made
a detailed study of the determination. His micro-method is lengthy and is essentially
a research procedure. Von Fellenberg (1937) employed the Denigbs reaction. A
detailed colorimetric procedure for-amounts of methyl alcohol as small as 0.01 % to
0.02 % was described. Ant-Wuorinen and Kotonen (1935-1937) made a careful study
of the factors influencing the accuracy of the colorimetric determination. Cerutti and
Vedani (1951) described a modification of Ant-Wuorinen’s procedure. A simple
qualitative colorimetric procedure for detecting amounts of 0.5% or more of methyl
alcohol in wines was presented by Brune (1948). Bertrand and Silberstein (1948)
studied the DenigAs reaction and suggested several improvements including use of a
photoelectric colorimeter. Mariani (1950) also studied the colorimetric procedure.
Exact details for the preparation of the reagents and timing of the procedure are given.
Although the reports of Guymon (1951) and St. Mokranjac and Radmic (1951)
were primarily directed toward spirits, their results are applicable to the determination
of methyl alcohol in wines. They investigated the factors influencing the development
of color with Schiff’s reagent using the Denigbs procedure. As little as 1 mg. of methyl
alcohol in 100 ml. of 0.95% ethyl alcohol could be determined. Hall et al. (1962)
used the DenigAs reaction but used chromotropic acid for the development of
the color.
A micro-method was proposed by Leon061 (1946). The methyl alcohol is first con-
centrated by fractional distillation and then carefully oxidized with a dichromate-
sulfuric acid mixture, and the carbon dioxide produced is purified and measured.
Results accurate to 2% with 2 mg. of methyl alcohol were reported. By careful control
of oxidation conditions, Zeglin (1952) reported a procedure sensitive to 0.02%. Akiya
and Sasao (1951) neutralized and distilled the wine, oxidized the diluted distillate with
permanganate, and determined the color with chromotropic acid.
Hunkk’s (1938) procedure for oxidation of methyl alcohol to formaldehyde by
bromine in the presence of sugar is capable of detecting only 5 % of the alcohol. A
colorimetric procedure for methyl alcohol in vinegar was developed by Piccoli (1933).
Jauker (1937) tested various procedures and used a semiquantitative method said to
be sensitive to 0.0001%.
The volume of sodium hydroxide required to produce cloudiness in 80% ethyl
alcohol is proportional to the methyl alcohol content and can be used for its deter-
mination, according to Charles (1938). Since only two alcohols may be present at
once, the procedure does not appear to be of interest in alcoholic beverages. A micro-
method based on formation of methyl p-bromobensoate mas described by Goldbach
and Opperschaum (1950).
Source. The general conclusion is that methyl alcohol is not produced

in alcoholic fermentation. Cerutti (1951), for example, found no methyl
alcohol after fermenting an artificial must. He also showed that glycine
is not a source of methyl alcohol, which is a theoretical possibility accord-
ing to Antoniani (1951b). Bertrand and Silberstein (1949-1952, 1950-
1952) have shown that methyl alcohol does not result from the alcoholic
fermentation of sugars but that it is probably derived from the hydrolysis
of the naturally occurring pectins of the grape skins. Peynaud (1952),
using Bertrand and Silberstein’s (1950-1952) data on white and red
wines, calculated that 0.023%, to 0.066% pectin would have to be
hydrolyzed in white musts and 0.082% to 0.112% in reds to account for
the methyl alcohol. Kilbuck et al. (1949) and Hall et aE. (1952) reported
that use of pectolytic enzymes increased the methyl alcohol content.
The latter found the increase greatest when added to the pomace. This
does not agree with Flanzy’s (1934b) concept that methyl alcohol is one
of the minor products of alcoholic fermentation. Jauker (1937) found
little or no difference in the amounts of methyl alcohol formed in various
substrates using pure cultures or wild yeasts. Flanzy (1951) fermented
all his wines on the skins, and he questions whether wines fermented out
of contact with the skins would contain much methyl alcohol.
Amounts. Bertrand and Silberstein (1949-1952) considered Flanzy’s
(1935a, b) values only approximately correct because of the procedure
used. Their range and averages, however, were approximately the same
as Flanzy’s. Jauker (1937) found only 0.5 t o 10 mg. per liter in four wines
and none in three others. Data on the methyl alcohol content of other
alcoholic beverages were also given. Col$escu et al. (1941) reported as
much as 0.3 g. of methyl alcohol per 100 ml. in wines of hybrid vines.
Data on the methyl alcohol content of various wines are given in

Table I.
TABLEI
Methyl Alcohol Content of Various Wines
(Grams per liter)
No. of Mini- Maxi- Aver-
Region Type of wine Source of data samples mum mum age
France Table Bertrand and Silberstein 21 0.038 0.188 0.103
(1949-1952)
France Vitis vinijera Flanzy (1951) 121 0.050 0.282 0.089
France Hybrids Flanzy (1951) 15 0.066 0.200 0.136
France White table Bertrand and Silberstein 9 0.041 0.114 0.071
(1949-1952)
France Red table Bertrand and Silberstein 7 0.062 0.200 0.146
(1949-1952)
Greece Vitis vinijera Flanzy (1951) 4 0.177 0.185 0.181
Italy White table Cerutti (1951) 17 0.000 0.182 0.040
Italy Red table Cerutti (1951) 43 0.000 0.635 0.103
U. S. White Hall et al. (1952) 10 0.004 0.018 0.016
U. S. Red Hall et al. (1952) 7 0.014 0.026 0.021
3. Higher Alcohols
The alcohols above ethyl in the series are generally spoken of as
higher alcohols. An extensive literature on their presence in brandy has
developed because of their importance to the organoleptic character of
brandy. Much less information is available for wines. The chief higher
alcohols found are isoamyl (3-methyl-l-butanol), active amyl (( -)-2-
methyl-1-butanol), n-propyl (1-propanol), isobutyl (2-methyl-1-propa-
nol), n-butyl (1-butanol), and ( - ) sec-butyl (2-butanol). Others doubtless
occur and will be identified as better methods for their separation are
developed. Buscar6ns (1941) fractionated (under vacuum) a fusel oil
from wine pomace and identified amyl, propyl, isobutyl, butyl, and
isopropyl (2-propanol)alcohols as esters and higher alcohols up to decyl.
No higher secondary alcohols were found. The residue consisted of esters,
fatty acids, furfural, cylic bases, and hydrocarbons. Only acids with an
even number of carbon atoms were demonstrated. The unsaturated acids
oleic and linoleic were present in small amounts, presumably from the
seeds. Ethyl esters were more important in amount than amyl esters.
There was 3% furfural, 5.5% fatty acids (free and esterified), 30.9%
ltlcohols (free and esterified), and 1.6% hydrocarbons (terpene). Dupont
and Dulou (1935) demonstrated sec-butyl alcohol in a technical propyl
alcohol that had been produced from fusel oil.
The presence of isopropyl alcohol in wines has been the subject of
some controversy. Bodendorf (1930), for example, demonstrated iso-
propyl alcohol in brandy, but his procedure was rather insensitive.
Flaney and Banos (1938) also isolated isopropyl alcohol from the fusel
oil of a wine. They calculated that wines contain a t least 6.6 mg. per liter.
Metra et al. (1938), however, were unable to detect this alcohol in samples
from wines or from the heads using a colorimetric procedure reputedly
sensitive to 0.01 % of this alcohol. They concluded that isopropyl alcohol
may be present only in traces, if at all. I n the fraction of the heads boiling
at 27.8" to 30" C. (82" to 86' F.) only sec-butyl alcohol was found in a
wine distillate by Durodie and Roelens (1942) ; and isopropyl alcohol was
not detected.
More recently Webb et al. (1952) reported no isopropyl alcohol in a
wine fusel oil sample. They did find 4.1% n-propyl, 1.9% n-butyl, 4.9%
(-)-sec-butyl, 18.3 % isobutyl, 9.6 % (-)-2-methyl-l-butanol, 54%
isoamyl, trace of n-amyl, 1.5% n-hexyl, 5.6% esters, and traces of acetic
and butyric acids and acetal. The esters included 0.19% ethyl caproate,
0.60% ethyl caprylate, 0.52% isoamyl caprylate, 1.32% ethyl caprate,
0.38 % isobutyl caprate, 0.58 % ethyl laurate, 0.25 % ethyl palmitate, a
trace of butyrate ester, 0.06 % myristate ester. Probably present were
methyl salicylate, isoamyl caprate, active amyl caproate, isoamyl
caproate, active amyl caprylate, isobutyl caprylate, active amyl caprate,
active amyl laurate, and isoamyl laurate.
Methods. The higher alcohols are usually determined by a colorimetric procedure
based on a condensation reaction with an aromatic hydroxy aldehyde such as p-di-
methylaminobenzaldehyde, vanillin, or salicylaldehyde-the Komarowsky reaction.
Von Fellenberg (1929)studied the details of time and color intensity b y this pro-
cedure, using three different mixtures of higher alcohols. Siirgi (1932) revised the
method to a micro-procedure and gave exact details for its execution. Trost (1935)
also studied the reaction and gave information on it. I n this country the Penniman
el al. (1937)modification is used. A slightly different variation was used by Guymon
and Heitz (1952)and Osborn and Mott (1952).A spectrophotometric modification of
the Komarowsky-Fellenberg procedure was proposed by Federico and Cioffi (1947).
Salicyaldehyde is used for development of the color. Guymon and Heitz (1952)sug-
gested that differences in procedure may have accounted for the somewhat high values
reported b y Cioffi (1948).A systematic study of the influence of concentration, per
cent alcohol and sulfuric acid, temperature, time, etc., of the Komarowsky reaction
was made by Gierer and Hoffman-Ostenhof (1951)and a micro-procedure developed;
with this they obtained results to &2% with concentrations of 0.2 to 2 mg. per 100 ml.
using only 0.5to 1 ml. A similar study was made b y Rosenthaler and Vegezzi (1953).
Moreno (1934)and Clavera and Moreno (1936)showed that methyl alcohol inter-
feres in the determination of higher alcohols by the procedure of Rose (which depends
on the increase in volume of chloroform with higher alcohols under specified condi-
tions). They obtained a curve for correcting for this error. Bonaterra (1949)developed
a procedure based on the color developed b y higher alcohols in alcoholic furfural-
sulfuric acid mixtures.
A procedure to determine isopropyl alcohol in alcoholic beverages is given by
Alessandrini (1933).It is based on Noetzel's procedure of oxidation to acetone and the
reaction of acetone with hydroxylamine hydrochloride.
Formation. The mechanism of the formation of higher alcohols has

been studied by Zalesskaya (1938), Cioffi (1949), and Castor and Guymon
(1952). Zalesskaya (1938) also reported that higher alcohol formation
coincided with the beginning of the main fermentation and that the
amount produced increased as the amount of fermented sugar or the less
the initial number of yeast cells present. Cioffi studied the influence of
various fermentation inhibitors on amino acid utilization during fer-
mentation. He found less higher alcohols were produced in the presence
of sodium arsenite (either on an absolute basis or relative to ethyl alcohol
production). Sodium sulfite reduced the production relative to ethyl
alcohol production in comparison with hydrogen cyanide, but the
amounts produced were about the same, and the quantity of the amino
acids deaminated was greater; Castor and Guymon showed that the
formation of higher alcohols paralleled ethyl alcohol formation. Contrary
to expectation, amino acid disappearance preceded higher alcohol appear-
ance. They explained this by suggesting that the enzyme systems of the
latter stages of the conversion of sugar to ethyl alcohol and of amino
acids to higher alcohols are identical. Their finding that more higher
alcohols were formed than could be accounted for by amino acid disap-
pearance was accredited to formation a t the expense of the yeast% own
proteins. Luers (1948) found higher alcohol formation in beer to occur
during the later stages of fermentation and to be independent of amount
of yeast. An increase in fermentation temperature resulted in an increase
in amount of higher alcohols produced. I n potato and molasses fermenta-
tions, Kilp and Deplanque (1934) found higher alcohol formation to
occur generally before ethyl alcohol production.
Genevois (1952) suggested two mechanisms for conversion of amino
acids to higher alcohols: an “endogenous” metabolism of yeast protein,
in which for each kilogram of sugar fermented 2 to 4 g. of amino acids
were utilized, and an “exogenous” metabolism on the amino acids of the
media, in which 10 to 12 g. of amino acid were utilized per kilogram of
sugar fermented.
Yamada (1932) found higher alcohol production to depend on the
presence of amino acids. However, in the presence of glutamic acid,
asparagine, or ammonium phosphate or sulfate only traces of higher
alcohols were produced. Addition of nitrogenous materials during fer-
mentation reduces the deamination of amino acids ‘and also the pro-
duction of higher alcohols, according to Antoniani (1951s). Antoniani
and Cioffi (1949) found the higher alcohol content of a Saccharomyces
ellipsoideus fermentation of grape juice to be as follows when glycine
was added:
COMPOSITION O F WINES 37 1
Amino N, Ammonia N, Higher
mg./lOO ml. mg./100 ml. alcohols,
Alcohol Before After Before After mg./100 ml.
Control-200 ml. 7.8 8.6 2.8 0.14 0.00 0.052
+170 mg. glycine 7.8 22.2 3.4 0.14 0.00 0.217
+250 mg. glycine 7.8 30.0 9.1 0.14 0.00 0.242
+420 mg. glucine 7.7 42.1 23.5 0.14 0.00 0.251
+670 mg. glycine 7.7 62.4 46.5 0.14 0.00 0.315
Gobis (1950) also showed that addition of 50 mg. of ammonium sulfate

decreased the higher alcohol from 0.060 ml. for 100 ml. t o 0.025 ml. Castor
(1950) reported 30% less higher alcohols when ammonium phosphate was
added, and Vogt (1952) reported a 50% reduction when the chloride and
asparagine were added.
Amounts. Casale (1934) reported 0.85 and 0.73 g. per liter of higher
alcohols in two wines fermented a t a starting temperature of 0" to 3" C.
(32" to 37.4" F.). Mestre and Mestre (1939), however, reported 0.86 and
0.95 g. per liter of higher alcohols in musts fermented a t room tempera-
ture, and 1.00 and 1.00 when fermented a t 4" to 12" C. (39.2"t o 53.6"F.).
In seventy California dessert wines Filipello (1951) reported 0.125 to
0.685 g. per liter.
Other data on higher alcohol content of wines have been summarized
by Guymon and Heitz (1952), who add extensive analysis of California
wines (see Table 11). They reported lower results than Cioffi (1948)
TABLEI1
Higher Alcohol Content of Various Types of Wines
(Grams per liter)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
Argentina Unknown Pettigiani (1943) 24 0.15 0.30 -
California White table Guymon and Heita (1952) 120 0.162 0.366 0.250
California Red table Guymon and Heitr (1952) 130 0.140 0.147 0.287
California Dessert Guymon and Heita (1952) 117 0.156 0.90 0.374
Italy White table Cioffi (1948) 12 0.44 1.37 0.87
Italy Red table Cioffi (1948) 22 0.14 1.72 0.77
(Italian wines) and less in white table wines than red-also contrary t o
his findings. Their results are similar to those of Pettigiani (1943) (Argen-
tinian wines). I n California dessert wines, very variable results were
obtained by Guymon and Heitz, probably because of the varying amounts
of higher alcohols in the fortifying brandies employed.
Filipello (1951) found a high degree of negative correlation,
T = -0.480
between the amount of the higher alcohol present in the wine and the
organoleptic score of the wine. Pool and Heitz (1950) also reported that
high-fuse1 brandies produced poor dessert wines. Genevois (1952) and
Guymon and Heitz (1952) suggested that higher alcohols may play a
role in the odor of wines. The latter reported a red table wine with an
obvious higher alcohol odor which was confirmed by analysis.
4. Glycerol
While glycerol usually constitutes from 0.5% to 1.5% of wines, it is
seldom determined, particularly in sweet wines, because the procedures
for its determination are long and tedious and often inaccurate. However,
next to alcohol, it is the major product of fermentation present in wines.
It has a measurable influence on taste, and it is sometimes used in the
enological ratios to detect sophistication. Whereas it figures in most of the
schemes depicting the several stages of fermentation, its exact genesis has
not been evaluated, according to Antoniani (1951a).
Methods. Peynaud (1947b) has reviewed the primary problems of glycerol deter-
mination in wine: separation from sugars, 2,3-butylene glycol, and other oxidizable
organic materials, and prevention of loss by heating and clarification or during extrac-
tion. Pritzker and Jungkunz (1930a, b) apparently were the f i s t to point out that
2,3-butylene glycol might interfere in the glycerol determination but that, as the
determination is normally conducted, no correction is necessary. The standard United
States procedure (Association of Official Agricultural Chemists, 1950) removes the
sugars by treatment with a freshly prepared, carbonate-free calcium hydroxide solu-
tion. Although this is fairly satkfactory for table wines, it is seldom so for dessert
wines. Pritzker and Jungkuna (1930a, b) demonstrated that when the official German
evaporation procedure is used (for table wines) practically pure glycerol, uncon-
taminated with 2,3-butylene glycol, is found. Espil's (1936) lead subacetate-lime-
copper procedure involves a voluminous precipitate from which it is difficult to extract
glycerol. The FerrbMichel (1938) method removed 2,3-butylene glycol as an inter-
fering substance but is long and tedious. Fleury and Fatome's (1935) and Fatome's
(1935) alkali or alkali-plus-lead-acetate clarification may not remove all interfering
substance. Peynaud (194713) objected to this procedure, used by Amerine and Dietrich
(1943), as leading to the production of methylglyoxal from sugars. However, this
would apply only to sweet wines and only when the sugar-alkali mixture is heated
excessively. Vasconcellos (1946) used lead subacetate in an ammoniacal solution for
removal of acids and sugars. Satisfactory results were also obtained by Thaler and
Roos (1950) using a basic clarification. They attributed the variable results obtained
in most methods to the volatility of glycerol in aqueous solutions and recommended
that glycerol solutions should not be boiled. They modified the Fleury and Fatome
(1935) procedure so as to avoid mannite interference by using a methanol/acetone
(1/8)mixture rather than ether/alcohol (1/3). Barium hydroxide was used for removal
of acids and sugars. The 2,a-butylene glycol was not removed. Whereas most pro-
cedures proposed for the determination of glycerol depend on purification by alkali
treatment, Sfimichon and Flanzy (1930a) proposed distilling the glycerol. They and
von Fellenberg (1931) entrained the glycerol with steam a t 115" C. (237" F.). Chroma-
tography separation is also practical. Purified ether or isopropyl acetate were recom-
mended by Jacquin and Tavernier (1952) for extraction.
A certain polemic has developed over the question of the possible loss of glycerol
during evaporation. Von Fellenberg (1931), for example, found none, but it may be
significant that the standard United States procedure cautions against local over-
heating during evaporation. Further data were given by Jaulmes (1951), who reported
a very low volatility.
Once the glycerol is separated, various methods have been used for determining it.
The Association of Official Agricultural Chemists (1950) procedure requires weighing
the extracted glycerol before and after ashing and subtracting the ash; or for oxidation
with dichromate. When dichromate oxidation is employed, an aliquot should be tested
for sugars, except in the case of the S6michon and Flanay (1930a) and von Fellen-
berg (1931) procedures, where the glycerol is separated by steam. Oxidation with
dichromate was also employed by Ferr6 and Michel (1938), Pavolv-Grishin (1940),
and Korotkevich and Arbuzova (1949). Determination of glycerol in wines b y
periodic acid oxidation is now common. The basic reaction is that of Malaprade:
+
CH20HCHOHCHZOH 2HIO4 -+ 2HCOH + HCOOH + 2HIOs. Various meth-
ods of determining the extent of the reaction have been devised-Fatome (1935),
Fleury and Fatome (1935), Amerine and Dietrich (1943), Vasconcellos (1946),
Peynaud (1948b), and Thaler and Roos (1950). Peynaud (1948b) made a notable
improvement in removing 2,3-butylene glycol, which causes a 4% to 7% error.
Lambert and Neish (1950) used the periodic oxidation, the excess iodate or periodate
being reduced with sodium arsenite, and the formaldehyde colorirnetrically deter-
mined with chromotropic acid.
A number of colorimetric procedures have been devised. Bertram and Rutgers
(1938) used the fixing of cupric ion b y glycerol as the basis of their procedure. The
color produced by methylglyoxal with morphine and its derivatives was used by
Legkov (1931). The glycerol was oxidized t o dioxyacetone and then converted to
methylglyoxal. Greater sensitivity and accuracy compared to the usual procedure was
claimed. A similar micro-procedure was proposed by Prado (1934)-the bromine
oxidation was used to prepare dioxyacetone and colored derivatives were then pre-
pared. As an alternative he also converted the dioxyacetone to formaldehyde with
sulfuric acid. Compared t o the usual weighing procedures high values were obtained.
A micro-colorimetric procedure was developed by Ghimicescu (1935f). The glycerol
was oxidized by bromine and the product determined colorimetrically by pyro-
catechine. He obtained good recovery of glycerol added to a white wine. A colorimetric
procedure in sugar-free wines was developed by Diemair et al. (1940), and a semi-
quantitative procedure was developed by Cunha (1939)4ecolorizing the wine,
treating with bromine, resorcinol, and sulfuric acid, and matching the color produced
with that of similarly treated standard glycerol solutions.
In Dessert Wines. Pritaker (1940a), considering the problem of accurate glycerol
determination in sweet wines, recommended determining the amount of reducing
sugar in the glycerol extract and subtraction of the amount found. Using calcium oxide
and magnesium carbonate for clarification, he found from 2 to 11 mg. of sugar in the
extract with 47 to 141 mg. of glycerol. Von Fellenberg’s (1943) micro-method avoids
the coprecipitation of the alkaline-treated sugars and glycerol by use of large amounts
of acetone. He distills to separate 2,3-butylene glycol. The possibility of methyl-
glyoxal production in the hot methanol-barium hydroxide-sugar mixture should be
considered. Recovery from sweet wines was 95% to 106%. Hog1 (1952) proposed
separating the glycevol from the sugars in dessert wines by paper chromatography.
The solvent was a miiture of ethyl alcohol, butyl alcohol, and water in the proportions
1.1/4.0/1.9. Silver nitrate was used for developing the chromatogram. The wine was
first run through an anion exchanger to remove acids. The method can be made quan-
titative to about fO.O1 g. per 100 ml. by the use of standard glycerol solutions. He
recommended the method as a means of distinguishing fortified musts from dessert
wines which had been fortified after some fermentation. Hydroxymethylfurfural,
various types of sugars, and sorbite can be similarly determined. For a colorimetric
method in connection with the determination of lactic acid, see p. 404.
Source. Hickinbotham and Ryan (1948) believed glycerol imparts
smoothness to wines and in particular ameliorates the burning taste of
alcohol. They review previous work on the factors influencing the pro-
duction of glycerol in wines. They found less in laboratory-prepared
wines than in larger trials and questioned whether the low carbon dioxide
pressure in small fermenters may not have an effect. Analysis of variance
of their data showed more glycerol in fermentations a t 13" and 22" C. and
less a t 28" and 36" C. (82.4 and 96.8" F.). They believe that high tem-
perature of fermentation is one reason Australian wines have less glycerol
than European. Brockmann and Stier (1948) also found less a t higher
temperatures and suggested as a cause lower phosphatase activity, which
would reduce hydrolysis of glycerol-1-phosphate. However, Uchimoto
(1951) found the glycerol content increased as the fermentation tem-
perature was increased from 12.8' to 32.2" C. (55" to 90" F.). This might
be associated with the greater number of yeast cells present a t the higher
temperatures of fermentation. Amerine and Webb (1943) showed that
the per cent glycerol in California dessert wines was lower than what
would be expected, even taking dilution and the shorter period of fer-
mentation into account.
Hickinbotham and Ryan (1948) found a significant difference in the
ability of different yeasts to produce glycerol. Some musts and seasons
led to significantly higher yields than others. Using eight yeasts, Beck-
with (1935) obtained 0.54 to 0.71 g. of glycerol per 100 ml. in duplicate
fermentations. Anibal (1935) found glycerol formation to be faster a t the
start of the fermentation. The ratio of succinic acid to glycerol also
100 X g. glycerol/l
decreased during fermentation. He used the ratio
g. alcohol,l. * to
determine addition of alcohol or water. When this ratio falls'below 6,

sophistication may be suspected. In connection with their calculations
of the balance of fermentation products in Bordeaux red and white wines,
Genevois et al. (1949a) have shown that the high glycerol content of
certain sweet wines resulted from its production by Botrytis cinerea
growing on the grapes. I n some cases the glycerol produced by fermenta-
tion was less than that found in the grapes. Ferr6 and Michel(l938) using
their relatively foolproof procedure found 0.60 to 1.1 g. per 100 ml. of
glycerol in sound Burgundy wines, and 1.04% to 1.51% in wines made
from moldy grapes. The glycerol, as per cent of alcohol, varied from 6.5
to 9.6 in the sound grapes, and from 10.1 to 12.5 in moldy grapes. These
results substantiate earlier investigations. Schumakov (1930) studied the
influence of sulfur dioxide on the production of glycerol. Daily additions
of 21 to 63 mg. of sulfur dioxide per liter increased production by 0.42%
to 0.61 %. To increase glycerol production he recommended daily addition
of sulfur dioxide in amounts just short of that needed to stop fermenta-
tion. This would also increase the “fixed” sulfur dioxide content by fixing
acetaldehyde as it is produced by fermentation. Venezia and Gentilini
(1941) studied the possibility of using poor-quality grapes for the produc-
tion of glycerol by sulfite fermentation. They obtained yields of as much
as 7.8% using 10% yeast at 15” C. (59” F.) with very high sodium sulfite
and a pH of 9.1. As much as 35% of the sugar fermented was converted
to glycerol. Addition of a “2” factor (from boiled yeast) stimulated
glycerol production according to Venezia (1939-1940) , but the effect
decreased as the sulfite content increased. Changes in glycerol due to
bacterial action are discussed by Osterwalder (1952).
Amounts. A summary of the glycerol content of a variety of types of
wines is given in Table 111. Whereas there is much variation between
averages, there is also a wide range within the types-owing no doubt to
the varying amounts of glycerol in moldy grapes, to diperences resulting
from fermentation practices, to varying amounts of sugar, and of course
to dilution owing to fortification or watering.
Alcohol/Glycerol Ratio. This ratio has been thought to be of diagnostic
value, a low ratio indicating higher quality. The data of Table IV indicate
that this has limited validity. Von Fellenberg (1943) found normal
alcohol/glycerol ratios for wines prepared from raisins-9/20. In fer-
mented apple and pear juice Jacquin and Tavernier (1952) found less
glycerol formed than in grape fermentations-ratios varying from 10 to
18.7. I n order to calculate the ratio of fortified wines on the basis of the
glycerol and alcohol produced by fermentation Amerine and Webb (1943)
(100 - A ) F * 1375 - 13.754 - 55E
proposed the formula __ 1 where F =
100 - A - 0.55E
’
A is the per cent alcohol, E the extract of the fortified wine. G, the
glycerol corrected for dilution by fortification, is defined as equal to
G(1OC) - F )
1 where G is the glycerol in grams per 100 ml., F* is in grams
(1UO - A )
per 100 ml., and F in milliliters per 100 ml. Using this formula and
assuming an initial soluble solids content of 25” Balling, they found
slightly below normal ratios for California dessert wines. Post-Repeal
California wines seemed to have more uniform and normal alcohol/
glycerol ratios than pre-Prohibition wines.
TABLE I11
Glycerol Content of Table and Dessert Wines
(Gramsper 100 ml.)
Table wine
Algeria Table Bertin (1931) 25 0.67 1.57 1.07
(Mascara) (unsugared)
Argentina Table Prado (1934) 46 0.37 1.33 0.87
Argentina Table Anibal (1935) 39 0.61 1.05 0.77
Australia Table Hickinbotham and 12 0.47 0.86 0.63
Ryan (1948)
California Wh. table Amerine (1947) 79 0.63 1.68 0.96
California Red table Amerine (1947) 60 0.46 1.43 1.06
Czechoslovakia Table Kopal (1938) 650 0.51 1.39 0.83
France Table Ferr6 and Michel 25 0.57 1.51 1.00
(1938)
France Table Genevois et al. 31 0.70 1.04 0.86
(1948b)
France Table Peynaud (1950a) 6 0.72 0.85 0.80
France Table* Peynaud (1950b) 6 0.79 2.60 1.65
France Sparkling Hennig (1952) 14 0.58 0.95 0.81
Germany Table Heiduschka and 29 0.42 0.85 0.59
Pyriki (1930)
Germany Table Alfa (1932) 47 0.43 1.40 0.65
Germany Wh. table Remy (1932) 10 0.51 0.87 0.68
Germany Table Alfa (1933) 46 0.45 0.92 0.72
Germany Wh. table Heide and Zeissett 24 0.49 1.38 0.90
(1935)
Germany Table Mader (1936) 11 0.76 1.42 1.01
(sugared)
(unsugared)
Germany Table Hennig (1952) 11 0.56 1.08 0.73
Germany Sparkling Hennig (1952) 43 0.40 1.10 0.70
Hungary Wh. table Torley (1942) 10 0.47 1.10 0.74
Italy Table Sallusto (1936- 77 0.64 1.20 0.86
1937)
Italy Wh. table Dalmasso and 143 0.41 1.00 0.67
Dell’Olio (1937)
Italy Tablet Sallusto and 26 0.61 1.66 1.30
Sculco (1937-
1938)
1939b)
Italy (Sicily) Swt. table Sallusto and Di- 30 0.57 1.57 1.14
Natale (1938-
1939)
Italy (Falerno) Table Sallusto (1938- 16 0.75 1.53 1.03
1939a)
TABLEI11 (Continued)
Italy Table Lucchetti (1941) 12 0.44 0.86 0.66
Italy Pink table Cosmo (1950) 22 0.52 1.24 0.74
Portugal Table Correia and 114 0.30 1.00 0.70
Ribeiro (1942)
Portugal Table Vasconcellos 8 0.73 0.90 0.84
(1946)
Roumania Table Vumuleanu and 33 0.28 1.28 0.74
Ghimicescu
(1936)
Switzerland Table Berner (1952) 25 0.41 0.97 0.65
Tunisia Table Sallusto (1938- 25 0.92 1.28 1.10
1939a)
Dessert wines
California Dessert Amerine (1947) 124 0.34 1.671 0.75
France Dessert Peynaud (1950b) 8 0.54 0.71 0.58
Italy Dessert Pritzker (1940a) 5 0.36t 0.71 0.54
Portugal Red port Ramos and Reis 38 0.11 0.86 0.40
(1945)
Portugal White port Ramos and Reis 19 0.19 0.48 0.32
(1945)
Portugal Port Vasconcellos 17 0.33 0.81 0.54
(1946)
Spain Dessert Pritzker (1940) 6 0.24 0.71 0.49
Spain Fino Bobadilla and 15 0.379 1.023 0.663
Navarro (1952)
Spain Oloroso Bobadilla and 10 0.640 1.266 0.743
Navarro (1952)
* Sweet. High glycerol owing t o "botrytiaed" grapes.
t Average per cent alcohol 16.3: reducing sugar 1.1.
$ A mistell. which is aupposed to be a fortified must.
6 . 2,d-Butylene Glycol, Acetylmethylcarbinol, and Diacetyl

Another constant product of alcoholic fermentation is 2,3-butylene
glycol. Usually this is determined along with glycerol, though it is par-
tially lost in the usual evaporative procedures. It can, however, be
separated by fractional distillation. Its oxidation product, acetylmethyl-
carbinol, is of lesser importance. The oxidation product of the latter,
diacetyl, is present in even smaller amounts or is absent. Although
2,3-butylene glycol appears to have little organoleptic importance, the
other compounds may play a minor role. Diacetyl, for example, can be
detected organoleptically at only 2 to 4 mg. per liter.
Methods. Kniphorst and Kruisheer (1937) made an extensive study-of the deter-
mination of 2,3-butylene glycol, acetylmethylcarbinol, and diacetyl in wines. Their
procedure, the Legmoigne reaction, is based on oxidation of the two to diacetyl-
CHsCHOHCHOHCH3-+ CHaCOCHOHCHs+ CH8COCOCH8.-and its determi-
TABLEIV
Alcohol/Glycerol Ratios (by Weight) of Table Wines
Region Type of wines Source of data samples mum mum age
Algeria Table Bertin (1931) 25 8.0 16.4 11.7
(Mascara)
Argentina Table Prado (1934) 46 3.7 14.0 8.7
Argentina Table Anibal (1935) 39 10.7 16.7 12.5
Australia Table (wh) Hickinbotham and 6 11.0 15.6 14.2
Ryan (1948)
Australia Table (red) Hickinbotham and 6 13.0 21.7 15.6
Ryan (1948)
California Wh table Amerine and Webb 68 5.6 12.9 10.2
(1943)
California Red table Amerine and Webb 56 5.0 11.1 9.3
(1943)
Croatia Table Pereti6 (1950) 25 4.6 12.4 8.9
Czechoslovakia Table Kopal (1938) 650 7.7 16.4 11.1
France Table F e d and Michel 25 8.7 15.4 10.8
(1938)
France Table Peynaud (194813) 48 15.4 10.2 12.2
Germany Wh. table Heiduschka and 29 9.0 16.7 12.5
Pyriki (1930)
(sugared)
(unsugared)
1937)
Italy Table Sallusto and Sculco 26 7.8 17.5 9.7
(1937-1938)
Italy Pink table Cosmo (1950) 22 8.9 17.7 14.0
Italy (Falerno) Table Sallusto (1938- 16 7.7 15.1 10.7
1939b)
Italy (Sicily) Swt. Table Sallusto and Di 30 9.3 14.3 11.1
Natale (1938-
1939)
Portugal Table Vasconcellos 8 10.0 11.9 10.8
(1946)
Roumania Table _Sumuleanu and 33 9.3 27.8 13.3
Ghimicescu
(1936)
Tunisia Table Sallusto (1938- 25 8.1 12.7 10.4
1939a)
nation as nickel dimethylglyoxime. A similar method was used by Pritzker and
Jungkunz (1930a, b) and Garino-Canina (1933). The disadvantage is that the oxida-
tion is not complete, and a correction factor must be used. The advantage is the large
weight factor of the precipitate. Matignon et al. (1934) obtained a yield of only 75.6%.
This was a colorimetric procedure, as was that of Moureu and Dod6 (1934). Kniphorst
and Kruisheer (1937) increased this to 93%. The Lemoigne reaction was also used
successfully by Diemair and Kleber (1941) to determine 2,3-butylene glycol and
acetylmethylcarbinol. They used a special distillation flask t o prevent losses and
determined the nickel dimethylglyoxime colorimetrically.
Von Fellenberg (1932a) used chromic acid oxidation, but Kniphorst and Kruisheer
(1937) criticized this procedure as yielding high results when only small amounts were
present. Perdigon (1941) developed a semi-micro periodic acid procedure similar to
Brockmann and Werkman’s (1933) b u t applied specifically to wines. Peynaud (1947b)
criticizes Perdigon’s procedure as being inapplicable because more glycerol than
2,3-butylene glycol is present.
Peynaud (1947a) and Ribbreau-Gayon and Peynaud (1947a) separated the 2,3-
butylene glycol from most of the glycerol by steam distillation. The acetaldehyde was
removed by refluxing under a Vigreux column. The actual determination was madc
by periodic acid oxidation in which 2,3-butylene glycol yields acetaldehyde and
glycerol produces formaldehyde. Formaldehyde was fixed by adding asparagine and
the acetaldehyde then distilled.
Source. The production of acetylmethylcarbinol during fermentation

was studied by Antoniani (19514. Actively fermenting yeasts produced
only 2,3-butylene glycol. Moderately fermenting yeasts produced both
2,3-hutylene glycol and acetylmethylcarbinol, whereas weakly fermenting
yeasts produced only acetylmethylcarbinol. Antoniani and Gugnoni
(1941) reported that Pseudosaccharomyces apiculatus and P . magnus
produced only acetylmethylcarbinol and normal wine yeasts only 2,3-
butylene glycol. In mixed cultures they found the latter, as the normal
wine yeasts predominate. However, when both types were present and the
samples were stored in air acetylmethylcarbinol was produced. The
question of which yeasts produce 2,3-butylene glycol and which acetyl-
methylcarbinol is a complicated one. Peynaud and Lafon (1951b, 1952)
have shown t ha t a wide variety of yeasts produce both, but Antoniani
considers that some yeasts produce exclusively one and some exclusively
the other. Their differences are aired in a polemic (Antoniani, 1953).
The differences may be due to the fact tha t the analyses were made
immediately after the main phase of the fermentation in one case and
about 3 weeks later in the other. More data on the conditions in the
fermenter, oxidation-reduction potential, etc., the amounts present a t
various stages of the fermentation and data on the microflora are needed.
The formation of 2,3-butylene glycol and acetylmethylcarbinol during
fermentation was studied by Diemair and Kleber (1941). They found
variable amounts of 2,3-butylene glycol in wines, depending somewhat on
fermentation conditions. Using a Jerez strain of yeast, they obtained no
more 2,3-butylene glycol than with a regular strain of Saccharomyces.
This seems t o indicate that 2,3-butylene glycol is not produced by an
oxidative mechanism. I n the same fermentations, acetylmethylcarbinol
was produced in only one case, and in general they conclude th a t it is not
produced from 2,3-butylene glycol. In acetic fermentations they found a

simultaneous increase in acetylmethylcarbinol and a decrease in citric
acid but little change in 2,3-butylene glycol-another indication that the
production of the former is not from the latter. On the other hand,
Kobakhidze (1950) studied the formation of acetylmethylcarbinol during
acetification, which may or may not be the same as that of wines. He
concluded that 2,3-butylene glycol and not citric acid is the source of
acetylmethylcarbinol in the acetic acid fermentation.
Peynaud and Lafon (1951b) reported 2 to 20 mg. per liter of acetyl-
methylcarbinol in wines and considered it a product of fermentation.
Diacetyl was found in all red wines (2 to 3 mg. per liter), but it was not
present in some white wines. I n brandy distilled from wine, Peynaud and
Lafon (1951a, b) found 2,3-butylene glycol and acetylmethylcarbinol and
traces of diacetyl. They found these not to be related to the quality of
the brandy, except possibly diacetyl in some cases, but concluded that
they might prove useful in differentiating various brandies. For example,
the armagnacs contained about eight times as much 2,3-butylene glycol
as did the cognacs. In contrast to cognac the pomace brandies of Burgundy
did not contain diacetyl.
I n Italian wines Romano (1951) found varying amounts of 2,3-
butylene glycol (Table V) but no relation between the amounts of this
substance and the contents of alcohol, sugar, total or volatile acidity, or
organoleptic quality was noted. Garino-Canina (1933) reviewed the
previous work on the presence of 2,3-butylene glycol and acetylmethyl-
carbinol in wines and the methods for their determination. He showed
that grape juice fermented in the presence of acetaldehyde produced
acetylmethylcarbinol. He also reported less in fortified wines and in wines
of low alcohol content. Usually 0.2% to 0.66% of 2,3-butylene glycol was
found, and the amount reported was proportional to the per cent alcohol.
Kniphorst and Kruisheer (1937) also found no acetylmethylcarbinol or
diacetyl in twenty-one assorted European wines. The absence of the
former was probably due to analytical problems. They pointed out how
useful the data on 2,3-butylene glycol was in determining sophistication
in wines. Table wines of normal alcohol content and low 2,3-butylene
glycol content have been fortified. Dessert wines with no 2,3-butylene
glycol have been fortified before fermentation. Rib6reau-Gayon and
Peynaud (1947b) observed no relationship between the quantity of
2,3-butylene glycol and that of glycerol. The influence of 2,3-butylene
glycol on the organoleptic character of the wine was shown to be small,
as it has similar properties to the glycerol, which is present in 10 to 20
times greater amounts. Gobis and Farfaletti-Casali (1952) did not find
any direct relationship between the acetylmethylcarbinol content and the
age of the wine, although one might expect more in older wines with their
higher oxidation-reduction potential. Differences in the amount of oxygen
reaching the wines during aging might account for the lack of correlation.
According to Gvaladze and Rodopulo (1946) acetylmethylcarbinol is
formed in wines at the expense of 2,3-butylene glycol when Acetobacter
aceti begins to grow. Garino-Canina (1933) also showed the presence of
large amounts of acetylmethylcarbinol in wine vinegar. I n apple wines
TABLEV
2,BButylene Glycol in Various Types of Wines
(Grams per liter)
France Table Ribbreau-Gayon and 12 0.328 1.350 0.699
Peynaud (1947b)
France Red table Peynaud (1950a) 6 0.62 0.78 0.70
France Swt. table Peynaud (1950a) 6 0.44 1.11 0.86
Germany Wh. table* Diemair and Kleber 25 0.062 0.596 0.220
(1941)
Germany Wh. tablet Diemair and Kleber 35 0.094 0.973 0.365
(1941)
Italy Table Romano (1951) 14 0.080 0.980 0.320
Italy White Gobis and Farfaletti- 13 0.256 0.810 0.414
Casali (1952)
Italy Red table Gobis and Farfaletti- 51 0.259 0.986 0.489
Casali (1952)
Misc. Table Kniphorst and Kruisheer 9 0.379 0.983 0.575
(1937)
Misc. Dessert Kniphorst and Kruisheer 9 0.065 0.302 0.260
(1937)
Misc. Table Krauze (1933) 34 0.73 1.57 1.15
Spain Montilla Casares and Gonealee 51 0.269 1.397 0.696
(1953)
* Laboratory fermentations.
t Commercial fermentations.
Guittonneau et al. (1941a, b) found Aerobacter aerogenes and A . cloacae
to be primarily responsible for the high acetylmethylcarbinol and 2,3-
butylene glycol contents.
Amounts Reported. Genevois et al. (1948a) reported 5 to 10 millimoles
of 2,3-butylene glycol in commercial fermentations but only 3 to 6 milli-
moles in laboratory fermentations. The ratio of 2,3-butylene glycol to
glycerol normally varied from 4 to 9, though under different fermentation
conditions it was 7 to 12. The presence of 2,3-butylene glycol in amounts
up to 0.06% was demonstrated by Pritzker and Jungkunz (1930a, b).
Using the Pritzker-Jungkunz (1930a) technique Garino-Canina (1933)
382 MAYNARD A . AMERINE
reported 0.12 to 0.41 g. per liter of 2,3-butylene glycol in Italian wines-

and the amount was generally proportional to the alcohol content.
Balavoine (1943) reported 0.045 to 0.225 g. per liter (average 0.137) of
2,3-butylene glycol in 6 M4lagas but the best had 0.224; he proposed a
minimum value of 0.200. Ramos and Reis (1945) in fifty-seven red and
white ports found from none to 0.031% 2,3-butylene glycol, with more
in the reds than in the whites. There was a fair correlation between high
sugar and low 2,3-butylene glycol, which was to be expected.
Ribbreau-Gayon and Peynaud (194713) reported diacetyl in amounts
of 0 to 6 mg. per liter in Algerian wines and 0 t o 1 in Bordeaux wines. In
TABLEV I
Acetylmethylcarbinol in Various Types of Wines
(Milligrams per liter)
France Table RibBreau-Gayon and 12 2.0 20.0 7.0
Peynaud (194713)
France Wh. table Peynaud (1950a) 6 4.5 12.0 7.0
Italy Wh. table Gobis and Farfaletti-Casali 13 3 .O 14.3 7.0
(1952)
Italy Red table Gobis and Farfaletti-Casali 51 3.0 28.0 6.6
(1952)
the former it could therefore be a factor in the aroma. The amounts of
2,3-butylene glycol reported in other studies are given in Table V, and of
acetylmethylcarbinol in Table VI. The amounts of the former reported
by Rib6reau-Gayon and Peynaud are appreciably higher than in other
studies, but their results for acetylmethylcarbinol fall in line with the
others reported.
IV. ALDEHYDESAND RELATEDCOMPOUNDS
Although acetaldehyde is the primary aldehyde present in wines,
there are reports of hydroxymethylfurfural, furfural, and higher alde-
hydes. Acetal and acetone are also found in some wines.
1. Acetaldehyde
Acetaldehyde is a normal by-product of alcoholic fermentation, but
it may increase during aging owing to oxidation of ethyl alcohol or to the
activity of film yeasts. It is easily fixed by sulfur dioxide, so that much
of that present is “bound.”
Methods. The total absence of aldehyde from some wines reported by Nelson and
Wheeler (1939) was probably due to analytical errors. The primary methods employed
are the colorimetric procedure with Schiff’s reagent and the distillation procedures in
which the aldehyde distilled is received in sulfite or hydroxylamine solution. Teixeria
JGnior (1940)showed t h at low results were obtained if the distillate from the alcohol
determination was employed. Charles (1930)used a distillation procedure and titrated
with alkali after adding hydroxylamine hydrochloride or used Schiff’s reagent. Com-
parable results were obtained. Tarantola (1934)made a study of factors influencing
the development of color with Schiff’s reagent, using a Pulfrich photometer and a
Lange photoelectric colorimeter. He pointed out that the procedure must be followed
exactly in order t o obtain satisfactory results. He did not use carbon dioxide during
distillation.
The procedure of Jaulmes and Espezel (1935)is now widely used. It depends on
fixing the aldehyde in a buffered sulfite solution, titrating the excess sulfite in acid
solution, hydrolyzing the aldehyde-bisulfite complex under alkaline conditions, and
then titrating the bisulfite released-which is equivalent to the aldehyde present.
Joslyn and Comar (1938)studied this and other procedures. They noted that less than
100% of the aldehyde present is recovered. This procedure can be improved by con-
trolling the p H when hydrolyzing the aldehyde-bisulfite complex, e.g., with a less
alkaline solution such as bicarbonate. To improve the sensitivity of Jaulmes’s pro-
cedure Roche (1948)recommended a more dilute iodine solution and a standard color
to determine the end point. He proposed a dilute mixture of basic fuchsin and indigo
carmine for this.
A comparison of procedures for the determination of acetaldehyde was made by
Iribarne, (1941).The colorimetric Schiff, the bisulfite procedure, and the hydroxylamine
method were compared. I n general, the bisulfite procedure gave a better and more
regular recovery, but on 49 wines it showed only 101 mg. per liter of aldehyde, and the
hydroxylamine, 104.
A polarographic procedure for determining the acetaldehyde content of port wines
was developed by Almeida (1950).His results (Table VII) are generally lower than
those obtained earlier in port wines by Teixeira Jdnior (1940)using the Jaulmes and
Espezel procedure. A summary of the methods for determination of aldehydes was
given by Mestre and Campllonch (1942).
While much of the aldehyde present in wines is bound to bisulfite, etc., some is
free. Villforth (1940)outlined a procedure for its determination which has also been
used by Koch and Bretthauer (1950-1951).It is based on treating ice-cold wine with
calcium carbonate and calcium hydroxide, decarbonating with ice-cold barium
chloride, and filtering. The filtrate is distilled with calcium carbonate and the alde-
hyde determined by the usual procedures. Probably more than the free aldehyde is
determined.
Formation. The experiments of Mestre and Campllonch (1942) sup-
port the view that acetaldehyde is an intermediary in the process of
alcoholic fermentation. I n some cases they report high volatile acidity
arising from oxidation of aldehydes, even in the absence of bacteria.
Uchimoto (1951) found a positive correlation between temperature of
fermentation and amount of acetaldehyde formed. Many investigators
have demonstrated a rapid rise in the aldehyde content of wines under a
film yeast. Ter-Karapetian (1952) has studied the formation of acetalde-
hyde in 16% alcohol wines. Aldehyde formation occurred only when the
wine was aerated and yeast cells were present. Fornachon (1953) has
384 MAYNARD A. AMERIKE
shown that alcoholic suspensions of Saccharomyces beticus accumulate

acetaldehyde when shaken in air. Added acetaldehyde is consumed under
anaerobic conditions but in nonalcoholic suspensions under either
anaerobic or aerobic conditions. The alcohol/aldehyde equilibrium under
aerobic conditions varies with the concentration of alcohol, temperature,
and the composition of the substrate. (See p. 464 for the relation of acet-
aldehyde to the other by-products of alcoholic fermentation.)
Amounts. The presence of acetaldehyde is usually objectionable in red
wines. Charles (1930) finds it possible to recognize acetaldehyde in
amounts not exceeding 0.1 g. per liter and considers the wine unmarket-
TABLEVII
Aldehyde Content of Various Wines
(Milligrams per liter)
Region wines Source of data samples mum mum age
Argentina Table Iribarne (1941) 49 9 310 101
California White table Amerine and Joslyn (1951) 480 9 92 54
California Red table Amerine and Joslyn (1951) 170 5 491 46
California Dessert Joslyn and Amerine (1941) 142 15 217 87
France Dessert Peynaud (1950b) 8 62* 264 * 128*
France Red table Peynaud (1950a) 6 22 53 36
France White table Peynaud (1950a) 6 44 84 68
France Sparkling Hennig (1952) 13 10 47 20
Germany Sparkling Hennig (1952) 40 3 100 50
Portugal Port Almeida (1950) 30 20 90 38
Portugal Port Teixeira Jtnior (1940) 135 15 166 95
Spain Sherry Bobadilla and Navarro 25 90 500 218
(1952)
* Total includes that bound t o polyphenols.
able a t 0.5 g. per liter. Although there is usually a simultaneous decrease
in acetaldehyde, color, and tannin contents of red wines during aging,
Joslyn and Comar (1941) found that this varied from wine to wine.
Furthermore, the higher the initial acetaldehyde content, the more rapid
its disappearance during storage. As expected, they found that sulfur
dioxide reduced the rate of disappearance of aldehyde. It is of practical
interest to note that addition of tartaric acid decreased the rate of acet-
aldehyde formation. This aspect of these experiments might be further
investigated. Teixeira J6nior (1940) found that in high-quality ports
there was slightly greater variation in acetaldehyde and volatile acidity
than in poor-quality ports, and the wines of higher quality had a some-
what higher acetaldehyde content. The amounts in several types of
wines are given in Table VII.
I n untreated German fruit dessert wines Koch and Bretthauer (1950-
1951) reported 22.2, 24.9, 76.7, and 99.3 mg. per liter of free aldehyde. A
vermouth contained 129.9 and a white table wine, 31.3. Peynaud (1950b)
calculated that 0 to about 60% of the total aldehyde in eight French
dessert wines was free.
2. Acetal
Little quantitative information is available on the acetal content of
wines. RibEreau-Gayon and Peynaud (1947~)gave a procedure based on
the fact that acetal distills without decomposition at a pH of 9. The
distillate, which also contains the acetaldehyde, is brought to volume, and
aliquots are distilled a t pH 9 and 1 to give the acetaldehyde and acetalde-
hyde plus acetal contents, respectively. Out of twelve wines they found
no acetal in eight, traces in three, and 3 mg. per liter in a port. I n brandy
the content is much higher, 24 to 118 mg. per liter. Joslyn and Comar
(1941) reported 6 and 8 mg. per liter of acetal in two red California wines.
Sisakhi et al. (1950a) reported 35 mg. per liter of acetal during fermenta-
tion but only 2.7 mg. at the time of the first racking. During growth of a
sherry film this rose to 35.4 mg. in six months. They consider the ratio of
acetal to acetaldehyde to be an important factor in the quality of sherry-
a lower ratio indicating poor quality. This appears to be a fruitful field for
further investigation. Sapondzhian and Gevorkian (1953) found that the
larger the amount of acetal and the lower the per cent alcohol, the more
acid and the longer periods of distillation required to obtain complete
recovery. A more specific procedure for acetal is greatly to be desired.
3. Acetone and Benzaldehyde
Chelle et al. (1936) developed a colorimetric procedure for the deter-
mination of acetone in wine distillates. They were able to secure good
checks with this procedure, even in the presence of considerable amounts
of higher alcohols. Acetone was reported in wines to which commercial
ethyl alcohoI had been added. No benzaldehyde was found in grapes by
Mathers and Schoeneman (1952), but cherries contain significant
amounts. During fermentation benzyl alcohol, benzyl, and benzoin are
produced from benzaldehyde. A polarigraphic procedure for its determi-
nation was developed.
4. Hydroxymethylfurfural
Methods. Von Fellenberg (1935)reported on three methods for the determinatiou
of hydroxymethylfurfuratl. He preferred precipitation by phloroglucin and oxidation
of the precipitate by dichromate. Kruisheer et al. (1935) criticized t h e phloroglucin
oxidation procedure as giving high results and weighed the precipitate directly. Cunha
(1947) developed a semiquantitative colorimetric phloroglucin method. He distin-
guished wines with 0, traces, below 60 mg. per liter, and above 60 mg. per 1.
Botelho (1938)and Amerine (1948)used Fiehe’s reagent for a qualitative test. I n
Fiehe’s reaction resorcinol in concentrated hydrochloric acid reacts with the hydroxy-
rnethylfurfural previously ether-extracted from wine. The amount of red precipitate
is roughly proportional to the color. Levulinic acid does not interfere. Michel (194813)
recommended use of very clean equipment and freshly distilled (over dichromate) and
washed ether.
Maaskant (1936) suggested p-nitrophenylhydrazine for its qualitative or quantita-
tive determination. Amerine (1948) also used this for preparing the derivative. A
somewhat different procedure for identifying hydroxymethylfurfural in sweet wines
was developed by Huntenburg (1936). He criticized the phloroglucin procedure as
lacking specificity. His rather lengthy procedure is based on heating in acid solution
to convert the hydroxymethylfurfural to levulinic acid. This then forms a derivative
with l-phenyl-3-methyl-6-oxo-1,4,5,6, tetrahydropyridazine. It is less useful for
quantitative studies, as only a 40% yield waa obtained.
Amounts. Kruisheer et al. (1935) first recognized that hydroxymethyl-
furfural in wines would not occur in wines which had not been heated or
which had not been prepared by heating or by addition of caramel.
Genuine ports, for example, had only 0 to 24 mg. per liter.
Botelho (1938) confirmed these results, finding none in 28 genuine
port wines of the vintage of 1888 to 1933. Prosstosserdov and Taranova
(1949) used phloroglucin for detecting hydroxymethylfurfural in 16 of
24 port-type wines and in 6 of 8 madeira-type wines produced in Russia-
indicating rather general heating of these wines during preparation.
Amerine (1948) found the same true of 154 California dessert wines-all
except 44 contained hydroxymethylfurfural using Feihe's test. Using the
phloroglucin procedure 32 to 305 mg. per liter (average 161) were found
in 6 California sherry-type wines. Only traces or no hydroxymethyl-
furfural occurred in experimental dessert wines that had not been heated.
Kniphorst and Kruisheer (1937) found hydroxymethylfurfural in 2
Hungarian Tokay wines, which they therefore believed to have been
prepared with concentrate. Spanish wines of Mtilaga and Tarragona were
also high in hydroxymethylfurfural, indicating heating or the use of con-
centrate. Using this rather specific procedure, Huntenburg (1936) found
5 to 1149 mg. per liter in 7 dessert wines. The largest was in a "dark"
Mtilaga to which presumably reduced must had been added. Michel
(194813) reported that commercial pasteurization of grape juice for 30
minutes a t 75" C. (167"F.) did not result in formation of hydroxymethyl-
furfural, but that use of concentrators and desulfiters did lead to its
formation.
6 . Acrolein
A procedure for the quantitative determination of a c r k i n in musts
and spirits was developed by Wilharm and Holz (1951). It occurs rarely,
and then apparently as a by-product by bacterial attack on glycerol.
V. ACIDS
The organic acids of wines play an important role in their flavor,
color, and keeping quality. Three are derived from the grape: tartaric,
malic, and citric; others are produced during alcoholic fermentation :

formic, acetic, lactic, and succinic; and some may be formed by bacterial
action: acetic, lactic, butyric, and possibly other acids. Markley et aZ.
(1938) identified linoleic, oleic, palmitic, stearic, and higher saturated
fatty acids of the series CZOto C32 in the saponified bloom of Concord
(Vitis Zabrusca) grapes, and oleanolic acid was identified in the ether
extract. The enologist is primarily interested in the total titratable acidity,
the pH, and the volatile acidity.
I n the grape, of course, the total titratable acidity is almost entirely
composed of fixed or nonvolatile acids. The relative and total amounts
of the two important acids, tartaric and malic, vary greatly from variety
to variety, from region to region, or from season to season (for the same
variety), and according to the amount of crop, vine diseases, etc. I n the
wine the fixed exceed the volatile acids five to twenty times. However, the
volatile acids are important from the regulatory point of view. See also
Ribdreau-Gayon (19384.
General methods for analyzing the organic acids of wine were studied by Gadzhiev
(1940) and Torley (1942). Gadzhiev (1940) separated the tartrates (as the acid
tartrate) and lactic, malic, and succinic acids by their differential solubilities and
varying percentages of alcohol-not an altogether happy solution. A review of the
difficulties of determining the organic acids and a comparison of the various methods
was given by Saburov et al. (1938). Konek and Wettstein (1934) used benzoyl chloride
to separate lactic, malic, and tartaric acids from acetic and succinic acids. At 150’ C.
(302” F.) the former are benzoylated and treatment with hot water precipitates the
former as insoluble benzoyl malic and dibenzoyl tartaric acids. The acetic acid is
distilled in the usual way and the succinic acid determined as iron succinate. The
oxyacid precipitate is treated with potassium hydroxide and when reacidified, the
benzoic acid precipitates. The tartrate was determined as insoluble potassium bi-
tartrate. Deviations of 8% to 10% for malic and 7% to 9% for succinic with known
mixtures of the four acids were reported. Paper chromatography was used by Rodopulo
(1952a) to identify tartaric, malic, citric, and oxalic acids in musts. I n wines succinic
and fumaric acids were also reported. He believed formation of fumaric acid to occur
by oxidation of succinic acid when wines were stored for a long period under aerobic
conditions in contact with a yeast deposit. Mathers (1951) noted t h a t ammonium
vanadate was useful in paper chromatography studies on the acids of wines-tartaric
acid turning red, citric and malic, yellow, and oxalic, blue. An electrical conductivity
method was used by Lakkopoulos (1939) for the determination of citric, malic, and
tartaric acids. The procedure (being long) appears to offer little of interest-the values
for citric were high (up to 35%), and the tartaric/malic ratio must be 3.7 or greater.
A review of the acids in wines was made by Genevois (1951). Pey-
naud’s (1947~)thesis on the organic acids of grapes and wines is also
important. Peynaud has developed a number of new methods or has
modified older procedures for determining the various organic acids
present in musts and wines. These are also summarized by Genevois et al.
(1938b), Peynaud (1946), Rib6reau-Gayon and Peynaud (19474, and
Genevois (1949a). Peynaud has determined the amount of organic acids

present in musts and wines and made a balance to show that all the acidic
elements were accounted for. The average balance in forty-seven Bor-
deaux red wines (average pH 3.40) was given by Peynaud (1947~) as
follows (in milliequivalents) :
A. Cations
Titratable acidity 75.00
Alkalinity of the ash 23.50
Ammonia 1.20
___
Sum of cations 99.70
B. Anions
Tartaric acid 26.9
Malic acid 2.8
Citric acid 1.4
Acetic acid 17.4
Succinic acid 15.5
Lactic acid 25.5
Acid esters 3.1
Phosphoric acid 4.5
Sulfurous acid 0.8
__
Sum of anions 97.9
Similar data for other regions of France have been reported by Peynaud
(1950a, b).
Two methods of calculating the acid balance were given by Peynaud
(1947~).The first procedure is to determine all of the organic and inor-
ganic anions such as tartrates, malates, chlorides, and sulfates. This should
equal the summation of the cations and titratable acidity (potassium,
sodium, ammonia, etc.). The second procedure involves only a balance
of the organic acids. The alkalinity of the ash, ammonia, and titratable
acidity should just balance the organic anions, the acid esters, one func-
tion of the phosphate, two functions of the free sulfur dioxide, and one-
half of the bound. An example of this type of calculation has already been
given. Bremond ( 1937a, b, 1937c, 1938a) made a complete balance of the
acid constituents of three Algerian wines and Marcilla (1934) for a young
Montilla wine.
It is also possible to make a balance of the acids only in musts using
acid salt salt
the equations pH = pK1 + acid and pH = pKz -k acid salt
for the
acids. From the amount of free acid and acid salt present the titratable
acidity may be calculated. This should, of course, equal that actually
determined. This method has also been used by Amerine and Winkler
(1943). An application of the methods of calculating the percentage of the
organic acids present as free acid and as acid salts and salts was made by
Pato (1932). Using this data the pH may be approximately calculated.
Pato then calculates t,he necessary amount of acid to add to various
musts in order to secure a satisfactory pH. The corrections reported are
somewhat greater than those used in practice but are less than those
recommended by some earlier enologists. Pato (1935) recommended
simultaneous use of tartaric t o combine with the potassium translocated
to the fruit after the fruit was ripe and enough malic to replace that which
had been respired following full maturity. Finally, he recommended a
preliminary experiment on a sample of the must to determine the exact
amount to add. Kondareff (1940) made a complete analysis of the acids
in two Bulgarian wines and made a balance of them. The correction of the
must acidity for Elparkling wines was discussed by Ferr6 (1943). He noted
also (1945) that the wines made from the last juice coming from the press
were more subject to a malo-lactic fermentation than was the free-run.
For the correction of the wine acidity prior to the bottle fermentation he
recommended lactic acid. No experimental results are available.
During ripening of the grapes three factors operate to decrease the
acidity: dilution owing to an increase in the size of the fruit, continuous
translocation of bases into the fruit, and respiration. Peynaud (1947~)
showed that tartaric as well as malic acid decreased during ripening.
Genevois and Gatet (1940) have also studied the changes in the acids
during maturation. They note that the problem of methodology is impor-
tant. Peynaud used water to extract the acids from the pomace, while
they used the pressed juice and added 50% alcohol. However, they found
the usual regular decrease in tartaric and malic acids as the grapes
ripened.
The difference in the organic acid content of different varieties is
quite clear from the work of Peynaud (1947~)and Amerine and Winkler
(1943). Peynaud classified grape varieties into high-acid varieties, if they
were high in malic acid, and low-acid varieties, if they were low in malic
acid. Because of the variable influence of climatic conditions on the
acidity Peynaud believes the Balling degree/acid ratio is of limited value
in determining the harvest date for Bordeaux conditions. Amerine and
Winkler (1941), however, found it a useful means of distinguishing
varieties under the more uniform climatic conditions of California.
Peynaud (1939-1940) has also studied the changes in organic acids
during fermentation. Acetic, citric, lactic, arid succinic acids are formed,
whereas malic and tartaric acids decrease: malic appears to undergo a
genuine fermentation and tartaric is precipitated as the bitartrate.
Rib6reau-Gayon and Peynaud (1938a, b) have summarized the evidence
in favor of permitting acid-reducing bacteria t o reduce the total acidity
in the production of fine table wines, particularly reds. They therefore

indicated that both pasteurization and addition of sulfur dioxide might
be injurious to the development of some wines and beneficial to others.
I n contrast to wines, a malic acid fermentation does not occur in fer-
mented pears, according to Jacquin and Tavernier (1951-1952), but is
common in fermented cider. However, a notable decrease in malic acid
takes place during the alcoholic fermentation.
1. Titratable Acidity
Methods. Usually a simple titration is employed. The primary problem is in select-
ing an indicator for red wines. Use of neutral litmus as a n outside indicator is illogical,
because it is difficult t o standardize and because it changes color a t a p H of 7, whereas
the theoretical end point is about 8.2. If red wines are diluted and the indicator added
after the color of the natural red pigments has changed to grey or green, then phenol-
phthalein may be employed. However, phenolphthalein was reported to give too high
results in the total acidity titration of red wines by JimBnez (1937). Phenol red was
recommended b u t not litmus. Jaulmes and Slisewica (1943) showed the theoretical
and actual differences in determining total acidity that may result from use of different
indicators. Melcher (1947) proposed using bromothymol blue. Bromocresol purple was
recommended by Miconi (1948).
A simple procedure for the rapid determination of the total acidity in wines and
fruit juices was patented by Holabach (1936). He prepared alkaline pieces of filter
paper corresponding to 0.05% to 0.01 % acid, which he added to the solution. Bromo-
cresol was used as a n indicator, and from the number of pieces of alkaline paper added
the acidity was calculated. A simple blue patented German indicator, useful for all
except the medium-to-dark red wines, was employed by Larsen (1951). Indicators,
such as acridine and umbelliferone, which are fluorescent in ultraviolet light for the end
point in the acidity determination, were proposed by Volmar and Clavera (1931). One
advantage of these was that they can be used in red wines.
Acetic acid is quantitatively produced from sodium acetate in the presence of fixed
acids and acid salts. This was used by Fontanelli (1941) for determining the total
acidity. The results were higher than those obtained by titration. The distillation was
accelerated by saturating the solution with sodium chloride.
An iodometric procedure, based on the reaction IOs- + 51- + 6Hf 3 3 1 2 + 3H20
was proposed by Vihles (1939). Slightly high results were obtained, and the reaction
took 24 hours. The main advantage claimed was the end point. Because of the diffi-
culty in detecting the color change at the end point, SBmichon and Flanzy proposed
(1930b, 1932d) a method based on the amount of carbon dioxide released when wine
is placed on calcium carbonate. The procedure is empirical and not suited to routine
analysis, as Ferr6 (1931) indicated. He preferred titration to a pH of 7, although this
is not the true end point for the titration of a weak acid with a strong base. An adapta-
tion of the old gasimetric procedure of Bernard (see Jaulmes, 1951) for determining the
total acidity was reported by Voskoboinikov and Negenzev (1930), who recommended
the procedure for dark-colored hybrids. But, although speedy, this procedure offers
nothing useful.
Gunts (1950) proposed using an ion exchanger. The diluted wine was slowly run
through Amberlite IR lOOH to remove hydrogen ion. The hydrogen ion was then
washed off the column with water and titrated. If the wine is run through both an
anion and cation exchanger, only nonionized materials such as glycerol, sugars, and
coloring material appear in the final solution. To selectively remove acetate he recom-
mended fist saturating t h e column with tartrate, malate, succinate, and lactate
anions. The necessity of heating wines to remove carbon dioxide before titrating for
the total acidity was emphasized by Flygare (1949).
The ability of Botrytis cinerea to reduce the total acidity is well known.
For a review of the literature see Schanderl(l950). Ventre (1936) reported
permanent differences in the acid-producing properties of several strains
of wine yeasts.
Deacidification of wines by various materials was reviewed by
Genevois (1934b). Sodium carbonate gave a disagreeable taste. Ammo-
nium bicarbonate, potassium carbonate, or potassium tartrate can be
used, but their effect varies markedly with the composition of the wine
treated. Calcium carbonate has a regular action, but carbon dioxide is
evolved and the precipitate is voluminous. Magnesium salts give an un-
pleasant taste.
The total acidities of a large number of Argentinian wines were
reported by Ruspini (1936). Reports on the empirical relationships
between the total acidity and the sugar content of the grapes have been
given by Amerine and Winkler (1941) and Korotkevich (1948). The
former classified the varieties of grapes into three groups on the basis of
their sugar/acid ratios. The latter showed that the ratio also varied with
the region and season and classified grape varieties into eight groups.
2. Tartaric Acid
Potassium Acid Tartrate Procedure. lmprovements in this procedure for deter-
mining tartaric acid were given by Seiler (1932, 1943b). At least 0.5 mg. of potassium
acetate must be added for every milligram of free tartaric acid present. Methyl alcohol
may be substituted for ethyl. A similar modification of the standard German pro-
cedure was made by von der Heide and Hennig (1934) and Steuart (1934) by varying
the amount of potassium acetate added according t o the total titratable acidity.
Podkletnov (1951) and Amerine (1952) also used this method. A modification of the
procedure was presented by Berg and Schmechel (1932). No correction factor for the
solubility of the acid tartrate was used, yet rather accurate values were reported,
particularly with the lower percentages of tartrate. Dubaqui6 (1932) reviewed the
methods for tartaric acid and potassium but gave a minimum of analytical data.
Hartmann and Hillig (1930) found the potassium acid tartrate procedure unsatis-
factory because when alcohol is added pectins and other colloidal materials are pre-
cipitated and occlude other acids. To avoid this they recommend adding alcohol first
to remove the pectins and to precipitate the acids in the filtrate with lead acetate.
The precipitate is then dissolved and treated with hydrogen sulfide to remove the lead.
Recovery of 97% to 100.2% of the tartaric acid from mixtures with malic and citric
acid was reported. The procedure of Beis (1934) for the determination of tartrates and
tartaric acid is based on the relative insolubility of the tartrates in alcohol. Titration
before and after removal of the tartrates was used to obtain the tartrates by difference.
Other procedures appear t o be more suitable.
Racemute Procedure. Precipitation as calcium racemate, usually called the Kling

procedure, has been frequently employed, but the cost and availability of levotartaric
acid or its salts have discouraged investigators. S6michon and Flanzy (1933a) used
calcium sulfate in place of calcium acetate to prevent coprecipitation of lev0 calcium
tartrate and recommended other changes in Kling's procedure. Their results showed
the racemate method to give the best recovery. Peynaud (1936b) also employed the
racemate procedure but used a double precipitation, increased the time allowed for
the first precipitation, and made other changes t o improve the method. Kling (1936)
has supported his original procedure.
The method of determining the precipitated calcium racemate has also been a
subject of debate. Some have ashed the precipitate to the carbonate; others have
oxidized it with permanganate. Hartmann and Hillig (1930) reported that the tem-
perature for the permanganate oxidation should be a t least 80" C. (176" F.). Pato
(1944) gave an excellent review of the various procedures and the criticisms of each,
preferring the racemate method. A correction curve was given for determining the
equivalence of permanganate and tartaric acid. He also studied methods of avoiding
coprecipitation of lev0 calcium tartrate and proposed a modified racemate procedure.
A micro-method using periodic acid to oxidize the racemate was developed by Poux
(1949) and Poux and Ournac (1949). Compared to the macro-procedure it showed an
error of about +4%. Peynaud's modification of the Kling procedure was applied to
port wines by Guimar&es(1950). Sugars did not interfere, the temperature of oxidation
must not exceed 70" to 75" C. (158" to 167" F.), and errors of only 0.5% to 2.6% were
obtained. Peynaud (1951a), noticing the difficulty and expense of securing levo-
tartaric acid, recommends using an aqueous extract of Bauhinia reticulata D. C.,
which is rich in the lev0 acid.
Other Methods. Various procedures for determining tartrates were tested by
FAbregues and Mestre (1948). They preferred a conductometric method for tartrates
in lees, based on diluting with ammonium hydroxide to the point of minimum con-
ductivity, adding a known amount of tartaric acid, and repeating the dilution. For
wines they favored the racemate procedure.
Apparently the first to use the red color developed by tartaric acid and sodium
metavanadate in dilute acetic acid for the determination of tartrates in wines was
Ghimicescu (193513, c). However, he separated the tartrates by precipitation as the
acid tartrate. This method was also employed by workers a t the Western Regional
Research Laboratory (1943), by Iljin (1943), and by Bergman and Magoon (1945).
Because there is an inflection in the titration curve a t pH 4.05, the pH2 for tartaric
acid, Canals and Vergnes (1940) suggested this as a means of quickly estimating the
tartaric acid content of wines.
Sophistication of berry wines with grape wines is a frequent problem of law enforce-
ment agencies in this country. A sensitive test for tartaric acid is necessary, since most
authorities agree that little or no tartaric acid occurs in berries. Marsh and Kean
(1951) recommended chromatographic methods. In a genuine berry wine they found
no tartaric acid; by the Association of Official Agricultural Chemists (1950) procedure
they found it in three. The colorimetric procedure of Mathers (1949) is also applicable
but subject to interference. It gave, for example, a positive test on one of the samples
in which paper chromatography showed no tartaric acid. In Brazil the problem is
apparently the opposite-sophistication of grape wines with fruit wines. Buhrer (1950)
has developed an empirical test bases on the voIume of precipitate obtained when
mercuric oxide solution is added to hot wine-the precipitate being greater with grape
wine. Further tests based on presence of citric acid are not applicable, as this acid is
frequently added to grape wines.
COMPOSITION OF W I N E S 393
Mathers et al. (1951) have developed a polarographic procedure for determining
srnall amounts of tartaric acid. They precipitated the tartaric acid with bismuth and
complexed the tartrate with antimony. The polarographic waves were best in a p H
range of 3 to 4. Although recovery was only about 90%, this is satisfactory at low
roncentrations compared t o other methods. Tartrates were not found in genuine
loganberry, gooseberry, currant, blackberry, peach, chrrry, apple, or elderberry wines.
Changes during Ripening or Aging. The tartrate content of ripening

Australian grapes was studied by Swaby (1943). He found the usual
decrease during maturation. He also reported appreciable amounts in the
skins and seeds. Amerine and Winkler (1943) showed that malic acid
decreases relatively more rapidly than tartaric during ripening but that
the varieties differ markedly and characteristically from each other in the
total and relative amounts of each present at maturity. In the studies of
Babadilla and Navarro (1949) the tartrate in Spanish grapes decreased
during ripening, increased when the grapes were drying in the sun,
decreased during fermentation, and underwent little change during aging
under the "flor."
Cambitzi (1947) found optically inactive racemic calcium tartrate in
old wine deposits. Since wine normally contains only dextrotartaric acid,
he interpreted this as indicating autoracemization. The calcium salt of
racemic tartaric acid is only about one-eighth as soluble as the salt of the
dextro acid. Storage at room temperature favored the racemization, and
hence the precipitation. A review of the solubility of potassium and
calcium tartrates in wine and musts and the factors influencing their
solubility was given by Genevois (193413). Malic acid increases the solu-
bility of potassium tartrate and acid tartrate. The decreased solubility of
calcium tartrate as the pH is raised is undoubtedly of importance in wine
aging, where the pH generally increases. Tables giving the solubility of
potassium acid tartrate at temperatures at 5" to 25" C. (41" to 77" F.), a t
alcohol percentages of 0 to 25, and at various pH values with tartaric,
citric, malic, succinic, lactic, and acetic acids were reported by Ivanov
(1947). Marsh and Joslyn (1935) have studied the factors influencing the
deposition of potassium acid tartrate from wines. The process is slow, even
a t low temperatures, but is much more rapid at -3.9" C. (25" F.) than
at other temperatures-as much tartrate was removed at -3.9" C.
(25" F.) in about 2 days as at 25" C. (77" F.) in 225 days. Mursaeva and
Brailovskai; (1952) have shown that the deposition of tartrates is a com-
plicated process depending on many factors. They studied the removal of
tartrates by cooling to between 2" to -30" C. (35.6" to -2220" F.). The
more rapid deposition of tartrates took place at -5" C. (22" F.). At this
temperature the frozen phase is principally water with a minimum of
dissolved material.
A decrease in tart,aric acid owing to bacterial action occasionally
occurs and was discussed by Osterwalder (1952). Sallusto (1935) studied

the rather variable ratio of potassium acid tartrate to per cent alcohol;
the average is 1/48.
Genevois (1951) considered the aging of wines as partially a problem
of the aging of a solution of tartaric acid. He envisages a system in which
ferrous ion plays a part and tartaric acid yields dihydroxymaleic acid,
TABLEVIII
Tartrate Content of Various Wines
(Grams per 100 ml. as tartaric acid)
Alsace Wh. table Lobstein and Schmidt 19 0.09 0 . 3 6 0 . 2 2
(1931)
France
(Nante) Wh. table Huriez (1948) 11 0.15 0.37 0.28
Germany Wh. table Heiduschka and Pyriki 29 0.08 0.33 0.17
(1930)
Germany Wh. table Hennig (1952) 19 0.14 0.37 0.20
Germany Wh. table Seiler (1944) 128 0.12 0.45 0.26
Italy Table Violante (1950) 40 0.17 0.29 0.21
Portugal Table Correia and SBrgio (1943) 113 0.07 0.47 0.19
Portugal Red table Correia and Vilas (1943) 27 0.16 0.32 0.19
Spain Fino Bobadilla and Navarro 15 0.084 0.221 0.144
(1952)
Spain Oloroso Bobadilla and Navarro 10 0.118 0.214 0.144
(1952)
Spain Montilla Casares and Gonzales 51 0.066 0.195 0.100
(1953)
whose ketonic form, oxytartaric, by oxidation-reduction produces
dioxytartaric acid-which, in turn, by decarboxylation, gives carboxy
glyoxal (oxypyruvic acid). This, in turn, by hydration yields tartronic
(hydroxymalonic) acid, which finally by decarboxylation results in
glyoxal. These reactions are summarized as follows:
COOHCHOHCHOHCOOH+ COOHCOH=COHCOOH+
COOHCHOHCOCOOH+COOHCOCOCOOH+COOHCOCHO+
COOHCHOHCOOH+CHOCHO.
Gatet and Genevois (1941) reported 10 to 20 mg. per liter of dihydroxy-
maleic acid in wines containing 3 to 7 mg. per liter of iron, and 60 to 100
mg. in wine containing 20 to 40 mg. of iron. They explained its formation
on the basis of oxidation of tartaric acid by ferric salts. Apparently
dioxytartaric acid accumulates first and is converted to carboxy glyoxal
acid by the action of ferric salts. Rodopulo (1951) has also studied the
changes in tartaric acid. He found that prolonged action of iron on the

acid resulted in an iron-tartrate complex salt. This catalyzes the oxidation
of tartaric acid. Prolonged storage led to formation of the complex.
During racking it particularly catalyzed the oxidation of tartaric acid.
The products of the oxidation were glyoxyalic acid and finally oxalic acid.
During bottle storage, however, diketosuccinic and dihydroxymaleic
acids were formed. Dihydroxymaleic acid also accelerated the oxidation
of tartaric acid in the presence of iron salts. The dihydroxymaleic,
diketosuccinic, glyoxylic, and oxalic acids are genetically related, as is
glyoxal, which is also present. The tartrate contents of several types of
wines are given in Table VIII.
3. Malic Acid
Interest in malic acid is primarily directed toward controlling its
utilization by microorganisms. The central position of malic acid in
enology has been most clearly documented by Ribbreau-Gayon and
Peynaud (1938~).Not only do the varieties of grapes differ markedly in
malic acid (see Peynaud 1938b, 1947c; Amerine and Winkler, 1943;
Korotkevich, 1948; and Amerine, 1950-1951), but the per cent present is
markedly affected by maturity and climatic conditions. Finally, malic
acid usually undergoes a malo-lactic fermentation during the later stages
of fermentation as well as during aging (CharpentiB, 1950).
Methods. The most common procedure for determining malic acid is that of per-
manganate oxidation in a buffered solution to produce acetaldehyde as developed for
musts and wines by Peynaud (1938b, 1946). By slow introduction of very dilute
permanganate a recovery as high as 98% to 99.2% was reported, although Joslyn and
Comar (1938) obtained lesser recovery from pure aldehyde solutions. The disadvan-
tage of this procedure is the correction needed for the presence of tartaric acid.
Amerine and Winkler (1943) and Amerine (1950-1951) were unable to obtain a con-
stant acetaldehyde production from tartaric acid and preferred to remove most of it
as potassium acid tartrate. Some coprecipitation of malic acid probably takes place.
Malic acid was determined colorimetrically with Pinerua’s reagent (@-naphthol in
sulfuric acid) by Ghimicescu (1935a). The malic acid was separated in 70% alcohol.
Nitschk6 (1952) also proposed a colorimetric procedure based on the red color
produced in alkaline solution with a-naphthol. A standard curve is prepared; the color
must be extracted with isooctyl alcohol and is sensitive to light. Sugars, tannins, and
tartaric and lactic acids should be absent.
A polarographic procedure for determining malic acid was developed by Hennig
and Burkhardt (1951). I n i t they heated with alkali t o convert the malic to fumaric
acid. The polarographic procedure depends on the reduction of the double bond of
fumaric acid using a dropping mercury cathode. Since this method may be used only
on tannin- and sugar-free wines, Tanner and Rentschler (1953) removed these by
passing the wine successively through a cation and anion exchanger. The acid was then
washed from the anion exchanger. This appears to be a practical solution to many
clarification problems.
bntschler (1949) proposed using a pure culture of Bacterium gracile for its deter-
mination by measuring the carbon dioxide produced. Only 90% of the malic acid
could be fermented and even a t constant temperature there was a 6% to 8 % error.
Each determination took about 3 to 4 hours.
Amounts Found. I n green grapes Amerine (1950-1951) reported to

50% of the titratable acidity was due to malic acid and a t maturity only
10% to 40%. The varieties differed markedly in malic acid from each
other. Casale (1934) found the ratio tartaric acid/malic acid in Piedmont
grape musts to vary from 0.5 to 3.0, whereas in California grapes Amerine
and Winkler (1943) reported ratios of 1.3 to 4.4 a t Balling readings of 20"
t o 25". Flanzy (1948) showed the intense oxidation of malic acid that
occurs in the fruit when the temperature goes above 29" C. (84.2" F.).
Considerable tartaric acid was also oxidized; however, the acid tartrate
resisted oxidation. Hennig and Burkhardt (1951) found less malic acid
in the grapes in warm than in cool seasons and higher amounts in the
varieties Miiller-Thurgan, Sylvaner, and White Riesling. Bobadilla and
Navarro (1949) also studied the changes in malic acid content during
maturation, during and after fermentation of sherry wines. They found
the usual decrease during maturation, particularly when the grapes were
exposed to the sun for a day after picking and prior to crushing-the
usual Andalusian practice. During fermentation they observed the same
decrease as had earlier investigators. Further decreases due to the malo-
lactic fermentation were observed after fermentation and preceding
fortification. A slight decrease occurred during the "flor " stage. Placing
grapes in an atmosphere of carbon dioxide before crushing results in a
decrease in the malic aicd content, according to Garino-Canina (1949).
Disappearance of 10% t o 30% of the malic acid during alcoholic fer-
mentation a t a pH of 2.6 to 3.6 was demonstrated by RibBreau-Gayon
and Peynaud (1946b). Peynaud (193810) showed a decrease of 9.9% to
14.5%. At a pH of above 3.6 destruction decreased. This decrease is
attributed by Peynaud (1947~)to the splitting off of two hydrogen atoms
and decarboxylation of the resulting oxalacetic acid to acetaldehyde,
which, in turn, acts as a hydrogen acceptor and is reduced to alcohol.
RibBreau-Gayon and Peynaud showed that the lactic acid formed is
approximately equal to one-half the malic acid destroyed and that the
destruction of citric acid does not start until the malic acid has prac-
tically disappeared. Both glycerol and tartaric acid are attacked by cer-
tain bacteria but not a t first by those causing the malo-lactic fermenta-
tion. Rippel (1949) reported that the activity of Bacterium gracile was
dependent on a catalyst of unknown nature derived from the grape. For
historical and other information on the malo-lactic fermentation see
Garino-Canina (1943) and Schanderl (1950). I n Alsatian wines Lobstein
and Schmidt (1931) showed th at the wines high in malic acid were low
in lactic and vice versa. They indicated th at the malo-lactic fermentation
was frequently necessary in order to produce balanced wines. An example
of the changes in acids during storage of a German white table wine in
which a malo-lactic fermentation is occurring is given in Fig. 1.
: : %2.0 *----o Malic Acid
---
0- -0
Lactic Acid
Tartaric Acid
I. I 0.. ....... O
Volatile Acidity
O*'O
10 20 30 40 50 60 70 00 90 100 110
TIME IN DAYS
FIG.1. Changes in the malic, tartaric, lactic, and volatile acids during storage of a
German white table wine (Kramer and Bohringer, 1940).
Rib6reau-Gayon and Peynaud (1938~)reported the normal range of

malate in French wines was 0.067 to 0.268 g. per 100 ml. The malic acid
content of various types of wines is given in Table IX.
4. Citric Acid
It is now generally agreed th at citric acid is a normal though minor
component of grapes and wines. Its accurate determination in musts and
wines is difficult.
Methods. The quantitative determination of citric acid is occasionally important,
ns several European countries limit the citric acid content. Cerasari (1950) has re-
viewed the legal problems and has given a qualitative and quantitative procedure.
The two procedures now commonly used for the determination of citric acid are the
pentabromacetone and the permanganate oxidation. Von Fellenberg (1933) used the
pentabromacetone procedure. His method was not free of empirical details, and the
yield was only about 90% of the theoretical. Reichard (1934a) used it but reported
that the temperature, amount of bromine, relation of bromide to citric acid, and
amount of permanganate to add must be standardized. No interference with other
acids was noted, but sugars must be fermented out (objectionable) or the citric acid
separated as barium citrate. Good recovery of added citric was reported with a
variety of wine types. See also Reichard and Bleyer as reported in Bleyer (1938).
Other studies of the methods for the determination of citric acid in wines were
made by Pfab (1931) and Tarantola (1937a, b). The former used a semiquantitative
TABLEIX
Malic Acid Content of Various Musts and Wines
(Grams per 100 ml.)
Algeria Wh. table Fabre and BrBmond (1932) 13 0.18 0.59 0.37
Algeria Table BrBmond (1937b) 3 0.11 0 . 1 8 0.14
Alsace Table Heiduschka and Pyriki 19 0.01 0 . 0 3 0 . 0 2
(1930)
Bordeaux Wh. table Peynaud (1938b) 29 0.03 0.40 0.14
Bordeaux Red table Peynaud (1938b) 21 0.00 0.05 0 . 0 2
Bordeaux Musts Peynaud (1947~) 14 0.10 0 . 6 0 0 . 3 0
California Musts Amerine (1950-1951) 32 0.12 0.57 0.29
France
(Nante) Huriez (1948)
Wh. table 11 0.12 0.41 0.20
Hungary Wh. table
Torley (1942) 10 0.15 0.29 0.21
Portugal Table Correia and SBrgio (1943) 107 0.01 0.38 0.15
(1952)
(1 952)
Switzerland Wh. table NitschkB (1952) 16 0.01 0.73 0.20
procedure, and the latter employed the Reichard (1934a) method with recoveries of
95% to 105%. Schindler and Koz&k (1934) studied several procedures for the deter-
mination of citric acid in wines, criticizing features of all but recommending a rather
complicated one based on conversion to pentabromacetone. The method of Cartier
and Pin (1949) was developed with biological materials in mind. They converted the
citric acid to pentabromacetone which was colorimetrically measured using the
KonBtiani reaction. An accuracy of f 2 % with 0.1 to 1.2 mg. was claimed. Taufel and
Mayr (1933) also showed that special attention must be paid to buffering, etc., in
order to keep the acetonedicarboxylic acid produced in the keto form, as the enol form
is more likely to undergo undesirable oxidation. They recovered 98.4% to 99.6% of
the citric acid added. Mohler (1937) and von Fellenberg (1933), however, obtained a
low recovery with the pentabromacetone method. Hargreaves el al. (1951) perfected
the pentabromacetone procedure so as to be able to determine citric and d-isocitric
acids.
The Kogan (1930) permanganate procedure, based on the oxidation of citric acid
to produce acetone, was modified by Taufel and Mayr (1933) and Peynaud (1938a)
and later by Godet and Charrihre (1946). The latter introuced the permanganate a t
a more regular rate to secure a slower and more even oxidation. They reported a n
accuracy of 2 % and preferred the procedure t o the pentabromacetone method, from
which they could get only 91% to 92% recovery. Bartels (1933) also studied Kogan’s
procedure and reported t h at while organic acids and sugars do not interfere, it was
necessary to separate the citrate as the calcium or barium salt in the presence of
alcohol and ammonia in order to avoid interference from glycerol. Recovery of 92%
to 102% of the added citric was reported. Peynaud (1938a) reported a n accuracy of
better than 4%.
The method of Heiduschka and Sommer (1935) is longer, including heating with
barium hydroxide and making t o 50% alcohol to precipitate barium citrate, treating
the precipitate with sulfuric acid, and eventually distilling off the acetone formed into
iodine. A qualitative test using mercuric sulfate was also developed b y Reichard
(193413). A simple and rapid procedure for the detection of citric acid was given by
Roleff (1952). This depended on the cloudiness developed in wines containing citric
+
acid when bromine water and 1 % vanadium pentoxide (dissolved in 1 3 sulfuric
acid by warming) was added to charcoal-treated wines. Only 0.02% in 5 ml. was
needed for a positive test, and the best results were obtained in the range 0.02% to
0.1%. Sugar, up t o 7%, did not interfere.
Occurrence. Citric acid, although present in only small amounts in
musts, is of considerable biochemical interest. RibBreau-Gayon and
Peynaud (1938d) found more in certain varieties than others. During
ripening, citric acid decreased in four Bordeaux varieties and increased
in one (Sauvignon blanc) (Peynaud, 1938a). On a per berry basis there
was a decrease in two, no change in one, and an increase in two varieties.
The average content a t maturity varied from 0.02% t o 0.03%.
Formation of 1 to 2 meq. per liter of citric acid at the expense of 166 g.
of sugar during alcoholic fermentation was demonstrated by RibBreau-
Gayon and Peynaud (1946b). Since only small amounts of citric acid are
present in musts, this may result in a 40% increase during fermentation.
Peynaud (1947~)attributes the increase in citric acid during fermentation
to the direct action of the yeast on sugars. Saccharomyces ellipsoideus and
Kloeckera apiculata were found by Gobis (1950) to produce citric acid by
an oxidative mechanism. She noted that citric acid production, like
higher alcohol and succinic acid production, may be associated with the
protein metabolism of yeasts. For example, addition of ammonium
sulfate decreased production of higher alcohols and citric acid. Addition
of tartaric and formic acids increased citric acid formation. Decreases in
citric acid during fermentation were shown by Tarantola (1937a) to
depend on the organism present rather than on the absence of sulfur
dioxide. I n numerous Italian wines he found 0 to 0.478 g. per liter. Wines
of low total acid also contained less citric acid.
RibBreau-Gayon and Peynaud (1938d) reported white wines to con-
tain over four times as much as the reds-owing to bacterial utilization
in the nonsulfited reds. The citric acid content of nonpasteurized wines

was always less than that of the pasteurized checks. RibBreau-Gayon
(1938a) noted that sulfur dioxide should always be added along with
citric acid. Luthi (1949) reported that it increased the volatile acidity of
Swiss red table wines 0.15% and should not be employed. Charpenti6
et al. (1951) find acetylmethylcarbinol, acetic acid, diacetyl, lactic acid,
and 2,3-butylene glycol as the products. They propose a mechanism by
which the degradation of citric acid proceeds through an oxalacetic-
pyruvic acid stage. The pyruvic acid then reacts as follows to give the
above products in the order listed:
2CHoCOCOOH + CHsCOCHOHCHs 2CO2 +
+
CHsCOCOOH HnO + CHsCOOH 2COt 2H + +
2CHsCOCOOH + CHsCOCOCHs 2C02 2H + +
+
CHsCOCOOH 2H -+ CHsCHOHCOOH
+
2CHsCOCOOH 2H + CHsCHOHCHOHCHs 2CO2 +
They calculated the theoretical and actual amounts of the products
formed according to this scheme and found good agreement.
The protective effect of citric acid against ferric phosphate cloudiness
is well known. Verda (1940) claimed that addition of citric acid to wines
decreased the loss of vitamins, particularly ascorbic acid.
A summary of the regulations of various countries limiting the use of
citric acid was published by Berg and Schulze (1934). Mohler (1937) has
reviewed the earlier work on the citric acid content of wines and con-
siders up to 0.05% normal for wines. This is the French limit. As of 1937
Germany, Hungary, and Austria prohibited use of citric acid; Rumania,
Yugoslavia, and Switzerland had the 0.05% limit; Italy and Spain a
0.10% limit; and Greece had no limit. The present United States limit
is 0.05%, but under a more liberal section of the regulations almost un-
limited amounts can be added. Buogo (1938) has reviewed the literature
and finds no biochemical reason why wines containing this acid should be
considered sophisticated, but believes that it should not be added in
excessive amounts because of its flavor. This aspect of its use needs to be
investigated further.
Amounts Reported. The amounts of citric acid reported in various
wines are given in Table X. It is noteworthy that Tarantola (1937a) was
unable to detect any citric acid in 18 of the 57 wines which he analyzed.
This was obviously owing to bacterial decomposition during aging.
Rodopulo (1952a) reported 0.023% citric acid in a Russian must and
0.035% in the fermented wine. Von Fellenberg (1933) found from 0.07 to
0.38 g. per liter in 4 authentic Rheinpfalz wines. In wines made from
Spanish, Greek, or California raisins he reported 0.02 to 0.63 g. per liter.
In 14 Swiss wines 0.04 to 0.44g. per liter were found, but fruit wines con-
tained up to 2.94 g. per liter. I n 60 samples from Bari, Italy, Buogo and
Picchinenna (1938) reported 0.03% to 0.05% with an occasional sample
up to 0.1%. Bobadilla and Navarro (1949) reported 0.03% to 0.30% in
sherry wines. Little changes occurred during fermentation or during
TABLEX
Citric Acid Content of Various Wines
(Grams per 100 ml.)
Alsace Table Lobstein and Schmidt 19 0.00 0.02 0.01
(1931)
France Wh. table Peynaud (1938a) 34 0.009 0.058 0.033
France Red table Peynaud (193%) 26 0.003 0.038 0.0076
France
(sugared)
(unsugared)
Germany Wh. table Reichard (1936) 70 0.000 0.030 0.008
Germany Table Reichard (1934a) 15 0.000 0.010 0.003
Germany Red table Reichard (1936) 30 0.000 0.045 0.009
Germany Wh. table Wirthle (1931) 22 0.01 0.02 0.02
Italy Table Tarantola (1937a) 57 0.000 0.045 0.015
(1952)
(1952)
Switzerland Table Godet and CharriAre 5 0.02 0.05 0.03
(1946)
Various Wh. table Berg and Schulze 11 0.00 0.03 0.01
(1934)
Various Red table Berg and Schulze 5 0.00 0.03 0.01
(1934)
Various Swt. table Berg and Schulze 7 0.00 0.27 0.11
(1934)
Various Dessert Reichard (1936) 26 0.000 0.103 0.024
aging under the “flor.” Piatkowska and Smreczyriska (1950) noted that
the high citric acid content of current, bilberry (Vacciniummyrtillus), and
gooseberry wines makes it possible to detect their addition to grape wines.
However, as much as 30 % of fruit wine would have to be added, and since
some countries do not control the citric acid addition t o grape wine, the
addition of apple, cherry, or blackberry wines cannot be so detected.
6. Succinic Acid
This acid is found in all fermented beverages, and in wine constitutes
an important fraction of the fixed acidity. Generally about 1% as much
succinic acid as alcohol by volume is found in unfortified wines.
Methods. Espil (1937) separated the barium tartrate, malate, and
succinate in 80% alcohol. These salts were then decomposed with hydro-
chloric acid by evaporating to dryness with potassium sulfate. The
tartaric and malic acids were separated by ether extraction and the
succinic acid in the residue titrated. SBmichon and Flanzy (1932~)deter-
mined succinic acid by oxidizing the other constituents and weighing the
residual succinate after purification. No data on accuracy were given.
The best study of the determination of succinic is that of Marignan
(1944). (See also Jaulmes, 1951, and Amerine, 1952.) It is applicable
primarily to sugar-free wines, but by extracting with ether prior to the
permanganate oxidation it can be adapted t o sweet wines. He oxidized
the other acids with permanganate, extracted the residual succinic acid
with ether, precipitated as silver succinate, and determined the residual
silver.
Source. I n pure cultures RibQeau-Gayon and Peynaud (1946b)
showed that more succinic acid was formed a t the start of the fermenta-
tion than a t the finish and that the amount formed per gram of sugar
fermented varied with the must and the yeast. Peynaud (1947~)found
about 0.47 g. of succinic acid formed per 100 g. of sugar fermented and
noted a fairly regular production during fermentation. He believes that
it originates from sugars with acetaldehyde or pyruvic acid as the inter-
mediate and not from amino acids. Bobadilla and Navarro (1949) found
no changes in succinic acid during aging under a “flor.”
Amounts. Vasconcellos (1940b, 1941) has reported data on the per-
centage of succinic acid in the sweet dessert wine, port. Although this
author’s procedure had a maximum error of about lo%, in over 70%
of the samples it was less than 5%. I n eight samples where the original
sugar content of the must was known, no direct relationship between the
quality of sugar fermented and the amount of succinic acid formed was
observed. In the studies of Genevois et al. (1948~)succinic acid in Bor-
deaux wines varied greatly-from 48 to 106 millimoles per liter, average
75. The ratio of acetic acid to succinic acid, according to the yeast strain,
varied from 0.4 to 4.3, but usually from 0.4 t o 2.0. Anibal (1935) calcu-
lated the ratio of succinic acid per liter times 100 divided by the glycerol
content (grams per liter) for forty-three Argentinian table wines to vary
from 0.46 to 2.79. Correia and Ribiero (1942) found the succinic acid,
as per cent of the weight of alcohol produced, to vary from 0.68 to 2.25 in
114 Portuguese table wines. This was too variable to make the succinic
acid : alcohol relationship of use for identification purposes. The succinic
acid content of various types of wines is given in Table XI.
TABLEXI
Succinic Acid Content of Various Wines
(Grams per 100 ml.)
Argentina Red and Wh. Anibal (1935) 43 0.035 0.195 0.069
table
Argentina Table Velazquez (1936) 39 0.04? 0.181 0.061
France
Italy (Cirb) Red table * Sallustro and Sculco 26 0.041 0.073 0.054
(1937-1938)
Portugal Table Correia and Ribeiro 114 0.06 0.225 0.121
(1942)
Portugal Table Correia and Shrgio 105 0.05 0.67 0.103
(1943)
Portugal Red port Ramos and Reis (1945) 38 0.005 0.127 0.035
Portugal Wh. port Ramos and Reis (1945) 19 0.005 0.050 0.014
(1952)
(1952)
* Av. alcohol 15.3.
6. Lactic Acid
Lactic acid is not only a product of alcoholic fermentation but of the
bacterial activity in wines. I n some red table wines it is the pre-
dominant acid. Its accurate determination is therefore of considerable
interest.
Methods. The two most common procedures for the determination of lactic acid
are the Moslinger procedure, or a modification of it, which depends on the solubility
of barium lactate in 70% to 80% alcohol, or variations of the Espil (1936) technique,
which is based on its oxidation in a buffered solution with permanganate and measure-
ment of the acetaldehyde formed. Those standardizing lactic acid solutions should not
forget that saponification is necessary. Hickinbotham’s procedure (1948) is conven-
ient. Von Fellenberg (1931, 1932a) gave further details on hie earlier steam-distillation
procedure for lactic acid, which may yield high results with red wines. Benvegnin and
Capt (1932) compared several procedures for the determination of lactic acid, pre-
ferring the Moslinger procedure for wines of normal lactic acid content. They obtained
good results with the von Fellenberg (1932a) technique, especially for wines of low or
high lactic acid and with sweet wines, but the technique is difficult. Modifications of
the MGslinger-Bonifazi method were proposed by Fabre and Brdmond (1931rt),
Velazques (1936), and Michel (1931). Errors of less than 0.2 g. per liter were reported
by the former. BoFinjak (1938) compared various methods for lactic acid. The Mas-
linger procedure could not be used on sweet wines nor with wines high in lactic acid,
and losses occurred with Espil’s (1936) method. His best results were secured by
extraction.
The micro-procedure of Ghimicescu (1935e) can also be used for larger amounts.
It too is based on isolation of barium lactate in 80% alcohol and conversion to barium
carbonate. Seiler (1941) pointed out that the Moslinger technique was not applicable
to sweet or fruit wines. He questioned if a calculation of the original malic acid content
based on determination of the final lactic acid content was justified.
The latest modifications of the Espil technique are given by Peynaud (19461, by
Peynaud and Charpentid (1950) and by Koch and Bretthauer (1951). They reviewed
the problem and suggested several modifications. They removed 2,3-butylene glycol
by bringing the pH to 6 and evaporating t o dryness. By keeping the acidity low they
kept aldehyde formation from glycerol to a minimum. Amerine (1950-1951) compared
several procedures for the determination of lactic acid. Both oxidation and precipita-
tion procedures gave satisfactory results.
A dichromate oxidation procedure for table wines was proposed by Sdmichon and
Flanzy (1932b). No data on the accuracy of the procedure were given. A new pro-
cedure by Krause (1948) is based on the solubility of calcium lactate and acetate in
anhydrous methyl alcohol. The lactate and acetate were precipitated as silver salts
and ashed, and the alkalinity of the ash was determined. By subtracting the acetic
acid, the lactic acid content can be calculated. The procedure is said to be satisfactory
for red and white wines, even if they contain sugar.
Colorimetric procedures may be subjected to errors from impurities.
Hillig (1937) extracted, purified with charcoal, developed a color with
ferric chloride, and estimated it photometrically a t 450 to 460 mp. The
colorimetric procedure of Strohecker et al. (1938) employing veratrole is
not satisfactory for wines, although Diemair et al. (1940) used it. It is
based on evaporation of the volatile acids, precipitating most of the other
organic acids in alcohol as barium salts. The glycerol and lactic acid
remaining in solution are separated after removal of barium, with ether
or an ether-alcohol mixture. The lactic acid was determined colorimetri-
cally with veratrole and the glycerol measured after oxidation to formal-
dehyde with fuchsin. This procedure seems too complicated for ordinary
use, and Beer’s law is applicable for the veratrole-lactic acid color only
up to 0.1 g. per liter.
The special problem of lactic acid determination in sweet wines was
studied by Berg and Schulze (1931) using barium carbonate for removal
of sugars. The results were only fair. A qualitative test for lactic acid was
developed by Widmar (1931). He neutralized the volatile acid distillate
with barium hydroxide and added alcohol under specific conditions. A
small amount of precipitate which settled slowly indicated lactic.
Source. Rib6reau-Gayon and Peynaud (194613) and Peynaud (1947~)

showed that 5 to 7 meq. per liter of lactic acid were formed at a pH of
2.6 to 7.0 with twelve different yeasts. The constant occurrence of lactic
acid in fermentation products was also demonstrated by Durmishidze
(1938). He showed that sugar not malic acid was its probable precursor
in fermentation. The formation of lactic acid was intensively studied by
Hohl and Joslyn (1941b). With a variety of yeasts they found lactic acid
formation to parallel alcohol formation. While aeration showed no appre-
ciable effect on lactic acid production, variations resulted from the use
of different yeasts and media. Uchimoto (1951) found less lactic acid
produced at lower temperatures than at high. Bobadilla and Navarro
(1949) considered this acid more important to the flavor and body of
sherry wines than tartaric or malic. During the period under the “flor”
the lactic acid increased. The changes during storage were studied by
Seiler (1943a). Lactic acid increased during the first four to seven months
and then decreased; after bottling there was little change.
For detailed historical information on the malo-lactic fermentation,
see Kramer (1941b), Vogt (1945), and Schanderl (1950). Reviews of the
lactic acid content of wines are given by Anonymous (1940), Genevois
and RibQeau-Gayon (1947), Amerine (1950-1951), and Fabre and
Bremond (1932). The classification of the last-named authors on the basis
of lactic acid content is interesting; young wines which have not under-
gone a malo-lactic fermentation contain less than 0.1% and a pH of
about 3.1; diseased wines have over 0.2% lactic acid and a pH of 3.3
or higher.
The amounts reported in various types of wines are indicated in Table
XII.
TABLEXI1
Lactic Acid Content of Various Wines
(Grams per 100 ml.)
Algeria Table Fabre and 13 0.072 0.46 0.19
Br6mond (1932)
Algeria Table Br6mond (1937b) 3 0.074 0.128 0.092
Alsace Table Lobstein and 19 0.03 0 . 4 3 0.17
Schmidt (1931)
Argentina Table Velazquez (1936) 58 0 . 1 2 0.41 -
California Table Hillig (1937) 17 0.11 0.49 0.29
California Wh. table Amerine (1950- 17 0.08 0.43 0.19
1951)
California Red table Amerine (1950- 4 0.17 0.25 0.21
1951)
Czechoslovakia Wh. table Kopal (1938) 300 0.09 0.41 0.24
TABLHXI1 (Continued)
France Table Hugues and 80 0.06 0.29 0.14
Chevalier (1930)
France Table * Hugues and 20 0.12 0.37 0.23
Chevalier (1930)
France Wh. table Rib6reau-Gayon 28 0.04 0.25 0.10
and Peynaud
(1937a)
France Red table Rib6reau-Gayon 31 0.06 0.38 0.20
and Peynaud
(1937a)
Germany Wh. table Heiduschka and 29 0.06 0.46 0.28
Pyriki (1930)
Germany Wh. table Alfa (1933) 46 0.03 0.50 0.19
Germany Wh. table Heide and Zeissett 24 0.12 0.39 0.22
(1935)
Germany Wh. table Wirthle (1931) 22 0.05 0.41 0.18
(sugared)
(unsugared)
Germany Wh. table Seiler (1936b) 71 0.02 0.32 0.20
Hungary Table Torley (1942) 10 0.09 0.44 0.25
Italy (Cirb) Red table Sallusto and Sciulco 26 0.14 0.30 0.19
(1937-1938)
Italy Table Violante (1950) 40 0.23 0.37 0.32
Italy Table Garino-Canina 9 0.08 0.34 0.15
(1951a)
Roumania Table &muleanu and 33 0.09 0.44 0.21
Ghimicescu
(1936)
Russia Table Voskobohikov 117 0.037 0.33 0.15
(1931)
Spain Fino Bobadilla and 15 0.071 0.140 0.109
Navarro (1952)
Spain Oloroso Bobadilla and 10 0.085 0.189 0.134
Navarro (1952)
Switzerland Wh. table Nitschk6 (1952) 16 0.08 0.45 0.26
Turkey Table Akman (1951) 66 0.02 0.16 0.07
* Diseased wines..
7. Other Fixed Acids

SBmichon and Flaney (1933b) verified earlier work that glyoxylic acid
is a constituent of grapes (and even of wines). They used a Cannieearo
reaction to produce glycolic and oxalic acid, determining the latter,
assuming no oxalic acid was present originally. They reported 210 mg. per
liter-this needs to be confirmed. Rodopulo (1952a) reported 0.0085 %
oxalic acid in a must and 0.0087ojo in the wine. Ventre (1939) found up
to 0.13% glucuronic and 1.0% gluconic acids in musts of diseased grapes.
Sound vintages contained none or no more than 0.03 % of glucuronic acid.
The acids were formed st the expense of glucose and fructose. Peynaud
and Charpenti6 (1953) developed a colorimetric procedure for gluconic
acid. I n red wines from sound grapes they found only traces; however,
thirty-seven Bordeaux white wines, all of which were produced from
grapes with more or less Botrytis cinerea, contained 0.29 to 2.46 g. per
liter (average 1.02). Other organic acids probably will be disclosed by
application of suitable techniques. See also p. 465.
8. Volatile Acidity (Acetic Acid)
The primary volatile acid present in wines is acetic. The term volatile
acid, a rather loose one, refers to their volatility with steam. Usually
formic, butyric, and possibly other fatty acids are included as well as
acetic, but lactic is not.
The determination is important, since the amount permitted in com-
mercial wines is limited by law. The limits in this country are lower than
those of most European countries. The Federal limits are 0.120 g. per
100 ml. as acetic and exclusive of sulfur dioxide for white table and dessert
wines, and 0.140 for red table wines. Michel (1948a) proposed a limit for
free and esterified volatile acidity of 0.187% (as acetic). Whether it is the
acetic acid or ethyl acetate which gives a spoiled taste will be discussed
under esters (p. 430). Under any circumstance a high volatile acidity
usually means a high percentage of ethyl acetate. A review of the volatile
acidity as a quality factor was given by Kramer (1941a).
Methods. The determination is complicated by the presence of carbon dioxide and
sulfur dioxide and by the volatility of lactic acid. The volume of work is due not only
to the analytical difficulties but t o the desire to secure a simple and rapid procedure.
Hugues (1930) and Ketelbant (1936) summarized the common methods used in
France. Fonzes-Diacon and Jaulmes (1932) discussed the merits and defects of four
common procedures, preferring their special rectification column procedure to S6mi-
chon and Flanzy’s (1931a) method. They stressed that changes after sampling must
be avoided and recommended salicylic acid to prevent changes during storage. A
theoretical discussion of the distillation of various volatile fatty acids was made by
Foucy (1932). His empirical procedure possibly does not take into account the varia-
tions encountered in practice. Astruc and Caste1 (1932a) recommended for ordinary
control work no defecation with alkali and distilling lO/llths of the volume of the
diluted wine-possibly an oversimplification. Shmichon and Flanzy (1931a) found the
principal sources of error to be: production of acids by pyrolysis, entrainment of lactic
acid, and retention of free acid by the extract of the wine. Jaulmes (1935b) could not
confirm the latter. However, individual components of the extract may reduce or
increase the volatility of acetic acid. For example, organic acids and sugars increase it,
whereas glycerol decreases it. He attributed the decrease in acidity in SBmichon and
Flanzy’s experiment to growth of molds on the wine. Defecation with calcium oxide
does not change the volatile acidity, or with wines it may increase it, owing to produc-
tion of acetic acid. Addition of starch or gum arabic did not change the volatile
acidity. The disadvantages of defecation with calcium oxide prior to the determination
of volatile acids were studied by Jaulmes (1935a). High results owing to formation of
formic and acetic acid from sugars were noted. A sidelight on the current methods for
the determination of volatile acidity by steam distillation is Pozzi-Escot’s (1938)
claim to priority.
Jaulmes (1934, 1951, 1952) has made a valuable theoretical and experimental
study of the distillation of volatile acids. He considered the volatile acidity to consist
of formic, acetic, propionic, butyric, and its higher homologs, such as isovaleric and
capric. With the usual steam-distillation procedure, the results are frequently low,
owing to dilution of the sample during distillation. He recommended adding a crystal
of tartaric acid.
A detailed study of the influence of various factors affecting the rate of distillation
of acetic acid from wines was made by Ionescu et al. (1933) and Ionescu and Popescu
(1934). The speed of distillation is important; with a fixed speed the concentration of
acid, the volume being distilled, and the amount distilled are also important. They
recommended steam distillation of 10 ml. of wine, collecting 120 ml. of distillate in
50 minutes. The 50-minute limit was proposed owing to the slight volatility of lactic
acid. The amount of lactic acid distilled depends on its concentration and the speed
of the distillation. For the same speed of distillation and concentration in the wine, the
volume of wine and the volume distilled governs the amount of lactic acid distilled.
Acetic acid increased the volatility of lactic acid. Usually less than 0.005% of the
volatile acidity is due to the lactic acid entrained during distillation. The conditions
of distillation should always be stated-they prefer a modified Saunier-Gazenave tube.
Mechanical entrainment must, of course, be prevented.
A collaborative study of the Association of Official Agricultural Chemists pro-
cedure by Joslyn (1938b) revealed considerable discrepancies, probably mainly from
differences in technique. Various theoretical problems were discussed. Further reports
on collaborative analysis by Joslyn (1939) also showed variable results, but variation
in the type of still did not seem to be important. He recommended further study of
distillation methods. In a study of Peynaud’s (1937a) procedure, Joslyn (1940b)
showed that the difference between the methods was not due to lactic acid. He sug-
gested that the extent of neutralization and the period of contact may influence
results by Peynaud’s procedure with wines of high sulfur dioxide content. Bringing
the pH to 8 rather than 7 might be useful in decomposing more of the bound sulfur
dioxide.
The micro-procedure of Ghimicescu (1935d) utilized steam distillation of 15ml. of
wine. A feature of his apparatus, found in recent American models also, is a method
for removing the spent sample. See also Egorov (1951). Pavelka and Montini (1948)
used a modified Pregl apparatus distilling 65 ml. from 10 ml. of wine. This appears
to be too small an amount of distillate; however, the results obtained were very similar
to those with the method used in Europe where 200 ml. are distilled from 50 ml. of
wine. The American practice of distilling 100 ml. from 10 ml. appears better. Use of
superheated steam (260' C.) (500" F.) for distilling was proposed by Amadio and
Paronetto (1936). They obtained values close to those of the official Italian distillation
procedure.
Addition of sodium chloride prior to distillation was recommended by Jeanp6tre
(1931) in order to increase the percentage distilled in the first 100 ml. Modified
Caaenave apparatus for the determination of the volatile acidity were developed by
G a d (1941) and Miconi (1952b). The latter's improvement is the large trap with pro-
vision for rectification. A simplified formula for correcting for sulfur dioxide is also
given. An all-glass apparatus and a n adequate rectifying and scrubbing device t o
retain lactic acid were employed by Colombier and Clair (1938). They recommended
a-napthalein as an indicator. Picozzi's (1947) apparatus was inadequately described
but apparently allowed sufficient rectification to prevent appreciable distillation of
lactic acid and had a special anti-spray trap.
Rather than distilling 100 ml. from 10 ml. of wine several short cuts have been
proposed. Foucy (1932) used the 100-ml. distillate from 50 ml. of wine for determining
the volatile acidity. Although he found a fairly constant fraction of the acids to distill
from synthetic solutions, this is by no means true of wines of very variable composi-
tion. Direct distillation of 72% of the wine and doubling the titration value has been
tested by Violante and Imbrici (1949). On 34 of 35 samples their results were within
f0.007% of those by steam distillation and the usual difference was f0.003. Procopio
(194%) also distilled a fixed amount of the wine and applied an appropriate correction
factor. Rentschler and Simmler (1949) proposed distilling 5 ml. of wine after addition
of tartaric acid and tannin. The apparatus is simple, and the determination requires
only 7 to 10 minutes.
Palieri (1952) distilled two 7.5-ml. portions from 20 ml. o iwine. The second
7.5-ml. portion was titrated and contained about 60% of the acetic acid, largely free
of sulfur dioxide or lactic acid. I n general, the amount of acetic acid present in the
distillate was a function of the volume distilled, and a curve, y = $2, where y is the
total amount present and z the per cent distilled, could be constructed.
The importance of carbon dioxide in the volatile acid determination was em-
phasized by Marsh (1936). He recommended boiling the distillate for 1 minute and
titrating hot. Removal of carbon dioxide is more difficult from sparkling wines. Pato
and Salvador (1949) titrated the volatile acid distillate to a p H of 6.33 and then to 9.
At a p H of 6.33, n (ml. 0.1 N NaOH) = 0.79A +0.5C, where A is the volatile
acidity expressed as acetic acid and C represents the bicarbonate content. At a p H of
9, N (ml. 0.1 N NaOH) = A + C. From this A =.",.,":'- They used bromocresol
and phenolphthalein as indicators.
An early suggestion t h a t entrainment of lactic acid may occasionally introduce
errors was made by Fonzes-Diacon and Jaulmes (1932), who recommended a rectifying
column and rapid distillation to avoid the error. In order to prevent interference of
lactic acid Ferr6 and Archinard (1935) made a double distillation. To reduce the lactic
acid error Jaulmes (1951, 1952) employed a rectification tube, a 60-cm. Vigreux
column (two theoretical plates). To reduce the carbon dioxide error, he used carbon-
dioxide-free water.
Marcille (1934, 1935) proposed correcting for the sulfur dioxide in the distillate
by iodine titration before and after desulfiting and gave an empirical equation for this.
He admits that Jaulmes's (1934, 1951) procedure of determining both the free and
410 MAYNARD A . AMERINE
combined sulfur dioxide is more practical and yields better results. Farrugia (1942)
also recognized the dual nature of the correction to be made on the volatile acid distil-
late owing to free and bound sulfur dioxide, but Fabre and Brkmond (1931b) in com-
paring several methods corrected for only the free sulfur dioxide, as did Ferr6 and
Archinard (1935). Bertin’s (1934) procedure of refluxing the wine to destroy the sulfur
dioxide has been criticized as too long (30 minutes). Peynaud (1937a) neutralized to
pH 8.5 with barium hydroxide, distilled, and corrected the acidimetric titration by
subtracting one-half the sulfur dioxide found in the distillate. Carbon dioxide was
removed by shaking under vacuum. Tarantola (1949) tested various procedures and
recommended that of Jaulmes (1934), since those of Peynaud (1937b) and Marcille
(1934) gave high values. Although the interference of sulfurous acid in the determina-
tion of the volatile acidity is recognized by all, many analysts fail to make a correction
for it or do not realize how much influence it may have. Procopio (1950a, b) has
demonstrated how serious this omission may be with Italian wines. He distilled a por-
tion of the wine and titrated. This distillate contains most of the sulfurous acid. The
remaining liquid is brought to volume and distilled, and the distillate is titrated. The
difference between the two represents, roughly, the sulfurous acid. At best the method
seems too empirical, though simple and rapid. Mestre and Campllonch (1935) com-
pared the results of several methods. They recommended dividing the volatile acid
distillate into three parts, determining the total acidity, free sulfur dioxide, and total
sulfur dioxide. The free sulfur dioxide plus one-half the total less the free is calculated
as acetic acid and subtracted from the acetic as grams per liter. To reduce the sulfur
dioxide Politova-Sovzenko and Dikhtyar (1948) recommend passing about 10 1. of
air through 25 ml. of wine within an hour. Oxidizing the sulfurous acid was proposed
by Perazzo and Arbecchi (1933) and Leggieri (1951). The latter treated the wine with
hydrogen peroxide and barium chloride.
Origin. The most rapid period for formation of acetic acid during
fermentation appears to be during the initial stages, according to Sal-
varezza (1935-1937). Joslyn and Dunn (1941) and RibBreau-Gayon and
Peynaud (1946b) confirmed this. The latter noted that after passing
through a maximum more than half disappeared, varying with the yeast
strain employed. Peynaud (1938~)observed that when acetic, propionic,
or butyric acids were added to musts before fermentation only a slight
amount of higher alcohol was present with acetic and considerable with
the other acids-indicating their reduction during fermentation to their
corresponding alcohols. The amount of acetic acid reduced was also
shown by Peynaud (1939-40) to vary with the rH, the phosphate content,
temperature, and the acidity of the media. He envisages the following
scheme :
CHsCHO
Uchimoto (1951), however, found no correlation between volatile acid

formation and the temperature of the fermentation. Joslyn and Dunn
(1941) reported that the yields of formic and acetic acid in various
atmospheres were as follows (grams per 100 ml.) :
Gas Formic Acetic
Air 0.0489 0.0686
Oxygen 0.0273 0.132
Nitrogen 0.0162 0.0532
Carbon dioxide 0.0308 0.0492
Control 0.0289 0.0292
The effect also varied with the stage of fermentation. Activity of yeast
dehydrogenase in oxidizing aldehydes, alcohol, and acetic acid are related
to the changes in volatile acidity, but association or competitive action on
these substrates and the influence of other acids may be important. The
hypothesis of Joslyn and Dunn (1941) that acetic acid is formed by
oxidation of ethyl alcohol was questioned by Peynaud (1947~).He
postulated that acetic acid is formed by dismutation of acetaldehyde
through the action of aldehydomutase. He also found that if dimedon was
added to combine with the aldehyde, very little acetic acid was formed.
High volatile acidity is not always due to Acetobacter. Cappucci (1948)
found volatile acidities of from 0.104% to 0.369% in new 1946 and 1947
wines. He attributed this to Saccharomyces apiculatus (KloeckeraQ)and
Zygosaccharomyces sp. The rapidity with which volatile acidity can
develop during fermentation under warm weather conditions was shown
by Cruess (1936). He recommended 100 p.p.m. of sulfur dioxide to control
the spoilage. Burdzhanadze (1951) showed that the alcohol lost during
storage (owing solely to oxidation to acetic acid) was greater than could
be accounted for by the acetic acid formed. Evaporation, decomposition
of acetic acid, and oxidation of alcohol to other products probably
account for the difference.
Since excess volatile acidity is prohibited by law, its reduction is of
great interest. Two procedures have been proposed : refermentation or
use of film yeasts. Ventre (1937), Peynaud (1938c, 1939-1940), and others
have shown that acetic acid is reduced to alcohol during fermentation and
that the lower the r H value, the greater the amount reduced. Various
yeasts had different reducing properties. SBze (1938), Jaulmes (1952), and
earlier workers had suggested the addition of high volatile wines to
fermenting musts. An objection to the first procedure is the possible
spread of infection and the spoiling of a potentially good wine.
The other procedure, use of a film-forming yeast, has been recom-
mended by Marcilla et al. (1936), Cruess et al. (1938), Schanderl (1943),
Bobadilla (1943), and more recently by Florenzano (1952), who found
that both white and red wines could be so treated. The concomitant loss
of alcohol was decreased by reducing the surface/volume ratio, but color
and tannin were lost during the process. The claim that the organoleptic
quality was increased should probably be considered only in relation to
the removal of acetic acid. Peynaud (1938c), for example, objected to the
use of film yeasts because of accumulation of aldehydes.
Both procedures assume that a wine once unfit for human consump-
tion can thereafter be made fit for human consumption. Generally public
health authorities forbid or frown on such practices. Possibly ion-exchange
procedures might escape the legal and flavor problems. Patterson and
Bawtenheimer (1930) found that deacetization with magnesium car-
bonate occurred when all the acids, fixed as well as volatile, were neutral-
ized. Magnesium acetate is only slightly volatile in the absence of a
hydrolyzing agent.
9. Other Volatile Acids
SCmichon and Flanzy (19314 developed methods for formic, butyric,
and propionic acids based on Duclaux’s method.
Formic Acid. Hohl and Joslyn (1941a) concluded that formic acid was
“not a final by-product of alcoholic fermentation of sugar” by the
strains of yeast tested by them. The indications from their data are that
little would be formed at the high pH of California grape juice and that
utilization during fermentation was greater than formation.
Seifert and Ulbrich (1930) found 0.023 to 0.089 g. of free formic
acid per liter in 24 Austrian and Hungarian wines, a variable percent-
age being esterified. Some of the formic acid may have originated from
the long storage on the lees-decomposition of leucine being a possible
source: CH(CH3)&H2CHOHCOOH = CH(CHJ2CH2CHO HCOOH. +
The ratio of free formic acid to the volatile acidity was very variable.
Villforth (1950-1951) reported about 50 mg. per liter in nine normal
German table wines but found 270 and 460 mg. per liter in two Italian
red wines. His procedure, based on a Duclaux separation, steam-distilling
250 ml. from 50 ml. of wine, however, gave a low but constant recovery
of 44 to 46% with from 12.5 to 36.4 mg. Only dry wines can be used, as
there is evidence for the formation of formic acid from invert sugar
during steam distillation. He found the mousy taste of a gooseberry wine
to be associated with a high formic acid content. Actually he stated that
a polymerization of formaldehyde and acetaldehyde would produce the
mousy odor. Hohl and Joslyn (1941a) used a steam-distillation procedure
for formic acid but found lactic acid to interfere. They therefore pre-
ferred the total-extraction mercury-reduction procedure of von Fellen-
berg (1936).
Butyric Acid. The presence and methods of determination of butyric

acid in biological materials was reviewed by Grossfeld and Battay (1931).
Detection of this acid in dilutions of 1/12,500 was reported. I n five wines
they reported 0 to 0.213 mg. per liter, average 0.088, using a potassium
perchlorate procedure. Miermeister and Battay (1931), using the same
procedure, found 0 to 0.60 mg. per liter, average 0.033, in seven authentic
Greek sweet wines. I n six Greek muscat wines, in which sophistication
with the carob bean (St.-John’s Bread, Ceratonia siliqua) was suspected,
0.080 to 0.250 mg. per liter, average 0.143, were found. Berg and Schulze
(1932) reported relatively large amounts of butyric acid in fermented
beverages containing carob bean. Only small amounts were found in
various European dessert wines. However, they considered it a normal
constituent of sweet wines. A sensitive procedure for butyric acid was
developed by Klinc (1935). In fifteen Jugoslavian wines 10 to 20 mg. per
liter were reported. The wines were sound table wines of 9% to 11%
alcohol and 0.06% to 0.08% volatile acidity. Wine vinegars contained
up to 290 mg. per liter.
10. p H
The color, taste, clarity, and resistance to disease of a wine are closely
related to its pH.
Methods. Morani (1930),Geloso (1931),Genevois and RibBreau-Gayon (1933),and
RibBreau-Gayon (1938b)have given reviews of methods for determining the pH, the
latter preferring the glass electrode. Von der Heide and Miindlen (1930)gave a good
review of the methods available at t h a t time. A comparison of the results obtained on
four wines by five procedures follows:
Wine
Method 1 2 3 4
Quinhydrone electrode 3.51 3.35 2.83 3.29
Hydrogen electrode 3.54 3.43 2.88 3.34
Diazoacetic acid ester
hydrolysis 3.48 3.37 2.82 3.31
Sucrose inversion 3.52 3.38 2.84 3.31
Colorimetric 3.38 3.26 - 3.26
Sulfur dioxide interfered with the hydrogen electrode procedure. They considered the
quinhydrone method best, but, of course, the glass electrode was not easily available
to them. Joslyn and Marsh (1935)showed that the hydrogen electrode was poisoned
unless protected by an atmosphere of hydrogen. Measurements of the p H on de-
alcoholized samples were less than on the original wine.
Fornachon (1946)has also reviewed the various procedures and found differences
of 0.1 pH unit and more between the quinhydrone and glass electrodes, with the glass
electrode values being more nearly correct. The quinhydrone electrode gave lower
values because of the alcohol, but rsducing substances, such as sulfites and tannin,
may result in high values. The compensation of the two errors results in approxi-
mately correct results in some wines. RibBreau-Gayon and Peynaud (1937b) and
Peynaud (19370) have applied the influence of pH on the hydrolysis of acetal as a

method for the determination of the pH. The results for 26 samples by the acetal
procedure averaged 3.19 and by the glass electrode 3.17.
An ingenious method for the determination of pH was devised by Bremond
(193813). It consisted of two freshly cleaned platinum electrodes connected with an
electrometer and with the two solutions by a salt bridge. One electrode was placed
in wine plus quinhydrone and the other in potassium acid phthalate with quinhydrone.
Sodium hydroxide was then added until the sensitive millivoltmeter, or capillary
electrometer, showed no current. A table showing the pH for various amounts of
sodium hydroxide added to the potassium acid phthalate solution was given.
Whereas the glass electrode gives results of adequate sensitivity in fortified wines,
the influence of alcohol on the activity of the hydrogen ion is appreciable. For wines of
19% to 22% (vol.) of alcohol Ribeiro (1938) found the true pH to be about 0.1 pH
unit below the observed value.
A method of capillary analysis was adapted to determine the pH by Boutaric and
Bouchard (1935). The method is only an approximate one, since they diluted the
wines 1/100. A pH indicator consisting of 2 volumes of saturated (aqueous) benzyl
orange, 2 volumes of saturated (aqueous)a-dinitrophenol,and 5 volumes of 0.26 %
(alcoholic) bromophenol blue was proposed by Marcilla and Feduchy (1943). The
",:,,
range waa from 2.8 to 4.6 with colors from orange to brown to gray to blue violet and
a sensitivity of about 0.2 pH unit.
Roussopoulos (1930) used the following formula for calculating the pH based on
inversion of sucrose: pH = 2.19 - log A'(A - a t 35" C., where A is
log A / ( A - X') - -
the invert sugar (10.256) equivalent to 10 g. of sucrose (the amount used), X and X'
the amount of invert sugar formed in equal times, and 100/94 a correction factor (2.19
is the pH of 0.1 N tartaric acid a t 35" C.).
In Grapes and Wine. Crisci (1931) studied the increase in pH during
ripening of grapes and the changes in the juice during pressing-the first
press was lower in pH than the second. Gerasimov (1931) also studied
the pH changes during ripening. He reported that addition of amino acids
during fermentation decreased the pH, but amides and ammonia were
without influence. He also correctly interpreted the changes in pH and
titratable acidity caused by the malo-lactic fermentation.
The importance of pH in the fermentation and stability of the cham-
pagne wines of France was studied by Franpot (1945). He found no direct
relationship between pH and titratable acidity, but he found that the
relatively low pH of these wines was an important quality factor. The
optimum pH for fermentation for a number of wine yeasts was studied
by Burgvits and Hochberg (1936). They found 3.0 to 3.61 to be the best
for speed of fermentation. With low pH musts too much sulfur dioxide
should not be added lest the pH be reduced so low as t o delay fermenta-
tion. The influence of the pH on the rate of fermentation was also investi-
gated by Casale (1930b, c). Whereas a pH of 3.3 to 4.0 favored the rate
of fermentation, the best utilization of sugar occurred a t pH's as low
as 2.8. Lactic acid formation was favored by higher pH values. Decreases
in pH commonly occurred during fermentation, the decrease being

greater the higher the original pH. At pH’s below 2.5 Casale (1930a)
found cell division and alcohol formation to be restricted, possibly be-
(’&useof changes in enzyme activity. When 60 strains of Saccharomyces
ellipsoideus were inoculated into samples of the same must, the resulting
wines had pH’s of from 3.08 to 3.30 in the study of Goes and Correia
(1942).
The spoilage bacteria of dessert wines isolated by Fornachon (1943)
were very sensitive to a pH of below 3.5; this is probably the primary
reason why so many European wines are relatively resistant to bacterial
spoilage. A decrease of pH during growth of the film yeast in sherry pro-
duction was reported by Marcilla et al. (1936). The changes in pH during
the malo-lactic fermentation are summarized by Schanderl (1950).
Influence on Taste. Crisci (1930) showed that pH and acid taste were
more related than pH and total acidity, but that the presence of other
substances, alcohol, sugar, tannin, etc., modified the acid taste. The un-
dissociated organic acids also have an influence-when they are added to
wine the acid taste increases, but the pH remains nearly constant. Like-
wise when a wine is diluted, the pH changes little, but the acid taste is
greatly reduced. Wines vary from one to the other in this respect. Al-
though Gerasimov (1931) and others stated that the acid taste is related
more to the pH than t o the total acidity, in his study this was limited to
old Crimean wines of the same total acidity. The influence of alcohol,
sugar, tannin, etc., made the relation less exact. The general relationship
of acidity and pH to taste and disease resistance is indicated in the follow-
ing data of Biedermann (1951) for Swiss table wines:
Total acid,
% as tartaric pH Taste Disease resistance
1 0 -1 8 2 7 Very acid Very good
0 6 -1.2 3.1 Acid Good
0 4 -0 9 3 5 Average Normal
0 35-0 6 3 8 Flat Poor
0 25-0 4 4 1 Very flat Very poor
Gerasimov (1931), Genevois and RibBreau-Gayon (1933), and Berg
(1939) showed that wines are so buffered that there is no simple relation
between the pH and the titratable acidity. Bremond (1937~)has also
amply demonstrated this. The importance of a relatively low pH to the
appearance and flavor of French sparkling wines was emphasized by
Frangot (1945). The possible influence of pH on various chemical changes
in wines during aging was stressed by Gentilini (1952). The importance
of the pH to the color and disease resistance of port was stressed by
Ribeiro (1938). The pH of a potassium acid tartrate-10 %
’ alcohol solution
is about 3.5. Biedermann (1951) has shown that in wines of below 3.5,
precipitation of potassium acid tartrate causes a decrease in the pH,
whereas in wines of above 3.5, an increase in pH results. Dissolving the
salt in wines causes the opposite changes. Under winery conditions the
actual amount of pH variation with bitartrate precipitation is small and
of little importance to the taste and biology of either grape juice or wines.
TABLEXI11
pH of Various Types of Wines
Alsace Table Fabre and Br6mond (1932) 13 2.96 3.30 3.11
California Wh. table Amerine (1947) 187 3.03 3.89 3.44
California Red table Amerine (1947) 282 3.05 3.97 3.54
California Dessert Amerine (1947) 315 3.14 4.40 3.82
Croatia Table Peretid (1950) 104 3.03 4.33 3.07
Croatia Table * Suc6vid (1950) 136 3.10 4.40 3.67
France Table Peynaud (1947~) 103 2.83 3.73 3.28
Italy Wh. table Dalmasso and Dell’Olio 145 2.90 3.50 3.10
(1937)
Portugal Dessert t Ribeiro (1938) 13 3.60 3.90 3.70
Portugal Dessert $ Ribeiro (1938) 20 3.50 3.85 3.64
Portugal Dessert 0 Ribeiro (1938) 17 3.50 3.75 3.61
Portugal Table Correia (1943) 264 2.94 3.91 3.56
(1952)
(1952)
* Wine grape varieties.
t Experimental dessert wines.
$ Red ports for export.
0 White ports for export.
Crisci and Michielini (1932) demonstrated that small changes in pH
had an influence on the efficiency of decolorizing charcoals and the
clarifying properties of gelatine. In both cases the efficiency was greater
a t higher pH’s of about 3.4 to 3.8 compared to lower (to 2.73) and higher
(to 5.97) values. The influence on the pH of the wines of addition of 0 to
520 p.p.m. of sulfur dioxide to musts was indicated by the results of Pato
and Sousa (1938). They found the pH of the resulting wine to vary from
2.87 to 2.54 following addition of 0 to 520 p.p.m. Such low pH values are
not common in northwestern Europe or in this country.
Gerasimov (1931) reported in 202 Crimean wines a pH range of 3.0
to 3.93 with 84% of the samples between 3.2 and 3.6. In 14 wines of 27
or more years of age the pH was higher, 3.40 to 3.88. The pH of 42 Italian
wines was reviewed by Morani (1930) and varied from 2.92 t o 3.60. Other
European wines were as low as 2.65 and as high as 3.78. The pH’s of
various types of wines are summarized in Table XIII.
Bufer Capacity. The buffer capacity, dB/dpH, where B is the milli-
liters of 0.1 N sodium hydroxide added per liter, was used by Morani
(1930) in his studies to detect addition of mineral acids. I n Bordeaux red
wines it varies from 35 to 45, according to Genevois and Ribereau-Gayon
(1935a). In sweet wines at pH’s of 9 and over alkaline glucosates are
formed. They concluded that the buffer capacity was of little analytical
interest but noted that the buffer capacity to a pH of 4 was a measure of
the organic acid content of the wine. I n red wines the coefficient to a pH
of 10.5 is a measure of the phenolic compounds and appears to increase
with the amount of color. In white wines it is a function of increasing
sugar content. In Bordeaux wines with pH’s of 2.7 to 3.5 no relationship
of the pH to the quality or origin of the wine was found.
The extraordinary buffer capacity of wines has been observed by
many investigators. Rippel (1949) noted the difficulty in changing the
pH, either by biological or chemical means, but Schanderl (1950-1951)
believed his results with calcium carbonate might be in error, since
Schanderl found an increase in pH from 3.4 to 4.2 when the titratable
acidity was reduced from 0.78 % to 0.38 %. Rippel (1949) showed that the
buffer capacity increased during the alcoholic fermentation-owing to
absorption of buffer substances by the yeast. Similar changes due to the
malo-lactic fermentation were also reported, but no detailed studies on
different types of wines have been made. Differences between wines of
warm and cold seasons suggests that such studies should be made.
The buffer capacity of wines is largely owing to the buffer capacity of
the organic acids they contain. The following data of Kramer and
Bohringer (1940) are for 0.5% solutions of four acidic substances and a
mixture; they show the change in pH with dilution with water:
48.5 mole % tartaric
Dilution, Potassium acid + 51.5 mole %
% Tartaric acid tartrate potassium acid tartrate Mnlic Lactic
0 2.25 3.52 2.97 2.40 2.51
5 2.25 3.50 2.97 2.40 2.51
10 2.26 3.51 2.93 2.42 2.52
15 2.28 3.51 2.91 2.44 2.54
20 2.30 3.53 2.90 2.46 2.56
30 2.33 3.5L 2.91 2.48 2.58
50 2.42 3.56 2.96 2.55 2.65
75 2.61 3.63 - 2.73 2.86
87.5 2.81 3.68 - 2.94 3.02
These results show that the pH changes least with a mixture of tartaric
acid and potassium acid tartrate. Addition of sugar, alcohol, or gelatine
generally raised the pH values and on dilution, even up to 50 %, the pH's
remained higher. Morani (1930) correctly noted the influence of strong
acids on the buffer capacity of wines. Morani and Marimpietri (1930) gave
10 W 30 40 50 Bo 70 0 90 100110 120 130140 I 5 0

Mllliequlv. NaOH adCd
FIG.2. Titration curve and buffer capacity of a wine. The total acidity is equal to
about 107 meq. (after Vergnes from Jaulmes, 1951).
further data on this point. They found the buffer capacity to be very
variable between various wines. The buffer capacity of grape juices and
wines was studied by Berg (1939). The relation of the titration curve and
the buffer capacity is indicated in Fig. 2. The increase in buffer capacity
above a pH of 9 is owing to tannins, coloring material, and other poly-
phenolic compounds.
VI. CARBOHYDRATES
Fructose and glucose are the chief sugars present in grapes. Pentoses
and sucrose have also been found. Related compounds present in wines
include pectins, dextrins, caramel, and mannite. The sum of the dissolved
non-volatile material in wine is called the extract.
1. Hexoses
The alcohol of wines is derived from the fermentation of sugars. Little
new information on the types and concentrations of sugars present has
been reported in the period under review.
Methods. The modified Lane and Eynon procedure with Soxhlet's reagent is now
widely used for the determination of reducing sugars in wines. The main problem is
the removal of interfering nonsugar materials. More experiments on recovery of added
sugar should be employed, as suggested by Joslyn (1950). Ribeiro (1946) recommended
addition of lactic acid and use of subacetate of lead as the clarifying agent. He recog-
nized that basic lead acetate alone gives low results (owing to fructose absorbed in the
basic lead precipitate); hence the addition of lactic acid, which lowered the p H and
did improve the recovery to nearly 100%. The SBmichon and Flanzy procedure of
defecating the wine with a mercuric salt was criticized by Pluchon (1937) as giving
high results with wines containing sucrose or pectins. He clarified with mercuric
sulfate and then with alkali and zinc powder. The procedure seems not to be used
nowadays though the results reported were good. Jaulmes (1951) prefers mercuric
acetate. Cunha and Ribeiro (1945) studied various volumetric and gravimetric pro-
cedures for determining reducing sugars in dessert wines. The Lane and Eynon pro-
cedure gave the best results, and since there is usually more fructose than glucose in
dessert wines expressed the results as fructose. A modification of the Lane and Eynon
procedure for the rapid determination of reducing sugars in musts was proposed by
Gentilini (1950). Satisfactory agreement with gravimetric procedures was obtained
on eighty-five wines.
A volumetric copper procedure and the usual copper gravimetric method on wines
gave comparable results, providing details of the volumetric procedure were exactly
followed, according to Kalberer (1930). Geiss (1938) proposed using a Pulfrich
photometer to measure the blue color remaining after the reaction of Fehling's solution
and the sugar. Lehmann (1940) used volumetric and gravimetric procedures with
Fehling's solution, and Ghimicescu (1937) gave a micro-method. Von Fellenberg
(1932b) found differences between his iodometric procedure and the gravimetric
copper method, where wines contained reducing substances not precipitated by lead
acetate.
JiinEnez and Mendivelz6a (1939) used a mixed reagent of alkaline copper sulfate
and potassium ferricyanide. The sugar solution was added while boiling until the
color turns green, methylene blue being used as the indicator. The method is probably
overly sensitive, as are most procedures employing ferricyanide. For example, while
GCorgacopoulos and Costopoulos (1952) favored the ferricyanide method for deter-
mining the sugar in musts and wines, it gave appreciably higher results than the copper
p r o c d i r e s when the sugar content was above 6%. Vartan'gn (1951) also employed
alkaline ferricyanide and used methylene blue as the indicator. Amerine (1952) used
ferrous orthophenanthroline as the indicator. Alexis (1933) proposed determining the
residual sugar of red wines by heating to 100" C. (212" F.), adding sodium chloride and
acetic acid, making alkaline with carbonate, adding iodine and titrating the remainder
after 2 hours. Just what was determined by this procedure is not clear but certainly
more than the residual sugar. Kielhofer (1953) calculated the approximate sugar
content of normal wines from data on the specific gravity of the must and wine and
gave tables to facilitate the calculation. A qualitative procedure was proposed by
Trauth (1949). This consisted of extracting with ether and petroleum ether and
tasting the residue.
Sidersky (1942) determined the total reducing sugar and the polarization. He
reported the polarizing value times 0.1122 as “active” fructose and the total sugar
less the “active” times 0.3548 as “inactive” fructose. The difference between the total
sugar and sum of the “active” and “inactive” fructose was reported as glucose. A
simple procedure for determining fructose in grape juice before and during fermenta-
tion and in sweet wines was presented by Prillinger (1952~).It is based on heating
2 hours at 50” C. (122” F.) in an alkaline copper sulfate solution. He added sodium
chloride to precipitate coloring material and other suspended material that might
react with the iodine-added to determine the excess copper.
The value of the Zeiss refractometer for determining the soluble solids content of
must in comparison with a Mohr specific gravity balance or with the Ochsle hydrom-
eter was indicated by Buxbaum (1932) and Gerum (1932). Kramer (1935) stated that
a t least ten single berries should be used in determining the sugar content of the must
by means of a hand refractometer. Actually many more berries should be used (see
Amerine, 1952). The utility of the Zeiss refractometer in determining per cent sugar
in the grapes was shown by Zweigelt (1938). He also indicated that single berries were
not suitable samples. The conversion of refractometer readings to Ochsle degrees is
explained by Bohringer (1943). The equation x = 4.25y, where y is the refractometer
reading and z the degree Ochsle, fits the data quite well. He recommends the hand
refractometer for field studies, particularly with hybrids. Palieri (1951) has proposed
it also as a rapid means of determining the sugar content of sweet wines when the
alcohol content is known. Although his formula apparently works fairly well for the
sweet table wines of Castelli Romani (near Rome), it is not applicable to California
dessert wines.
The hydrometer designed by the Klosterneuberg station, near Vienna-commonly
called the Babo hydrometer-reads per cent sugar. Gentilini and Carli (1950) used this
on 390 Venetian musts on which the sugar content was determined by Fehling’s
solution. The degree Babo read about -0.13 below the true value but was so variable
that they recommended that it not be used. Barini-Banchi (1952) has also shown that
the Babo hydrometer usually gives low results in comparison with copper-reduction
procedures-no doubt owing to the variable amount of nonsugar solids in the musts.
He also found that the refractometer reading less 3 is also usually lower than the
chemical results.
Calculation of the sugar content of dessert wines when the alcohol and specific
gravity are known was reported by Berg (1932). A table of the Wine Institute (see
Amerine, 1952) gives similar data. Procopio (194813) has developed nomograms to
facilitate calculating the sugar content from the density, alcohol, and extract con-
tents. On seventeen wines the calculated sugar varied from -0.17 to +0.90 from that
determined.
The problems of quantitative analyses of sugar in concentrate were indicated by
Garoglio and Barini-Banchi (1940). In general there was slightly more fructose than
glucose in the concentrate, Good agreement between the soluble solids calculated
from the refractive index and those calculated from the density were obtained.
Source. The fructose and glucose content of ripening French grapes

wm measured by Alexis (1936) at intervals of three to four days and the
COMPOSITION OF WINES 42 1
glucose/fructose ratio decreased from 1.40 to 1.01. In an extensive study

of the fructose and glucose contents of Hungarian grapes Szab6 and
Rakcsinyi (1935) showed that a t the beginning of the ripening of the
grapes glucose predominated. During ripening fructose increased more
rapidly than glucose, until in overripe grapes it amounted to 56% of the
total reducing sugar. Fructose equaled or exceeded glucose when the total
sugar was about 17%. Drying or cooking of grapes did not change the
ratio. This is surprising when one considers the heat sensitivity of fructose.
In unripe grapes Sidersky (1942) reported a glucose/fructose ratio of
above 1, but this varies with variety and origin. Espinosa (1943) showed
the glucose to decrease from 86 % of the total reducing sugar to about 47%
during ripening. In thirty varieties the ratio a t maturity varied from 0.87
to 0.96, average 0.90.
The increase in per cent sugar in grapes attacked by Botrytis cinerea
is well known. SisakGn and Marutfan (1948) reported that the glucose/
fructose ratio generally decreased during ripening, but their data were
variable. For a recent summary of the process see Schanderl (1950).
I n Fermentation. While most yeasts ferment glucose more rapidly
than fructose, not all do so. The fact that the Sauternes’ strain of yeast
ferments fructose faster than glucose has been known for some time.
Normally glucose ferments more rapidly, even though fructofuranose has
an affinity for hexokinase twice that of any form of glucose. The explana-
tion, according to Gottschalk (1946), is that the cell walls of this strain
are more permeable t o fructose than to glucose. It is strange therefore
that Kniphorst and Kruisheer (1937) should find a genuine French
Sauternes with 72.4% of its residual sugar as fructose. Straub (1934)
reported the fructose content of seven Sauternes to amount to 66.7%
of the total sugar. In the drier neighboring wines of the Graves region
seven samples averaged 55.7% fructose. In nine fortified wines, such as
port, the fructose amounted t o 50% to 65% of the total sugar content.
Clavera and Oro (1932) reported the glucose to exceed the fructose in
Mhlaga wines. Dubaquie and Debordes (1935) noted that with certain
yeasts when the must was treated with monobromacetic acid nearly all
the fructose fermented before the glucose. A yeast isolated from concen-
trate or previously treated with monobromacetic acid behaved in the same
way. They also reported data on the production of alcohol and the per
cent residual glucose and fructose for several yeasts.
The most extensive study of the changes in fructose and glucose
during fermentation appears t o be that of Szab6 and Rakcsinyi (1935).
They showed with Hungarian musts that the glucose was fermented much
more rapidly than the fructose in musts of 17% to 20% total reducing
sugar. Both sugars fermented a t the same rate a t 20% to 25% sugar, and
glucose fermented less rapidly than fructose a t higher per cent sugars.
This accounts for the fact that there is more glucose than fructose in the
very sweet Tokay essence type of wine. The presence of added alcohol did
not influence the rate of fermentation. Since concentrate has an equal or
higher fructose than glucose, imitation wines, prepared by adding con-
centrate to the dry wine, did not show the expected high dextrose content.
Espinosa (1943) obtained similar results in Argentina. During fermenta-
tion the glucose/fructose ratio declined from about 0.9 to 0.2.
Although grapes seldom contain enough sugar to slow down fermenta-
tion, Mathieu (1938) reported such an instance for the RhBne region of
France.
I n Wines. The dangers of excessive residual reducing sugar in red
wines were considered by Dubaqui6 and Debordes (1931). A rapid pro-
cedure for the detection of excess sugar was developed. They considered
0.2% the permissible maximum limit and recommended chemical analyses
of the new wines rather than tasting.
An increase in residual sugar during the first six years of aging was
noted by Perre and Michel (1947). This they attributed to hydrolysis
of glucosidesmainly of the oenin. The enzymes responsible were
reported more prevalent in some years than in others. In two dry Swiss
table wines Godet and Martin (1946) reported 0.026% and 0.024%
glucose and 0.036 % and 0.022 % fructose, respectively.
I n three ports Muttelet (1930) reported the following percentages of
fructose and glucose, respectively: 5.15 and 3.97, 5.38 and 4.02, and 5.40
and 4.00. The excess of fructose indicates either that the grapes were
harvested late, when fructose was in excess, or that the yeasts employed
fermented glucose more rapidly than fructose, or both. Ribeiro (1938)
found the glucose to exceed the fructose in other Portuguese dessert
wines, as the following data for the glucose/fructose ratio indicate:
No. of samples Minimum Maximum Average
13* 0.3 0.6 0.4
20 t 0.6 0.8 0.7
17$ 0.4 0.9 0.7
* Experimental dessert wines-lower sugar than oommeroial samplea.
t Red ports for export.
t White ports for export.
No differences in flavor, aroma, or carbonation of the dry finished

sparkling wine were observed by Goresline and Champlin (1938) in fer-
mentations with sucrose (cane and beet), glucose, anhydrous glucose, and
commercial invert sirup. When used for the final sweetening dosage,
differences in flavor between the sugars did occur in the wine.
2. Pentoses and Related Compounds

The compounds generally regarded as being present include pentoses
such as arabinose and methyl pentose; rhamnose; pentosans such as
araban and methyl pentosans; and acids such as ascorbic and dihydroxy-
maleic. I n a study of the composition of the residual sugar in seventeen
“dry” table wines von der Heide and Burkard (1935) showed that the
total reducing sugar varied from 0.07% to 0.567%, of which 0.025% to
0.5% was due to fructose and the remainder was largely arabinose.
Actually arabinose appears to have been determined in only four wines,
but the values do agree well with those obtained by subtracting the
fructose from the total reducing figure.
By dealcoholizing and refermenting, Tarantola (1948, 1950b) found
about 0.05% to 0.18% nonfermentable reducing substances which he
found to be pentoses. I n 89 samples of Italian red and white wines he
reported from 0.036% to 0.199% pentoses (as arabinose). White wines
averaged lower than reds, and some pentoses were extracted from the
stems and skins when the must was fermented with them. I n 8 wines
0.020% to 0.038% pentosans were reported. The fermentable residual
sugar in “dry” wines of moderate alcohol content was found to be fruc-
tose. Canals and Collet (1939) on the basis of polarimetric studies con-
sidered that invert sugar, not fructose, was the residual sugar in dry red
table wines. No reference t o the possible arabinose content was made. A
very slight residual rotation was due to tartrate. Torricelli (1941) re-
ported 0.05% t o 0.11% pentoses (calculated as arabinose) in 41 natural
white wines. When contact of the pomace with the wine was prolonged,
an unidentified glucoside was found, called the “ Tresterfaktor.” Using
both the arabinose content and the “Tresterfaktor,” he was able to
identify 20 or 21 wines made from pomace. A micro procedure for the
determination of pentose and colorimetric procedure for the unidentified
glucoside were given. Torricelli (1943) has also determined the arabinose
content of natural Swiss white wines. The range was from 0.43 to 1.14 g.
per liter, the average 0.62. Genevois el al. (1949a) reported only 0.05 to
0.09 g. per 100 milliliter of pentose in 4 sweet Bordeaux table wines.
Peynaud’s (1950a) values for pentose indicate that from a quarter to a
half or more of the residual sugar is pentose. I n 6 red table wines he
reported 0.05 to 0.07 g. per 100 ml. and in 6 sweet white table wines 0.05
to 0.10, average 0.8. Finally, Jaulmes (1951) reported 0.04 to 0.13 g. per
100 ml. of arabinose in 5 red wines of Southern France.
Torricelli (1943) reported less than 0.05% rhamnose in wines and
Jaulmes (1951) gave a range of 0.015% to 0.03%.
3. Sucrose
Sucrose is added of necessity to musts in eastern United States and in
northern European countries to increase the sugar enough so that the
alcohol content of the resulting wine will fall in an acceptable range. The
presence of considerable amounts of unfermented sucrose in sweet wines
is usually considered sophistication. Details of the official German pro-
cedure for the detection of sucrose in wines were given by Jahr (1931).
The particular point made was that the clarification should be the same
for the sucrose determination as for the reducing sugar determination.
He used animal charcoal and lead acetate. A method for detecting sucrose
was devised by Krauze (1933) based on destruction of hexoses with
peroxide and conversion of sucrose to hydroxymethylfurfural. This gives
a blue color with diphenylamine.
In Russian grapes S i s a k h and Marutkn (1948) reported 0.2% to
1.5% sucrose in 26 varieties. They used both acid hydrolysis and ester
hydrolysis. The sucrose decreased in Isabella grapes during ripening, but
the results with the Malbec variety were variable. No specific test as to
the identity of the hydrolysis product appears to have been made. Small
amounts of sucrose was reported in both immature and mature grapes by
Venezia and Gentilini (1935).
Although sucrose may be found in wines made of non-Vitis vinifera
grapes, it is usually found only in traces in wines made of Vitis vinifera
varieties. Muttelet (1930) has confirmed this for three authentic samples
of port wine. I n these he reported a very slight increase in reducing power
following inversion-amounting to 0.024% to 0.038% as sucrose. The
colorimetric diphenylamine test was negative, and there was little change
in polarization following inversion. He concluded that those samples had
not had sucrose added. Botelho (1935) also found no sucrose and normal
glucose and fructose contents in an authentic Portuguese port. I n eight
genuine MBlaga wines Clavera and Oro (1932) found no sucrose. Godet
and Martin (1946) reported 0.011% and 0.015% sucrose in two Swiss
wines. Lobstein and Schmidt (1931) reported 0.0 to 0.15 g. per 100 ml.
(average 0.06) in nineteen Alsatian table wines.
4. Related Compounds
Dextrins. Reducing sugar was determined before and after acid
hydrolysis under pressure by Herrero (1943) and the difference X 0.9
expressed as dextrins. The amounts found were less than 0.1%. Bucci
(1940) gave a technique for demonstrating dextrins in wine.
Caramel. Presence of caramel, a product of sucrose dehydration, is
usually considered evidence of sophistication in wines and its addition is
prohibited in some countries. Many tests for detection of caramel have
therefore been devised. Mastbaum (1933)rejected the resorcinol test for
detecting caramel in wine, as pure muscat wines of Portugal gave the test
(possibly because they contained hydroxymethylfurfural, see p. 385).
Various qualitative tests for caramel were reported by Fetzer (1938). A
procedure for detecting addition of caramel to wine, based on precipita-
tion of the caramel with an ether-alcohol mixture, was developed by
Milos (1942).The test is not quantitative, as some caramels do not pre-
cipitate quantitatively in such mixtures. A quantitative colorimetric
procedure more specific than that of Milos based on use of Lovibond
slides was developed by Mallory and Love (1945).Coal tar and vegetable
dyes, wool extractive matter, and natural colors did not interfere. The
chief advantage of their procedure is that the caramel is isolated in a
relatively pure form. 7 he use of Lovibond slides is a disadvantage. A
modification of the Marsh test for caramel was developed by Clapp and
was reported on by Valaer (1945).The use of cyclohexanol plus methyl
propyl ketone as the solvent is new. Valaer (1948) has reviewed the
available procedures for the detection of caramel in wines. If care is used
the Milos or Mallory-Love tests should not give positive results when
caramel is absent. Since, in certain cases where oak chips or certain dyes
have been used, the Marsh reagent may show a positive test, confirmatory
tests should be employed. The Mathers (see Valaer, 1948) test is simpler.
After separating the caramel the 2,4-dinitrophenylhydrazinetest is used
to confirm its presence. Scott (1946) preferred the Love and Mallory
procedure. He reported that owing to aging or heating caramel-reacting
materials are present in some normal California wines. The possi-
bility that the Maillard reaction may be responsible is also to be
considered.
Torricelli (1945) decolorized grape juices with animal charcoal and
then examined them under a quartz-filtered ultraviolet light. Those pre-
pared from raisins or concentrate had a strong luminescence. Pasteuriza-
tion of grape juice at 85" C. (185"F.) for 30 minutes did not result in a
positive test.
Mannitol. Whereas mannitic fermentations are seldom a problem
where sulfur dioxide, pure yeasts, and temperature control are employed,
Martucci (1941)has reported them in Argentina. He recommended con-
trol of the must acidity, since a high pH also favored such spoilage. A
complicated polarimetric procedure for mannitol (a sugar alcohol) in
wines was presented by Salani (1937). Formation of mannite during
dialysis of musts a t low temperatures (8" to 10" C. (46.4"to 50" F.)) in
the presence of chloroform was reported by Barbera (1933b) (possibly
owing to enzyme action).
6. Pectins
Methods. Barbera (1933a) extracted the pectin material of dried grapes with boiling
water and then fractionated it into an alcohol-insoluble fraction (xylan-araban?) and
an alcohol-soluble fraction. Methyl alcohol was rapidly produced by pectase. His
terminology is now somewhat outdated. To differentiate the pectins and the gums
Peynaud (1952) made the alcohol content of the acidified must 80%. The impure
pectin and gum precipitate was dissolved in hot water and reprecipitated. After a
second purification it was dissolved. An aliquot sample was then placed in a platinum
dish, dried, and weighed. The contents were ashed and the weight subtracted to get
that of the pectins plus gums. Another aliquot of the purified pectin was titrated to
determine the pectic acid and was then subtracted to determine the amount of esteri-
fied pectin. These two subtracted from the net pectin plus gum weight gave the
amount of gum. The milliequivalent weight of pectic acid was taken as 176 and of
esterified pectic acid as 190. In a comparison with the usual calcium precipitation
method, he found this procedure to give lower results. The methods used by NBgre
et al. (1947) for pectic material has been shown by Peynaud (1952) to yield results two
to nine times too high, owing to entrainment of insoluble calcium salts. Solms et al.
(1952) have given a procedure for determining pure pectin and per cent esterification
of grape pectin. They found two water-soluble polyuronides in grapes-one with the
usual equivalent weight and a low degree of esterification and the other of much
higher equivalent weight, though of the same components. The latter was not readily
precipitable as the calcium salt.
Composition. SQmichon(1933) reported 12.86% methoxy in a purified

pectin from grapes and 5.71 % mineral material-possibly calcium
pectate. He reported a large increase in pectin content when grapes were
allowed to become overripe and maintained that the increase in pectin,
per se, was related to the increase in quality. This is difficult to believe
from what we know of the solubility of pectin in wines. The dextro-
rotatory substance in wines (insoluble in methyl alcohol) yielded uronic
acids, galactose, and glucose in Parisi and Della Barba’s (1931) study-
possibly from pectin. Barbera (1933~)separated from wines a methyl
alcohol-soluble and -insoluble substance. The first he found to be pectin,
and the latter was a uronic acid. A study of the pectins of wines was made
by von Fellenberg (19444. He reported polygalacturonic and galactu-
ronic acid (5.9 and 10.0 mg. per 100 ml.), araban (12.8 and 6.9), and
arabinose (26.8 and 69.0) in a white dry wine and a wine prepared from
“botrytised ” grapes. Using paper chromotography Solms et al. (1952)
found galacturonic acid, galactose, mannose, arabinose, and rhamnose in
pectin from Swiss musts. Von Fellenberg (1944b) gave a method of
separating “furfuroids” (from protopectin?). The “furfuroid ” content
was higher in wines made from moldy grapes.
Amounts. Peynaud (1951b) reported the following fractionation of
pectins and related materials in grapes (grams per liter) :
Total pectin Pectic acid Gums or

Variety material Free Esterified arabans
Merlot 1.77 0.07 0.19 1.51
SBmillon 4.43 0.02 0.14 4.27
Cabernet franc 1.22 0.08 0.37 0.77
The degree of purity of the total pectin material is much less in grapes
than in peaches and other fruits. Hauptmann (1952a) reported 97 mg.
per 100 ml. of calcium pectate in Riesling musts in September and 78.2
mg. a month later. Further data were reported by Peynaud (1952), who
found the following average amounts of pectins and gums in Bordeaux
musts and wines (grams per liter) :
N ~ of. Pectic acid
samples Free Esterified Gums
Musts 8 0.07 0.43 1.32
Musts* 2 0.02 0.13 3.72
White table 5 0.02 0.04 0.59
Red table 5 0.02 0.13 0.93
* From grapes attacked by Bofrytis cinersa.
He concluded that pectins played no role in the mellowness of wines,
since most of the pectin was lost during fermentation or during the first
two months after fermentation. This agrees with the conclusion reached
by Amerine and Joslyn (1951). The value, 0.05% for total pectic acid, is
somewhat lower than most of the values reported by them or by Solms
et al. (1952), who found 0.056% to 0.355% pure pectin (average 0.171) in
thirteen Swiss musts of which an average of 31.4% was esterified. The
gums, however, constitute about 0.1% of the wine. Peynaud calls these
osanes or polysaccharides such as araban or galactan. More work needs
to be done as to their identity and the changes which they undergo during
aging. Franpot and Geoffroy (1951) also separated the pectins and the
gums. They found three to four times as much in the press as in the free-
run juice. During fermentation from 30% to 90% of the pectins were
precipitated.
Removal. Use of pectolytic enzymes as an aid to the clarification of
grape juice has been studied by Willaman and Kertesz (1931) and
Mehlitz and Scheuer (1934). Hickinbotham and Williams (1940) and
Besone and Cruess (1941) used commercial preparations successfully for
clarifying the must. The resulting wines also clarify more rapidly. Cruess
and Kilbuck (1947) added the enzyme to the crushed grapes to increase
the juice yield. They obtained 6.5% more free-run juice when one part of
the enzyme was added to 1000 of the crushed grapes, which were then
allowed to stand overnight. They claimed the wines cleared more rapidly,
were fruitier in bouquet, and yet matured more than twice as rapidly as
untreated wines. Kilbuck et al. (1949) used amounts as low as 200 to 300
p.p.m. They reported the wines matured more rapidly. Little increase in
methyl alcohol content was found. Finally Cruess et al. (1951) confirmed
that in some cases treated samples were darker than untreated, particu-
larly if the enzyme was added to the crushed grapes. They also found that
the addition of galacturonic acid hastened darkening. Since galacturonic
acid is a product of pectic enzyme activity on pectin, this would suggest
a reason for the darkening of some treated wines. Considerable variability
in results was obtained. This agrees with Testa and Maveroff’s (1949)
results in Argentina, where they found regional and varietal differences.
The galacturonic acid content of six white wines was determined by
Cruess et al. (1951). Before treatment with pectic enzymes it varied from
3.1 to 8.0 mg. per 100 ml., average 5.3, and after from 6.0 to 14.1, average
10.1.
The value of pectolytic enzymes in aiding filtration was demonstrated
by Geiss (1939) and Procopio (1949). Paronetto (1948) reported 50 t o
100 g. per hl. of a pectin-splitting enzyme (Biozim) reduced the pectin-
gum content of two wines from 0.640 g. per liter to 0.580 and 0.504,
respectively. The organoleptic data were conflicting.
Hauptmann (1952b) recommended a preparation low in pectase for
red wines so that the extract and ash content of the wine would not be
too high. With the proper enzyme he found the red wines of better color
and flavor. For white wine clarification no protease should be present.
Better clarification and freedom from protein cloudiness in the bottle was
reported for the treated wines.
6. Extract
The meaning of the term “extract” has been debated since 1890.
While it obviously means the nonvolatile soluble solids, it is difficult to
specify analytically. Godet (1949) and Jaulmes (1951) have summarized
the arguments and clarified the issue. At present extract is usually deter-
mined by formula based on the specific gravity, by evaporation, or by
hydrometer readings on the dealcoholized wine. None of the procedures
is entirely satisfactory.
Formulas. The indirect determination of extract content may be represented by
s.g. wine + s.g. water = s.g. alcoholic distillate +
s.g. of the extract solution, where
8.g. is the specific gravity. This formula reduces to
s.g. extract = (s.g. wine + 1) - s.g. distillate

and has been modified by many later workers. Roussopoulos (1933) recommended a
specific gravity procedure based on drying 48 hours under vacuum over PzOF,.
His
formula is the usual one. Ghimicescu (1938) used the formula (D - D1)2.685to
calculate the extract, D being the density of the wine and D1, the density of the
distillate (both by pycnometer). On 45 Roumanian table wines, he obtained an
average extract value of 2.55 as compared to 2.60 for the drying procedure. The Rome
convention of 1935 (Anonymous, 1935) defined extract as based on the densimetric
method, using the specific gravity of the wine and the alcoholic distillate at 15O C.
(59" F.). The actual formula used may be Schermann, Houdart, Dujardin-Salleron,
or Roussopoulos, etc. The Dujardin-Salleron method sometimes gives higher and
sometimes lower results than the dry extract obtained after 6 hours on a boiling water
bath. Houdart's formula is (D - D ) K , where D is the density of wine at 15' C., D',
the density of a water-alcohol mixture of the same per cent as wine, and K is 2.062.
For 170 Alsatian wines Percher (1938) found K to vary from 1.079 to 2.591, average
2.198. For MBlaga wines Clavera and Oro (1932) found it to vary from 2.50 to 2.68.
The best direct procedure in their study was i n aacuo at a low temperature. Laganne
(1938) discussed the various formulas and proposed (q - 8 , x 2.062 -k l, where is
2.663
the extract in grams and s the sugar content. Percher (1938) preferred the Houdart
value.
In determining the extract content based on density measurements the contraction
is most difficult to determine. Jaulmes (1951) and Godet and Deuel (1947) have given
approximate formulas for making this correction. Godet and Deuel (1947) noted that
to account for the contraction which takes place when water and alcohol are mixed
this formula should read
Dextrsat = (Dprine - Daloohad +1 - (Cwine - [cdaohol + CaxtraotI).
e
Jaulmes (1951) expressed the extract in grams per liter as 1,000- (De - l),
e-1
where e is the density of the extract. This takes into account the water of hydration
which reduces the volume but not the increase in volume when a substance is dissolved
in water. This means that the final contraction, D f , equals the contraction of the water
less the increase in volume due to dilution and that the Jaulmes formula should read
e
1,000 -(De - 1 - c). Godet and Deuel (1947) established that the Cf equals
e-1
(De - 1) - ( p - v ) for volume of extract in 1ml. From this they find the extract in
D
grams per liter to equal (De - 1 - C), where D is the density of the dry
extract. This is essentially the same as the corrected Jaulmes formula. Unfortunately,
e, D , or the contraction of water or the final contraction cannot be measured. Grivas
(1952) proposed making two dilutions and calculating from these the extract. For-
mulas are given.
Tarantola (1947a) used the Windisch tables in preference t o those of the Acker-
mann. The errors in direct procedures using sweet wines or wines of high volatile
acidity are outlined. The indirect extract determination is usually higher than the
direct, according t o von der Heide and Zeisset (1935), but with very high glycerol
contents the opposite may occur, and they determined both in studying possible
sophistication. A useful table for calculating the extract from the specific gravity was
published by von der Heide and Mandlen (1933). Very useful tables relating the
density and the per cent alcohol of wines were given by Pato (1938). These include
temperature corrections and the relation between the density of the wine, its per cent
alcohol, and its extract content. A similar table is given by Anonymous (1944).
Evaporation. Von Fellenberg (1944a) using artificial extract solutions showed con-
stant weight losses up to 7 hours when drying at 103' to 105" C. (217.4" to 221" F.) ;
however, he believed the direct procedure to be better than the indirect and proposed
+
a formula a 2(a - b), where a is the weight after 4 hours drying at 105"C. (221"F.)
and b is the weight after 6 hours drying a t 105" C. (221' F.). He neutralized the extract
before heating and subtracted the weight of the potassium hydroxide used for neu-
tralization. Boinjak's (1936) study of the methods of determining the extract content
has received little attention because of its isolated place of publication. He reported
good results if the samples were dried in vacuum over sulfuric acid, providing the
sugar content is below 0.8%. Indirect procedures were best for sweet wines. Percher
(1938), however, did not find drying at 100" C. (212' F.) to yield consistent results
owing to the influence of the per cent moisture in the air and the loss of glycerol.
Drying 48 hours in a vacuum at 50" C. (122' F.) gave slightly better results than 60 to
120 hours. Fischl's (1942) direct method gave results to within 0.8%. Moreover, the
empirical direct heating procedures involved more or less changes in various con-
stituents: tartaric to metatartaric, malic to malonic, caramelization of sugars, etc.
The various methods for determining the per cent dissolved solids were discussed
by Malvezin (1934) and Ay (1936). The latter described the various hydrometers and
indicated their defects. The densimetric procedures were preferred to the direct
evaporation methods of extract determination by Krombach (1948). The official
French 6-hour evaporation procedure gave low results, and the Association of Official
Agricultural Chemists procedure of evaporating 235 hours and the densimetric
method gave concordant results. The value of the refractometer for determining the
extract content is emphasized by Vetscher (1947), Bohringer (1951a), and Jilke (1951).
Miscellaneous. A review of his work showing that living wine yeasts
form triose phosphates which are intermediaries in the fermentation
process was made by Kiessling (1949). He reported that addition of a
mycelium extract of Aspergillus sp. increased formation of triose phos-
phates. At the beginning of fermentation the triose phosphate content
decreased.
Markley et al. (1938) found nonacosane, hentriacontane, and sitosterol
in the saponified bloom of Concord (Vitis Zabrusca) grapes. Gatet (1939a)
identified a substance in grapes whose hydrazone was insoluble in hot
alcohol but soluble in sodium carbonate and which is believed to be a
furfural derivative. He also reported an aldehydic substance whose
hydrazone was insoluble in hot sodium carbonate but soluble in alcohol.
The presence of iodine-reducing substances other than sulfur dioxide,
aldehydes, or sugars in wines has long been known. Muth (1933) shows
that most of these are present in the grape but that some are produced
by fermentation.
VII. ESTERS
Very little work on the esters of wines had been done since the classic
studies of Berthelot in the last century until the Bordeaux enologists
began their studies in the early 30's. The result of this work seems to
relegate the esters to a secondary role in wine quality, except for ethyl
acetate as a spoilage product. This is contrary to the rather loose state-
ments often found in the literature.
Methods. Espil el al. (1933) proposed extracting esters from wines in a continuous
liquid-liquid extractor. Direct saponification of wines in the presence of sugars leads
to formation of volatile acids. Esters of the fatty acids are easily distilled, but esters
of polyhydroxy acids are more difficult. Because of the unfavorable partition coeffi-
cient, esters are more difficult t o quantitatively extract from wines. Espil and Peynaud
(1936), however, obtained satisfactory extraction of neutral esters using freshly
distilled neutral petroleum ether. Ethyl ether is unsatisfactory, since it extracts
colored saponifiable materials. Dangoumau and Debordes (1937) obtained less satis-
factory results, but Espil and Peynaud (1937) and Genevois (1937) indicated tha t this
was due t o a n insufficient extraction. For the accurate determination of total esters and
neutral esters the extraction procedure should be used. However, the determination
of total esters, as outlined b y Peynaud (1937b), is rather lengthy, and normally only
the neutral esters are determined (Amerine, 1944). The neutral esters may be frac-
tionated into volatile and nonvolatile neutral esters. Peynaud (1937b) has determined
in the latter the ethyl tartrate, malate, and citrate, a s distinguished from the ethyl
lactate and succinate. The acid esters may be calculated by difference between the
total esters and the neutral esters. It is also possible to determine the ethyl acid
tartrate b y determining the tartrate content before and after saponifying the acid
ester by the racemate procedure (Jaulmes, 1951).
Because extraction procedures require special equipment and considerable time,
the simpler distillation procedures have been investigated for the neutral esters and
particularly for the neutral volatile esters. Peynaud (1937b, 1938d), Grandchamp and
Vollaire-Salva (l939), Archinard (1939), and Peynaud (1939a) each have developed a
procedure. I n the first Peynaud (1937b) procedure the wine was distilled at a p H of
7.5, the distillate neutralized, excess base added and later back-titrated. In the second
(1938d) the wine was buffered t o a p H of 6.3 and about 15% distilled into base.
After saponification the solution was acidified and distilled. The Archinard (1939)
method is similar t o the former except the distillate is received in base and the saponi-
fication carried out at a temperature of not over 50" C. (112' F.) After saponification
a n equivalent amount of acid was added and the excess titrated with base. The speed
of distillation should be slow and care taken to remove carbon dioxide. Archinard
(1940) developed a distillation procedure which Jaulmes (1951) considers the most
rapid and accurate of the direct distillation methods. I n it 50 ml. of wine were neu-
tralized, buffered with 25 ml. of a p H 7.5 phosphate buffer, and distilled into a 150-ml.
volumetric containing 20 ml. of 0.1 N alkali and 115 ml. of water. Excess sulfuric acid
was added to an aliquot, the solution boiled and back-titrated with alkali. Grand-
champ and Vollaire-Salva's (1939) method is admittedly approximate, being based on
fist determining the volatile acidity. Another aliquot is saponified, then acidified and
distilled. The difference in the two represents the neutral esters. None of these pro-
cedures have given satisfactory recovery in the present author's laboratory, although
Peynaud (1938d) obtained similar results by his second distillation procedure and by
ether extraction. Reis (1946) favored distillation over extraction procedures, since he
obtained low results on pure solutions. Peynaud (193713) and Amerine (1944), possibly
with longer and better extraction, did not find this to be the case. Roubert (1951)
determined amyl acetate in the vapors condensed during fermentation b y a nephelo-
metric procedure.
Diethyl tartrate is rapidly saponified by lead acetate, but ethyl acid tartrate
resists saponification even when heated, according to Hartmann (1939).
Source. Espil et al. (1933) studied the speed of esterification of the
various acids present in wine at 10% alcohol and in 0.1 N solution at
100" C. (212" F.). The K X lo4 a t pH 3 and 4 were: acetic, 12 and 3.5;
propionic, 10 and 5.0; butyric, 6 and 2.5; lactic, 250 and 60.0; succinic,
120 and 50.0; malic, 110 and 45.0; and tartaric, 40 (pH 3). The primary
esterification is by bacteria, although some results from yeast activity.
Even so, equilibrium is seldom reached, even in very old wines. For
example, Espil et al. (1933) obtained the following results:
Acetic Ethyl
acid, acetate, Coefficient of esterification
Year Alcohol, % pH millimoles/l. millimoles/l. Actual Calculated
1893 14.8 3.94 29.4 3.1 10.9 14.6
1900 11.0 3.60 17.5 1.7 9.4 11.7
1907 10.2 3.30 20.6 1.8 8.5 11.2
1914 11.1 3.50 19.4 2.0 9.9 11.8
1925 10.1 3.63 19.2 1.9 9.6 11.1
1930 9.2 3.30 17.8 1.4 7.5 10.4
1933 10.0 3.77 15.6 1.4 8.4 11.0
1934 11.5 3.20 19.0 1.8 8.9 12.1
1936 10.6 3.60 11.2 1 .o 8.7 11.4
Formation of neutral and acid esters of the polyhydric acids is essen-
tially a chemical reaction and proceeds very slowly. The following data
of Espil et al. are indicative of this:
Neutral Ethyl acid
esters, tartrate,
Year Alcohol, % pH millimoles/l. millimoles /l.
1893 12.6 3.35 0.75 1.60
1914 11.1 3.60 0.60 1.60
1926 10.7 3.23 0.40 1.25
1933 11.4 3.68 0.30 -
1934 12.2 3.50 0.20 trace
1935 10.6 3.60 0.00 0.00
The rate of esterification of acetic, propionic, citric, butyric, malic,
succinic, lactic, and tartaric acids at three different pH's over a 30- to
60-day period was determined by Espil and Peynaud (1936). They
showed that none of the ethyl esters of the polyhydric alcohols contribute
to the aroma and would not even if present at ten times their normal
amounts. Tomaghelli (1937) studied the rate of esterification of the
system acetic acid-ethyl alcohol and that of saponification of the system
ethyl acetate-water at 100" C. (212" F.) for 500 hours and a t 150" C.
(302" F.) for 320 hours.
The most extensive investigations of the ester content of wines were
made by Peynaud (193713).Although other alcohols than ethyl are present
in wines, they are present in very small amounts. Glycerol is, of course,
present in larger amounts but forms very little esters. As a matter of fact,
only ethyl acetate was found to be an important ester from the organo-
leptic point of view. I n wines the limits of esterification were never
reached. Citric, for example, was only very slightly esterified a t the pH
of wine. Lactic and succinic approached the limits of esterification most
nearly and rayidly. Esterase was a negligible factor in wine esterification.
Pasteurization did not increase ester content but esterification was very
slow at 15" C. (59" F.) compared to 30" C. @So F.). Prolonged oxidation
did not increase the ester content. The esters of simple acids formed
during fermentation may have some slight effect on bouquet. More likely
oxidizable petroleum-ether-soluble materials, such as tannoids or coloring
matter, are responsible. Peynaud's work has been briefly reviewed in
Italian by Procopio (1949). Peynaud (1938d) proposed a limit of 220 mg.
per liter of ethyl acetate, in lieu of the present limits on the volatile
acidity. (See also Peynaud, 1936a, 1939a, and Michel, 1948a.) This was
based on the demonstration (1) that addition of pure acetic acid to wine
does not produce a spoiled character; (2) that subjecting a moderately
spoiled wine to a vacuum reduces the spoiled smell and the ethyl acetate
content without changing the per cent acetic acid; (3) that adding pure
ethyl acetate gives a spoiled character; and (4) that heating a wine con-
taining acetic acid (no spoiled odor) in a sealed tube gives a spoiled odor
while check samples a t room temperature show no such odor. Cellar
samples with a badly acescent odor, a low volatile acid content, and a
high ethyl acetate content were also observed. Grandchamp and Vollaire-
Salva (1939) also reported wines of high acetic acid content which did not
smell spoiled owing to their low acetate content and, vice versa, wines of
low volatile acidity which had an acescent character. Gentilini (1947),
however, found no correlation between ethyl acetate content and the
volatile acidity in forty Italian wines. Moreover, the organoleptic tests
were more in agreement with the volatile acidity than the ethyl acetate
content, and he rejected an ethyl acetate limit as an indication of quality
in Italian wines. He did find that wines high in ethyl acetate were usually
also high in acetic acid.
West et al. (1951) have made a similar study on beer, where they
reported a range of 23 to 55 mg. per liter of volatile esters, as ethyl
acetate, average 40.7. They found no relationship between ester content
and aroma, flavor, or foaminess.
Most of the neutral esters formed in wine result from biological
activity. Uchimoto (1951), however, showed that neutral esters were
produced in slightly higher amounts at lower fermentation temperatures
(less loss?). Even after many year's aging (up t o 50) only about 7501, of
the theoretical ester content was reached in the studies of Peynaud
(193713) and Rib6reau-Gayon and Peynaud's (1936). The acid esters are
formed by slow chemical activity, and esterification is more rapid a t the

lower pH’s. Approximately half the total esters found by Peynaud in
Bordeaux wines were acid esters.
Yeasts as producers of esters have been of interest to those who believe
esters contribute to wine quality. Gomes (1945), for example, isolated
Hansenula anomala var. Spjarerica (Naegeli) Dekker from port grapes.
This yeast produced 959 p.p.m. of esters in 45 days but only 3.5% of
alcohol. Unfortunately the volatile acidity and aldehyde content were
greatly increased. Tabachnick and Joslyn (1953) showed that ethyl
acetate was the ester produced by this yeast, that ester production was
favored a t a pH of below 3, that the nitrogen source was relatively un-
important, and that under aerobic conditions the ester is accumulated in
the fermentation of glucose and is itself utilized. They suggest that the
high ester yields indicate an energy-consuming reaction and not the
reversal of a simple esterase hydrolysis. Popova (1948) found that
esterase preparations from Oidium lactis, Botrvtis cinerea, and Aspergillus
oryzae show high synthesizing and hydrolyzing ability. The synthesizing
power is enhanced in oxidizing media. Wines were found to contain a
weak esterase activity. Addition of mold esterase increased the syn-
thesizing activity, as shown by an increase in ester content in young and
aged wines. Wines treated with mold extracts also showed an increased
enzymatic activity. Popova claimed that the principal role of the extracts
belongs to the esterases because, owing to their action, glycerol and esters
among other substances are formed, and these enhance the quality of
well-aged wines. This is doubtful, to say the least.
A polemic has developed in Russia on the formation and importance
of the carbonic acid ester diethylpyrocarbonate in sparkling wines.
Parfent’ev and Kovalenko (1951, 1952) refuted Rosenfeld’s (1952) con-
cept that diethylcarbonate is only condensed carbon dioxide in alcohol.
They note that it has been synthesized and its physical properties
determined. Kozenko (1952) reported the amount of carbonic acid ester
to be 0 in musts and to increase during fermentation. In still wines about
9 mg. per liter were found, whereas in sparkling wines 125 mg. per liter
were noted. I n a bottled sparkling wine 53 mg. were reported before
opening, 42 mg. at 18 days after opening, 32 a t 60 days, and 26 a t 90 days
after opening. The subject is, however, by no means settled; see, for
example, Merzhanian (1951, 1952).
Amounts. The part which esters play in the olfactory quality of wines
inspired the classical studies of Berthelot on esterification. Some claim
that esters are important, but others admit only that ethyl acetate is a
spoilage product and detracts from the quality. Reis’s (1946) results (see
Tables XIV and XV) lend some support to those who believe quality and
higher esters are related-the four-year-old and special quality ports

have the highest ester content. But most of the work reported does not
substantiate any such claim. I n six Russian wines Oparin and Manskaya
(1939) reported 1.1 to 3.6 meq. per liter of neutral esters (average 2.2)
and 1.8 to 4.5 of acid esters (average 3.6). The neutral esters increased
TABLEXIV
Total Neutral Esters in Various Types of Wines
(Milligrams per liter as ethyl acetate)
California Wh. table Amerine (1947) 79 80 553 209
California Red table Amerine (1947) 60 170 725 400
California Dessert Amerine (1947) 124 93 730 333
France Swt. table Peynaud (1950a) 6 141 273 21 1
France Dessert Peynaud (1950b) 8 112 336 209
France Sparkling Hennig (1952) 13 155 292 246
Germany Sparkling Hennig (1952) 40 103 351 207
Portugal Port (export) Reis (1946) 33 114 334 195
Portugal Port (old) Reis (1946) 4 598 924 693
Portugal Port (young) Reis (1946) 12 229 422 332
(1952)
TABLEXV
Total Volatile Neutral Esters in Various Types of Wines
(Milligrams per liter as ethyl acetate)
California Wh. table Amerine (1947) 79 8 234 73
California Red table Amerine (1947) 60 61 228 112
California Dessert Amerine (1947) 124 17 201 104
France Swt. table Peynaud (1950a) 6 141 274 211
(1952)
more during ten years aging of a dessert wine (from 2.0 to 4.7 meq.) than
did the acid esters (2.0 to 3.4). Reichard (1951) reported 15 to 350 mg.
per liter of volatile esters in German wines. The total and volatile neutral
esters of a number of different wines are given in Tables XIV and XV.
VIII. POLYHYDROXYPHENOLS
The tannins and coloring matter are the primary heterocyclic poly-
phenolic compounds found in wines. They are similar in their chemical
and oxidation-reduction properties. The tannins, furthermore, play an
important role in the taste of wines, particularly of the reds. Their anti-
biotic properties are less well established. The color of wines is one of their
most important attributes from the consumer’s point of view. In grapes
the tannins may be conaidered as phlobatannins-the phlobaphene-
producing tannins. These are polyhydroxy flavinacoles and are thus
basically derivatives of the heterocyclic benzopyrylium chloride nucleus
from which the color pigments-anthocyanidins, flavones, and flavanols-
are likewise derived.
1. Tannins
The phlobatannins of grapes are usually classified as pyrocatechol
tannins because they give a green color with ferric chloride. Gatet (193913)
pointed out that the polyphenols are colorless at an rH of below 14 and
highly colored a t an rH of above 23.
Methods. A large number of procedures for determining the tannin content of
wines have been proposed. Reviews of the various methods were made by Collier
(1935), Ghimicescu and Gheorghiu-Vieriu (1938), and Diemair et al. (1951). The
primary problem is to distinguish the tannins from the other polyhydroxyphenolic
compounds. The standard procedure is to determine the amount of permanganate
reduction before and after treatment with charcoal. The charcoal removes the tannins
and color; hence the difference between the two titrations represents the reduction due
to them. This is known as the Neubauer-Loewenthal method. Durmishidze (1948b)
and others have noted that the substances determined by the Neubauer-Loewenthal
method are frequently erroneously reported as “tannins,” even though other oxidiza-
ble substances are included, particularly in high oenidin red wines. As a correction he
introduced the term “coefficient of oxidation of anthocyanin.” From the total oxidiza-
ble substances the ether-soluble oxidizable substances were deducted. A correction
coefficient for oenotannin was also introduced. The formula is:
E
T = 0.00588 ( 2 - b.006y5 - p )
where T is the amount of tannin in grams per liter, Z: the milliliters of 0.1 N perman-
ganate used for oxidation of tannins and coloring matter in the first titration of the
wine according to the Neubauer-Loewenthal procedure, E the amount of oenidin ih
grams per liter, and p the amount of 0.1 N permanganate required for the oxidation
of that portion of ether extract which is absorbed by bone charcoal. Data obtained
with this formula are in Table XVI. Vasconcellos (1940a) uses a factor of 0.00173
instead of the more commonly employed factor of 0.00416 for converting perman-
ganate titration values to tannin in the Neubauer-Loewenthal procedure. He also uses
gelatine or casein to remove tannin, claiming that charcoal also removes non-tannin-
oxidizable substances. NBgre (1939a, b) has differentiated the oenotannin content-
that precipitated by zinc-from the total tannoid-that precipitated by lead. The
difference between the two he calculated as the polyphenol-non-tannin material.
Sampaio (1946) precipitated the tannins with ammoniacal zinc acetate and oxidized
the washed precipitate with permanganate using sodium sulfoindigotate as an indi-
cator. The results obtained are naturally lower than when titrating all the perman-
ganate-oxidizable material.
A rather lengthy procedure, based on determining the copper-reducing value
before and after removal of tannins, was devised by Astruc and Caste1 (1932b). The
macro-method of Ghimicescu and Gheorghiu-Vieriu (1938) is based on the reducing
ability of tannin for Fehling’s solution before and after detannising a sample of wine,
the cuprous oxide being determined. It is not certain that the basic lead acetate used
does not remove some sugar, which would vitiate the determination.
The Folin-Denis reagent was used by Rosenblatt and Peluso (1941) for the colori-
metric determination of tannin in wines. Careful attention to details of temperature,
time, and concentration is necessary, but the procedure is reported accurate t o within
TABLEXVI
Tannins in Red Wines*
Tannin and
Ether- color
Neubauer- soluble (Neubauer-
Sample and Loewenthal Oenidin substances Oenidin, Loewenthal), Tannin, t
date ml. 0.1 N KMnO4/1. t3.A g.P. g.A.
Saperavi, 1946 930 130.6 100 0.908 3.869 4.112
Saperavi, 1916 360 65.0 85 0.452 1.500 1.234
Saperavi, 1915 305 74.4 75 0.519 1.270 0.915
Saperavi, 1894 365 57.6 75 0.400 1.520 1.366
Saperavi, 1891 310 51.3 80 0.357 1.290 1.050
Saperavi, 1887 460 70.8 95 0.492 1.920 1.739
Kaberne, 1942 560 80.0 80 0.556 2.330 2.352
Kaberne, 1917 480 70.3 65 0.489 2.000 2.026
Kaberne, 1898 285 70.3 85 0.489 1,190 0.762
* Durmishidze (1948b).
t Corrected for oenidin and ether-soluble substances.
0.5%. A similar procedure was given by Diemair el al. (1951). They proposed pre-
cipitation of the tannins with lead acetate and gelatine, ,dissolving the precipitate in
phosphoric acid and sodium phosphate, adding sodium tungstate and sodium car-
bonate. The amount of blue color produced by the reduction of the tungsten by
tannin (etc.?) was used as a measure of the tannin present (1-cm. cell with an 5-68
filter). Results were from 20% to 60% lower than by the Neubauer-Loewenthal
method. Lang (1951) also used phosphotungstic acid. Pro (1952) modified the similar
Association of Official Agricultural Chemists (1950) method t o increase speed and
accuracy so that the color density follows Beer’s law more closely.
The polyphenolic compounds of musts and wines rapidly fix bromine, and bromine
titration may be used to measure them. Gatet and Genevois (1938) used this property
to show t h a t the amount of polyphenols was three times greater in the press juice than
in the free-run. The method has been developed further by RibBreau-Gayon and
Mauri6 (1942) and MauriE (1942). A substance extractable by ethyl acetate which
consumes 8 bromine atoms per molecule is found in wines and grape berries. Genevois
(1951) reported that the bromine consumption indicated 1.2 to 2.5 millimoles of
catechol per liter and 4 to 6 millimoles of oenidin were present. The “catechol” may
represent demethoxylated oenidin. Tarantola (1951) also used ethyl acetate for the
separation of tannin from color. He reported that only 9.4% to 24.6% of the per-
manganate required for oxidizing the tannin plus coloring matter of red wines was due
to the ethyl acetate-extractable (tannin) material. For white wines this was 18.0%t o
61.9%. I n the red wines the tannin content was not correlated with the color. The
permanganate index, for example, varied from 4.7 to 12.9 with colors of 41 to 330.
RibBreau-Gayon and MauriB (1942) reported that wines with a high permanganate/
color ratio were more astringent than those with a low ratio.
A photoelectric colorimeter was used by Faure and Pallu (1936) for the color of the
tannin-iron complex. Kretzdorn (1949) also tested a n iron chloride procedure, but
citric acid (1% t o 3%) interfered. The hide-powder method was applied to dealcohol-
ized wines by Ponte and Gualdi (1931). Feigl and Feigl (1946) proposed use of a,a’-
ferrous dipyridyl sulfate or a,a’-ferrous phenanthroline sulfate for the colorimetric
identification of tannin. Synthetic tannins did not react and seven red and white wines
gave positive tests.
Brugirard and Tavernier (1952) used the property of true tannins of flocculating a
solution of 1% gelatine in 10% sodium chloride for their tannin determination in
cider. They determined the total tannoids ( T )by the Neubauer-Loewenthal procedure,
then precipitated the true tannin in a buffered solution with 2.0% cinchonine sulfate,
and determined the non-tannin polyphenols (t’) in the centrifugate. T - t’ thus
equals the true tannins. Since the tannin content decreases with time, Fessler (1947)
noted that the age of the sample should be stated along with the tannin results.
Rentschler and Hauser (1950) precipitated the catechin tannins in a hot acid solution
with formaldehyde, filtered, washed with alcohol and ether, dried and weighed.
The anthocyanin pigments were reported not to interfere.
Source. Diemair et al. (1951) found the tannins of grape leaves and stems to be
catechin tannins. In wines resorcinol and pyrocatechol were demonstrated in the
tannin fraction. During the growing season, May to October, the tannin content of
the leaves increased from 0.1% to 0.5%, and in the stems there was a sharp rise from
0.1% to 0.6% in July followed by a continuous decrease. In the berries, after a n
increase of from 0.2% to 0.4% in May and June, the tannin content decreased to
0.05%. When the seeds were crushed with the fruit, an increase occurred in August;
and they showed this to be due to a large increase in tannin occurring a t this time in
the seeds. During fermentation of white grapes the tannin content decreased. Dur-
mishidze (1950d) also followed the changes in tannin content. During ripening he
reported a decrease in the tannin fractions in the clusters, in the seeds, and particu-
larly in the flesh and skin of the fruit. The composition of the tannin complex also
changed, the ratio of water-soluble fractions to polyphenol catechols decreasing.
However, the decrease in total tannin was not due to transformation of one type of
tannin to another. The varieties differed in total tannin as well as in the ratios of the
tannin fractions. Tannin diminution continued during storage of grapes.
The investigation of tannins by Durmishidze (194813) has shown that

the ether-soluble substances (polyphenols and catechols) pass from the
pulp into the fermenting must during the early stages of fermentation and
that the amount in the wine is independent of the duration of fermenta-
tion. The ether-soluble tannins remained essentially unchanged during
the aging of wines. The amounts of tannin and oenidin increased with the
duration of fermentation (contact with the skins). The accumulation of
these substances is quite regular. There is a reduction in oenidin and
ta.nnin during the aging of wine. They showed that the main part of the
tannin complex in Rkatsiteli, Mtsvane, and Saperavi wines are the
tannins of the catechin group, 63.1% to 76.0%. The data in Table XVII
are typical.
Politova-Sovzenko (1947) reported the isolation of seven different
catechin tannin materials from a white wine, but the original article is not
available as to details. Ponte and Gualdi (1931) on the basis of compara-
tive analysis by three different methods also concluded that the tannins
of wines are exclusively of the catechin type. Durmishidze (1950a)
isolated d-catechin by extracting with dry ethyl acetate, drying with
sodium sulfate and under carbon dioxide a t reduced pressure, and pre-
cipitating three times with chloroform and purifying with lead. The
TABLE XVII
Catechin, Mixed, and Precipitable Tannins*
Variety Catechin Mixed Precipitablef
of grape Material tannin, % tannin, % tannin, %
Rkatsiteli Leaves 66.2 17.8 16.0
Rkatsiteli Skin 76.0 15.8 8.3
Rkatsiteli Stems 74.0 18.4 7.6
Rkatsiteli Seeds 76.0 20.8 3.2
Mtsvane Leaves 72.4 10.6 17.0
Mtsvane Skin 68.3 23.2 8.5
Mtsvane Stems 72.3 20.7 7.0
Mtsvane Seeds 74.5 23.8 1.7
Saperavi Leaves 63.1 20.9 16.0
Saperavi Skin 72.5 20.5 7.0
Saperavi Stems 70.4 23.4 6.2
Saperavi Seeds 70.6 27.1 2.3
* Durrnishidze (1948h).
t According t o the Fisher-Bergman procedure.
d-catechin was identified from its acetyl and methyl derivatives. Later,
Durmishidze (1951) isolated Z-gallocatechin, which he found to constitute
45% to 54% of the total tannin. This compound and d-catechin do not
make up the whole of the tannin complex, since the summation of the
optical activities of the two do not add up to that of the crude material
(+75"). It is particularly interesting to note that the skins are lower than
the seeds in Z-gallocatechin. Durmishidze also showed that whereas
1-gallocatechin increased during ripening, d-catechin decreased. He sug-
gested oxidation of the former to the latter.
Using his differential methods for distinguishing tannins from poly-
phenol non-tannins, NBgre (1942-1943) found that the latter were mainly
dissolved at the beginning of the fermentation and the former a t the end.
In ciders the tannins constitute 32% of the total tannoids and 77.5%
in perrys, according t o Brugirard and Tavernier (1952). It is not often
that sufficient tannins are found in grape juice to cause clouding. Such a
case occurred in Austria in 1951 according to Prillinger (1952a). Pectin-
splitting enzymes were of no value but gelatine-fining prevented reoccur-

rence. Qualitative tests showed catechin tannins to be responsible.
According to Schanderl (1950), 0.2%t o 1.0% tannin is slightly inhibiting
on fermentation, but it was less toxic t o normal wine yeasts than t o
Kloeckera types. However, the use of small amounts of tannin in spar-
kling wine production is common and occasions no difficulty. NBgre
(1939a) reported that 200 p.p.m. of sulfur dioxide in the must did not
increase the tannin content of the wine but that addition of gallotannin
did result in better-colored wines. It is not hydrolyzed during fermenta-
tion. Use of tannin did not inhibit wild yeast or bacterial growth in
fermentation in the study of Turbovsky et al. (1934).
TABLEXVIII
Tannin and Coloring Matter Content of Various Types of Wine
California Wh. table Amerine (1947) 399 0 . 0 2 0.13 0.052
California Red table Amerine (1947) 282 0 . 1 0 0.38 0.21
California Wh. Dessert Amerine (1947) 407 0.01 0.11 0.05
California Red Dessert Amerine (1947) 81 0 . 0 4 0.16 0.08
Italy Table Lucchetti (1941) 12 0,011 0.310 0.129
Germany Wh. table Diemair et al. (1951) 90 0.004 0.064 0.022
Spain Montilla Casares and Gonzalee 51 0.005 0.174 0.067
(1953)
Various Red table Diemair et al. (1951) 42 0.019 0.100 0 . 0 5 8
Various Dessert Diemair et al. (1951) 51 0.008 0.072 0.028
Decrease in tannin during aging owing to combination with protein

was claimed by Beridze (1948). Since the protein content of wines is small
and the decrease in tannin rather large, this hardly seems t o be an ade-
quate explanation. Joslyn and Comar (1941) showed that there is a
gradual decrease in tannin as aldehydes are formed and that sulfur
dioxide inhibits this decrease. Removal of tannins with casein or activated
charcoal in the preparation of pharmaceutical wines was recommended
by Ross (1942). Removal of excess tannin in white wines by means of
gelatine-fining was recommended by Moreau and Vinet (1937). Rent-
schler and Hauser (1950) reported that addition of 0 to 1000 mg. of
gelatine per liter reduced the tannin content from 0.165% t o 0.025%.
Zlataroff and Poppoff (1937) studied the inhibitory effect of red and white
wines on enzymes such as lipase, oxidase, peroxidase, invertase, and
urease. The effect is attributed to the tannins. NBgre (1939a) also con-
sidered tannin of some value in determining the resistance of wines to
disease. Turbovsky et al. (1934) found tannin did retard the growth of
spoilage bacteria (Lactobacillis) and considered that 0.05 g. of tannin per
100 ml. of wine improved low-tannin white and red wines. Nowadays
tannin is seldom used. Cruess (1935) called attention to the influence of
COMPOSITION O F W I NES 44 1
tannin in the flavor of red wine and noted that low-tannin wines are more
subject to disease than high-tannin wines. Fornachon (1943) reported
similar results with fortified wines in Australia.
Amounts. I n fifty-six genuine red ports Vasconcellos (1940a) found
0.01 % to O.l6y0 tannin. I n sixteen white ports the range was from 0.01%
to 0.04%. These values should be multiplied by 2.4 t o be comparable to
those of the Neubauer-Loewenthal procedure. Other values are reported
in Table XVIII.
2. Color-Red Wines
The color of red wines has been studied much more than that of
whites. The importance of color to the commercial value of red wines was
stressed by Vogt (1935), Winkler and Amerine (1938), and others.
Methods. Many methods for measuring the color of red wines have been employed.
The best are those based on optical transmission. However, other procedures have
been more popular because of the labor involved in spectrophotometric procedures.
Roos (1930) stated that the color density of a red wine is more important commer-
cially than the tint. He used a standard of permanganate and dichromate and diluted
the wine with 0.5% sulfuric acid until the colors matched, the number of milliliters of
acid solution added being a measure of the color. The procedure has several faults-
many wines are less acid than 0.5% and the tint will change. Using a color standard,
Vogt (1935) reported on the color value of various German wines. Comparisons with
the color standard were made a t six different depths and the average value calculated.
Color values from 5 to 1470 were reported. The limitations of this method of expression
of color were indicated b y Amerine and Joslyn (1951) and Winkler and Amerine
(1938). Kielhofer (1944) modified the Vogt color standard to give a better comparison
as follows: 78 mg. alizarinastroviolet B, 14 mg. brillantcrocein MOO, and 4 mg. Griin
P L X made up t o a liter. Winkler and Amerine (1038) compared the Lovibond slides,
color standards, the Dujardin-Salleron “vino ” colorimeter, and spectrophotometric
data. They appear to have been the first to use transmission-curve data for the calcu-
lation of brightness (luminance), dominant wavelength, and purity of wines. For data
on the principles involved see Mackinney and Chichester (1954). Nedeltscheff and
Kondareff (1941-1942) used 1 % Bordeaux red for specifying the color. The ratio of the
optical density measured a t 480 mp and 640 mp was found by Boutaric et al. (1936)
to be a good measure of the color of red wines. They gave transmission curves on six
red wines. Mixtures of two wines gave values close to the calculated optical density
a t three different wavelengths. However, with dilution the color density does not obey
Beer’s law, particularly in highly colored wines. During neutralization the optical
density increases and then decreases. Boutaric et al. (1937) also studied tthe effect of
dilution on the optical density and found t h a t it decreased more rapidly than would
be indicated by Beer’s law. This they interpret as indicating that the color is not in
true solution but present in a colloidal condition as micelles. On dilution the equi-
librium between the micelles and the intermicellar liquids is disturbed (e.g., the
micelles are broken up) and the optical density decreases.
Faure and Pallu (1935) determined the optical density at 460 mp with a photo-
electric colorimeter and proposed the color unit ROB, which is K times the optical
density, where K is chosen so that the 100 ROB is a standard red wine color. There is
no assurance that i t would be satisfactory for a variety of wines of varying purity,
luminance, and dominant wavelength. Korotkevich (1951) measured the color formed
by making a diluted wine basic as a measure of the intensity of the original color
present. A Pulfrich photometer with an 843 filter for whites and an S53 for reds was
employed.
Genevois (1951) reviewed the methods used for color. He suggested bromination
or titration with titanium trichloride at a high pH. Zinc has been used for anthocyanin
reduction. However, oxidation of acidulated wine by means of permanganate is the
commonest procedure.
Components. The primary anthocyanins of V . vinifera grapes appear
to be the monoglucoside oenin and some diglucoside. Levy et al. (1931)
found a small amount of delphinidin and its 3’-methyl ester in the
Fogarina (i.e., Fogaruna?) grape. The absorption curve of natural and
synthetic oenin chloride (malvidin) from 420 mp. to 600 mp. and the
distribution numbers in a special solvent were identical for the synthetic
and natural product. The picrate consisted of only about 45% oenin-
the remainder may be petunidin (delphinidin 3’-methyl ester) and
delphinidin.
In V . labrusca it may be a mixture of oenin and monomethoxy-
delphinidin monoglucoside or, as Brown (1940) pointed out, a mixture of
oenin, delphinidin monoglucoside, and monomethyoxydelphinidin mono-
glucoside. In V . rolundifolia Brown (1940) reported the red color is
probably a 3,5-diglucoside of 3’-O-methyldelphinidin, which he named
muscadinin. The anthocyanin isolated by Cornforth (1939) from an
Australian wild grape vine, Vitis hypoglauca F.v.M. was oenin and con-
tained little or no delphinidin or its methyl esters. It therefore resembles
the European species more than the American. Sastry and Tischer
(1952b) identified chlorophyll, water-soluble yellow pigments, and
carotene in addition to anthocyanins in the skins of Concord ( V . labrusca)
grapes. The anthocyanin was oenidin 3-monoglucoside. In paper chroma-
tography studies, anthocyanidin and the diglucoside were identified.
Tannins, especially those from grapes or grape stems, had a protective
influence on the destruction of the anthocyanins by ultraviolet light.
The color of hybrids of American and European species of grapes was
studied by Violante (1948). Although his procedure was not critical, he
believed that his transmission curves justified concluding that hybrids
contained a mixture of anthocyanins of both parents. He noted particu-
larly the blue color in V . rupestris hybrids. Sudario (1953) reported the
red color of American species of grapes and of their hybrids to have a
lower methoxy content than that of V . uinifera grapes. The methoxy con-
tent of the color of the wines was lower and the differences between species
less. Variable rates of loss of color from wines by different varieties of
grapes were reported by Amerine and Winkler (1947), who concluded
that a mixture of pigments were probably present.
Separation of the anthocyanins by partition chromatography was

reported by Spaeth and Rosenblatt (1949, 1950), who utilized columns of
silicic acid with 10% phosphoric acid as the immobile aqueous phase and
a mixture of phenol and toluene as the nonaqueous phase. Following
separation the anthocyanidins were quantitatively determined by means
of optical density measurements. Mathers (1951) purified prior to chro-
matographing. He used calcium chloride-methyl alcohol to precipitate the
color, dissolving in phosphoric acid-isopropyl alcohol, adding ether, and
passing the solution through a silicic acid column.
Vil'Lrns and Taranova (1950) separated oenin and oenidin by a modi-
fied picrate method (oenidin is precipitated but oenin is not) and the
extinction coefficient was determined at 520 mp. The sum of the extinction
coefficients was only a little less than that of the wine itself. Durmishidze
(1948a) measured the oenidin content by dealcoholizing, and freeing of
other pigments and of glycerol, pectin, and mannitol. The color and tannin
are then separated as lead salts and the methoxy content determined by
the Zeisel procedure. He reported 4 % to 6 % oenidin in the skins of red
grapes and 0.06% in red wines. Durmishidze and Khachidze (1952) pre-
pared pure oenidin b y the picrate method from the skin of the Saperavi
grape and constructed a table for the conversion of photometer values t o
concentration. They used a 50% alcohol solution at a p H of 1.2 to 1.4.
Oenin and oenidin were separated with amyl alcohol. By this method only
anthocyanins and anthocyanidins which have not changed color can be
determined. Vil'gms and Taranova (1951) also absorbed the wine pig-
ments on absorbent cotton previously washed with ether and dried in
vacuum. On passage of red wine through cotton columns two zones were
formed. Elution with acidified water (pH 1 to 2) removes the oenin and
with acidified 50% ethyl alcohol (pH 1 to 2) removes the oenidin. At
higher pH values the pigments are unstable. Both eluates were made u p
to volume with alcohol, and the extinction values were determined with
the Pulfrich spectrophotometer a t 530 mp (filter S53).
The oenin and oenidin contents of several wines were as follows:
Oenin, Ocnidin,
Wine and date Treatment W.P. mg./l.
Saperavi, 1946 hTone 288 260
Saperavi, 1949 None 317 300
Saperavi, 1949 Steam, 1 hr. at 95" C. (203" F.) 245 190
Saperavi, 1949 Dry air, 7 days at 28" C. (82.4' F.) - 165
Saperavi, 1949 3 months storage 81 134
Saperavi, 1949 Dry air, 15 days at 48" C. (118.4" F.) - 39
Dneprooskoe, 1949 None 115 50
Dneprooskoe, 1949 Dry air, 7 days at 28" C. (82.4" F.) 101 44
Dneprooskoe, 1949 Dry air, 15 days at 48" C. (118.4' F.) 83 45
Source. The source of white and red wines is derived from grape pig-
ments. Ducellier (1935) found the pigment in a state of granulation or in
solution in the cells of the hypodermis of white-juice varieties. Amerine
and De Mattei (1940), in order to release the color, used very hot water
or steam to destroy the semipermeability of the cells. Quantitative data
on the time-color relationship for temperatures of 70" to 90" C. (158" t o
194" F.) were reported by Joslyn et al. (1929). Sastry and Tischer (1952a)
studied the influence of heating Concord grapes for varying periods a t
170", 210", and 250" C. (338",410", and 482" F.). After 63 minutes a t the
highest temperature there was pigment destruction but less under nitro-
gen or in the dark than in air. Both the mono- and diglucosides of the
anthocyanidin decreased. Using pure monoglucoside solutions even
greater effects of temperature and oxidation were observed.
Removal of the aglucon fraction results in cessation of oxygen absorp-
tion, according t o Chogovadze (1948). At high fermentation temperatures
the aglucon fraction increased-not only from the skins but from hy-
drolysis of tannins. This also occurred during heating, the source in this
case being leuco compounds.
Kaczmarek and Weise (1942) compared the transmission curves of
musts, fermenting musts, and wines. They found differences in the shape
of the curves between varieties, and in the height of the curves between
grapes of different maturity or grapes grown in different regions. The
changes in color with variations in pH were studied by Casale (1930d).
He found the isoelectric point for the color change of oenin to be p H 5.4
to 5.8.
Aging and Fining Egects. The function of the anthocyanins in the
aging of wines has been stressed by Genevois (1951). He reported pro-
gressive demethoxylation during the first three years ; the pigments
gradually become colloidal and are no longer reversibly reduced by hydro-
sulfite. Genevois postulated that the latter was a result of demethoxyla-
tion which gives an acidic orthodiphenol compound that condenses with
other compounds. According to Heide (1940) precipitation of red wine
color depends on the fining agent used, amount of sulfur dioxide and iron,
and other factors.
Detection of Sophistication. I n Europe where red wines of full color are
desired, and the lack of maturity tends to prevent full color development,
addition of foreign coloring matter is occasionally practiced. Various
methods have been devised for the detection of such additions.
The absorption spectra of dye solutions and colored wines were
compared by Casamada (1931). The differences found were too small to
allow conclusions as t o adulteration. The color absorption curve from
430 to 750 mp. of wines and of artificial colors was determined by Mon-
tequi (1933), and the possibility of detecting Bordeaux red by this means
was considered. On dilution he did find that the optical density of wine
obeys Beer’s law. Violante and Bemporad (1937) determined the spectral
absorption curves of red wines and various artificial colors. They showed
that the log of the molar-extinction coefficient is independent of concen-
tration and might be used to detect adulteration.
Use of elderberry (Sambucus nigra and S. Ebulus) juice for coloring
red wines, particularly port, is an ancient practice. Waser et al. (1932)
determined the absorption curve from 200 to 560 mp and demonstrated
that elderberry wine could be detected from the shape of the absorption
curve in the region of 250 to 350 mp. A chromatographic technique with
aluminum oxide was developed by Popov (1947-1948) for detecting
elderberry pigments. Whortleberry, kermes, beet, cochineal, and twenty-
four acidic, basic, or substantive dyes could also be detected, either alone
or in admixture with elderberry juice.
To detect hematin, basic and acid fuchsin, scarlet red, and Bordeaux
B and R in wines Gentilini (1939, 1941) absorbed on magnesium oxide
and used acetone as the washing liquid. Ajon (1943) used aluminum
hydroxide for detecting artificial color. Mohler and Hammerle (1935)
separated the natural pigments of red wines from artificial colors by
passing them through aluminum oxide. The artificial colors passed
through easily, whereas the anthocyanins were strongly adsorbed. The
absorption curve showed a maximum a t 513 to 518 mp and a minimum
a t 410 to 413 mp for the wine pigments, and a maximum a t 517 mp and
minimum a t 438 mp for the artificial colors. Mohler and Hammerle (1936)
also used chromatographic columns to detect white wine in red. The
eleutant from zones 1 and 2 were colorimetrically and spectrophoto-
metrically compared for untreated wines and untreated wines to which
50% decolorized wine had been added. Both indicated about 50%
dilution. A sort of primitive paper chromatography was used by Venezia
(1940) for detecting foreign colors, such as Bordeaux red, in wines. Ruf
(1952b) used paper chromatography to detect red beet, whortleberry,
blackberry, elderberry, Bordeaux red (synthetic and vegetable), and
synthetic raspberry color in a red wine.
To determine whether a white wine was made even partially from red
grapes or had been decolorized by charcoal, von der Heide (1932) mixed
10 ml. of wine and 3 to 5 ml. of 10% hydrochloric acid, a pink or red
color so indicating. The procedure is not new.
Various procedures for detecting addition of organic dyes were studied
by Conceiggo (1942). The Arata procedure was sensitive to 0.002 mg. per
liter, and a simplified modification could detect 0.01 mg. According to
Ferrari (1939, 1942) addition of artificial colors can be detected by adding
silver nitrate to a wine containing potassium bromide-a brown silver

bromide precipitate indicated a pure wine and a red precipitate a basic
artificial color. T o detect artificial colors in wines, Maravalhas (1935)
acidified, then shook them with xylene. The xylene is then absorbed on
filter paper and examined. A capillary luminescence procedure for detect-
ing added color in red or white wines was developed by Kocsis el al. (1941).
This consists of allowing 2 ml. of wine to spread onto a Schleicher and
Schull No. 602 filter paper, then examining the paper under ultraviolet
light. By this means they could demonstrate addition of cochineal, elder-
berry, or mallow leaves in red wines and of caramel or safflower (an
extract of saffron) to white wines. Karamboloff (1932) tested for the
presence of added colors by adding 3 to 5 ml. of 30% hydrogen peroxide
to 5 ml. of wine on a watch glass. Natural wines turn yellow within 15
minutes, whereas those t o which artificial colors have been added remain
red. However, hybrids (such as Othello) were decolorized only in 24 hours.
A new procedure for the detection of elderberry color was developed
by Wobisch and Schneyder (1952). It is based on the fact that there is
only one OH group on the benzopyrylium portion of oenidin, whereas
"c"=~=o
there are two on the elderberry chrysanthemin pigment. With the
dihydroxy colors borate forms a red addition compound as follows:
OH
0
vc\
C o-c~H~~o~
H
+
Therefore when borates are added to red wines in alkaline solutions, the
color remains blue, but if the wine contains elderberry color, there is a
shift to the red.
The special problem of detecting grape wine in berry wine has arisen
in this country because of the cheaper price of the former. The per cent
transmittance of authentic acidified and unacidified grape and fruit wines
in the ultraviolet and visible regions of the spectrum was determined by
Beyer (1945). When the curves from acidified and unacidified samples
were plotted on the same graph, the curves cross to the right of 590 mp
in red grape wines (except for wines of the variety Pinot noir) and t o the
left in red berry wines. If one calculates the ratio of the spectral trans-
mittance a t 590 rnb before and after addition of acid, a ratio of over 1
indicates grape wine and less than 1 indicates berry wine (except for the
low-color variety Pinot noir). With berry juices and wines Yang and
Wiegand (1950) reported a marked change in the shape of the absorption
curve during aging, as indicated in Fig. 3. This shows a marked reduction
in the ratio of the extinction coefficient at 515 mp and 340 mp.
1.0 1 3 E5'5'E340
Months 1.039 1
I Year 0.586
2 Years 0.424
3 Mo.
3 Years 0.280
1 I I
0
400 500 600 M I
FIG.3. Ratio of the extinction coefficients Es~s/Etrofor a loganberry wine (Yang
and Wiegand, 1950).
3. White Wines
No carotene or xanthophyll was found in white wines by Peyrot
(1934). From a study of the spectral absorption curve he concluded that
white musts contain a red pigment, which decreases rapidly during
vinification and rarely appears in the finished wine. The absorption
spectra of a white wine and of solutions of quercitin and quercitrin were
found by Peyrot not to be similar. He did find that a 0.008% solution of
cyanine (a reduction product of quercitin) gave a curve similar to that of
wine. The properties of the flavonols and other pigments of white wines
are reviewed by Genevois (1951). I n normal white wines they are stable,
but in old white wines exposed to air there is a darkening and precipita-
tion of the color. Genevois (1934a) attributed the slight fluorescence
of white wines to be due to 0.04 to 0.1 mg. per liter of flavine and lumi-
flavine. The fluorescent material was extracted by trichlorethylene. He
believed the flavine might arise from the musts or yeasts.
Chromatographic analysis was used by Williams and Wender (1952)
to identify isoquercitrin in Vitis vinifera. Isoquercitrin is the glucoside
of quercetin, and its isolation appears to be new. Previously quercitrin,
the rhamnoside of quercetin, had been reported. To detect foreign coloring

matter in white wines, Ajon (1943) added potassium alum and sodium
hydroxide t o bring to the isoelectric point. A colored supernatant liquid
after centrifuging was considered a positive test.
IX. NITROGENOUS COMPOUNDS
Much of the available information concerning the nitrogenous sub-
stances in wines has been obtained during the period under review. With
chromatographic techniques even more rapid progress can be anticipated.
Since most of the papers have been on the nitrogen complex as a whole,
the various compounds will be discussed together, except in the sum-
maries of the analyses.
1. Methods
Methods for determining total (by Kjeldahl) protein, ammonia, phosphotungstic-
precipitable, basic and amino nitrogen in musts and wines were developed by Muth
and Malsch (1934). Peynaud (1939b) and Shcherbakov (1940) reviewed the pro-
cedures. For total nitrogen Reichard (1943) used hydrogen peroxide and sulfuric acid.
Total nitrogen in musts and wines was rapidly determined by Garcfa and Freyre
(1951) by wet ashing with selenium and perchloric acid and distilling into Nessler’s
reagent. Results comparable t o those obtained by the regular Kjeldahl procedure
were obtained on 1 ml. in about 10 minutes.
Amino Acids. Castor and Guymon (1952) and Castor (1953b) used microbiological
assay techniques. Ribbreau-Gayon and Peynaud (1947a) gave a colorimetric tech-
nique for tryptophane and utilized the Sbrensen procedures for amino nitrogen. Muth
and Malsch (1934) used the Willstiitter technique for amino nitrogen. Sisakian and
Bezinger (1949), Valaire and Dupont (1951), and Liithi and Vetsch (1952) used paper
chromatography for identifying and roughly determining the amino acids in grapes
and wines.
Proteins. Almeida (1948) has used a polarograph t o identify certain proteins. From
the similarity of the curves obtained with the extract of a port wine and pure solutions
of albumins or globulins, he was able to estimate the protein content. A port wine
deposit showed a relatively high protein content. He did not find peptones.
2. In Grapes
The changes in total nitrogen, ammonia, and amide nitrogen in five
Bordeaux varieties during the ripening period of 1937 and 1938 was
reported by Peynaud (1939~).On a volume basis the total and amide
nitrogen increased during ripening. On a per berry basis all three in-
creased. Whereas Casale (1935-1937) believed the ratio total nitrogen/
ammonia t o be fairly unique for each variety, Peynaud did not find it so
specific. Casale reported 5 to 120 mg. per liter of ammonia and 56 to
670 mg. of total nitrogeii in Piedmont musts, whereas Peynaud found 19
to 144 of ammonia and 156 to 870 of total nitrogen. There was more
variation between localities than between varieties in the latter’s study.
A review of the nitrogen fractions present in musts and wines was given
by Genevois and Rib6reau-Gayon (1935b). The total nitrogen content of

Bordeaux wines varied from 300 t o 600 mg. per liter. New wines of
different regions have varying amounts of ammonia and the varieties of
grapes from the same region also differ in total nitrogen content. Wines
made from insect-injured grapes had abnormally high nitrogen contents.
Wines from the same variety growing on granitic soils contain more
nitrogen than those from schistic soils, according to Oliveira (1942). The
decrease in nitrogen content of grapes attacked by Botrytis cinerea is very
large, and this may favorably influence the stability of the resulting
sweet wine. Schanderl (1950) has summarized the data on the subject.
Barbera (1933~)believed the amino acids in wines were not primarily
derived from the yeast but from the fresh grapes.
3. I n Fermentation and Aging

The importance of the nitrogenous components of the must t o the
fermentation and clarification, aging, and character of the resulting wine
has been known for many years. However, quantitative data on many
of the nitrogenous compounds are lacking, particularly on their specific
relationship t o fermentation, clarification, aging, or the character of the
final wine.
Tarantola (194713) has made a contribution in this direction. He dis-
cussed the influence of time of pressing on nitrogen content and the
utilization of ammonia and amino acids in the early stages of fermentation
when proteolytic enzymes are also active. Later changes are outlined and
the results of other investigators evaluated. Valaize (1949) studied the
disappearance of ammonia during fermentation in musts to which 0 , 100,
and 300 mg. per liter of ammonium phosphate had been added. Typical
results, as milligrams per liter of ammonia, were:
Amount No. 1 No. 2 No. 3
added Must Wine Must Wine Must Wine
0 82.6 8.4 63.0 7.0 61.6 6.3
100 102.2 7.7 82.6 7.6 85.4 6.3
200 135.8 6.3 98.0 7.0 119.0 4.9
It is obvious t ha t the added ammonia is used up. No changes in the com-

position of the wine were noted. He recommends use of ammonium phos-
phate for cold musts.
The total nitrogen content of musts and their wines was not influenced
by fertilizing the vineyards, according to Niehaus (1938). Loss of nitrogen
occurred between 12 and 48 hours after fermentation started; 50.7% t o
58.5% of the total nitrogen was removed by fermentation. Yeast strains
had little influence on the amount lost, but aeration increased yeast
growth and nitrogen removal. Contact of the new wine with the lees
increased its nitrogen content.
Hennig and Oshke (1942) and Hennig (1944) reported that several
forms of nitrogen (total, protein, ammonia, amino, humic, and phospho-
tungstic acid) all decreased during fermentation and then increased. Only
amide and residual nitrogen decreased. While the nitrogen content is
generally lower in a good (warm) year in Germany, the relationship can
not be used for judging quality. Blue fining (addition of potassium ferro-
cyanide to remove copper and iron) and gelatine-tannin fining decreased
the nitrogen content slightly-8.4 to 26.6 mg. per liter less of total
nitrogen in blue-fined wines. The yeasts appear not to add any protein
material to young wines, but bacteria rapidly use up the amino acids.
Saenko (1951) showed that small amounts of nitrogen (194 mg. per
liter from ammonia, amino acids, or autolyzed yeasts) stimulated the
growth of the sherry film. Addition of ammonia (60 to 120 mg. per liter)
reduced the period of film formation by 3 to 4 days and increased the film
mass by 50% to 70%. The stimulated growth of the film also hastened
the appearance of the sherry taste. For a pH of 3.1 to 3.2 more ammonia
may be added, up to 120 mg. of nitrogen per liter (1.25 ml. of 20% am-
monia per liter). This will raise the pH to the optimum, 3.3 to 3.4, for the
film growth. For wines with a pH of 3.3 to 3.4 the amount of ammonia
must be reduced to 60 mg. of nitrogen per liter. He concluded that addi-
tion of ammonia would increase the production of finished sherry wine,
improve the quality, and reduce losses, because a rapid growth of the
sherry film reduces the danger of an acetic acid fermentation.
The changes in nitrogenous substances in sparkling wines during:
fermentation in bottles and subsequent storage on the yeast were in-
vestigated by Oparin et al. (1945, 1946, 1947). Fermentation is rapid
during only the first 20 to 30 days. They found that practically all the
yeast cells died about 80 to 90 days after fermentation started. The dead
yeast cells, however, released enzymes (proteases), and subsequent
changes in the nitrogenous substances occurred. These involved trans-
formations which caused the Kjeldahl-nitrogen content to decrease
between 110 and 255 days and after 370 days returned to normal. So,
apparently, nitrogen compounds of unknown structure were formed,
which escaped detection by the Kjeldahl method. Before fermentation
about half of the nitrogen present in the wine is not precipitated by
tannin and is also not determinable by the Van Slyke method. The nature
of these nitrogenous substances is unknown. Most of the nitrogen trans-
formation were a t their expense. As the fermentation proceeded there was
an increase of amino and basic nitrogen. After 300 to 350 days, the
proteases are practically inactive, and no further changes in the nitrog-
enous compounds were observed. Oparin and Bezinger (1949) found a

nitrogen compound in sparkling wines that was not hydrolyzed by
proteolytic enzymes. Its amino acid content was also higher than that of
proteins, and it gave a test for reducing sugar. They believed that during
aging in the presence of yeasts a high molecular weight nitrogen-carbo-
hydrate compound is formed. High molecular weight nitrogen compounds
from bottle-fermented sparkling wine were isolated by concentrating
under vacuum, dialyzing, reconcentrating, and precipitating in 88 %
alcohol. These were apparently formed from carbohydrates and amino
acids and polypeptides in the presence of yeast autolysate. They also
reported free amino acids and polypeptides in sparkling wines and found
that as much as 40% of the total nitrogen content could be accounted for
by these. Independent confirmation of the changes in nitrogen content
during the aging of sparkling wines comes from Schanderl (1950). He
reports that after nine to thirteen months of aging, about one-third of the
nitrogen content of the yeast is found in the wine. This he naturally
associates with autolysis of the yeasts, as very large decreases in the
amount of yeast in the deposit in the bottles occurred during the same
period.
Castor (1950) reported a rapid decrease in amino acids during the
early stages of fermentation, but several of the amino acids increased in
the latter stages of fermentation, probably from autolysis. Addition of
ammonium ion showed a “sparing” effect for arginine and leucine but
not for glutamic acid or the other amino acids. Schaefer (1939) believed
the amino acids to be important to the character of the wine. On the
basis of a study of the nitrogen fractions Oliveira (1942) considered that
a relation between quality and the per cent nitrogen had been demon-
strated with port wines.
4. In Clouding
The colloidal nature of the nitrogenous material in Bordeaux white
wines was noted by Rib6reau-Gayon (1932). As much as 26 mg. per liter
of heat-coagulatable, negatively charged colloidal material, called
albuminoids, was found. These can be eliminated by heating, by adding
acids or tannin, by aging, by ultrafiltration, or by adsorption on kaolin.
A cloudiness followed by a precipitate in bottled wines, 80% of which
was organic, was described by Kielhofer (1942). A more detailed study
of the protein-clouding of 47 German wines produced in 1947 was made
by Kielhofer (1948-1949). Organic fining agents were not effective, and
silicic acid was recommended. Aging and blue fining were of some value.
This clouding was not associated with heavy metals and did not occur
until a temperature of about 20” to 25” C. (68” to 77” F.) was reached.
Clouding reoccurred only when the wine was raised to a higher tempera-
ture. Kielhofer (1949) reported bentonite reduced the nitrogen content
49 to 59 mg. per liter, but increased the mineral content 240 mg. per liter,
probably owing to the absorption of exchangeable sodium, calcium, potas-
sium, and magnesium. Ultrafiltration through a protein-tight membrane
also reduced the nitrogen content but did not prevent turbidity when the
wine was warmed. Heat- and tannin-precipitable, alkali-soluble material
containing 6.4% to 11.8% nitrogen was found in certain German wines
by Kielhofer (1951). He considered it a protein hydrolysis product, as i t
could not be removed by ultrafiltration. He has also reported (1948) that
the turbidity probably includes tannins. Roleff (1948) believed the tur-
bidity of the 1947 wines due to high calcium content-possibly from the
glass. Roleff (1949) noted that the changes in pH during heating might
influence the proteins and that protein turbidity could be caused by a
protein-tannin precipitate. However, Schmid and Nestle (1949) did not
find the calcium differences in the 1947 wines t o be great enough to cause
cloudiness. Furthermore, they reported turbidity in wines which were
not in contact with glass.
Archinard (1937, 1939) did not consider ammonia to be a measure of
spoilage in wines, although French law sets a maximum of 200 mg. per
liter.
5 . Amounts
One of the first to fractionate the nitrogen compounds of wine was

Barbera (1933~). He reported only three-ammoniacal, amide and amine,
and total nitrogen. His data indicate the possible amount of amino acids
in wine and also how much is removed by fermentation. The first exten-
sive fractionation of the nitrogenous material in wines was that of Muth
and Malsch (1934). The following indicates some of their results (as
milligrams per liter) :
Amino
before and
Phospho- after Total by
Protein Ammonia tungstic hydrolysis Humin Sum* Kjeldahl
Geisenheimer
(1931) 13.3 45.0 313.6 223.3 274.5 9 . 2 655.6 665.8
Geisenheimer
(1932) 10.5 19.1 145.4 - 194.2 9 . 8 379.0 376.8
Geisenheimer
(1930) 11.2 35.9 221.0 203.7 251.1 8 . 6 530.8 536.5
Geisenheimert
(1904) 7.3 15.9 120.9 120.2 148.0 9 . 1 294.2 301.2
* Without amino before hydrolysis.
t An audese, a wine of late-piaked grapes.
In Swiss wines Godet and Martin (1946) reported 0.0656%, 0.0531%,

and 0.1425% of nitrogen as protein. Mestre and Mestre (1935) frac-
tionated the organic nitrogenous materials in musts and wines. The pro-
cedures used were not very specific, but tryptophane, glycine, cystine,
leucine, and arginine were tentatively identified in musts, and the last
two also in wines. When wine spirits are added prior to fermentation
much of the nitrogen is precipitated; this had an adverse effect on the
quality of the wines. However, when pure alcohol was used, no harmful
effect on quality was noted. Venezia (1935) determined the amino nitrogen
content of musts (116 to 358 mg. per liter) and of wines (56 to 245 mg. per
liter). Red musts and wines were generally higher than whites. He sug-
gested that the association of amino nitrogen and the yeast might explain
differences between the odor of wines of different origins (and varieties?).
Gentilini (1937) reported a total nitrogen of 0.0360% in a Cabernet franc
wine. Of this 0.0015% was ammonia nitrogen, 0.0019% amide nitrogen,
0.018% amino nitrogen, and 0.015% protein nitrogen. The methods used
may account for some of the results.
Using paper chromatography Sisakzn and Besinger (1949, 1950)
identified glutamic acid, alanine, valine, and proline as the amino acids
present in wines to the greatest degree. Lesser amounts of aspartic acid,
serine, and threonine were found and identification of phenylalanine was
doubtful. Later Sisakgn et al. (1950b) reported tryptophane in eight
wines, varying in amount from a trace to 6.4 mg. per liter (average 1.6).
In two Kakhetinskikh wines they reported a total nitrogen of 207 and
330 mg. per liter, of which 45% and 46.6%, respectively, were amino
nitrogen. The Van Slyke diazotation procedure for amino nitrogen gave
lower results than the determination of the free amino acids. The poly-
peptide content of one sample was 13% of the total nitrogen.
Castor and Guymon (1952) reported 7 to 10 mg. per 100 ml. each of
leucine, isoleucine, and valine in a grape juice of Vitis vinifera and less
than 2 mg. in the resulting wine, Castor (195313) reported glutamic acid
in the largest amounts in five of seven varieties (26.5 to 160.7 mg. per
100 ml.). Arginine was found in higher amounts in two varieties (7.0 to
113.0). The amounts of the other fourteen amino acids reported by him
were usually below 10 mg. per 100 ml., and lysine, methionine, glycine,
and cystine were present a t 2 mg. or less. From 75% t o 90% of the
amounts originally present were removed by fermentation, except for
glycine, lysine, and cystine, which showed little change during fermenta-
tion. After fermentation five amino acids were returned in small amounts,
presumably owing to yeast autolysis, namely, valine, isoleucine, leucine,
tryptophane, and tyrosine. Durmishidze and Mosiashrili (1948) showed
that a triose and not pyrotartaric acid was the hydrogen acceptor in the
diaminization of glutamic acid. When glutamic acid was added t o musts,

the succinic and acetic acid and glycerol contents were proportionally
increased.
The importance of amino acids to Bacterium gracile, one of the bac-
teria in the malo-lactic fermentation, was stressed by Luthi and Vetsch
(1952). A summary of the amino acid reports t o date follows:
Sisaksn and Valaise and Liithi and
Reich* Besinger Dupont Vetsch Castor
Amino acid (1950) (1949) (1951) (1952) (1950, 1953b)
Alanine - - Yes Yes -
Arginine Yes - - - Yes
Aspartic acid Yes Yes - Yes Yes
Cystine Yes - - - Yes
Glutamic acid Yes Yes - Yes Yes
Glycine - - Yes Yes Yes
Histidine Yes - - Yes Yes
Isoleucine - - Yes(?) - Yes
Leuoine Yes - Yes ( 1 ) - Yes
Lysine Yes - - - Yes
Methionine - - - - Yes
Phenylalanine - ? - Yes Yes
Proline Yes Yes Yes Yes -
Serine - Yes - Yes
Theonine - Yes - Yes -
Tryptophane Yes Yest - - Yes
Tyrosine Yes - Yes - Yes
Valine - Yes Yes Yes Yes
* It is not clear from the text whether Reich actually determined these or not.
t Sisakzn et al. (1950b).
The amounts of total nitrogen, amino nitrogen, amide nitrogen,

ammonia nitrogen, and other nitrogen fractions are given in Tables XIX,
XX, XXI, XXII, and XXIII.
X. ENZYMES,
VITAMINS,AND AROMATIC
CONSTITUENTS
Because the chemical composition of enzymes is poorly defined, they
will be omitted from this review. For a summary of information on the
enzymes in grapes and wines see Delp (1932), Requinyi and S o b (1935),
Baglioni et al. (1935-1937), Casale and Garino-Canina (1935-1937),
Venezia (1937), Manskaya (1939), Manskaya and Emel’yanova (1939),
Oparin and Manskaya (1939), Hussein and Cruess (1940a, b), Hussein
el a2. (1942), Cruess (1943), Ribdreau-Gayon (1943), Osterwalder (1945),
Garino-Canina (1945-1946), Rodopulo (1948), Durmishidze (1950b, c),
Popova and Puchkova (1950), Rentschler (1950), Rodopulo (1950a, b),
and Tarantola (19504.
TABLEXIX
Total Organic Nitrogen Content in Various Types of Wine
(Grams per liter)
Bordeaux Red table Peynaud (1939b) 35 0.143 0.666 0.330
Bordeaux Wh. table Peynaud (193913) 36 0.077 0.247 0.185
Bordeaux Table Peynaud (1950a) 12 0.168 0.394 0.246
France Table Valaize and Dupont (1951) 72 0.100 0.952 0.473
Germany Wh. table Reichard (1943) 50 0.560 0.980 0.766
Germany Wh. table Remy (1932) 10 0.294 0.908 0.491
Italy Wh. table Dalmasso and Dell’Olio 145 0.046 0.201 0.102
(1937)
Portugal Dessert Oliveira (1942) 64 0.106 0.215 0.152
Roumania Table ,Sumuleanu and 33 0.23 0.71 0.39
Ghimicescu (1936)
(1952)
(1952)
Switzerland Table Godet and Martin (1946) 3 0.111 0.240 0.156
TABLE XX
Amino Nitrogen in Various Types of Wine
(Grams per liter)
France Table Valaize and Dupont (1951) 73 0.059 0.165 0.083
TABLEXXI
h i d e Nitrogen in Various Types of Wine
(Grams per liter)
TABLE XXII
Ammonia Nitrogen (as Ammonia) in Various Types of Wine
France Table Valaize and Dupont (1951) 73 0.017 0.201 0.101*
France Table Archinard (1937) 8 0.007 0.069 0.020
Portugal Dessert CJliveira (1942) 35 0.011 0.030 0.016
Roumania Table Vumuleanu and 33 0.002 0.048 0.009
Ghimicescu (1'336)
(1952)
(1952)
Switzerland Table Godet and Martin (1946) 3 0.012 0.026 0.014
* There is no clear explanation for these high values.
TABLE XXIII
Other Nitrogen Fractions in Various Types of Wine
(Grams per liter)
Fraction Region Source of data samples mum mum age
Humin Germany Hennig (1944) 19 0.005 0.018 0.009
Peptone * France Peynaud (1950b) 8 0.008 0.018 0.014
Peptonet France Peynaud (1950a) 12 0.013 0.032 0.020
Phosphotungstic Germany Hennig (1944) 19 0.006 0.034 0.018
Protein France Peynaud (1950b) 8 0.001 0.011 0.003
Protein Germany Hennig (1944) 19 0.008 0.025 0.013
Protein Switzerland Godet and Martin 3 0.009 0.023 0.014
(1946)
Nitrates Roumania Vumuleanu and 33 0.003 0.010 0.005
Ghimicescu
(1936)
Residual Germany Hennig (1944) 19 0.007 0.016 0.010
*Listed as peptone and polypeptide and does not include protein.
t Probably includes protein and polypeptide.
Among those reported are : oxidase, tannase, invertase, pectinase,
ascorbase, catalase, peroxidase, dehydrase, polyphenoloxidase, esterase,
and a proteolytic enzyme.
1. Vitamins
Merzhangn (1930) reviewed what little was known of the vitamin
content of grapes and wines in 1930; these were mainly qualitative
studies of ascorbic acid content. A summary of Randoin's work on the
vitamin content of French wines was given by Hugues (1934). Minz and
Sirianni (1934) concluded that grapes contained only small amounts of
vitamin A, thiamin, and riboflavin, moderate amounts of ascorbic acid,
and no vitamin D. Parro (1948) summarized Portuguese studies on the
vitamin content of wines and recommended further studies in which the
ecological factors would be considered. On the basis of her studies Morgan
(1941) concluded that grapes supply (on the average) only 3% of the
daily nutritive requirements (of vitamins, calcium, and iron) for adults.
Addition of vitamins to wines has occasionally been recommended.
Ascorbic acid added t o wines was retained rather poorly according t o
Randoin and Gallot (1941)-only 7.5% after two months. Addition of
tannin did not affect the retention of ascorbic acid. Vetscher and Losa
(1947) added 100 to 200 mg. per liter of ascorbic acid and 50 to 100 mg.
per liter of thiamin to champagne cuv6es before bottling. Little influence
on odor or composition was noted with ascorbic acid, but thiamin resulted
in a deterioration in flavor.
Use of vitamins as accessory growth factors was studied by RibBreau-
Crayon and Peynaud (1952). They added 25 micrograms per liter of
biotin, 10 mg. of inositol, and 0.5 mg. of thiamin, and a mixture of all
three. Each accelerated the growth of yeast, but thiamin alone was as
good as the combination. RibBreau-Gayon et al. (1952a) isolated a sub-
stance from an extract of Botrytis cinerea or Aspergillus sp. which favored
yeast growth and increased the amount of glycerol and succinic acid
formed. Rib6reau-Gayon et al. (1952b) also found fermentation inhibitors
in the Botrytis extract. Use of thiamin when the fermentation starts
slowly was recommended by them; however, it would seem that more
data on its influence on quality should be obtained.
Ascorbic Acid. The practice of adding green walnut hulls or their
pressed juice t o wines is an ancient one. It has usually been considered
that this was done to add color and flavor, but Jeroch (1947) suggested
that the high ascorbic acid content of the hulls might be the basis of the
practice.
To determine the total ascorbic acid, Genevois (1938) recommended
reduction of dehydroascorbic acid with cystine, fixing the cystine with
formaldehyde, and titration with dichlorophenolindophenol. Gerasimov
and Vinogradova (1931) measured the ascorbic acid content of 150
Crimean grapes and 40 wines, using the colorimetric Bezsonov procedure.
Onokhova (1937) tested 74 varieties in Russia for ascorbic acid. T h e
cultivated sorts had only 2 to 5 mg. per 100 ml. of juice and the wild sorts
7.5 t o 12.5. Cahill (1933) reported that the hexuronic acid isolated from
grapes had no antiscorbutic value and might be 5-ketogluconic acid. More
likely i t was dehydroxyascorbic acid.
The reducing property of the juice of nearly ripe grapes is largely due
to ascorbic acid, according to Genevois (1938) and Genevois et al. (1938b),
but another oxidation-reduction system is also present. This second
system, with a high rH, is responsible for the darkening of musts. Gatet
(193913) emphasized the influence of the ascorbic acid on the oxidation-
reduction potential of the grape berry. Arutunyan (1939) reported grape
leaves to be an excellent source of ascorbic acid-4,000 I.U. per kilogram
-and recommended them as a possible source for the food industries.
Kramer and Satterfield (1942) reported no ascorbic acid in a red and
white wine of Scuppernong grapes after one year storage. Verda (1940)
showed that addition of citric acid did not improve the retention of
ascorbic acid in wine. Lesnovskaya and Vecher (1939) recommended
storage, filtering, and bottling wines without access of air t o prevent loss
of ascorbic acid.
Schon et al. (1939) reported 1.8 mg. per 100 ml. of ascorbic acid in
Portuguese red wines-slightly less than in the fruit. However, Scheurer
(1944) reported 135 mg. per kilogram of grapes of ascorbic acid but only
3.6 mg. per kilogram in the wine. Ascorbic acid decreased from 134 mg.
per kilogram of grapes to 3.6 mg. per liter of wine in Lodi’s (1943) investi-
gation. This was probably due to enzymic oxidation. Rather extensive
data on the ascorbic acid content of Italian grapes were reported by
Venezia (1938, 1944), who believed that fresh grapes might have thera-
peutic value. The ascorbic acid contents of a number of varieties of grapes
from several other authors have been recently summarized by Amerine
and Joslyn (1951). They showed quantities of 1.2 t o 18.3 mg. per liter.
Vitamin A . Daniel and Munsell (1932) in rat tests reported small
amounts of vitamin A in fresh grapes but none in commercial grape
juices. Unsulfured, soda-dipped, dehydrated raisins retained their
vitamin A and B contents rather well in the tests of Morgan et al. (1935).
Sun-drying or storage, even in air and when frozen, allowed rapid loss,
and if they were treated with sulfur dioxide prior to drying, the B was also
lost. Schon et al. (1939) found little vitamin A or carotenoid pigment
(about 0.005 mg. per liter) in a Portuguese red wine. Lodi (1943) reported
0.715 mg. per kilogram of carotene in grapes and none in the wines. The
skins and seeds contained most of the carotene. Loss during fermentation
is due to precipitation and enzymatic oxidation.
Thiamin. Lane et al. (1942) reported raw grapes to contain 0.67 mg.
per kilogram and raisins, 1.06 mg. Flavier (1939) reported 0.5 to 0.8 mg.
per kilogram in green Cabernet and Sauvignon blanc grapes, 0.2 to 0.4 in
ripe grapes, and 0.05 to 0.11 after fermentation. Daniel and Munsell
(1932) reported fair amounts of vitamin B in two varieties of grapes but
little or none in commercial grape juices. When thiamin was added to
grape juices or wines, i t was retained according to Perlman and Morgan

(1945). About 50% of the thiamin was destroyed by sulfiting the must
before fermentation or filtering through bentonite after fermentation,
according t o Perlman and Morgan (1945).
The thiamin decreased from 0.316 mg. per kilogram to 0.007 in one
case and from 0.134 to 0.07 in another in Lodi’s (1943) study. Esterifica-
tion to cocarboxylase may be a cause of the decrease. Genevois and
Flavier (1938-1939) reported 3d to $iof the thiamin present as the free
base and the remainder as the phosphoric ester, although in a n occasional
wine as much as % was esterified. They reported less than 0.2 mg. per
liter in most wines but two of thirty has 0.3, which seems very high.
Thiamin (B,) was shown by Schanderl (1950) to increase rapidly for
8 weeks in sparkling wine production. For 44 weeks thereafter it de-
creased. As much as 0.23 mg. per liter were present at 8 weeks but only
about 0.05 mg. at 52 weeks. Schanderl(l950) has summarized some of the
German results as follows: fresh grape juice, 0.120 to 0.268 mg. per liter;
commercial grape juice, 0.027 to 0.135; white wine 0.013 to 0.150; and
red wine, 0.133 to 0.266. Morgan et al. (1939) reported 3 t o 4 I.U. of
thiamin per 100 ml. in red and white table and dessert wines, or about
to as much as in the original grape juices. Randoin (1936) found the
B vitamins were preserved best in wines prepared as simply as possible
and with a minimum of manipulation. Pasteurization was particularly
harmful t o their retention. Cailleau and Chevillard (1949) determined
the thiamin of nine red and white French wines. They reported 0.008 to
0.086 mg. per liter, with an average of 0.034. Sisakgn et al. (1950a)
reported 0.043 mg. per liter of thiamin in a new wine and 0.025 after
growth of a film yeast.
RiboJEavin. Flavier (1939) found little riboflavin (about 0.05 mg. per
kilogram) in green or ripe Cabernet and Sauvignon blanc grapes, but 0.3
to 0.5 mg. per liter after fermentation. The high percentage of yeast in the
small Bordeaux barrels may explain the variable amounts. Riboflavin is
easily destroyed by light, and about 50% is lost by sulfiting the must
before fermentation or by filtering through bentonite, according t o
Perlman and Morgan (1945). Since white Bordeaux wines are usually
marketed in clear glass bottles, this may also account for some of the
variation. Riboflavin was found by Morgan et al. (1939) to be higher in
white (1.00 t o 1.50 mg. per liter) than in red wines (0.27 to 0.90). Perlman
and Morgan (1945) also reported added riboflavin to be retained by both
grape juices and wines, except when they were stored in clear glass bottles
exposed to the light (for sixteen months), where it was largely destroyed.
No difference in the riboflavin content of diploid and tetraploid grapes
of the same variety was found by Smith and Olmo (1944). I n Lodi’s
(1943) report the riboflavin content in grapes was 0.141 mg. per kilogram
and 0.098 mg. per liter in the wine. Cailleau and Chevillard (1949) found
0.08 t o 0.145 mg. per liter (average 0.210) in eleven French wines.
Sisakgn et al. (1950a) reported the riboflavin to decrease from 0.182 mg.
per liter before fermentation to 0.167 mg. after fermentation, and 0.087
mg. after six months under a film. An older sherry had only 0.063 mg.
The changes in several vitamins during alcoholic fermentation are
shown in Fig. 4.
Other Vitamins. Pyridoxine was found by Perlman and Morgan (1945)
in amounts of 0.83 to 1.82 mg. per kilogram of grape juice and 0.66 t o
$60-
U
J
50-
0
0
LL 40-
I-
-
y 30
w
LT
a 20-
I0 -
0 % I I I I
0 3 6 10 17 24 31 45 60 75 86 97 106
DAYS OF FERMENTATION
FIG. 4. Changes in carotene, thiamin, riboflavin, and ascorbic acid during fer-
mentation (Lodi, 1943).
0.72 mg. per kilogram of wine. Pyridoxine added to grape juices or wines
was well retained during storage. In eleven French wines Cailleau and
Chevillard (1949) reported 0.2 t o 1.2 mg. per liter (average 0.67).
Pantothenic acid was reported in grape juices and wines by Perlman
and Morgan (1945) in amounts of 0.007 t o 0.100 mg. per liter. When
added to either, it was retained during storage. Smith and Olmo (1944)
found significantly higher amounts of pantothenic acid in the juice of
tetraploid compared t o diploid varieties. Labrusca X vinifera interspecific
hybrids were also higher in this vitamin than vinifera hybrids.
Cailleau and Chevillard (1949) found 0.70 to 1.90 mg. per liter (aver-
age 1.08) of nicotinic acid in eleven French wines. Teply et al. (1942)
found 0.28 mg. per 100 g. of nicotinic acid in fresh Sultanina grapes (1.84
mg. per 100 g. on a dry basis). Popova and Puchkova (1948) used Lacto-
bacillus arabinosus for assay of nicotinic acid as did Sisakfan et al. (1950a),
who reported 2.34 mg. per liter in a must, 1.69 during fermentation, and
0.72 t o 0.90 after growth of a film yeast on the wine. They reported,
however, that higher alcohols and tryptophane interfered by reducing
growth of the bacteria. Castor (1953b), however, reported a n unknown
inhibitor factor present in musts where little tryptophane and no higher
alcohols are present.
Considerable vitamin Bg was reported by Perlman and Morgan (1945)
in wines after seven months storage. Castor (1953a) also reported biotin
and p-aminobenzoic acid. Inositol has been identified by German workers
and recently also by Castor (1953a).
The need of these vitamins for yeast growth was questioned by Wik6n
and Richard (1951-1952), who made 120 successive transfers of the
Fendant strain of yeast on a strictly chemical medium with no sign of
diminished growth or viability. They also reported th a t (+)-biotin and
meso-inositol when added in amounts of 0.0025 to 0.025 mg. per liter and
0.25 t o 2.5 mg. per liter, respectively, only slightly stimulated growth.
They concluded that the Fendant strain of yeast employed in Switzerland
is a n auto-autotrophic strain capable of synthesizing its cellular con-
stituents from glucose-mineral-salt media. Similar results were obtained
with other strains except that meso-inositol stimulated growth more.
However, Ribkreau-Gayon and Peynaud (1952) believe that the
thiamin stimulates growth sufficiently to be considered of some use in
winery practice. Considering the rate of fermentation actually observed
in wineries, a t least in California. Its use may be rather limited.
Wines have a component which strengthens capillary resistance,
according t o Lavollay and Sevestre (1944). This they attributed to the
so-called vitamin P. A summary of the vitamin content of wines is given
in Table XXIV.
2. Aromatic Constituents
While enologists have been interested in the odorous constituents of
grapes and wines for many years, Hennig and Villforth (1942) and
Hennig (1943, 1950-1951) seem t o have made the first systematic studies
on the subject. They extracted wine with pentane and after hydrolysis
identified the alcohols and acids. They reported the following aldehydes,
ketones, and related compounds (formaldehyde, acetaldehyde, propion-
aldehyde, cinnamaldehyde, vanillin, acetone, methyl ketone, acetyl-
methylcarbinol, and acetal-caproaldehyde and higher members of the
series, benzaldehyde, and furfural were not positively identified but
probably also occur) ; alcohols (methyl, ethyl, isopropyl, isobutyl,
isoamyl, and a-t,erpineol-n-propyl, n-heptyl, and sec-nonyl (2-nonanol)
are also probably present) ; acids (formic, acetic, propionic, n-butyric,

caproic caprylic, capric, and lauric-isobutyric, isovaleric, and oenanthic
are probably present). The alcohols and acids are present mainly as esters,
since little free acid or alcohol was found.
S o b et al. (1948) have also undertaken studies in this field. They
reported three aldehydes (oenanthaldehyde (heptanol), acetaldehyde, and
TABLEXXIV
Vitamin Content of Various Types of Wines
Must, Wine,
micrograms/ micrograms/
Vitamin 100 g. 100 g. Source
30-57 * 0-12 Morgan et al. (1939)
0-24 Perlman and Morgan (1945)
....... 15-30 Genevois and Flavier (1938-
Thiamin. . . . . . . . . . . . . . 1939)
7.4 Scheurer (1944)
0.8-8.6t Cailleau and Chevillard (1949)
. . . . . . . . . . Watt and Merrill (1950)
....... 0-27.2 11 SisakGn et al. (1950a)
p 4
l5 t 25-122t
27-50 Morgan et al. (1939)
Perlman and Morgan (1945)
6-22 8 Perlman and Morgan (1945)
....... 5-40 * Genevois and Flavier (1938-
1939)
Riboflavin. . . . . . . . . . . . .
14.1 9.3 Scheurer (1944)
1....... 7.5 Schon et al. (1939)
8.0-45.0* Cailleau and Chevillard (1949)
40 $ . . . . . . . . . . Watt and Merrill (1950)
....... 0-12.5** SisakGn et al. (1950a)
Pyridoxin . . . . . . . . . . . . . 83, 182 88, 70, 72 Perlman and Morgan (1945)
89, 150 40-110 Perlman and Morgan (1945)
Pantothenic acid. . . . . . . . . . . . . . 7-45 I Perlman and Morgan (1945)
....... 20-120* Cailleau and Chevillard (1949)
85-210* Cailleau and Chevillard (1949)
Nicotinic acid. . . . . . . . . .
. . . . . . . . . . . . Watt and Merrill (1950)
* Per 100 ml.
t In experimental samples after 1 month’a storage.
a
$ Per 100 g. of edible portion.
**
In commeraial wines.
-
Average of wines = 5.34.
Average of wines 4.10.
propionaldehyde), nine acids (acetic, lactic, propionic, butyric, valerianic,

caproic, caprylic, capric, and pelargonic) , and six alcohols (ethyl, propyl,
butyl, amyl, hexyl, and heptyl). More recently Holley (1951) reported
isolation of nine volatile components from Concord grapes; details are
lacking.
Haagen-Smit et al. (1949) extracted the constituents in a volatile oil
from fresh Zinfandel grapes. This variety was not a happy choice for the
study, as the aroma of this grape is not distinctive and only after fer-
mentation can its berrylike character be identified. They isolated ethyl
alcohol (244 g. per 1000 g.), acetaldehyde (1.8 g.), acetic acid (0.0053 g.),
n-butyric acid (0.003 g.), n-caproic acid (0.0015 g.), glyoxylic acid
(0.118 g.), n-butyl phthalate (2.25 g.), leaf aldehyde (0.327 g.), sulfur
(0.004 g.), acetylmethylcarbinol (0.013 g.), waxy substances (0.024 g.),
and a carbonyl compound (0.024 g.).
The identification of lauric acid as one of the characteristic con-
stituents of wine distillates was probably first made by Grossfeld and
Miermeister (1928). They reported 5.8, 19.0, and 20.0 mg. per liter in
three table wines (or 47.5, 163.8, and 183.4 mg. per liter of alcohol). They
also reported 20.5 mg. per liter of caprylic acid (or 198 mg. per liter of
absolute alcohol).
Chauvet (1950) has given a useful summary of the sources of the
odorous constituents in wine and their importance to the quality. Chauvet
considered the vinous odor to be due mainly to esters of lauric acid-
concentrations of 1 in 40,000 being found in new wine. The propionic and
butyric acids he believed might be attributed to enzymatic action on the
oils of the grape seed or to enzymatic action on the amino acids. The
tannins may be involved in the odor of red wines, particularly of wines
stored in new oak casks, and their antioxidant effect may prevent oxida-
tion of desirable aromatic principles already present. Esterification of
fixed acids, such as tartaric or succinic, he does not consider important
in the development of the “aged ” bouquet-because they develop slowly
and because they have very little odor. The roselike odor of certain
Beaujolais wines he attributed to p-hydroxyphenylethyl alcohol,
derived from phenylalanine. Valaize and Dupont (1951) reported that
Paris added phenylalanine to a must and secured the roselike odor.
Presence of p-hydroxyphenylethyl alcohol in fermented rice wine was
reported by Shimamoto and Sugayama (1951).
Schanderl’s (1938) experiments showed that the character of Rhinkand
Moselle wines was derived entirely from the grape. Use of various strains
of yeast to produce the characteristic flavors of these wines was unsuccess-
ful. However, the sherry wine odor or “Sudweinbukett” can be obtained
by using the proper type of yeast. The distinctive odor produced by film-
forming yeasts has been noted by all who use them. The dry Spanish
sherry wines owe most of their characteristic odor to these yeasts. Consult
Shcherbakov (1941), Saenko (1947, 1948), Creuss (1948), and Fornachon
(1953). Joslyn (1938a) showed that electrolysis produced a sherrylike
flavor although not equal in all respects to the Spanish product. He also
suggested simultaneous electrolysis and heating. Neubauer (1941)
pointed out that production of aroma by yeasts could be hastened by

aeration, particularly in producing a ‘‘sherrylike ” odor.
Ravaz (1935) has indicated, and most enologists agree, that the pre-
dominant factor in determining the characteristic quality of a wine is
variety. Mezzadroli et al. (1931), however, reported a case where Barbera
grapes fermented with “ Barbera” and “Chianti ” strains of yeast pro-
duced wines with different bouquets. Peynaud (193713) believes the bou-
quet is probably due to substances in the skin which are conditioned by
variety as well as soil and climatic conditions.
Prillinger (1952b) believed that the odorous substances produced
during fermentation should not be lost. He absorbed them on activated
charcoal; they appeared to be esters and aldehydes and were not char-
acteristic of the variety of grape. Smaller amounts were obtained in long,
slow, low-temperature fermentations. Although these fermentation odors
were considered favorable t o the character and balance of young wines,
they were not important in aged wines or in wines to which oxygen had
been added. T o prevent covering up the delicate fermentation odor he
recommended keeping the sulfur dioxide content low.
In studies with synthetic solutions Kutal’ova (1931) showed that
addition of amino acids or ammonium salts gave distinctive odors in the
fermented product. I n the presence of lactic acid or ammonium salts a
winelike character was produced, with glycine a yeast odor, with leucine
a fruitlike odor, and with alanine a wine or yeast odor. Procopio (19488)
has proposed a process for increasing the aroma of wines and brandies by
addition of a yeast hydrolysate. This was said to accelerate aging changes
and t o result in wines of greater intensity of odor. Similar results during
aging appear to be inherent in Russian work on tank fermentations.
Rib6reau-Gayon (193th) showed that the development of bouquet in
wines took place only in the bottles in the absence of oxygen, i.e., under
reducing conditions. Lehmann (1943) has indicated the effect of alcohol,
acids, and the soluble solids on flavor. The paper by Nelson (1937) on the
flavor of alcoholic beverages adds little information.
XI. SUMMARY
The relatively constant relation between the major by-products of
alcoholic fermentation was apparently first recognized by Genevois
(1936). He found that the equation g = 5s + + 2a h, where g is the num-
ber of moles of glycerol, s the number of moles of succinic acid, a those
of acetic acid, and h of acetaldehyde, theoretically expressed the relation.
The actual analysis showed g to be less than the equation indicates and
it is usually calculated as 0.9 g. Genevois et al. (1946; 1947a, b; 1948a, b, c;
1949a, b, c) , Peynaud (1948a) , Peynaud and RibBreau-Gayon (1947), and
Genevois (194913, 1950) have reported on the subject. Their final equation
-5s + + + +
2a b 2m h = 0.9g(Z), where s, a, h, and g are as above, b
is the moles of the 2,3-butylene glycol, and m is the moles of the acetyl-
methylcarbinol-has been shown to be valid for various yeasts and for
musts of varying composition. A large number of wines have been tested
and found to conform fairly well with this equation. With different yeasts,
however, the relative amounts of acetic acid, succinic acid, 2,3-butylene
glycol, and glycerol varied considerably. Also by changing the fermenta-
tion conditions the ratios between the by-products could be varied. These
relations were also shown to apply to fortified wines by Genevois et al.
(194%). They showed Z/g to vary from 0.88 to 0.94 and b / g X 1,000 from
108 to 128. In laboratory and commercial fermentations s varied from 5
to 9 millimoles per liter; a was 10 to 12 in commercial fermentations and
4 to 12 in laboratory fermentations; and b varied from 5 to 10 in commer-
cial fermentations and from 3 to 6 in laboratory fermentations. The ratio
a/s was 0.4 to 2 under normal fermentation conditions and 1 to 3 when
the fermentation conditions were varied. Likewise b/g was usually 4 to 9,
but by changing the fermentation conditions varied from 7 to 12. Many
of the results of their research may be of practical importance in winery
operation. They found, for example, that the acetic acid/succinic acid
ratio is very different for grapes fermented on the skins compared to
those fermented off the skins. Genevois (194913) has also suggested the
possible formation of 1,4 butanedial and the following acids as minor
secondary by-products of alcoholic fermentation: d l citramalic, 7-hy-
droxybutyric (4-hydroxybutanoic), isocitric, aconitic, oxyglutaric, and
glutaric. See also p. 407.
Of more theoretical interest has been the study of the quantity of water
fixed by various secondary products of alcoholic fermentation. Genevois
et al. (1949b) showed that the amount fixed, e, equals a + + 2s 4c, where
c is the citric acid.. With 29 French yeasts on grape juice e varied from
15 to 26, average 19. With the same yeasts on a sucrose media e varied
from 26 to 35, average 29. Another interesting theoretical relationship
was their discovery that e/g is nearly constant at 1/3 with a variety of
yeasts on two media. I n this respect, at least, the yeasts appear limited.
Finally they have shown that Aa = a + 2s - g/3. When A is between
3 and 10 there is a presumption of formation of acetic acid by anaerobic
bacteria and not as a by-product of alcoholic fermentation. Even with
various types of yeasts Genevois et al. (1948a) found the fermentation
balance was maintained. These included succinogenic, glycologenic, and
acetogenic yeasts. Addition of acetaldehyde during fermentation in-
creased the succinic acid and 2,3-butylene glycol content, but the effect
varied with the yeast.
Perhaps the outstanding advances made in this period have been in

the greater quantization of data and the welcome biochemical interpre-
tation which has been applied to it. This has been true particularly of the
new and better data on a number of the minor constituents. The applica-
tion of new analytical techniques will certainly aid in future work. Many
important gaps exist in our knowledge. The relation between the various
acids, cations, and the buffer capacity of musts and wines should be
delineated. Data on the substances which contribute to the characteristic
odor of many wines are almost completely lacking. And finally, there is a
pressing need for more specific and accurate methods for the determina-
tion of practically all of the organic constituents of wines-both methods
for control purposes and accurate research procedures.
ACKNOWLEDGMENTS
I am greatly indebted to members of the library staff for their patience and dili-
gence in locating and checking the references, particularly Patricia L. Golton, Shirley
Hopkinson, Mrs. Aileen R. Jaffa, Sara B. Schreiber, and Louise B. Wheeler. I am also
grateful to my colleagues, Mr. Harold W. Berg and Professors J. G. B. Castor, M. A.
Joslyn, and A. D. Webb who have read all or portions of the manuscript and to Mrs.
Angelo Arnold who has faithfully typed the various “editions l 1 of the manuscript.
The errors which remain, however, are the author’s.
REFERENCES
Ajon, G. 1943. Determinazione qualitativa della sostanza colorante dei vini. Chim.
ind. agr. biol. 19, 180-181.
Akademiix Nauk-Institut Biokhimii. 1947-1950. Biokhimix vinodelig. A kademiia
Nauk S.S.S.R., Moscow, Vol. 1, p. 208, 1947; Vol. 2, p. 183, 1948; Vol. 3, p. 246,
1950.
Akiya, S., and Sasao, H. 1951. Determination of methanol in wine by chromotropic
acid. J . Pharm. SOC.Japan 71, 1325-1326; C . A . 46, 2234 (1952).
Akman, A. V. 1951. Die Weine Ankaras. Deut. Vein-Ztg. 88, 101-102.
Alessandrini, E. 1933. Dosage de l’aloool isopropylique dans les boissons alcooliques
(transl.). Atti 4th eongr. naz. chim. pura ed appl., Rome, 1933, 611-614.
Alexis, E. 1933. Le dosage du sucre restant dans les vins rouges. Ann. fals. et fraudes
26, 509-610.
Alexis, E. 1936. Influence de la date des vendanges sur la qualit6 des vins d’Aramon
de plaine. Ann. fals. el fraudes 29, 29-30.
Alfa, J. 1932. Ergebnis der amtlichen Weinstatistik. V e i n u. Rebe 14, 181-191.
Alfa, J. 1933. Ergebnis der amtlichen Weinstatistik. Wein u. Rebe 16, 175-183.
Almeida, H. de. 1948. Estudo polarogrhfico de algunas protefnas puras. Anais inst.
vinho P8rlo 9, 17-43.
Almeida, H. de. 1950. Estudo polarogrhfico do aldefdo ac6tico no vinho do Parto.
Anais inst. uinho Pdrto 11(1), 23-31.
Amadio, G., and Paronetto, L. 1936. La determinazione dell’acidith volatile nei vini
col nuovo acidacetimetro Jozzi. Ann. chim. appl. 26, 173-178.
Amerine, M. A. 1944. Determination of esters in wines: liquid-liquid extraction. Food
Amerine, M. A. 1947. The composition of California wines a t exhibitions. Wines &
Vines 28(2), 24-26; (3), 23-25, 42-46.
Amerine, M. A. 1948. Hydroxymethyl furfural in California wines. Food Research 13,
264-269.
Amerine, M. A. 1950-1951. The acids of California grapes and wines. I. Lactic Acid.
11. Malic Acid. Food Technol. 4, 177-181; 6, 13-16.
Amerine, M. A. 1951. The educated enologist. Proc. Am. SOC.Enologists 1961, 1-30.
Amerine, M. A. 1952. Laboratory Procedures for Enology (rev.). University of
California, Davis.
Amerine, M. A., and De Mattei, W. 1940. Color in California wines. 111. Methods of
removing color from the skins. Food Research 6(5), 509-19; see also Wines &
Vines 22(4), 19-20 (1941).
Amerine, M. A., and Dietrich, W. C. 1943. Glycerol in wines. J . Aesoc. Ofic. Agr.
Chemists 26, 408-413.
Amerine, M. A., and Joslyn, M. A. 1951. Table Wines: The Technology of Their Pro-
duction in California. University of California Press, Berkeley.
Amwine, M. A., and Webb, A. D. 1943. Alcohol-glycerol ratio of California wines.
Food Reseaech 8, 280-285.
Amerine, M. A., and Wheeler, L. B. 1951. A Check List of Books and Pamphlets on
Grapes and Wine and Related Subjects, 1938-1948. University of California
Press, Berkeley.
Amerine, M. A., and Winkler, A. J. 1941. Maturity studies with California grapes. I.
The Balling-acid ratio of wine grapes. Proc. Am. SOC.Hort. Sci. 38, 379-387.
.4mcrine, M. A., and Winkler, A. J. 1943. Maturity studies with California grapes. 11.
The titratable acidity, pH, and organic acid content. Proc. Am. SOC.Hort. Sci.
40, 313-324.
hmerine, M. A., and Winkler, A. J. 1947. Relative color stabilities of the wines of cer-
tain grape varieties. Proc. Am. SOC.Hort. Sci. 49, 183-185.
Anibal Burgoa, P. 1935. Estudio comparativo del contenido normal en glicerina de 10s
vinos argentinos. Rev. fac. cien. q u h . univ. nacl. La Plata 10, 95-104.
.4nonymous. 1933. Roumanie. MBthodes officielles d’analyses des vim. Bull. ofice
idern. vin 6(67), 54-92.
Anonymous. 1934. Minister0 dell’agricoltura e delle foreste. Metodi ufficiali di analisi
per le materie che interessano l’agricoltura. I1 (Part I). Mosti, vini, birre, aceti,
sostanze tartariche, materie tanniche. Instituto Poligrafico dello Stato, Roma.
Anonymous. 1935. Convention international pour l’unification des methodes d’analyse
des vins dans le commerce international, 1935. Ann. ferment. 1,310-319; see also
Ann. fals. et fraudes 28, 418-426 (1935.)
Anonymous. 1937. Schweizerisches Lebensmittelbuch. Zimmermann & Cie., Bern,
4 Ed.
Anonymous. 1940. Note sur l’acide lactique dans lcs vins. Ann. ferment. 6, 290-301;
see also Rev. uiticult. 91, 242-248 (1939.)
hnonymous. 1944. Convenient chart for vintners. Food Inds. 16, 85.
Anonymous. 1052. Unification des mBthodes d’analyse. Bull. inst. natl. appel. orig.
vins et eaux-de-vie 40, 18-24.
Antoniani, C. 1951a. Cib che si sa e cib che si ignora sul chimismo della fermentazione
alcolica. Riv. viticolt. e enol. (Conegliano) 4, 183-196.
Antoniani, C. 1951b. I processi accessori della fermentazione alcolica del mosto d’uva.
Ann. sper. agrar. (Rome) (N.S.) 6, 481-488.
Antoniani, C. 1953. Levures apiculBes, elliptiques e t formation d’ac6to‘ine e t de
2,3-butanediol. Inds. agr. et aliment. (Paris) 70, 397-398.
Antoniani, C., and Cioffi, R. M. 1949. Transaminazione in vivo nei lieviti alcolici.
Boll. S O C . ital. biol. sper. 26, 7-8.
Antoniani, C., and Gugnoni, S.1941. L’acetilmetilcarbfnolo nei vini. Ann. chim. appl.
31, 417-422.
Ant-Wuorinen, O . , and Kotonen, E. 1935-1937. Bestimmung des Methylalkohol-
gehaltes in Alkohol und alkoholhaltigen Getranken. 2. Untersuch. Lebensm. 69,
59-67; 74, 273-281.
Archinard, P. 1937. La limite de 20 mgr. d’NHs admise par la loi pour caracteriser les
vins impropres h la consommation. Est-elle justifibe? Rev. viticult. 87, 339-343.
Archinard, P. 1939. Le dosage de l’ester acetique et les vins alteres impropres B la
consommation. Ann. fals. et fraudes 32, 254-260.
Archinard, P. 1940. MBthodes rapides de dosage des esters neutres dans les boissons
fermentbes. Ann. fals. et fraudes 85, 76-85.
Arutunyan, L. A. 1939. Antiscorbutic property of grape leaves (transl.). Voprosy
pitaniz 8, 54-57.
Association of Official Agricultural Chemists. 1950. Official and Tentative Methods of
Analysis of the Association of Official Agricultural Chemists. Association of
Official Agricultural Chemists, Washington, 7 Ed.
Astruc, H., and Castel, A. 1932a. Contribution B 1’6tude du dosage des acides volatils
dans les vins. Ann. chim. anal. et chim. appl. 14, 145-152.
Astruc, H., and Castel, A. 1932b. Le dosage pratique du tannin dans les vins. Ann.
fals. et fraucles 26, 477-480.
Ay, H. 1936. Die Methoden zur Bestimmung des Mostgewichtes und des Saure-
gehaltes im Most. Wein u. Rebe 18, 161-172.
Azevedo, M. P. de. 1942. 0 Auxiliar do Analista. Edi@o do Instituto do Vinho do
PBrto, PGrto.
Baglioni, S., Casale, L., and Tarantola C. 1935-1937. L’attivitA proteolitica del succo
d’uva. Annuar. R. staz. enol. sper. Asti (2)2, 203-219.
Balavoine, P. 1939. Sur la mustimbtrie. Mitt. Gebiete Lebensm. u. Hyg.30, 331-335.
Balavoine, P. 1943. Contribution B l’appr6ciation des vins de MAIaga. Mitt. Gebiele
Lebensm. u. Hyg. 34, 365-367.
Balestrazzi, A. 1952. Descrizione di un apparecchio per la correzione del grado alcolico,
determinato con ebulliometri tip0 Malligand, in sequito a spostamenti dello
“zero” per variazioni di pressione. Riv. viticolt. e enol. (Conegliano) 6, 264-266.
Barbera, G. 1933a. Ricerche sperimentali sulla pectina dell’uva. Ann. tern. agrar. 6,
229-245.
Barbera, G. 193313. Sulla formazione della mannite nei mosti dializzati a bassa tem-
peratura in presenza di cloroformio. Ann. chim. appl. 23, 470-473.
Barbera, G. 1 9 3 3 ~.Sulla sostanze estrattive dei vini. I. Pectine e gornme. 11. Corn-
ponenti nitrogeni. Ann. chim. appl. 23, 95-99, 115-120.
Barini-Banchi, G. 1948. Determinaeione dell’alcool nei vini mediante il metodo di
ossidazione cromica. Ann. chim. appl. 38, 423-428.
Barini-Banchi, G. 1949. Le applicazione della colorimetria fotoelettrica all’analki ed
alla tecnica enologica. Riv. viticolt. e enol. (Conegliano) 2, 365-370.
Barini-Banchi, G. 1952. Confront0 fra il metodo rifrattometrico e quello densimetrico
e chimico nella determinazione delle sostanze zucherine nei mosti. Riv. viticolt. e
enol. (Conegliano) 6, 21-25.
Bartels, W. 1933. Bestimmung der Citronensaure im Wein. 2.Untersuch. Lebensm. 66,
1-37.
Bassler, K., and Trauth, F. 1934. Ein Beitrag zur Frage der Beziehungen zwischen
Mostgewicht und Alkoholgehalt und deren Nutzanwendung bei der Verbesserung
der Moste. Wein u. Rebe 16, 107-118.
Beckwith, J. 1935. Pure cultured yeasts. J. Dept. Agr. S. Australia 38, 858-867.
Beis, C. 1934. Determination of potassium acid tartrate and free tartaric acid
in musts and wines (transl.). Prakt. Akad. Athenon 9, 201-205; C . A . 30,
3938 (1936).
Benvegnin, L. 1934. DBchets dans la manutention des vins. Ann. agr. Suisse 1934,
1173-1 184.
Benvegnin, L., and Capt, E. 1932. Contribution A 1’6tude des quelques mkthodes de
dosage de l’acide lactique dans les motits e t vins. Mitt. Gebiete Lebensm. u. Hyg.
23, 46-64.
Benvegnin, L., Capt, E., and Piguet, G. 1951. Trait6 de Vinification. Librairie Payot,
Lausanne, 2nd Ed.
Benz, G. 1931. Berechnung der ursprunglichen Oechslegrade im angegorenen Trauben-
most. Wein u. Rebe 13, 63-66.
Berg, P. 1932. Die Bestimmung der Susse in Dessertweinen nach Baum6-Graden.
Wein u. Rebe 14, 175-180.
Berg, P., and Schmechel, S. 1932. Weinsaurebestimmung im Wein. 2. Untersuch.
Lebensm. 64, 348-357.
Berg, P., and Schulze, G. 1931. Bestinimung der Milchsiiure im Wein bei Gegenwart
von Zucker. Wein u. Rebe 12, 433-445.
Berg, P., and Schulze, G. 1932. Uber das Vorkommen von niederen Fettsiiuren
(Buttersiiure) in Siissweinen. 2. Untersuch. Lebensm. 63, 62-68.
Berg, P., and Schulze, G. 1934. Ein neues Pentabromaceton-Verfahren zur Bestim-
mung der Citronensilure im Wein. 2. Untersuch. Lebensm. 67, 605-613.
Berg, V. A. 1939. Buffer properties of fruits and berry juices and of wines (transl.).
Konserven-, Obst- u. Gemtiseind. 10, 17-19.
Bergman, H. F., and Magoon, C. A. 1945. Tartrate content of Maryland-grown
American grape varieties. Proc. Am. Sac. Hort. Sci. 46, 253-254.
Beridze, G. I. 1948. The aging of Kakhetinski wine on the pomace (transl.). Vinodelie a’
Vinogradarstvo S.S.S.R. 8(8), 11-14.
Berner, C. 1952. ConsidCrations analytiques sur quelques vins genevois et r6sultats de
l’application de quelques regles oenologiques. Mitt. Gebiete Lebensm. u. Hyg. 43,
388-4 13.
Bertin, C. 1931. La teneur in glycerine des vins de Mascara. Ann. fals. et jraudes 24,
350-352.
Bertin, C . 1933. Influence du sucre sur le degrk kbulliom6trique des vins. Ann. fals.
et fraudes 26, 103-105.
Bwtin, C . 1934. Dosage de l’acidit6 volatile des vins sulfitks. Ann. fals. et jraudes 27,
172-174.
Ikrtram, S. H., and Rutgers, R. 1938. The estimation of glycerol and some other
hydroxylated compounds. Rec. trav. chim. 67, 681-687.
Hertrnnd, G., and Silberstein, L. 1948. MBthode ameliorie de recherche e t de dosage de
petites quantities de mkthanol. Ann. agron. 18, 139-144; see also Compt. rend.
226, 365-367 (1948.)
Bertrand, G., and Silberstein, L. 1949-1952. Sur la teneur des vins en m6thanol.
Compt. rend. 229, 1281-1284; 234, 491-494.
rertrand, G., and Silberstein, L. 1950-1952. Sur la teneur des vins en m6thanol.
Compt. rend. acad. agr. France 36, 59-61, 191-192; 38, 162-163; see also Ann.
inst. Pasteur 82, 668-675 (1952).
Besone, J., and Cruess, W. V. 1941. Observations on the use of pectic enzymes in wine
making. Fruit Products J . 20, 365-367.
Beyer, 0. F. 1945. Report on the spectrophotometric examination of wines. J. Assoc.
O&. Agr. Chemists 28, 471-475.
Bicalho, N. dos Santos. 1933. Analise de vinhos. (MBtodos sequidos no Laboratorio
Bromatolgico de Dep. Nacional de Sadde Publica.) Rev. SOC. brasil. chim. 4,63-73.
Biedermann, W. 1951. Die pH-hderungen bei der Abscheidung von Weinstein aus
Getranken. Mitt. Gebiete Lebensm. u. Hyg. 42, 476-482; see also Schweiz. 2. Obst-
u . Weinbau 61, 193-195 (1952).
Bleyer, A. 1938. Alkoholische Genussmittel. Handbuch der Lebensmittelchemie. J.
Springer, Berlin, Bd. 7.
Bobadilla, G. F. de. 1943. Aplicaciones industriales de las levaduros de flor. Agri-
cultura (Madrid) 12, 203-207.
Bobadilla, G. F. de, and Navarro, E. 1949. Vinos de Jerez. Estudios de sus Acidos,
desde el period0 de madurez de la uva hasta el envejecimiento del vino. Bol. inst.
nacl. invest. agron. Spain 9(21), 473-519.
Bobadilla, G. F. de, and Navarro, E. 1952. Vinos de Jerez. Contribuci6n a1 estudio de
sus caracetlsticas. An4lisis de varios tipos de vinos. Bol. inst. nacl. invest. agron.
Spain 12(27), 377-395.
Bodendorf, K. 1930. Ein einfacher Nachweis des Isopropylalkohols in Spirituosen.
Bohringer, P. 1943. tfber die Bestimmung des Refraktometerwertes von Pfiilzer
Trauben mittels des Zeiss'schen Handrefraktometers. Gartenbauwiss. 17, 505-520.
Bohringer, P. 1951a. Bestimmung des Alkohol- und Extraktgehaltes von trockenen
20"
Weinen aus der Refraktion bei 20" C. und dem Gewichtsverhiiltnis Y L 200. 2.
Lebensm.-Untersuch. u. -Forsch. 93, 65-75.
Bolcato, V., D'Orazi, F., and Pasquini, G. 1941. Alcune miscure sperimentali
sull'alcole trasportato dalla COa delle fermentazione alcolica. Zndustria saccar.
ital. 34, 357-364.
Bonaterra, R. 1949. Study of the method of evaluation of higher alcohols by colori-
metric determination based on the furfural-sulfuric acid reaction. Applied to
distilled beverages (transl.). Rev. fac. cienc. qutm. univ. nacl. La Plata 24,125-133;
C . A . 47, 3516 (1953).
Bordas, F., and Roelens, E. 1930. Corrections alcoomhiques pour les tempBratures
au-dessous de zBro. Ann. fals. el fraudes 23, 263-283.
Bordas, F., and Touplain, F. 1930. &Etude sur 1'alcoomBtrie. Ann. fals. et fraudes 28,
84-102.
Bognjak, I. 1935. Priifung und Vergleich der Methoden zur Analyse des Weines.
I. Ftir die Verwerfung des indirekten Extraktes bei der Analyse vergorener Weine
und seine Ersetzung durch das Vakuumextrakt. Jahresber. Oenol. Stat. Aker-
bauminist. Jugoslav 1, 175-235, 281-287.
Boinjak, I. 1938. Priifung und Vergleich der Methoden zur Analyse des Weines. 11.
Milchsiiurebestimmungsmethoden. Jahresber, Oenol. Stat. Ackerbauminist. Ju-
goslav 2, 131-216.
Botelho, J. C. 1935. gtudes sur le vin de porto. Ann. chim. anal. el chim. appl. [3] 17,
49-63, 179-180.
Botelho, J. C. 1938. Dosage de l'oxymbthylfurfurol dans le vin de porto. Son r81e dans
la sophistication. Ann. chim. anal. et chim. a p p l . [3] 20, 203-205; see also Rev.
viticult. 89, 202-205 (1938).
Botelho, J. C. 1939. Dosage rapide de l’alcool du vin au moyen du picnometre de
Mendelejeff. Ann. chim. anal. et chim. appl. [ 3 ] 21, 203-205.
Boutaric, A., and Bouchard, J. 1935. Sur une mBthode simple pour 1’Btude de I’aciditB
des vins exprimBe par la concentration en ions hydrogene. Compt. rend. acad. agr.
France 21, 737-743; see also Bull. ofice intern. vin 8(86), 44-49 (1935).
Boutaric, A., F e d , L., and Roy, M. 1936. Recherches spectrephotomBtriques sur la
couleur des vins. Compt. rend. 203, 1142-1144; see also Ann. fals. et fraudes 30,
196-209 (1937).
Boutaric, A., FerrB, L., and Roy, M. 1937. Recherches spectrophotomBtriques sur la
dilution et le melange des vins. Compt. rend. 204, 343-344.
BrBmond, E. 1937a. Bilan complet et repartition des substances ionisables contenues
dans les vins. Bull. SOC. chim. France [5] 4, 296-305.
BrBmond, E. 1937b. Bilan complet et repartition des substances ionisables contenues
dans les vins. Ann. fals. et fraudes 30, 136-146.
BrBmond, E. 1937c. Contribution ? l’fitude
i Analytique et Physico-chimique de 1’AciditB
des Vins. La Typo-Litho et Jules Carbonel RBunies, Alger.
BrBmond, E. 1938a. Qtude analytique et physico-chimique de l’acidit6 des vins. Ann.
ferment. 4, 86-102.
BrBmond, E. 1938b. Nouveau dispositif de mesure du p H des vins. Ann. agron. 8,
371-379, 557-558; see also Ann. fals. et fraudes 32, 85-95 (1939).
Brockmann, M. C., and Stier, T. J. B. 1948. Influence of temperature on the produc-
tion of glycerol during alcoholic fermentation. J . Am. Chem. SOC.70, 413-414.
Brockmann, M. C., and Werkman, C. H. 1933. Determination of 2,3-butylene glycol
in fermentations. Znd. Eng. Chem., Anal. Ed, 6, 206-207.
Brown, W. 1,. 1940. The anthocyanin pigment of the Hunt Muscadine Grape. J . Am.
Chem. SOC.62, 2808-2810.
Brugirard, A., and Tavernier, J. 1952. Sur le dosage des matieres tannoides dans les
cidres et les poirbs. Ann. fals. et fraudes 46, 108-110.
Brune, H. 1948. Der qualitative Nachweis von Methanol neben k h a n 0 1 in Tinkturen
und alkoholischen G e t r h k e n nach DenigBs. 2.Lebensm.-Untersuch. u. -Forsch.
88, 274-278.
Bucci, F. 1940. Sulla ricerca destrina nei vini. Ann. chim. appli. 30, 533-539; see also
Prog. vinic. olea. Ztal. 26, 37-40 (1949).
Biirgi, J. 1932. Zur Bestimmung der hoheren Alkohole nach Komarowsky-v. Fellen-
berg (Mikromethode). Mitt. Gebiete Lebensm. u. Hyg. 23, 94-95.
Buhrer, N. E. 1950. Mktodo prltico para diferenciar vinhos de uva de outros vinhos
de frutas, especialmente de laranja. Anais assoc. qubm. Brazil 9, 104-105.
Buogo, G. 1938. L’acido citrico costituente normale dei vini genuini. Probabile sua
funzione. Raffronto con l’acido tartarico. Atti 10th congr. intern. chim. Rome 4,
487-494.
Buogo, G., and Picchinenna, D. 1938. L’acido citrico nei vini genuini della provincia
di Bari. Ann. chim. appl. 28, 427-431.
Burdzhanadze, V. F. 1951. On acetic acid and sugar-alcohol balance in wine making
(transl.). Vinodelie i Vinogradarstvo S.S.S. R . 11(3), 12.
Burgvits, G. K., and Hochberg, R. B. 1936. Uber die Wirkung der Wasserstoffionen-
konaentration des Mostes auf Weinhefe (transl.). Arch. biologitscheskich Nauk.
43, 39-47.
Buscar6ns Ubeda, F. 1941. Sobre 10s componentes de alto punto de ebullici6n del fuse1
de orujo de uva y RU approvechamiento industrial. Anales jbs. y q d m . Madrid 37,
356-369, 371-383.
Buxbaum, W. 1932. Mostgewichtsbestimmung auf optischem Wege mit Hilfe des

Zeiss’schen Handzuckerrefraktometers. Wein u. Rebe 14, 171-174.
Cahill, W. M. 1933. Recherches sur le facteur antiscorbutique. Bull. soc. chim. biol.
16, 1462-1471.
Cailleau, R., and Chevillard, L. 1949. Teneur de quelques vim franpais en aneurine,
riboflavine, acid nicotinique et acide pantothhique. Ann. agron. 19, 277-281.
Cambitzi, A. 1947. Formation of racemic calcium tartrate in wines. Analyst 72,
542-543.
Campllonch Romeu, I. 1945. Ejercicios de AnBlisis de Vinos: mostos, mistelas, t&rtaros,
vinagres y alcoholes. Sociedad Enol6gica del PenadBs, S. A., Vilafranca del
PanadBs, 5th Ed.
Canals, E., and Collet, H. 1939. Essai polarimetrique des vins. J . pharm. et chim. 29,
385-390.
Canals, E., and Vergnes, P. 1940. Coefficient tampon des vins. Bull. SOC. chim. France
[5] 7, 231-234.
Cappucci, C. 1948. Osservazione sulle probabili cause della inconsueta aciditd volatile
di alcuni vini della regione Emiliano-Romagnola. Riv. viticolt. e enol. (Conegliano)
1, 386-388.
Cartier, P., and Pin, P. 1949. Microdosage de l’acide citrique. Bull. SOC. chim. biol. 31,
1176-1183.
Casale, L. 1930a. La fermentazione alcoolica nei liquidi ad elevata concentrazione
idrogenionica. 111. Ann. chim. appl. 20, 357-361.
Casale, L. 1930b. Influenza che esercitano i prodotti della fermentaaione sulla velocith
di moltiplicazione della cellula di li6vito. 11. Ann. chim. appl. 20, 353-357.
Casale, L. 1930c. Influenza del valore p H del mezzo sulla fermentazione alcoolica. I.
Ann. chim. appl. 20, 336-353.
Casale, L. 1930d. Ricerche fisico-chimiche sulle materie coloranti delle uve e dei vini
rossi. Ann. chim. app. 20, 559-566.
Casale, L. 1934. Esperienze de fermentazione a bassa temperatura. Congr. intern.
quim. pura apl. 9(V), 254-259; see also Annuar. R . staz. enol. sper. Asti (2) 2, 1-5
(1935-3 7).
Casale, L. 1935-37. Esame critico, tecnico e pratico della varietd delle uve d a vino
coltivate in Piemonte, in Lomhardia ed in Liguria. Annuar. R. staz. enol. sper.
Asti (2) 2, 67-104.
Casale, L., and Garino-Cnnina, E. 1935-37. Ricerche sugli enaimi del vino e del mosto.
Annuar. R. staz. enol. sper. Asti (2) 2 , 239-250.
Casamada Maurl, R. 1931. Espectros de absorci6n en el ultra-violeta y la investigaci6n
de colorantes artificialcs en 10s vinos. Mem. acad. cienc. artes Barcelona 22,
251-256.
Casares, R., and Gonzales de Rivera, C. 1953. Contribuci6n a1 estudio de 10s vinos de
la zona de Montilla y Moriles (Espaiia). Anales bromatol. (Madrid) 6, 23-67.
Castor, J. G. B. 1950. Biochemical events during vinous fermentation. Proc. A m . Soc.
EnoZogists 1960, 21-37, 104.
Castor, J. G. B. 1953a. B-complex vitamins of musts and wines as microbial growth
factors. Appl. Microbiol. 1, 97-102.
Castor, J. G. B. 1953b. The free amino acids of musts and wines. Food Research 18,
139-145, 146-151.
Castor, J. G. B., and Guymon, J. F. 1952. On the mechanism of formation of higher
alcohols during alcoholic fermentation. Science 116, 147-149.
Cerasari, E. 1950 (i.e. 1951). Sul contenuto e sulla determinazione dell’acido citric0
nei vini. Ann. triest. cura univ. Trieste Sez. 2 20, 127-139.
Ccrutti, G. 1951. Sull’originc e sul contenuto in alcool metilico dei vini italiani. Ann.
sper. agrar. (Rome) (N.S.) 6, 331-335.
Ccrutti, G., and Vcdani, A. 1951. Sulla detcrminaaione dcl nictanolo nei vini e nei
fermentatio ttlcolici. A n n . sper. agrar. ( R o m e ) (N.S.) 6, 419-421.
Charles, E. 1930. Sur la presence et le dosage dc l’ethanal dans les vins. Ann. fals. et
fraudes 23, 153-154.
Charles, E. 1938. Dosage de l’alcool iiibthylique ou dc l’alcool isopropylique en
prdscnce d’alcool 6thylique. 18th Congr. chim. ind. Nancy 1, 38C-48C.
Charpentie, Y. 1950. La fermentation malolactique des vins blancs de la Gironde.
Bull. ofice intern. vin 23 (231), 61-64.
Charpentie, Y., Ribdreau-Gayon, J., and Peynaud, E. 1951. Sur la fermentation de
l’acide citrique par les bacteries malo-lactiqucs. Bull. SOC. chim. biol. 33,1369- 1378.
Chauvct, J. 1950. L’arBrne des v i m fins. Bull. inst. natl. appel. orig. vins et eaux-de-vie
34, 8-17.
Chelle, L., Dubaquid, J., and Vitte, G. 1936. Sur la presence de l’acetonc e t son
dosage dans les alcools de vin. Bull. soc. pharm. Bordeaux 74, 112-126.
Chogovadae, S. K. 1948. Role of the aglucon fraction of coloring matter and tannin
materials i n the wine-maturing process (transl.). Vinodelie i Vinogradarstvo
S.S.S.R. 8(6), 14-17; C. A. 44, 9619.
Churchward, C. R. 1953. Spirit content of wine by ebullioscope. Australian Brewing
and W i n e J . 71(5), 4, 6, 8.
Churchward, C. R., and Johns, B. G. 1940. Use of the Dujardin-Sallcron ebulliometer
for the determination of the alcoholic strength of wines. Australian Chem. Inst.
J . & Proc. 7, 18-30.
Cioffi, R. M. 1948. Sul contcnuto in alcoli superiori dci vini Italiani. Riv. viticolt. e enol.
(Conegliano) 1, 341-343.
Cioffi, R. M. 1949. Studio dell’infiuenza di alcuni inibitori ensimatici sull andamento
del process0 di “fermcntazione alcoolica dcgli aminoacidi ” Riv. viticolt. e enol.
Clavera,.J. M., and Moreno Martin, F. 1936. Influencia del alcohol metflico en la
dosificaci6n de 10s alcoholes superiores en 10s aguardientes. Anales SOC. espa8. fis.
y quim. 34, 507-512.
Clavera, J. M., and Oro L6pca, M. 1932. Los az6cares y el extract0 seco en 10s vinos
de MAlaga. Anales SOC. espaii. f i s . y quini. 30, 140-144.
Collier, D. 1935. Sur le dosage des tanins du vin. A n n . fals. et fraudes 28, 208-224.
Colombier, L., and Clair, E. 1936. La dbtcrmination du degr6 alcoolique des vins.
Ann. fals. et fraudes 29, 411-416.
Colornbier, L., and Clair, E. 1!)38. Quclqucs observations sur la d6termination de
I’acidit6 volatile dcs vins. Ann. fals. et fraudev 31, 414-418.
Col$escu, I. H., Nichitovici, V., Dobrcscu, J., Iliescu, L., Tomescu, F., and Moldovan,
E. 1941. Beitrag zur Kenntnis der Zusammensctzung der Hybridenweine direkter
Eraeugung. Analelc Inst. Cercetari Agron. Ronimllniei 13, 3-1 1.
Conceipso, A. de B. F. da. 1942. Pcsquiza do corantes orgdnicos sintbticos nos vinhos.
A n a i s inst. super. agron., Univ. tic. Lisbou 13, 191-197.
Cordebard, H. 1939. Titrimetric determination of organic substances by chromic
oxidation. Use of stable nitro-chromic solutions (transl.). J. pharm. chim. 30,
263-272.
Cornforth, J. W. 1939. The anthocyanin of V i t i s hypoglauca F.v.M. J . Proc. Roy.
SOC.N . S . Wales 73, 325-328.
Correia, E. M. 1943. Riqueaa em Bcido fosf6rico e ferro de alguns vinhos de pasto
portugueses. Anais inst. super. agron., Univ. tic. Lisboa 14, 327-334.
Correia, E. M., and Ribeiro, E. C. 1942. Riqueza em &cido succhico e glicerina dos
vinhos comuns portugueses. Anais inst. super. agron., Univ. tbc. Lisboa 13,155-164.
Correia, E. M., and Sbrgio, R. J. de R. 1943. Riqueza em Acidos orghicos de alguns
vinhos portugueses. Anaiu inst. super. agron., Univ. t6c. Lisboa 14, 351-357.
Correia, E. M., and Vilas, M. A. 1943. Subsidio para o estudo das caracterfsticas
ffsicas, qufmicas e ffsico-quhicas dos vinhos da regiao demarcada de Colares.
Anais inst. super. agron., Univ. t b . Lisboa 14, 359-360.
Cosmo, I. 1950. Ulteriori indagini sui vini rosati e cerasuoli delle Venezie. Ann. sper.
agrar. (Rome) 4, 803-817; see also Annuar. staz. sper. viticolt. e enol. (Conegliono)
14(9), 1-15 (1950-51).
Crisci, P. 1930. Intorno alla pretesa proporzionalith fra il p H e il sapore acido delle
. soluzioni acquose con speciale riguardo ai vini. Ann. chim. appl. 20, 566-583.
&isci, P. 1931. I1 p H nell’uva-sua misura e sua interpretazione. Ann. teen. agrar.
4(1), 6-35.
brisci, P., and Michielini, L. 1932. Sul comportamento della forza acida dei vini e
mosti d’uva rispetto ad alcune operazioni tecniche. I. Chiarifacazione e de-
colorazione. Ann. chim. appl. 22, 663-666.
bruess, W. V. 1935. Some observations on tannin. Wines & Vines 16(1), 5-7.
h e s s , W. V. 1936. Observations of ’36 season on volatile acid formation in muscat
fermentations. Fruit Products J . 16, 198-200, 215, 219.
Cruess, W. V. 1942. Research in enology. Proc. Znst. Food Technol. 1942, 152-161.
Cruess, W. V. 1943. The role of microorganisms and enzymes in wine making. Ad-
vances in Enzymnl. 3, 349-386.
Cruess, W. V. 1947. The Principles and Practice of Wine Making. Avi Publishing Co.,
New York, 2nd Ed.
Cruess, W. V. 1948. Investigations of the flor sherry process. Calif. Agr. Ezpt. Sta.
Bull. NO. 710, 1-40.
Cruess, W. V. 1950. Minor constituents of vinous fermentation. Proc. Am. SOC.
Enologists 1960, 28-36.
Cruess, W. V., and Kilbuck, J. 1947. Pectic enzymes in wine making. Wines & Vines
28(8), 23-24; see also Rev. quim. ind. argentina 2 , 22-24 (1948).
Cruess, W. V., O’Neal, R., Chong, G., and Uchimoto, D. 1951. The effect of pectic
enzymes in wine making. Proc. Am. SOC.Enologists 1961, 59-75.
Cruess, W. V., Weast, C. A., and Gillilland, R. 1938. Summary of practical investiga-
tions on film yeast. Fruit Products J . 17, 229-231, 251.
Cunha, J. D. S. da. 1947. Mdtodo expedito para determinaggo do oximetilfurfural no
vinho do PBrto. Anais inst. vinho PGrto 8, 49-54.
Cunha Ramos, M. da, and Ribeiro, M. de B. 1945. A determinagfio dos agdcares
redutores no vinho do PBrto. Comun. 18th congr. Luso-espafi. par progress0 cienc.
Cdrdoba 1944, Supl. Caderno inst. vinho PGrto No. 63, 9-36.
Cunha, V. 1939. Tentativa para dosagem de glicerina nos vinhos. Arquiv. hig. sazide
ptiblica (Sdo Paulo) 4, 65-66; see also Ann. SOC.pharm. chim. Sit0 Paulo 3, 13-15
(1939).
Cusmano, I. 1949. Confront0 tra alcuni ebulliometri usati in pratica per la deter-
minazione del grado alcoolico del vino. Riv. vilicolt. e enol. (Conegliano) 3,335-340.
Dalmasso, G., and Dell’Olio, G. 1937. I vini bianchi tipici dei colli Trevigiani. Annuar.
staz. sper. vitocolt. e enol. (Conegliano) 7 , 3-115; see also Ann. sper. agrar. (Rome)
26, 4-93 (1937).
Dangoumau, A., and Debordes, G. 1937. Sur l’extraction, la separation et le dosage
des esters du vin. Bull. soc. ehim. France [5] 4, 911-918.
Daniel, E. P., and Munscll, H. E. 1932. The vitamin A, B, C, and G contents of
Sultanina (Thompson Seedless) and Malaga grapes and two brands of commercial
grape juice. J . Agr. Research 44, 59-70.
De Astis, G. 1933. Recherche5 BbulliomBtriques e t les lois de 1’6chelle alcooliques.
Bull. ofice intern. vin 6(58), 47-65.
De Astis, G. 1941. Variations du degrB alcool--6bulliomBtrique aux diverses pressions
et altitudes. Bull. ofice intern. uin 14(146), 91-96.
Delp. 1932. u b e r die Enzyme des Weines und die organischen Sauren im Wein und
Weinessig. Deut. Essigind. 36, 219.
Dicenty, D., Requinyi, G., Palinkas, S., Srabb, E., Sobs, E., Rakcsanyi, L., and
Wettstein, E. 1935. Le point de vue hongrois. 4th Congr. intern. vigne et uin,
Lausanne, 1935 I, 179-192; see also Magyar Ampelol. lhkonyv. 12, 1-88 (1938).
Diemair, W., Janecke, H., and Krieger, G. 1951. u b e r eine methode der Gerbstoff-
bestimmung in der Rebe und im Wein. 1-11. 2.anal. Chem. 133, 346-359.
Diemair, W., and Klcber, J. 1941. Beitrag zur Kenntnis der Bildung von Acetyl-
methylcarbinol und 2,3-Butylenglycol bei der Garung. 2. Untersuch. Lebensm.
81, 385-404.
Diemair, W., Riffart, H., and Mollenkopf, K. 1940. Die stufenphotometrische
Bestimmung von Milchsiiure und Glycerin im Wein. 2.anal. Chem. 118,189-201.
Dubaqui6, J. 1932. Potasse et compos-6startriques dans les vins. Ann. fals. et fraudes
26,280-285.
DubaquiB, J., and Debordes, G. 1931. D u sucre restant dans les vins rouges. Ann.
fals. et fraudes 24, 477-484.
Dubaquie, J., and Debordes, G. 1935. Antiscptiqucs et fermentation Blective. Ann.
ferment. 1, 33-40.
Dubrowskaja, V. P. 1946. Determination of alcohol (transl.). Vinodelie i Vino-
gradarstvo S.S.S.R. 6(4),40-41.
Duccllier, G. 1935. L’augmentation de la couleur des vins par le brassage automatique.
Rev. viticult. 83, 266-269.
Dujardin, J., Dujardin, L., and Dujardin, R. 1938. Notice sur les Instruments dc
PrBcision Appliques A l’oenologie. Dujardin-Salleron, Paris, 6th Ed.
Dupont, G., and Dulou, R. 1935. Sur la presence d’alcool butylique secondaire actif
dans certains fusels. Compt. rend. 200, 1860-1861.
Durmishidze, S. V. 1938. Formation of lactic acid in ordinary yeast fermentation
(transl.). Biokhimiya 3, 308-320.
Durmishidze, S. V. 1948a. Determination of oenidin in grapes and wines (transl.).
Biokhimiya 13, 16-22.
Durmishidze, S. V. 1948b. Quantitative determination of tannin in red grapes and
wines (transl.). Biokhimii?l vinodeli2 2, 169-176.
Durmishidze, S. V. 1950a. d-Catechin in the composition of grape tannin (transl.).
Doklady Akad. Nauk. S.S.S.R. 73, 987-990; C. A . 46, 719 (1951).
Durmishidze, S. V. 1950b. Oxidizing enzymes in wine grapes (transl.). Vinodelie i
VCnogradarstvo S.S.S.R.10(8), 29-30.
Durmishidze, S. V. 1950c. The polyphenoloxidase of grapes and its role in wine tech-
nology (transl.). Biokhimiya 16, 58-66.
Durmishidze, S. V. 1950d. Transformation of tannin compounds in grape clusters in
the process of ripening (transl.). BiokhimiG vinodeZi& 3, 7-24.
Durmishidze, S. V. 1951. l-Gallocatechin in the composition of tannins of grapes
(transl.). Doklady Akad. Nauk. S.S.S.R. 77, 859-862; C . A . 46, 8089 (1951).
Durmishidze, 8. V., and Khachidze, 0. T. 1952. Photometric determination of rcd
pigments in grapes and wine (transl.). Vinodelie i Vinogradarsfvo S.S.S.R. l 2 ( l ) ,

18-20.
Durmishidze, S. V., and Mosiashrili, G. I. 1948. Triose as a hydrogen acceptor in the
process of diaminisation of glutamic acid during alcoholic fermentation (transl.).
BiokhimiG vinode1iG 2 , 143-148.
Durodie, J., and Roelens, E. 1942. CaractBrisation de l’alcool butylique secondaire et
absence de l’alcool isopropylique dans les fractions de diverses huiles essentielles
de vin distillant au voisinage du point d’Ebullition de l’alcool isopropylique. Bull.
soc. chim. France [5] 9, 822-825.
Eckert, A. 1950. Sitsung des Ausschusses fur Weinforschung in Rudescheim. Z.
Egorov, I. A. 1951. New method for determining volatile acids in wine (transl.).
Vinodelie i Vinogradarstvo S.S.S.R. 11 (3), 44.
Emiliani, E. 1938. La determinazione ebulliometrica del grado alcoolico nei vini dolci.
Ann. chim. appl. 28, 409-412.
Errichelli, E. 1952. Un nuovo ebullioscopio. Riw. viticolt. e enol. (Conegliano) 6,389-391.
Errichelli, U. 1950. L’acidith volatile dei vini in rapport0 alla determinazione dell’alcool
col metodo per distillasione. Riv. viticolt. e enol. (Conegliano) 3, 99-105, 135-138.
Espil, L. 1936. Les constituants des vins. Glycerine et acide lactique. Bull. S O C . chim.
France [5] 3, 1286, 1334-1338.
Espil, L. 1937. Separation et dosage de l’acide succinique dans les milieux de fer-
mentation. Bull. SOC. chim. France 151 4, 1626.
Espil, L., Genevois, L., Peynaud, E., and RibBreau-Gayon, J. 1933. Sur la formation
des esters de l’alcool Bthylique. Enzymologia 4, 88-93.
Espil, L., and Peynaud, E. 1936. Dosage des esters neutres dans lea milieux de fer-
mentation. Bull. SOC. chim. France [5] 3, 2324-2325.
Espil, L., and Peynaud, E. 1937. Dosage des esters dans les milieux de fermentation.
Bull. SOC. chim. France [5] 4, 904-906.
Espinosa, N. A. 1943. Variaci6n de las proporciones relativas de glucosa y levulosaen
la uva y sus derivados. Industria y quim. 6 , 155-173, 213-219.
Etienne, A. D. 1950. Determination of alcohol in wines and liqueurs. J . Assoc. Ogic.
Agr. Chemists 33, 1015-1020.
Etienne, A. D. 1952. Determination of alcohol in wine and the effect of temperature
on the determination. J . Assoc. Ogic. Agr. Chemists 36, 66-67, 454-455.
Etienne, A. D., and Breyer, G. F. 1951. Determination of alcohol in wines and liqueurs.
Wines & Vines 32(4), 63-64; see also Soap, Perfumery & Cosmetics 26, 1284-
1286 (1952).
Ettinger, I. 1947. New micromethod for alcohol. Wines & Vines 28(11), 31.
Fabre, J.-H. 1936. ProcBdBes Modernes de Vinification. 11. Analyse des vins et inter-
pretation des rBsultats analytique. “La Typo-Litho ” et Jules Carbonel RBunies,
Alger, 3rd Ed.
Fabre, J.-H., and Bremond, E. 1931a. Le dosage de l’acide lactique dans les vins. Ann.
Fabre, J.-H., and Bremond, E. 1931b. Influence de l’anhydride sulfureux sur le dosage
de l’acidite volatile des vins. Ann. fals. et fraudes 24, 345-349.
Fabre, J.-H., and Bremond, E. 1932. L’acide lactique dans les vins d’AlgCrie. Ann.
fals. et fraudes 26, 99-110, 157-170.
Fabre, J.-H., and BrBmond, E. 1935. A study of different methods of estimating
alcohol in wines (transl.). Congr. intern. tech. chim. ind. agr., Bruxelles 3,308-315.
Fhbregues Soler, J. M. de, and Mestre Jane, A. 1948. MBtodos analiticos para el
Acido thrtarico de 10s vinos y residuos de 10s mismos, y ensayos realieados con un
procedimiento conductometrico. Bol. inst. nacl. invest. agron. Spain 18, 189-239.
Fantoni, P. 1949. Apparecchio per la determinaeione del grado alcoolico dei vini nella
pratica cnologica. Rend. ist. super. Sanitd 12, 740-747.
Farrugia, A. J. 1942. Influencia del anhidrido sulfuroso en el dosaje de la acidea
volhtil en 10s vinos y metodo para la determinacih de la acides volhtil libre do
SO2 en 10s vinos. Anales asoc. qutm. argentina 30, 47-48.
Fatome, M. 1935. Dosage du glycerol dans les vins. Ann. ferment. 1, 291-297; see
also Bull. ofice intern. vin 8(90), 47-53 (1935).
Faure, A., and Pallu, R. 1935. Mesure de la coloration des liquides, application aux
vins. Ann. fals. et fraudes 28, 5-9.
Faure, A., and Pallu, R. 1936. Le photo-densimhtre S.N.P. Ann. fals. et fraudes 20,
393-401.
Federico, L., and Cioffi, R. M. 1947. Sulla determinaeione degli alcoli superiori nei
vini e fermentati analoghi. Chimica e industria (Milan) 29, 298-299.
Feigl, F., and Feigl, H. E. 1946. Reagbo senslvel e seletiva para Acido thnico e outros
taninos. Brazil, Ministbrio agr., Dept. nacl. produciio mineral, Lab. produciio
mineral, Bol. No. 24, 25-30; see also Ind. Eng. Chem., Anal. Ed. 18, 62-63 (1946).
Fellenberg, Th. von. 1929. Zur kolorimetrischen Bestimmung der hoheren Alkohole
in Spirituosen. Modifikation der Methode Komarowski-von Fellenberg. Mitt.
Gebiete Lebensm. u. Hyg. 20, 16-29.
Fellenberg, Th. von. 1931a. Milchsaurebestimmung in Wein. Mitt. Gebiete Lebensm. u.
Hyg. 22, 141-145.
Fellenberg, Th. von. 1931b. Mikrobestimmungen von Glycerin, 2,3-Butylenglycol und
Milchsaure in Wein durch Verdampfen bezw. durch Dampfdestillation. Mitt.
Fellenberg, Th. von. 1932. Titrimetrische Zuckerbestimmung in Wein. Mitt. Gebiete
Lebensm. u. Hyg. 23, 77-81.
Fellenberg, Th. von. 1933. Citronensaurebestimmung in Wein. Mitt. Gebiete Lebensm.
U . Hyg. 24, 142-156.
Fellenberg, Th. von. 1935. Oxymethylfurfurol-Mikrobestimmung durch Chromsaure-
verbrennung; Oxymethylfurfurol in Sussweinen. Mitt. Gebiete Lebensm. u. Hyg.
26, 249-257.
Fellenberg, Th. von. 1936. Ameisensaure in Fruchtsaft und Sirup. Mitt. Gebiete
Lebensm. u. H y g . 27, 182-200.
Fellenberg, Th. von. 1937. Die Bestimmung des Methylalkohols in alkoholischen
Getranken. Compt. rend. 5th congr. intern. tech. chim. ind. agr. 1, 184-196.
Fellenberg, Th. von. 1943. Glycerinbestimmung in Sussweinen und Weinen. Mitt.
Fellenberg, Th. von. 1944a. Beitrag zur Untersuchung von Mistellen und Sussweinen.
Mitt. Gebiete Lebensm. u. Hyg. 36, 77-93.
Fellenberg, Th. von. 1944b. Nachpriifung des Verfahrens von A. Torricelli Bum
Nachweis von Tresterwein in Weisswein. Mitt. Gebiete Lebensm. u. Hyg. 56,
149-188.
Ferrari, C. 1939. Applicaeioni dell’analisi capillare. 1. Riconossimento di coloranti
artificiali nei vini rossi. Ann. chim. a p p l . 29, 169-183.
Ferrari, C. 1942. Nuevo metodo per il riconoscimento di tracce di coloranti artificiali
basici. Ilal. vinic. ed agr. 32(19), 238-239.
Ferrari, V. 1949. Contributo alla messa a punto di un metodo di dosaggio chimico
dell-alcool etilico. Ann. sper. agrar. (Rome) (N.S.) 3, 927-944.
Ferre, L. 1931. L’aciditB des vins e t la nouvelle reglementation. Ann. fals. et fraudes
24, 75-80.
F e d , L. 1943. Aciditat der Champagnermoste und -weine im Jahre 1942. Schweiz.
Weinztg. 61, 289-293.
FerrB, L. 1945. Fermentation malo-lactique. le Vigneron Champenois July, 1945; see
Bull. ofice intern. vin 16(171/174), 142-144 (1945).
Ferr6, L., and Archinard, P. 1935. A propos du dosage de l’acidit6 volatile des vins.
FerrB, L., and Michel, A. 1938. Dosage de la glycerine dans les vins. Ann. fals. et
fraudes 31, 85-94.
Ferrh, L., and Michel, A. 1947. Hydrolyse de la matiere colorante des vins. Casse
hydrolasique. Compt. rend. acad. agr. France 33, 239-241.
Fessler, J. H. 1941. Alcohol determination by dichromate method. Wines & Vines
22(4), 17-18.
Fessler, J. H. 1947. Some technical notes on tannin content in wines. Wine Rev. 16(5),
12-13.
Fetser, W. R. 1938. Analysis of caramel color. Znd. Eng. Chem., Anal. Ed. 10,340-353.
Filaudeau, G. 1933. Projet de methode officielle internationale pour l’analyse des vins.
Filipello, F. 1951. Correlation of fortifying brandy with wine quality. Proc. A m . Sac.
Enologists 1961, 154-156.
Fischl, P. F. 1942. The determination of alcohol and extract in wines. Food Manufac-
ture 17, 198-199.
Fischler, M. 1937. Mostgewicht-Alkoholgehalt bei badischen Weinen. Wein u. Rebe
19, 83-84.
Fischler, M. 1938. tfber die Besiehungen swisehen Mostgewicht und Alkoholgehalt
bei badischen Weinen. Wein u. Rebe 20, 193-194.
Flaney, M. 1934a. Nouvelle methode de microdosage de l’aleool mhthylique en prbs-
ence de quantites importantes d’alcools homologues. Compt. rend. 196, 94-97.
Flanzy, M. 193413. Presence de l’alcool methylique d a m les alcools de vin, de mare et
de fruit. Compt. rend. 198, 2020-2022.
Flaney, M. 1935a. Sur les mbthodes de recherche de dosage de l’alcool methylique dans
les liquidcs et les milieux naturels. Ann. fals. et fraudes 28, 146-158.
Flansy, M. 193513. Nouvelle m0thode de dosage de faibles quantites d’alcool methylique
in presence d’alcool bthylique. Ann. fals. et fraudes 26, 260-277.
Flansy, M. 1948. Les acides organiques dans les raisins et les vins. Ann. agron. 18,
60-64.
Flansy, M. 1951. L’alcool methylique dans les vins. Vignes et Vins 6(17), 28-30; (18),
27-30.
Flansy, M., and Banos, M. 1938. Presence du propanol-2 dans les alcools de vin.
Compt. rend. 206, 218-219; see also Ann. fals. et fraudes 31, 418-419 (1936).
Flansy, M., and Boudet, V. 1949. Les pertes d’alcool. Vitic. Aboric. 96, 104-107.
Flavier, H. 1939. Evolution des vitimines B1 et Bz au cours de la maturation du raisin
et de la fermentation alcoolique. Compt. rend. soc. biol. 130, 499-500.
Fleury, P., and Fatome, M. 1935. Dosage du glycerol en presence des sucres par
l’aeide periodique. J . pharm. et chim. 21, 247-266.
Florentin, D. 1948. MBthodes Actuelles d’Expertises Employees au Laboratoire
Municipal de Paris . . . 111. Boisons et derives immbdiates. Donod, Paris,
2nd Ed.
Florenzano, G. 1952. La corresione dell’acidith volatile nei vini per via biologica. Atti
accad. ital. vite e vino 3, 215-235.
Flygare, H. 1949. Zur Bestimmung der Gesamtsauren in Most und Wein. Wein u. Rebe
31, 19.
Fontanelli, G. 1941. Ricerche sull’aciditrl fissa e volatile dei vini. Ann. facoltd agrar.
univ. Pisa (N.S.) 4, 353-364.
Fonzes-Diacon, H., and Jaulmes, P. 1932. Sur I’acidit6 volatile des vins. Ann. falsif.
et fraudes 26, 149-152; see also Bull. ofice intern. vin 4(33), 95-101 (1931).
Fornachon, J. C. M. 1943. Bacterial Spoilage of Fortified Wines. Australian Wine
Board, Adelaide.
Fornachon, J. C. M. 1946. The p H of wines. Examination of glass and quinhydrone
values. Ind. Eng. Chem., Anal. Ed. 18, 790-793.
Fornachon, J. C. M. 1953. The accumulation of acetaldehyde by suspensions of
yeasts. Australian J . Biol. Sci. 6 , 222-223.
FOUCY, J. 1932. Sur le dosage de I’acidit6 volatile des vins. J. pharm. et chim. 16,
376-382.
Frangot, P. 1945. Acidit6 total et acidit6 d e l e des moiits et des vins de Champagne.
Bull. ofice intern. oin 18 (167/170), 114-118 (from le Vigneron Champenois
1945).
Frangot, P., and Geoffroy, P. 1951. Les pectines et les gommes dans les moats e t les
vins de Champagne. le Vigneron Champenois 72(2), 54-59; see also Bull. ofice
intern. vin 24(242), 94-96 (1951)A
Frolov-Bagreev, A. N., and Agabal’ianG, G. G. 1951. Chemistry of Wine (transl.).
Pischepromizdat, Moscow; for review see Vinodelie i Vinogradarstvo S.S.S.R.
12(5), 62-63 (1952).
Gahl, L. 1941. Verbesserung des Volatimeters von Cazenave-Saunier. Kisdrlet. Kozlem.
44, 147-150.
Gadzhiev, D. M. 1940. A new method for determining tartaric, malic andsuccinic
acids in wine. (transl.) Vinodelie i Vinogradarstvo S.S.S.R. 1(9/10), 21-22.
Garcia de Angulo, J. R., and Freyre, E. 1951. Notas de un metodo para determination
del nitrogen0 total en mostos y vinos por colorimetria. Bol. inst. nacl. invest.
agron. Spain, 11, 403-414.
Garino-Canina, E. 1933. I1 2-3 butilenglicole e l’acetilmetilcarbinolo nei vini e negli
aceti. Ann. ehim. appl. 23, 14-20; see alsoldnnuar. R. staz. enol. sper. Asti 1,235-
241 (1932-34).
Garino-Canina, E. 1943. La fermentation de l’acide malique dans la technique
oenologique. Bull. ofice intern. vin 16(159), 63-76.
Garino-Canina, E. 1945-1946. La catalasi nell’uva e nel mosto. Ann. accad. agr.
Torino 99, 95-97.
Garino-Canina, E. 1948. Etudes des mEthodes r6centes d’analyse des vins en vue de
leur unification internationale. 7th Congr. intern. inds. agr., Paris, 1948 1, (Q2),
B 1-9.
Garino-Canina, E. 1949. Fermentazione vinaria con dettagli biochimici del process0
della fermentazione del mosto d’uva. Ann. sper. agrar. (Rome) (N.S.) 3, 343-350;
see also Bull. oflce intern. vin 21(204), 55-61 (1948).
Garino-Canina, E. 1950. Unification des m6thodes d’analyse et d’apprhciation des
vins. 6th Congrks intern. oigne et vin, Athknes, I, 661-671; National Repts, pp.
671-717; see also Bull. ofice intern. oin 24(243), 4-60 (1951).
Garino-Canina, E. 1951a. I grandi vini da arrosto della Valtellina. Atti accad. ital. vile
c vino 3, 157-164.
Garino-Canina, E. 1951b. Per l’unificazione internazionale del servizio repressione
frodi nella preparazione e ma1 commercio del vino. Atti accad. ital. vite e vino 3,
124-137.
Garoglio, P. G. 1941-1942. Trattado de Enologia; enciclopedia vitivinicola moderna.

“Prog. vinic. ed olea., Firenze,” 5 v. (Vol. V. Analisi chimica, chimico-fisica e
merceologica dei mosti, vini, derivati e sottoprodotti.) (See also his Nuovo Trat-
tat0 di Enologia, Sansoni Edizioni Scientificke, Firenze, 3 v. 1953.)
Garoglio, P. G., and Barini-Banchi, G. 1940. Saccarimetria dei mosti concentrati.
Pros. vinic. ed olea. 16, 2782-2784.
Gatet, L. 1939a. Qtude du pouvoir r6ducteur des jus de fruits au cours de la matura-
tion. Enzymologia 6, 375-386.
Gatet, L. 1939b. Recherches biochimiques sur la maturation des fruits. Ann. physiol.
phys. biol. 16, 984-1064.
Gatet, L., and Genevois, L. 1938. Indice de brome du glucose et des substances
ph6noliques. Application 8. l’ktude de la maturation du raisin. Bull. SOC. chim.
France [5] 6 , 578.
Gatet, L., and Genevois, L. 1941. Sur le pouvoir rkducteur des vins. Bull. SOC. chim.
France [5] 8, 485-487.
Geiss, W. 1938. Zuckerbestimmung in Weinen und Siissmosten auf photometrischen
Wege mit Hilfe des Zeiss’schen Pulfrich-Photometers. Wein u . Rebe 20, 181-187.
Geiss, W. 1939. Versuche zur Klarung der filtrationsfordernden Wirkung des Filtragols
bei Apfel- und Traubensaft. Wein u . Rebe 21, 70-82.
Geloso, J. 1931. Relation entre le vieillissement des vins et leur potentiel d’oxydo-
reduction Ann. brass. el dist. 29, 177-181, 193-197, 257-261, 273-277.
Genevois, L. 1934a. Recherche de la flavine dans les vins blancs. Bull. SOC.chim.
France [5] 1, 1503-1504.
Genevois, L. l934b. La solubilit6 des tartrates de potassium e t de calcium dans les
solutions alcooliques acides, les modts et les vins. Ann. brass. et dist. 32,310-315,
326-328, 337-343.
Genevois, L. 1936. Acide succinique et glyc6rhe dans la fermentation alcoolique.
Bull. soc. chim. biol. 18, 295-300.
Genevois, L. 1937. Sur la saponification des esters en milieu neutre. Bull. soc. chim.
France [5] 4, 1506-1508.
Genevois, L. 1938. Les substances rkductrices au cours de la maturation du raisin.
Bull. 8oc. chim. France [5] 6 , 580-597.
Genevois, L. 1949a. Acides organiques et du vin. Rev. ferment. ind. aliment. 4, 67-76.
Genevois, L. 194913. Les Bquations de la fermentation alcoolique. Inds. agr. et aliment.
(Paris) 66, 329-335.
Genevois, L. 1950. Essais de bilans de la fermentation alcoolique due aux cellules de
levures. Biochim. et Biophys. Acta 4, 179-192.
Genevois, L. 1951. Les produits secondaires de la fermentation; acides organiques
des vins; matihres colorantes e t vieillissement des vins. Rev. ferment. ind.
aliment. 6, 18-25, 43-47, 88-96, 111-115.
Genevois, L. 1952. Formation des alcools supErieurs et de leurs ddrives aux cours de la
fermentation alcoolique. Inds. agr. et aliment. (Paris) 69, 27-32.
Genevois, L., Espil, L., Peynaud, E., and Rib6reau-Gayon, J. 1938a. Dosage des
acides organiques et des esters dans les modts et les vins. 18th Congr. chim. ind.
Nancy 1, 1320-1330.
Genevois, L., and Flavier, H. 1938-1939. Dosage des vitamines BI et Bz dans quelques
vine de la Gironde. PTOC. verb. d u n . SOC. sci. phy. nat. Bordeaux 1938-39, 72-73.
Genevois, L., and Gatet, L. 1940. Formation et 6volution biologique des acides
organiques dans le raisin Saint-amilion A Cognac. Rev. viticult. 02, 243-247,
259-263.
Genevois, L., Gatet, L., and Cayrol, P. 1938b. Analyse des substances rbductriees
par formation de complexes; acide ascorbique dans les moats e t les vins. 18th
Congr. chim. ind. Nancy 1, 93c-94c.
Genevois, L., Peynaud, E., and RibBreau-Gayon, J. 1946. Sur un bilan des produits
secondaires de la fermentation alcoolique. Compt. rend. 223, 693-695.
Genevois, L., Peynaud, E., and Ribkreau-Gayon, J. 1947a. Bilan sur produits sec-
ondaires de la fermentation alcoolique. Rev. ferment. ind. aliment. 2, 193-199.
Genevois, L., Peynaud, E., and RibBreau-Gayon, J. 1947b. Sur un bilan des trans-
formations de l’bthanal ajout6 8. une fermentation alcoolique. Compt. rend. 224,
762-763.
Genevois, L., Peynaud, E., and Ribbreau-Gayon, J. 1948a. Action du milieu sur lea
produits secondaires de la fermentation alcoolique des levures elliptiques. Compt.
rend. 226, 126-128.
Genevois, L., Peynaud, E., and Ribbreau-Gayon, J. 1948b. Bilans des produits
secondaires de la fermentation alcoolique dans les vins rouges de la Gironde.
Compt. rend. 226, 439-440.
Genevois, L., Peynaud, E., and Ribereau-Gayon, J. 194%. La fermentation alcoolique
des vins doux naturels. Compt. rend. 227, 227-228.
Genevois, L., Peynaud, E., and Ribbreau-Gayon, J. 1949a. La fermentation alcoolique
des vins blancs de la Gironde. Compt. rend. 229, 479-480.
Genevois, L., Peynaud, E., and RibCreau-Gayon, J. 1949b. Sur la quantit6 d’eau
fixbe par les produits de la fermentation alcoolique et sur une relation empirique
entre la quantit6 d’acides et de glycerol form&. Compt. rend. 229, 777-778.
Genevois, L., Peynaud, E., and Rib6reau-Gayon, J. 1949c. Sur une relation num6rique
entre les divers produits secondaires de la fermentation alcoolique. Bull. soc.
chim. biol. 31, 369-373.
Genevois, L., and RibQeau-Gayon, J. 1933. Les Bquilibres ioniques dans les mohts e t
les vins. Ann. brass. et dist. 31, 273-277, 289-294, 305-311.
Genevois, L., and Ribbreau-Gayon, J. 1935a. Coefficients tampons de quelques vins.
Bull. S O C . chim. France [5] 2, 1286-1287.
Genevois, L., and Ribbreau-Gayon, J. 1935b. Sur les substances azot6es des mohts
e t des vins. Ann. ferment. 1, 541-546; see also Bull. ofice intern. vin. 9(94), 36-41
(1936).
Genevois, L. and RibBreau-Gayon, J. 1947. Le Vin. Hermann & Cie., Paris.
Gentilini, 1,. 1937. Di alcuni carboni in enologia. Annuar. staz. sper. vilicolt. e enol.
(Conegliano) 7 , 301-337.
Gentilini, L. 1939. Contributo all0 studio dell’analisi cromatografica applicata alle
materie coloranti naturali del vino. I. Ann. chim. appl. 29, 169-183.
Gentilini, L. 1941. Studio dell’analisi cromatografica applicata a1 riconoscimento di
alcuni dei pih comuni sofisticanti del colore dei vini rossi. Annuar. staz. sper.
niticolt. e end. (Conegliano) 10, 59-72.
Gentilini, L. 1947. L’acetato di etile e l’aciditii volatile nei vini spunti e acescenti.
Annuar. slaz. sper. viticolt. e enol. (Conegliano) 13(1),2-8; see also Ztal. vinic. ed
agr. 37(10), 112-114; ( l l ) ,125-128; ( l a ) , 137-139 (1947).
Gentilini, L. 1950. Procedimiento cuprometrico rapido per il dosaggio degli zuccheri
riduttori nel inosto d’uva. Ann. sper. agrar. (Rome) (N.S.) 4, 213-217; see also
Annuar. slaz. sper. viticolt. e enol. (Conegliano) 14(2), 213-217 (1950-51).
Gentilini, L. 1952. I1 p H in enologia. Riv. viticolt. e enol. (Conegliano) 6, 423-429.
Gentilini, L., and Carli, E. 1950. Corrispondenza fra gradi Babo e contenuto zuc-
cherino reale nei pih comuni mosti d’uva del veneto. Ann. sper. agrar. (Rome)
(N.S.) 4, 1049-1067; see also Annuar. staz. sper. viticolt. e enol. (Conegliano)
14(4), 1-19 (1950-51).
Georgacopoulos, MM., and Costopoulos, J. 1952. Dosage des sucres par la methode
au ferricyanure. Bull. ofice intern. vin 26(258), 114-120.
Gerasimov, M. A. 1931. Die aktuelle Aeiditat des Traubensaftes und des Weines.
Magarach. K r v m s k a a zonal’naa o p g t n a s stanssiâ PO vinogradamtvu i vinodeliu,
Trudy 2, 1-42; see also Bull. ofice intern. vin 6(48), 22-27 (1932); Ann. fals. et
fraudes 26, 353-355 (1932).
Gerasimov, M. A., and Vinogradova, N. I. 1931. Der Gehalt des Vitamins C im
Traubensafte und im Wein. Magarach. K r y m s k a a zonal’naa opytna& stanszi&
PO vinogradarstvu i vinodeliu, Trudy 3, 1-21.
Gerum, J. 1932. Die Bestimmung der Oechslegrade mit Hilfe des Zeiss’schen Ein-
tauchrefraktometers. W e i n u. Rebe 14, 235-238.
Ghimicescu, G. 1935a. Microdosage colorim6trique de l’acide malique dans les vins.
Ann. sci. univ. Jassy 21, 321-325.
Ghimicescu, G. 1935b. Microdosage colorim6trique de l’acide tartrique total dans les
vins. Ann. sci. uniu. Jassy 21, 326-329.
Ghimicescu, G. 1935c. Microdosage colorim6trique du bitartrate du vin. Ann. sci.
univ. Jassy 21, 343-345.
Ghirnicescu, G. 1935d. Microdosage de l’acide volatile dans les vins. Ann. sci. univ.
Jassy 21, 306-314
Ghimicescu, G. 1935e. Une micromethode pour le dosage du acide lactique dans les
vins. Ann. sci. univ. Jassy 21, 315-320.
Ghimicescu, G. 1935f. Une nouvelle microm6thode pour le dosage colorim6trique de
la glycerine dans les vins. Ann. sci. univ. Jassy 21, 346-351.
Ghimicescu, G. 1937. Microdosage des sucres dans le vin. Mikrochemie 22, 201-207.
Ghimicescu, G. 1938. Sur le dosage de l’extrait sec du vin, du vinaigre, et de 1s bibre.
Ann. sci. univ. Jassy 24, 87-90.
Ghimicescu, G., and Gheorghiu-Vieriu, A. 1938. Microdosage de la tanin dans le vin.
Mikrochemie 26, 187-191.
Gierer, S., and Hoffmann-Ostenhof, 0. 1951. Eine kolorimetrische Mikromethodik
zur quantitativen Bestimmung der Fuselalkohole bei Garversuchen. Mikro-
chemie 38, 232-236.
Gobis, L. 1950. Richerche sul mecanismo di formaeione dell’acido citric0 nel corso
della fermentaeione alcoolica. Ann. sper. agrar. (Rome) (N.S.) 4, 569-578.
Gobis, L., and Farfaletti-Casali, P. L. 1952. I1 contenuto in acetilmetilcarbinolo e
glicol butilenico nei vini e il suo significato enologico. Riv. viticolt. e enol. (Cone-
gliano) 6, 297-300.
Godet, Ch. 1949. Sur l’unification des mbthodes d’analyse des vins. Bull. ofice intern.
vin 22(223), 60-66; see also 7th Congr. intern. ind. agr., Paris, 1948 1(&2), 1-7
( 1948).
Godet, C., and Charribe, R. 1946. Sur le dosage de l’acide citrique dans lea vins, jus
de fruits, concentrb, etc. Mitt. Gebiete Lebensm. u. Hyg. 37, 317-326.
Godet, C., and Deuel, H. 1947. Les poids sp6cifiques de l’extrait du vin. Mitt. Gebiete
Lebensm. u. Hyg. 38, 5-11.
Godet, C., and Martin, L. 1946. Contribution la connaissance des vins suisses. Mitt.
Goes, L. A. de A., and Correia, E. M. 1942. Influbcia de raga de levedura (Saech.
ellipsoideus) como elemento a considerar na correcpfio da concentraqb hidro-
geni6nica dos mostos. Anais inst. super. agron., Univ. tdc. Lisboa 13, 171-175.
Goldbach, N. J., and Opperschaum, R. Z. 1950. Zum toxikologischen Nachweis des
Methylalkohols als p-Brombenzosauremethylester. Z. Lebensm.-Untersuch.
u.-Forsch. 91, 179-181.
Gomes, J. V. M. 1941. Doseamento do Alcool por destilaqb nos vinhos do PBrto.
Anais inst. vinho Pdrto 2, 305-359.
Gomes, J. V. M. 1945. Microflora duriense; leveduras produtoras de esteres. Anais
inst. vinho Pdrto 6(2), 59-75.
Goresline, H. E., and Champlin, F. M. 1938. Sugars in champagne production. Ind.
Eng. Chem. SO, 112-113.
Got, N. 1947. Les Vins Doux Naturels. Chez I’Auteur, Perpignan, France.
Gottschalk, A. 1946. Mechanism of selective fermentation of d-fructose from invert
sugar by Sauternes yeast. Biochem: J. 40, 621-626.
Grandchamp, L.-E., and Vollaire-Salva, J. 1939. Dosage rapide des esters des acides
gras volatils. Ann. fals. et fraudes 32, 244-247.
Grivas, G. 1952. Le dosage de I’extrait du vin d’aprb le poids specifique de la solution
aqueuse. Evaluation de la contraction de la solution. Bull. ofice intern. vin
26(258), 121-123.
Grossfeld, J., and Battay, F. 1931. Versuche uber Nachweis, Bestimmung und
Vorkommen der Buttersaure in Lebensmitteln. Z. Untersuch. Lebensm. 61, 129-
161.
Grossfeld, J., and Miermeister, A. 1928. Vorkommen, Nachweis und Bestimmung von
Laurinsaure in alkoholischen Getranken. Z. [Jntersuch. Lebensm. 66, 167-187.
Guimarks, A. F. 1950. Doseamento do Acid0 tarthico no vinho do PBrto. Anais. insl.
vinho Pdrto 11(1), 59-72.
Guittonneau, G., BBjambes, M., and Tavernier, J. 1941a. Sur la presence et l’origine
de l’ac6tylmBthylcarbinol e t du butanediol 2-3 dans les cidfes normands. Ann.
ferment. 6, 159-167.
Guittonneau, G., Tavernier, J., and BBjambes, M. 1941b. Sur la presence et l’origine
de l’ac6tylm6thylcarbinol et du butanediol 2-3 dans les cidres normands. Les
Aerobacter en cidrerie. Compt. rend. 213,257-259.
Guntz, A. A. 1950. Echangeurs d’ions et oenologie. Chim. anal. 32, 246-248.
Guymon, J. F. 1951. Report on methanol. Colorimetric determination of methanol in
solutions containing ethyl alcohol. J. Assoc. Ofic. Agr. Chemists 34, 310-328.
Guymon, J. F. and Heitz, J. E. 1952. The fuse1 oil content of California wines. Food
Technol. 6, 359-362.
Gvaladze, V. Z., and Rodopulo, A. 1946. Acetylmethylcarbinol as indicator of the
beginning of acetic acid fermentation in wines. (transl.) Vinodelie i Vinogradarslzw
S.S.S.R. 6(5), 8-11.
Haagen-Smit, A. J., Hirosawa, F. N., and Wang, T. H. 1949. Chemical studies on
grapes and wines. I. Volatile constituents of Zinfandel grapes (Vitis vinifera var.
Zinfandel). Food Research 14,472-480.
Hall, N. A., Krupski, E., and Fischer, L. 1952. Methanol in Washington wines. Am.
J. P h a m . 124, 343-350.
Hanak, A. 1932. Prazisionsdestillations-apparat.Chem.-Zgt. 66, 984.
Hargreaves, C. A., Abrahams, M. D., and Vickery, H. B. 1951. Determination of
citric and d-isocitric acids. Anal. Chem. 23, 467-470.
Hartmann, B. G. 1939. Report on (the analysis of) wines. J. Assoc. 0,fir. Agr. Chem.ists
22, 210.
Hartniann, B. G., and Hillig, F. 1930. Determination of tartaric acid in fruits and
fruit products. J. Assoc. Offic. Agr. Chemists 13, 103-1 12.
Hauptmann, K. H. 1952a. Action des enzymes de filtration dans la fabrication des jus
de fruits. Inds. agr. et aliment. (Paris) 69, 577-580.
Hauptmann, K. H. 195213. Fermentierungsprobleme bei der Weinbereitung. Angew.
Chem. 64, 358.
Heide, C. von der. 1932. Erkennung von weissem Wein aus roten Trauben. Wein u.
Rebe 14, 3-4.
Heide, C. von der. 1934. Spezielle analytische Methoden. 2.anal. Chem. 97, 221-240.
Heide, C. von der, and Burkard, J. 1935. Vber den Restzucker im Wein. 2. Untersuch.
Lebensm. 69, 68-71.
Heide, C. von der, and Hennig, K. 1934. Die Bestimmung der Weinsaure in Most und
Wein. Bericht Lehr- u.Forschungsanstalt, Wein-, Obst- u. Gartenbau, Geisenheim
1934, 26-27.
Heide, C. von der, and Mandlen, H. 1930. Vergleichende Bestimmungen der Wasser-
stoffionen-Konzentration im Wein nach verschiedenen Verfahren. Wein u. Rebe
11, 495-516.
Heide, C. von der, and Mandlen, H. 1933. Tafeln zur Ermittlung des Alkohol- und
Extraktgehaltes des Weines aus den Spezifischen Gewichten von Alkohol-
Wassermischungen und von Saceharoselosungen bei 20°, bezogne auf Wasser von
4". Z. Untersuch. Lebensm. 66, 338-341.
Heide, C. von der, and Zeisset, W. 1935. Die Extraktbestimmung im Wein. 2. Unter-
such. Lebensm. 69, 138-145.
Heide, E. 1940. Etwas uber Farbstoffausscheidungen bei Rotweinen. Weinland 12,
96-98.
Heiduschka, A., and Pyriki, C. 1930. Untersuchung von 1928er Traubenweinen des
Weinbaugebietes Pillnitz-Lossnitz-Meissen-Seusslitz. Z . Untersuch. Lebensm.
69, 104-109.
Heiduschka, A., and Sommer, H. 1935. Citronensaurebestimmung im Wein. Pharm.
Zentralhalle 76, 593-595.
Heitz, J. E., Roessler, E. B., Amerine, M. A., and Baker, G. A. 1951. Certain factors
influencing the composition of California-type sherry during baking. Food
Research 16, 192-200.
Hennig, K. 1943. Neue Wege zur Qualitatsbestimmung der Weine: Die Aromastoffe
des Weine. Deut. Weinbau 22, 311-313.
Hennig, K. 1944. Einige Fragen zur Bilanz der Stickstoffverbindungen im Most und
Wein. 2.Lebensm.-Untersuch. u.-Forsch. 87, 40-48; see also Bull. oflce intern. win
16(159), 82-86 (1943).
Hennig, K. 1950. Chemische Untersuchungsmethoden fur Weinbereiter und Suss-
mosthersteller. Verlagsbuchhandlung Eugen Ulmer, Stuttgart, 3rd Ed.
Hennig, K. 1950-1951. Die Miiller-Thurgau-Weine im Lichte der neuzeitlichen
Forschung. Wein u. Rebe 26, 15-24.
Hennig, K. 1951-1953. Fortschritte und Arbeiten auf dem Gebiet der Garungs-
physiologie, Weinbehandlung und Weinchemie. Deut. Wein-Ztg. 87, 551-556 ; 88,
258-260, 281-283, 569-570; 89, 29-30.
Hennig, K. 1952. Der Deutsche Schaumwein. Deut. Wein-Ztg. 88, 341-342, 345-346.
Hennig, K., and Burkhardt, R. 1951. Die polarographische Bestimmung der xpfel-
saure im Wein. Z . Lebensm.-Untersuch. u. -Forsch. 92, 245-252.
Hennig, K., and Oshke, P. 1942. Bilanz der Stickstoffverbindungen in einem glirenden
Most und Jungwein. VorratspJlege u. Lebensmittelforsch. 6, 408-428.
Hennig, K., and Villforth, F. 1942. Die Aromastoffe der Weine. Ein Beitrag zur
Untersuchung der Duft- oder Bukettstoffe der Weine. I. Isolierung der Aroma-
stoffe. 11. Die Ester. Ihre Komponenten. VorratspJlege u. Lebensmittelforsch. 6,
181-199, 313-333; see also Bull. ofice intern. vin 16(158), 8-16 (1943).
Herrcro, A. 19-13. Sustancias hidrolizables, expresadas en dextrina, de vinos naturales
dcl Irruguay. Anales asoc. quim. argenlina 31, 66-67.
Hewitt, J. T. 1939. The chemistry of wines. Science Prog. 33, 625-644.
Heyrovskjr, J., Smoler, J., and gtastnjr, J. 1933. Polarographic examination of fer-
mentation products (transl.). V h t n i k Ceskoslov. Akad. Zemkd&lskd 9, 599-607;
C. A . 28, 5171 (1934).
Hickinbotham, A. R. 1948. Acidimetric titration of lactic acid. Analyst 73, 509.
Hickinbotham, A. R., and Ryan, V. J. 1948. Glycerol in wine. Australian Chem. Znst.,
J . and Proc. 16, 89-100.
Hickinbotham, A. R., and Williams, J. L. 1940. The application of enzymatic clarifi-
cation to winemaking. J. Agr. 8. Australia 43, 491-495, 596-602.
Hillig, F. 1937. The colorimetric determination of lactic acid in fruits and fruit
products. J . Assoc. Ogic. Agr. Chemists 20, 605-610.
Hogl, 0. 1952. Untersuchungen an Sussweinen und Mistellen rnit papierchromato-
graphischen Methoden. Z . Lebensm.-Untersuch, u. -Forsch. 96, 177-179.
Hohl, L. A., and Joslyn, M. A. 1941a. Formic acid formation in alcoholic fermentation.
Plant Physiol. 16, 755-769.
Hohl, L. A., and Joslyn, M. A. 1941b. Lactic acid formation in alcoholic fermentation
by yeast. Plant Phusiol. 16, 343-360.
Holley, A. D. 1951. Research on the volatile constituents of grapes. Canner 113(21),
10-11.
Holzbach, F. 1936. Die schnelle Bestimmung der Gesamtsauren in der Wein- und
Obstpraxis durch die sogenannte “trockene Titration.” Sbornik C?eukoslov. Akad.
Zemkdklskd 11, 271-276.
Hozumi, T., and SatB, M. 1950. Studies on alcoholic beverages by paper partition
chromatography. I. Detection of reducing sugars of sake and wines. J . SOC.
Brewing J a p a n 46,322-325.
Hugues, E. 1930. L’acidit6 volatile des vins. Progr. agr. et vit. 93, 300-301.
Hugues, E. 1934. Les vitamines dans le jus de raisin et dans le vin. Progr. agr. el vit.
101, 561-563.
Hugues, E., and Chevalier, R. 1930. L’acidit6 lactique de quelques vins de 1’Herault.
Ann. fals. el fraudes 23, 214-216.
Hunk&, B. 1938. Schnellmethode zum Nachweis von Verfalschungen geistiger
Getranke mit Methylalkohol. Kisdrletuggi KiizZernhyek 41, 111-112; Bee also
Chem.-Ztg. 62, 240 (1938).
Huntenburg, W. 1936. Exakter Nachweis des Oxymethylfurfurols in Sussweinen.
IIuriez, H. 1948. Le vignoble nantais (Muscadet de S h e et Maine). Ann. agron. 18,
528-546.
Husfeld, A., Stellwaag, F., and Kramer, 0. 1939. Sonderaufgaben des Forschungs-
dienstes im Weinbau und in der Weinkellerwirtschaft Deutschlands. Forschungs-
dienst 8, 339-345.
Hussein, A. A., and Cruess, W. V. 1940a. A note on the enzymic darkening of wine.
Fruit Products J . 19, 271.
Hussein, A. A,, and Cruess, W. V. 1940b. Properties of the oxidizing enzymes of cer-
tain vinifera grapes. Food Research 6, 637-648.
Hussein, A. A., Mrak, E. M., and Cruess, W. V. 1942. The effect of pretreatment and
subsequent drying on the activity of grape oxidase. Hilgardia 14, 347-357.
Iljin, W. S. 1943. Der Stoffwechsel bei der Weinrebe wiihrend der Kalkchlorose.
Gartenbauwiss. 17, 338-381.
Ionescu, M. V., and Popescu, 0. 1930. Contribution B 1’Btude de l’identification et du
dosage de l’alcool methylique en presence de l’alcool Bthylique. J . pharm. chim.
12, 63-79.
Ionescu, M. V., and Popescu, C. 1934. Contribution B l’btude de la determination de
l’acidit6 volatile des vim. Influence de I’acide lacti que. Analele Znst. Cercetliri
Agron. Roumdniei 6, 303-313.
Ionescu, M. V., Gad, L., and Popescu, 0. 1933. Contribution B 1’6tude de la deter-
mination de l’acidit6 volatile des vins. Analele Znst. Cercetari Agron. Roum&niei
6,215-240.
Iribarne, J. V. 1941. La determinaci6n de aldehidos en vinos. Anales asoc. q u h .
argentina 29, 129-138.
Ivanov, B. V. 1947. Loslichkeit des Weinsteins in verschiedenen Medien (transl.).
Vinodelie i Vinogradarstvo S.S.S. R . 7(10), 17-20.
Jacquin, P., and Tavernier, J. 1951-1952. Contribution B 1’6tude de la fermentation
-alcoolique des cidres et poir6s. Znds. agr. el aliment. (Paris) 68, 599-607; 69,
115-127.
Jacquin, P., and Tavernier, J. 1952. Emploi de l’ac6tate d’isopropyle comme solvant
dans le dosage du glycerol par oxydation periodique. Inds. agr. et aliment. (Paris)
69, 497-499.
Jahr, H. 1931. Beitrag zur Bestimmung des Rohrzuchers in Susswein. 2.anal. Chem.
83,321-338.
Jakovliv, G. 1952. Essai de caracterisation chromatographique des fruits dans les
aliments sucres et les boisson. Znds. agr. et aliment. (Paris) 69, 223-225.
Jauker, H. 1937. Vber das Vorkommen und die Bestimmung von Methylalkohol in
Giirungsprodukten. Dissertation, Pharmakognostischen Institut, Universitat
Tartu.
Jaulmes, P. 1934. La dktermination de l’acidit6 volatile des vins. Ann. fals. et fraudes
27, 547-550.
Jaulmes, P. 1935a. La defecation B la chaux et le dosage de l’acidit6 volatile des vins.
Ann. fals. et fraudes 28, 540-545; see also Bull. pharm. Sud-Est 39, 235-241
(1935).
Jaulmes, P. 1935b. L’influence des matikres extractives sur la distillation des acides
volatils du vin. Ann. fals. et fraudes 28,590-599; see also Bull. pharm. Sud-Est 39,
314-325 (1935).
Jaulmes, P. 1950. Appareil pour la separation des produits entrainables par la
vapeur d’eau: alcool, acidit6 volatile, ammoniaque, etc. Ann. fals. et fraudes 43,
110-117.
Jaulmes, P. 1951. Analyse des Vins. Librairie Coulet, Dubois et Poulain, Montpellier,
2 Ed.
Jaulmes, P. 1952. L’acidit6 volatile et son dosage. Rev. agr. Afrique du Nord 60,27-31,
4.2-50.
Jaulmes, P. 1953. Remarques sup la determination d6nsimetrique du degre alcoolique
des vins par la mbthode officielle. Ann. fals. et fraudes 46, 84-95.
Jaulmes, P., and Espezel, R. 1935. Le dosage de l’acetaldehyde dans les vins e t les
spiritueux. Ann. fals. et fraudes 28, 325-335.
Jaulmes, P., and Marignan, R. 1953. Consbquences de l’adoption de la temperature
de r6f6rence de 20’ pour la definition du degr6 alcoolique des vins et des spiritueux.
Jaulmes, P., and Slizewicz, L. 1943. Les rbgles oenologiques et la methode officielle de
dosage des acidit6s totale et h e des vins. Bull. SOC. chim. France [5] 10, 59-60.
Jeanpbtre, J. 1931. Un am6lioration du procBd6 Landmann pour la determination de
l’acidit6 volatile du vin. Mitt. Gebiete Lebensm. u. Hug. 22, 94.
Jeroch, 0. P. 1947. The use of walnut extracts for vitaminizing wines (transl.).
Vinodelie i Vinogradarstvo S.S.S.R. 7(1), 17-18.
Jilke, W. 1950. Beitrag zur Statistik der Rheingauer Moste des Herbstes 1948. 2.
.Jilke, W. 1951. Uber die Anwendbarkeit refraktometrischer Methoden bei der Unter-
suchung von Wein, weinahnlichen und weinhaltigen Getranken. 2. Lebensm-
Untersuch. u. -Forsch. 93, 357-362.
Jim6nez de Abeledo, M. E. 1937. Determinaci6n de acidez total en vinos. Anales asoc.
qufm.argentina 20, 81-93.
Jimbnea de Abeledo, M. E., and Mendivelaba, G. 1939. Determinaci6n de azucar
reductor en vinos. Anales asoc. qufm. argentina 27, 1-20.
Joslyn, M. A. 1938a. Electrolytic production of rancio flavor in sherries. Znd. Eng.
Chem. SO, 568-577.
Joslyn, M. A. 1938b. Report on volatile acids in ines. J. Assoc. O@c. Agr. Chemists
21, 166-174.
Joslyn, M. A. 1939. Report on volatile acids in wines. J. Assoc. O@c. Agr. Chemists
22, 210-220.
Joslyn, M. A. 1940a. The by-products of alcoholic fermentation. Wullerstein Lab.
Communs. 3(8), 30-43.
Joslyn, M. A. 1940b. Report on volatile acids in wine. J. Assoc. OJic.Agr. Chemists
23, 183-189.
Joslyn, M. A. 1949. Chemistry of alcoholic fermentation. Proc. Wine Tech. Conf.
(Davis) 1949, 86-88.
Joslyn, M. A. 1960. Methods in Food Analysis, Applied to Plant Products. Academic
Press, New York.
Joslyn, M. A., and Amerine, M. A. 1941. Commercial production of dessert wines.
Calif. Agr. Expt. Sta. Bull. 661, 1-186.
Joslyn, M. A., and Comar, C. L. 1938. Determination of acetaldehyde in wines. Znd.
Eng. Chem., Anal. Ed. 10, 364-366.
Joslyn, M. A., and Comar, C. L. 1941. Role of acetaldehyde in red wine. Znd. Ens.
Chem. 33,919-928.
Joslyn, M. A,, and Dunn, R. 1941. Acid metabolism of wine yeast. I. The relation of
volatile acid formation to alcoholic fermentation. J. Am. Chem. SOC.60, 1137-
1141.
Joslyn, M. A., Farley, H. B., and Read, H. M. 1929. Effect of temperature and time
of heating on extraction of color from red-juice grapes. Ind. Eng. Chem. 21, 1135-
1137.
Joslyn, M. A., and Marsh, G. L. 1935. Methods of wine analysis. I. Comparisons of
direct and indirect methods of determining alcohol, extract, and total acid in dry
wine, J. Assoc. Ofic. Agr. Chemists 18, 307-313.
Joslyn, M. A., Marsh, G. L., and Fessler, J. 1937. A comparison of several physical
methods for the determination of the alcohol content of wine. J . Assoc. OJic. Agr.
Chemists 20, 116-130.
Kacsmarek, A., and Weise, R. 1942. Physikalisch-chemische Untersuchungsmethoden
fur die Rebenzuchtung. I. Uber das Absorptionsspektrum als Hilfsmittel bei dcr
Rotweinzuchtung. Gartenbauwiss. 16,314-357.
Kalberer, 0. E. 1930. Die Anwendbarkeit der titrimetrischen Zuckerbestimmunds-

methode Bruhns fiir Obst- und Traubenweine. Mitt. Gebiete Lebensm. u. Hyg. 21,
114-115.
Karamboloff, N. 1932. Eine neue ( 1 ) Rbaktion, um das Vorhandensein von Kunst-
farben in den Weinen festzustellen. Wein u. Rebe 14, 5 (see original in Ann. Univ.
Sofia 10, 207-218, 1931-32.)
Katar’gn, T. G. 1951. Accomplishments of Magarach Institute (All-Union Scientific
Research Institute of Wine Manufacture and Viticulture) (transl.). Vinodelie i
Vinogradarstvo S.S.S.R. 11(6), 58-59.
Ketelbant, E. 1936. Dosage des acides volatils dans les liquidcs ferment&. Ann.
ferment. 2, 109-127.
Kielhofer, E. 1937. Die Kellerbehandlung des Weins. Erkenntnisse und technische
Fortschritte im letzten Jahrzehnt unter besonderer Berucksichtigung der mittel-
europakchen Weinbaugebiete. Forschungsdienst 4, 331-341, 382-391.
Kielhofer, E. 1942. Troubles albuminoxdes du vin. Bull. ofice intern. vin 16(152),
7-10.
Kielhofer, E. 1944. Die Farbmessung bei Rotweinen. Wein u. Rebe 26, 1-14.
Kielhofer, E. 1948. Die Warmetrubung der 1947er Weine. Deut. Wein-Ztg.84,293-294.
Kielhofer, E. 1948-1949. Die Eiweisstrubung der 1947er Weine. Der Weinbau.
Wissenschaftl. Beih. 2 , 233-240; 3, 10-17, 33-39.
Kielhofer, E. 1949. Erfahrungen bei der Behandlung von Weinen, dic zu Erweiss-
trubungen neigen. Weinblatt 43, 540-542.
Kielhofer, E. 1951. Die Eiweisstrubung des Weines. IV. Uber die Beschaffenheit des
Trubes und die Verminderung des N-Gehaltes des Weines durch die Trubab-
scheidung. 2.Lebensm.-Untersuch. u. -Forsch. 92, 1-9.
Kielhofer, E. 1952. Der Anteil der Weinbaulander an der oenologischen Forschung.
Deut. Wein-Ztg. 88, 57-58.
Kielhofer, E. 1953. Die Bestimmung des Gehalts an unvergorenem Zucker in der
Praxis. Deut. Wein-Ztg. 89, 163-164.
Kielhofer, E., and Gunther, P. 1937. Vber die Beaiehungen zwischen Mostgewicht,
Zuckergehalt, Zuckerzusatz und Alkoholgehalt beim unreifen 1936er Mosten des
Moselgebietes. Wein u. Rebe 19, 33-45.
Kiessling, W. 1949. Formation d’acides triosephosphoriques au cours de la fermenta-
tion due aux levures vivantes. Inds. agr. et aliment. (Paris) 66, 111-118.
Kilbuck, J. H., Nussenbaum, F., and Cruess, W. V. 1949. Pectic enzymes: invcstiga-
tions on their use in making wines. Wines & Vines SO@), 23-25.
Kilp, W., and Deplanque, R. 1934. I n welchem Stadium der Giirung beginnt die
Fuselolbildung. 2.Spiritusind. 67, 306, 308, 313-314, 316; see also Brennerei Ztg.
61, 207-200 (1934).
Klinc, L. 1935. Uber den Buttersauregehalt der Weine und Essige. 2. Untersuch.
Lebensm. 69, 363-369.
Kling, A. 1936. Au sujet du dosage de l’acide tartrique h 1’6tat de racemate de chaux.
Kniphorst, L. C . E., and Kruisheer, C. I. 1937. Die Bestimmung von 2-3-Butylen-
glykol, Acetylmethylcarbinol und Diacetyl in Wein und anderen Garungs-
produkten. 2. Untersuch. Lebensm. 73, 1-19; 74, 477-485.
Kobakhidze, M. C. 1950. Origin of acetylmethylcarbinol during acetic fermentation
(transl.). BiokhimiG vinodeli2 3, 138-146.
Koch, J., and Bretthauer, G. 1950-51. Der Einfluss von Ultraschallwellen auf das
Aroma verschiedener Weine. Wein u. Rebe 26, 85-98.
Koch, J., and Bretthauer, G. 1951. Uber eine zuverlassige Methode zur Bestimmung
der Milchsaure in Sussmost und Wein. 2. anal. Chem. 132, 346-356.
Kocsis, E. A., CaokBn, P., and Horvai, R. 1941. Weinfarbenuntersuchung mittels
Capillar- Luminescenz-Analyse. 2. Untersuch. Lehensm. 81, 316-321; see also
ibid. 84, 332-340 (1942).
Kogan, A. J. 1930. Uber die Bestimmung der Citronensaure bei Gegenwart anderer
organischer Substanzen. 2.anal. Chem. 80, 112-122.
Kondareff, M. 1940. Bilanz und Verteilung der organischen Sauren in dem sud-
bulgarischen Weinen. 2. Untersuch. Lebensm. 80, 527-542; see also 2. Landwirt.
Versuchsstat. Sofia 9 (1939).
Konek, F., and Wettstein, E. 1934. Verfahren zur quantitativen Trennung und
Bestimmung der im Wein enthaltenen organischen Sauren. Magyar Tudombmyos
Akaddmia, Budapest, Matematikai ds termdszettundomdnyi drtesto. 61, 305-324; see
also Magyar Ampelol. Ibkonyo 9, 429-439 (1935).
Kopal, S. 1937. Chemick6 SloIenE Ceskos1ovenskJ;ch v h . NBkladem ministerstva
zemtidtilsvi republiky Ceskoslovenski, Praha.
Kopal, S. 1938. Die slowakische Weinfabrikation und die chemische Zusammen-
setzung der Weine. Chem. Listy V2du Premysl 32, 266-375.
Korotkevich, A . V. 1948. Relative acidity of musts and wines (transl.). Biokhimiz?a
uinodeli; 2, 149-168.
Korotkevich, A. V. 1949. Determination of alcohol in table wines by ebulliometry
(transl.). Vinodelie i Vinogradarstvo S.S.S.R. 8(10), 40; C. A . 44, 273 (1950).
Korotkevich, A. V. 1951. Determination of wine color by a photometer (transl.).
Vinodelie i Vinogradarstvo S.S.S. R. 11(1), 20-22.
Korotkevich, A. V., and Arbuzova, E. M. 1949. The determination of glycerol in dry
wines (transl.). Vinodelie i Vinogradarstvo S.S.S.R. 9(7), 29-31.
Kozenko, E. M. 1952. On the contents of stable esters of carbonic acid in wines
(transl.). Vinodelie i Vinogradarstvo S.S.S. R. 12(4), 25-28.
Kramer, M., and Satterfield, G. H. 1942. Ascorbic acid content of four varieties of
raw wines. Food Research 7, 127-129.
Kramer, 0. 1935. Die Bestimmung des Mostgewichtes mittels des Zeiss’schen Hand-
zuckerrefraktometers. W e i n u.Rebe 17, 275-281.
Kramer, 0. 1941a. Enstehung und Bedeutung der fluchtigen Sauren im Wein. W e i n -
land 13, 41-44, 72-75.
Kramer, 0. 1941b. Die Milchsaure im Wein. Weinland 13, 93-94, 122-123, 134-137.
Kramer, 0. 1942. Neuere Probleme der Kellerwirtschaft. Forschungedienst 16, 525-537.
Kramer, O., and Bohringer, W. 1940. Untersuchungen und Versuche uber die Saure-
verhiiltnissc bei wiirttembergischen Weinen und Mosten. W e i n u. Rebe 22,97-110,
123-142, 230-245, 252-269.
Krause, H. 1948. Neues Verfahren zur Bestimmung der Milchsaurc. Weinbau.
Wissenschaftl. Beih. 2, 297-307.
Krauze, S. 1933. Zur Frage des Nachweises von geeuckerten Weinen. Mitt. Gebiete
Lebensm. u. HYg. 24, 79-90.
Kretzdorn, H. 19-19. Beitrag zum Gerbstoffnachweis nach Nessler und Barth. Deut.
Wein-Ztg. 86, 11-12.
Krombach, H. 1948. Essais comparatifs des mkthodes de dosage de l’extrait sec sur
des vins de la Moselle. 7th Congr. intern. inds. agr., Paris 1(&2),D 1-3.
Kruisheer, C. I., Vorstman, N. J. N., and Kniphorst, L. C. E. 1935. Bestimmung des
Oxymethylfurfurols und des Lavulosins in Portwein und anderen Sussweinen.
liutal’ova, T. 1931. Die Roue der Aminosauren bei der Gestaltung des Weinbouquetes.
Odessa. fientrol’na naukovo-doslidchot vinogrado-vinorotcha stanGi2, Pratsi 4,
116-120.
Luganne, H. 1938. Calcul de l’extrait sec reduit dans les vins liquoreaux. Ann. fals. el
fraudee 31, 221-224.
Lagneau, C. 1945. Le contrble legal en oenologie. Bull. ofice intern. uin 18(167-170),
54-74.
Lakkopoulos, A. A. 1939. The determination of citric acid in musts by the method of
electrical conductivity (transl.). Prakt. Akad. Athenon 14, 547-553.
Lambert, M., and Neish, A. C. 1950. Rapid method for estimation of glycerol in fer-
mentation solutions. Can. J . Research 28B, 83-89.
Lane, R. L., Johnson, E., and Williams, R. R. 1942. Studies of the average American
diet. I. Thiamine content. J . Nutrition 23, 613-624.
Lang, W. 1951. tfber die quantitative Bestimmung von Gerbstoffen. Pharmazie,
Berlin, 6, 137-140.
Larsen, H. J. 1951. Vereinfachte Siiure -Bestimmung. Weinbau. Wissenschajtl. Beih.
6, 381.
Lavollay, J., and Sevestre, J. 1944. Le vin consid6r6 comme un aliment riche en
vitamine P. Compt. rend. acad. agr. France 30, 259-261; see also Bull. ofice intern.
vin 17(163-166), 82-84 (1944).
Leggieri, C. L. 1951. Nuovo sistema di dosaggio dell’acidith volatile del vino con
diretta esculsione dell’anidride solforosa. Controllo del limite consentito d i Sot.
Chimica (Milan) 6 , 157-159.
Legkov, P. 1931. Die colorimetrische Bestimmung des Glyzerins. Odessa. !?sentral’na
naukovo-doslidchol vinogrado-vinorotcha slant&, Pratsi 4, 88-91.
Lehmann, E. 1940. Zur Zuckerbestimmung im Wein. 2. Untersuch. Lebensm. 79, 270.
Lehmann, E. 1943. Die Geschmacksstoffe des Weines. Schweiz. Wein-Ztg. 61, 89-90.
Leonoel, C. P. 1946. Microdosage de l’alcool mbthylique dans les milieux alcooliques
complexes. Bull. SOC. chim. biol. 28, 868-873.
Lesnovskaya, V. V., and Vecher, A. S. 1939. The content and preservation of vitamin
C in fruit-berry wines (transl.). Nauch. -Isoledovatel. Inst. Pishchevo’l Prom.,
Sbornik Molodykk Nauch. Rabotnikov 1939, 35-47; C . A . 36, 4280 (1942).
Levy, L. P., Posternack, T., and Robinson, R. 1931. Experiments on the systhesis of
the anthocyanins. VIII. A systhesis of oenin chloride. J . Chem. SOC.1931, 2701-
2715; see also ibid. 2715-2722).
Lherme, G. 1933. Les 6bullioscopes et le degr6 des vins blancs. Ann. fals. et fraudes
26, 546-552.
Lobstein, E., and Schmidt. 1931. Le vignoble Alsacih et ses vins. Ann. fals. el fraudes
24, 220-229.
Lodi, M. 1943. Untersuchungen tiber den Gehalt von Carotin, Vitamin BI, Bz,und C
bei der alkoholischen Giirung der Weinbeeren. Vitamine u . Hormone 4, 443-455.
Lucchetti, E. 1941. Un’inchiesta sui vini de Montecarlo (Lucca). Ann. facoltcl agrar.
univ. Pisa (N.S.) 4, 216-230.
Luckow, C. 1931. Nachprtifung von Alkoholometern. Wein u. Rebe 13, 267-269.
Luckow, C. 1935. Die Bestimmung des Alkohol- und Extraktgehaltes mit Hilfe des
Destillierapparates und der Spindel. Wein u. Rebe 17, 224-227.
Luers, H. 1948. Formation of flavour substances in fermentation. Brewers’ Digest 63,
45-59.
Ltithi, H. 1949. Zur Zitronensiiurebehandlung der Weine. Schweiz. 2. 0bst.-und
Weinbau 68, 315-319.
Luthi, H., and Vetsch, U. 1952. Papierchromatische Bestimmung von Aminosauren
in Weinen. Schweiz. 2.0bst.-und Weinbau 61, 390-394, 405-408.
Maaskant, L. 1936. Quantitative Bestimmung von Furfurol und Oxymethylfurfurol
mit p-Nitrophenylhydrazin. Rec. trau. chim. 66, 1068-1070.
Mackinney, G., and Chichester, C. 0. 1954. The color problem in foods. Advances in
Food Research, 6, 301-351 (1954).
Mader, 1936. Der Einfluss der Veredlung auf die chemische Zusammensetzung des
Mostes und des Weines. Wein u. Rebe 17, 250-258.
Mallory, G . E., and Love, R. F. 1945. Determination of caramel in wine, dis-
tilled spirits, vinegar, and vanilla extract. Znd. Eng. Chem., Anal. Ed. 17,
631-637.
Malvezin, P. 1931. Les methodes d’analyse et l’expertise 16gale. Bull. ofice intern. uin
4(33), 86-90.
Malvezin, P. 1934. A propos du dosage de l’extrait sec des vins. Ann. fals. et fraudes
27, 290-291.
Manskaya, S. M. 1939. Aging of wine (transl.). Priroda l989(9), 74-80; C. A . 86,3989
(1942).
Manskaya, S. M., and Emel’yanova, M. P. 1939. Oxidation processes in wine (transl.).
Biokhimiya 4, 581-592.
Maravalhas, N. 1935. Novo processo rapido para a determinaqb das substancias
colorantes artificiaes nos vinhos. Rev. chim. ind. (Rio de Janeiro) 4, 480.
Marcel, M., and Bastisse, E.-M. 1942. Sur la technique de determination du titre
alcoolique des vins, cidres, bihres, etc. Compt. rend. acad. agr. France 28, 382-383.
Marcilla Arrazola, J. 1934. Sobre la caracterizaci6n de vinos por el an&lisis qufmico.
C O T L ~intern.
T. quim. pura apl. 9(VI), 352-366.
Marcilla Arrazola, J., Alas, G., and Feduchy Marifio, E. 1936. Contribucih a1 estudio
de la levaduras que forman velo sobre ciertos vinos de elevado grado alcoholico.
Anales Centro Invest. Vinicolas 1, 1-230.
Marcilla Arrazola, J., and Feduchy Marifio, E. 1943. Una mezcla de indicadores de
pH, especialmente adecuada para la industria enol6gica. Inst. nacl. invest. agron.
Cuaderno (Spain), 24, 37-59.
Marcille, R. 1933. Observations sur les nouvelles mkthodes officielles de dosage des
acidites dans lea vins. Ann. fals. et fraudes 26, 286-292; see also Bull. ofice intern.
vzn 6(67), 39-44 (1933).
Marcille, R. 1934. Dosage de l’acidit6 volatile des vim sulfit6s. Ann. fals. et fraudes
27, 348-351.
Marcille, R. 1935. Dosage de l’anhydride sulfureux libre dans lea vins. Ann. falsif. el
fraudes 28, 93-96; see also Bull. ofice intern. vin 8(84), 49-52 (1935).
Marcille, R. 1937. Observations sur les mkthodes d’analyse des vins ayant fait l’objet
de la convention internationale de Rome. Ann. fals. et fraudes 30, 299-304.
Mariani, A. 1950. La determinazione di piccole quantith di alcole metilico nell etilico.
Rend. ist. super. SanitB 13, 154-166.
Marignan, R. 1944. Le Dosage de 1’Acide Succinique dans les Vins. Imprimerie A.
Quillet, Montpellier.
Markley, K. S., Sando, C. E., and Hendricks, S. B. 1938. Petroleum ether-soluble and
ether-soluble constituents of grape pomace. J . Biol. Chem. 123, 641-654.
Marsh, G. L. 1936. The peculiarities of carbon dioxide in wine. Wine Rev. 4, 17-18.
Marsh, G. L., and Joslyn, M. A. 1935. Precipitation rate of cream of tartar from wine.
Ind. Eng. Chem. 27, 1252-1257.
Marsh, G., and Kean, C. 1951. Applications of chromatographic methods to organic
acids in wine. Proc. Am. SOC.Enologists 1961, 157-159; see also Wines & Vines
33(3), 19-20 (1952).
Martucci, J. 1941. La manita en 10s vinos. Vinos, ViAas y Prutas 37, 165-166.
Mastbaum, H. 1933. Uber den Nachweis von Caramel in Sussweinen. Chem.-Ztg. 67,
959-960, see also 2. Untersuch. Lebensm. 66, 254-258 (1933).
Mathers, A. P. 1949. Detection of tartaric acid and tartrates in wine. J. Assoc. Oflc.
Agr. Chemists 32, 417-421.
Mathers, A. P. 1951. Report on chromatographic studies of wines and distilled spirits.
J . Assoc. Oflc. Agr. Chemists 34, 307-309.
Mathers, A. P., Beck, J. E., and Schoeneman, R. L. 1951. Polarographic determination
of tartrates in wines. Anal. Chem. 23, 1767-1770.
Mathers, A. P., and Schoeneman, R. L. 1952. Polarographic determination of benzalde-
hyde in wine. J . Assoc. Ofic. Agr. Chemists 36, 830-843.
Mathieu, G. 1938. Observations sur la fermentation alcoolique h. Chgteauneuf-du-
Pape. Compt. rend. acrcd. agr. France 24, 592-596.
Matignon, C., Moureu, H., and Dod6, M. 1934. Sur le dosage du butanediol-2,3. Bull.
SOC. chim. France 151 1, 411-419.
Mauri6, A. 1942. Lea compos6s phholiques du vin. (Mati6res colorantes et tannofdes).
McCharles, 0. H., and Pitman, G. A. 1936. Methods of wine analysis. Ind. Eng.
Chem., Anal. Ed. 8, 55-56.
Mehlitz, A., and Scheuer , M. 1934. Ober enzymatische Klarung von Fruchtsaften und
Siissmosten. Biochem. 2. 268, 345-354; see also ibid. 276, 66-85, 86-90 (1935).
Melcher, B. 1947. Zur Bestimmung der Gesamtsaure. Mitt. Gebiete Lebensm. u. Hyg.
38, 299-301.
Merzhanfan, A. A. 1930. Uber den Vitamgehalt von Trauben und Traubenweinen
Wein u. Rebe 11, 404-408; see also ibid. 12, 67-76 (1930).
Merzhan&n, A. A. 1951. On the behavior of diethyl esters of pyrocarbonic acid in
sparkling and gassified wines (transl.). Vinodelie i Vinogradarstvo S.S.S.I?.
11(3), 19-22.
Merzhanfan, A. A. 1952. Regarding some remarks on the theory of champagnization
(transl.). Vinodelie i Vinogradarstvo S.S.S. R. l2(2), 23-25.
Mestre Artigas, C., and Campllonch Romeu, I. 1935. Determinaci6n de la acidez
volatil real en 10s vinos. Bol. inst. nacl. invest. agron. Spain 1, 181-196.
Mestre Artigas, C., and Campllonch Romeu, I. 1942. La producci6n de aldehidos en
la fermentaci6n de mostos sulfitados y su influencia en 10s vinos. Bol. inst. nacl.
invest. agron. Spain 6, 1-16; see also Bull. oflce intern. vin 16 (155), 82 (1943).
Mestre Artigas, C., and Garcia Barcel6, J. 1947. Perfeccionamiento de 10s m6todos
electroqulmicos aplicables a 10s an&lisisde 10s vinos. Bol. inst. nacl. invest. agron.
Spain 16, 101-124.
Mestre Artigas, C., and Mestre JanE, A. 1935. Contribuci6n a1 estudio de la evoluci6n
de las substancias nitrogenadas del mosto en la fermentaci6n. Bol. inst. nacl.
invest. agron. Spain 1(2), 9-32.
Mestre Artigas, C., and Mestre Jan6, A. 1939. Estudio comparativo de fermentaciones
de mostos a moderadas y bajas temperaturas. Bol. inst. nacl. invest. agron. Spain
2(3), 31-43.
Metra, M., Lesage, L., and Descatoire, F. 1938. L'isopropanol. Sa recherche dans lea
alcools. Ann. fals. et fraudes 31, 218-221.
Meyerhof, 0. 1952. Recent advances in the study of metabolic reactions of yeast
preparations. Am. Scientist 40, 482-492, 517.
Mezzadroli, C., Amati, A., and Sgargi, L. 1931. Influenza della razza del fermcnto
alcoolico sulla qualitb e sul bouquet del vino. Giorn. biol. appl. ind. agr. 1, 161-
171.
Mical, L. 1949. La difficult6 d’ctablir une analyse precise du vin. Rev. belge des vins et
spiritueus 60, 209-211.
Michel, A. 1931. Dosage de l’acide lactique dans les vins. Ann. fals. et fraudes 24,
471-474.
Michel, A. l948a. Considerations sur le dccret du 28 juin 1938 fixant le taux maximum
de l’acidit6 volatile des vins propres S la consommation. Ann. fals. et fraudes 41,
48-52.
Michel, A. 194813. La reaction de Fiehe applique6 b la caracterisation de jus dc raisin
chauffcs (pasteurises, desulfites et concentres). Ann. f a k . et fraudes 41, 208-21 1.
.
Miconi, C. 1948. La . . dificile detcrminazione dell’acidith totale dei mosti e dci
vini. Riv. viticolt. e enol. (Conegliano) 1, 123-124.
Miconi, C. 1952a. Contenuto in zucchero del mosto ed alcool probabile. Riv. vitivolt.
e enol. (Conegliano) 6, 336-339.
Miconi, C. 1952b. La determinasione dell’acidith volatile dei vini. Riv. viticolt. e enol.
Miermeistcr, A., and Battay, F. 1931. Die Verfalschugen von Sussweinen und ihr
Nachweis durch Bestimmung der niederen Fettsiiuren (Buttersaure). 2. Unter-
such. Lebensm. 61, 161-171.
Milos, C. 1942. Detection of caramel in wines. Am. J . Pharm. 114, 138-140.
Minz, S., and Sirianni, E. 1934. Le vitamine dell uva e del vino. Riv. sintetica probl.
alimen. 4, 35-66.
Mohler, H. 1937. ubcr den Zitronensauregehalt des Weines. Mitt. Gebiete Lebensm. u.
Hug. 27, 27-40.
Mohler, H., and Hammerle, W. 1935. Chromatographischer und spektrophoto-
metrischcr Nac-hweis con kiinstlicher Fiirbung in Wein. 2. Untersuch. Lebensm.
70, 193-195.
Mohler, H., and I-Iammerle, W. 1936. Uber den Nachweis von Weisswein in Rotwein
2. Untersuch. Lebensm. 7 , 186-189.
Montequi, F. 1933. El color del vino. Anales soc. espafi. f;s. quim. 31, 663-668.
Morani, V. 1930. La. ricerca delle aggiunte di acidi minerali nel vino mediante il
potenziometro. Ann. chim. appl. 20, 30-48; see also Staz. chim.-agr. sper. Rome,
Pub. 268, 1-29 (1930).
Morani, V., and Marimpietri, L. 1930. Ulteriore studi sul potere tampone dei vini:
conservazione e gessatura. Ann. chim. appl. 20, 313-335; see also Staz. chim.-agr.
sper. Rome, Pub. 274, 1-23 (1930).
Moreau, L., and Vinet, E. 1937. Le tanin dans les vins blanc. Rev. viticult. 86, 65-69.
Moreno Martin, F. 1934. Influencia del alrohol metilico en la dosificaci6n de alcoholes
superiores en agardientes y licores. Congr. intern. quim. pura apl. 9(VI), 272-273.
Morgan, A. F. 1941. A nutritivp index of fruits. Fruit Products J. 21, 75-77.
Morgan, A.F., Field, L. K., and Nichols, P. F. 1935. The vitamin content of Sultanina
(Thompson Seedless) grapes and raisins. J. Nutrition 9, 369-382.
Morgan, A . F., Nobles, H. L., Wiens, A., Marsh, C. L., and Winkler, A. J. 1939. The
B vitamins of California grape juices and wines. Food Research 4,217-219.
Moureu, H.,and Dod6, M. 1934. Recherche et dosage du butylCneglycol2-3 dans les
jus de fermcntation. B d l . assoc. chim., sucr., dist. 61, 247-250.
Mursaeva, A. M., and BrailovskaG, A. E. 1952. Crystallization of atrtrates in grape
juices (transl.). Vinodelie i Vinogradarstvo S.S.S. R . 12(3), 21-23.
Muth, F. 1933. Vber die Bindung von Jod in ungeschwefelten Mosten und Weinen.
Bericht Lehr- u. Forschungsanstalt Wein-, 0bst.- u. Gartenbau, Geisenheim 1933,3-4.
Muth, F. 1934. Mostgewicht und Alkoholgehalt der Weine. Wein u. Rebe 16,239-244.
Muth, F., and Malsch, L. 1934. Versuche zur Aufstellung einer Stickstoffbilanz in
Traubemosten und -weinen. 2. Untersuch. Lebensm. 68, 487-500.
Muttelet, C. F. 1930. Les sucres des vins de Porto. Ann. fals. et fraudes 23,
205-207.
Nedeltscheff, N., and Kondareff, M. 1941-42. Untersuchung von Mavrudweinen und
Rotweinstocken. Ann. Fac. &on. Univ. Sofia 20, 41-49.
NBgre, E. 1939a. Contribution oenologique h 1’6tude des matiBres tannoides. Compt.
rend. acad. agr. France 26, 647-652.
NBgre, E. 193913. Dosage de l’oenotannin. Ann. jals. et fraudes 32, 175-178.
NBgre, E. 1942-1943. Les matiBres tannoides et la composition des vim. Bull. ofice
intern. vin 16(154), 20-52; 16(155), 25-56.
NBgre, E., Affre, J. P., and Marichal, M. 1947. Pour une am6lioration de l’industrie
du jus de raisin. Bull. Ofice intern. vin 20(195), 30-67; (196), 52-101.
Neish, A. C. 1950. Analytical methods for bacterial fermentations. Report No. 46-8-3,
rev., National Research Council of Canada, Praire Regional Lab., Saskatoon.
Neish, A. C. 1951. Analysis of mixtures of a simple aliphatic alcohols by partition
chromatopgraphy. Can. J . Chem. 29, 552-557.
Nelson, E. K. 1937. Flavor of alcoholic beverages. Food Research 2, 221-226.
Nelson, E. K., and Wheeler, D. H. 1939. Natural aging of wine. Ind. Eng. Chem. 81,
1279-1281.
Neubauer, A. 1911. Verfahren zur Aromatisierung von Beerensaften und Weinen.
0bst.-u. Cemuseverwert.-Ind. 28, 373-375.
Newton, W., and Munro, F. L. 1933. The determination of alcohol and extract in
wine by means of the density and refractive index. Can. Chem. Met. 17, 119-120.
Niehaus, C. J. 1934. The determination of alcohol by means of the ebulliometer. S.
African Wine Ann. (Special Ed. Wine and Spirit) 4, 1693-1697.
Niehaus, C. J. 1937. Sugar-alcohol ratio in South African musts and wines. Xci. Bull.
Dept. Agr. S. Africa 161, 1-11.
Niehaus, C. J. 1938. The nitrogen content of South African musts and wines. Sci. Bull.
Dept. Agr. S. Africa 172, 1-15.
NitschkB, E. 1952. MBthode de dosage colorimBtrique de l’acide malique dans les vins
et les modts. Mitt. Gebiete Lebensm. u. Hyg. 43, 50-57.
Noguero G6mez, E. 1942. Analisis de vinos. MBtodos empleados en el Laboratorio de
Bromatalogfa del Ministerio de sanidad y asistencia social. Rev. sanidid y asis-
tencia social (Venezuela) 7 , 573-590.
Oliveira, A. J . de 1941. Estudo estatfstico de algumas caracterfsticas q u b c a s do
vinho do Parto. Anais inst. vinho Pdrto 2, 389-442.
Oliveira, F. P. de 1942. Subsfdios para o estudo das substhncias azotadas nos vinhos
do PGrto. Anais inst. vinho Pdrto 3(1), 107-156.
Onokhova, N. P. 1937. Grapes as a source of vitamin C. Bull. A p p l . Bot., Genet., Plant
Breeding U.S.S.R. Suppl. 84(II), 195-200.
Oparin, A. I., and Bezinger, E. N. 1949. The nitrogen constituents of wine. I. Nitrogen
compounds of high molecular weight in champagne (transl.). Biokhimiya 14,
291-301.
Oparin, A. I., Besinger, E., and Batsyna, I. 1945. Transformation of nitrogenous sub-
stances in champagne during processing (transl.). Biokhimiya 10, 311-325.
Oparin, A. I., Kursanov, A. L., Saenko, N. F., and Bezinger, E. N. 1946. Biochemical
processes in champagne during the secondary fermentation period (transl.).
Vinodelie i Vinogradarstvo S.S.S.R. 6(5),12-19; C. A . 40, 1630 (1946).
Oparin, A. I., Kursanov, A. L., Saenko, N. F., and Bezinger, E. N. 1947.Biochemical
processes taking place in champagne during after-bottling maturation (transl.).
BiokhimiG vinodeliz’a 1, 134-157.
Oparin, A. I., and Manskaya, S. M. 1939.Oxidation in the fermentation of tea and the
aging of wine (transl.). Sbornik Akad. Nauk S.S.S.R. Prezidentu Akad. Nauk
S.S.S.R. Akad. V . L. Komarovu 1939, 588-600.
Osborn, G. H., and Mott, 0.E. 1952.The determination of higher alcohols in whiskey
and other potable spirits. Analyst 77, 260-262.
Osterwalder, A. 1945. Weitere Beitrage zur Kenntnis des Braunwerdens der Weine.
Ann. agr. Suisse 69,573-605.
Osterwalder, A. 1952. Uber die durch Bakterien verursachte Zersetzung von Wein-
saure und Glyzerin im Wein. Ann. agr. Suisse 66, 181-197.
Palieri, M.G. 1950. Aciditti volatile e distillato alcoolico. Riv. viticolt. e enol. (Cone-
gliano) 3, 354-356.
Palieri, M.G. 1951. Nuovo metodo per la determinazione rapida degli zuccheri nei
vini. L’ltalia Agricola 88, 781-783.
l’alieri, M. G. 1952.La determinazione dell’aciditti volatile con i metodi per distilla-
zione parziale a fiamma diretta. Riv. vitivolt. e enol. (Conegliano) 6 , 392-399.
Parfent’ev, L. N.,and Kovalenko, V. I. 1951. On the potential role of pyrocarbonate
esters in the formation of champagne characteristics in sparkling wines (transl.).
Vinodelie i Vinogradarstvo S.S.S.R. 11(3), 16-19.
Parfent’ev, L. N., and Kovalenko, V. I. 1952. On the question regarding the theory of
champagnization (transl.). Vinodelie i Vinogradarstvo S.S.S. R . 12(4), 28-29.
Parisi, E.,and Della Barba, L. 1931. Acido uronico, glucosio e galattosio, componenti
di alcune sostanze otticamente attive contenute nei vini completamente fer-
mentati. Giorn. biol. appl. ind. chim. 1, 95-102.
Parro, A. da C. 1948. Brhve esquisse d’un plan d’btudes sur la richesse en vitamines des
mohts et des vins du Portugal. Bull. ofice intern. vin 21(204), 83-88.
Paronetto, L. 1938. La determinazione dell’alcole nei vini liquorosi e vermut per
ossidazione cromica. Ann. chim. appl. 28, 164-169.
Paronetto, L. 1948. Alcune prove di vinificazione in bianco con un nuovo prodotto
enzimatico. Riv. viticolt. e enol. (Conegliano) 1, 377-380.
Paronetto, L. 1950. La determinazione ebulliometrica nei vini dolci. Riv. viticolt. e
enol. (Conegliano) 3, 223-228, 254-258.
Pato, M. d. S. 1932. Qufmica-fisica aplicada aos mostos e aos vinhos. Revista agronb
mica (see Estacao Viti-vintcola da Beira Litoral, SBr. A, Bol. 1, 1-59. 1932).
Pato, M.d. S. 1935. Considerations sur la correction de l’acidit6 des moQts au point de
vue technologique et au point de vue Bconomique. CongrPs International de la
Vigne et du Vin, Lausanne.
Pato, M. d. S. 1938. Tabelas para a determina@o do extract0 s&codos vinhos portu-
gueses, por densimetria. Direcqlo Geral Serviqos Agricolas, Repartipgo Estud.,
Infor. e Prop., Lisbo4.
Pato, M. d. S. 1944. Importance d’un dosage rigoureux de l’acide tartrique. Bull.
ofice intern. uin 17(161),57-74.
Pato, M. d. S., and Salvador, A. R. N. 1949. Mbtodo para a titulaggo dos 4cidos
vol4teis dos vinhos com deduptio do 4cido carb6nico. Anais junta nacional vinho
1, 107-122;see also Bull. ofice intern. vin 21(204), 70-82 (1948).
Pato, M. d. S., and Sousa, T. T. de 1938. Da influhncia do anidrido sulfuroso na
correc@o Bcida dos mostos. 5th Congr. intern. vinha e vinho, Lisboa; see also
ibid. rapports 11, 94 (1938).
Patterson, W. E., and Bawtenheimer, J. W. 1930. Volatility of magnesium acetatein
wines. Can. Chem. Met. 14, 287-288.
Pavelka, F., and Montini, L. 1948. Eine Bestimmung der fliichtigen Siiuren in Wein
mittels der Apparatur von Pregl-Parnas. Milcrochemie 33, 333337,
Pavolv-Grishin, S. I. 1940. Determination of glycerol in grape wines (transl.). Vino-
delie i Vinogradarstvo S.S.S.R. 1(5), 9-10; C. A. 37, 558 (1943).
Penniman, W. B. D., Smith, D. C., and Lawshe, E. I. 1937. Determination of higher
alcohols in distilled liquors. Ind. Eng. Chem., Anal. Ed. 9, 91-95.
Perard, J. 1939. Le bilan de la fermentation alcoolique est-il a reviser. Bull. assoc.
chim., sucr., dist. 66, 251-252.
Perard, J. 1940. A propos du bilan de Pasteur. Bull. assos. chim., sucr., dist. 87, 250-
251.
Perarso, A. A,, and Arbecchi, A. C. 1933. Acides volBtil de 10s vinos que contienen
anhidride sulfuroso. Anal. asoc. quim. argentina 21, 156-158.
Percher, G. 1938. L’extrait sec des vins. Ann. agron. 1, 63-66; see also Ann. fals. et
fraudes 31, 468-472 (1938).
Perdigon, E. 1941. Semi-microdosage du 2,3-butyl&ne glycol dans les liquides de
fermentation par oxydation periodique. Ann. ferment. 6, 143-158.
PeretiE, M. 1950. Les vins de la Croatie Nord, vendange 1948. Biljne proizvodnje 4.
Perlman, L., and Morgan, A. F. 1945. Stability of B vitamins in grape juices and
wines. Food Research 10, 334-341.
Petri, W. 1932. Neuere Forschungsergebnisse aus der Weinforschungsanstalt fur
Mosel, Saar und Ruwer. 2. Untersuch. Lebensm. 64, 177-205.
Pettigiani, A. E. 1943. Los alcoholes superiores en 10s vinos Argentinos. Rev. fac.
cienc. qu‘im. univ. nacl. La Plata 18, 95-104.
Peynaud, E. 1936a. L’acetate d’dthyle dans les vins atteints d’acescence. Ann.
ferment. 2, 367-384.
Peynaud, E. 1936b. Le dosage de l’acide tartrique dans les moats et les vins par les
mdthodes au racemate. Ann. fals. et jraudes 29, 260-273.
Peynaud, E. 1937a. Le dosage de l’acidite volatile des vins sulfites. Ann. fals. el
fraudes 30, 225-231.
Peynaud, E. 1937b. Etudes sur les phenomhes d’esterification. Rev. viticult. 86, 209-
215,227-231,248-253, 299-301, 394-396, 420-423, 440-444,472-475; 87, 49-52,
113-116, 185-188, 242-249, 278-296, 297-301, 344-350, 362-364, 383-385; for a
summary see Ann. ferment. 3, 242-252 (1937).
Peynaud, E. 19370. Methode chimique simple pour la determination du pH des vins.
Peynaud, E. 1938a. L’acide citrique dans les modts e t les vins de Bordeaux. Bull.
ofice intern. vin 11(118), 33-41.
Peynaud, E. 193813. L’acide malique dans les modts et les vins de Bordeaux. Ann.
fals. et fraudes 31, 332-347; see also Rev. viticult. 90, 3-12, 25-30 (1939).
Peynaud, E. 1938c. Comportement des acides gras volatils pendant la fermentation
alcoolique. Rev. viticult. 88, 88-90.
Peynaud, E. 1938d. Le dosage de l’ac6tate d’bthyle dans les vins. Ann. fals. et fraudes
31, 158-162.
Peynaud, E. 1939a. Acetate d’6thyle et acescence. Rev. viticult. 90, 321-327.
Peynaud, E. 193913. L’arote aminb et l’arote amid6 dans les vins de Bordeaux. Ann.
fals. et fraudes 32, 228-243; see also Bull. SOC. chim. France [5] 6, 784-785 (1939).
Peynaud, E. 1939c. Sur les variations de l’azote du raisin au cours de la maturation.

Rev. viticult. 90, 189-195, 213-225.
Peynaud, E. 1939-1940. Sur la formation et la diminution des acides volatils
pendant la fermentation alcoolique en anakrobiosc. Ann. ferment. 6, 321-337,
385-401.
Peynaud, E. 1946. Etude des mkthodes de dosage des constituants de l’aciditk des vins.
Chim. anal. [4] 28, 111-117, 127-135.
Peynaud, E. 1947a. L’acktylm6thylcarbinol e t le 2,3-butyl&n6glycol dans les milieux
de fermentation alcoolique. Rev. ferment. ind. aliment. 2, 150-158, 189-196.
Peynaud, E. 194710. Applications de l’oxydation pkriodique 8, l’analyse des vins
dosages du glyckrol, du 2,3-butylAn6glycol des sucres r6siduaires des vins rouges,
de l’acide tartrique. Bull. ofice intern. vin 20(196), 38-51.
Peynaud, E. 1947c. Contribution 8, 1’6tude biochimique de la maturation du raisin et
de la composition des vins. Inds. agr. et aliment. (Paris)64,87-95, 167-188,301-
317, 399-414; also printed by Imprimerie G. Sautai et Fils, Lille (1948), and sum-
marized in Rev. viticult. 92, 177-180, 271-272 (1946) and Bull. ofice intern. vin
20(191), 34-51 (1947).
Peynaud, E. 1948a. Bilans des produits secondaires de la fermentation alcoolique.
Inds. agr. el aliment. (Paris) 66, 99-108.
Peynaud, E. 1948b. Dosage du glycerol dans les vins par oxy8ation periodique. Ann.
fals. el fraudes 41, 384-402.
Peynaud, E. 1950a. Analyses complbtes de douze vins de Bordeaux. Ann. agron. Ser.
A, 1, 252-266.
Peynaud, E. 1950b. Analyses complbtes de huit vins doux naturels. Ann. agron. Ser.
A, 1, 382-388.
Peynaud, E. 1951a. Note sur le dosage de l’acide d-tartrique par la m6thode au
rac6mate 8, l’acide d’un extrait de Bauhimia reticulata D. C. Bull. SOC. chim.
France [5] 18, 911-912.
Peynaud, E. 1951b. Sur les matibres pectiques des fruits. Inds. agr. et aliment. (Paris)
68, 609-615.
Peynaud, E. 1952. Sur les matihres pectiques des modts de raisin et des vins. Ann.
Peynaud, E., and Charpentik, Y. 1950. Note sur le dosage de l’acide lactique dans les
boissons fermentkes. Ann. fals. et fraudes 45, 246-252.
Peynaud, E., and Charpentik, Y. 1953. Dosage de l’acide gluconique dans les modts et
les vins provenant de raisins attaques par le Botrytis cinerea. Ann. fals. et fraudes
46, 14-21.
Peynaud, E., and Lafon, M. 1951a. Note complkmentaire sur les corps acktoiniques
des eaux-de-vie. Ann. fals. et fraudes 44, 399-402.
Peynaud, E., and Lafon, M. 1951b. Presence et signification du diacbtyle, de l’ac6toYne
et du 2,3-butanediol dans les eaux-de vie. Ann. fals. et fraudes 44, 264-283.
Peynaud, E., and Lafon, M. 1952. Bilans des produits secondaires de la fermentation
alcoolique pour divers genres de levures et dans des conditions diffbrentes d’akra-
tion. Inds. agr. et aliment. (Paris) 69, 397-408.
Peynaud, E., and Ribkreau-Gayon, J. 1947. Sur les divers types de fermentation
alcoolique ddterminEs par diverses races de levures elliptiques. Compt. rend. 224,
1388-1390.
Peyrot, E. 1934. Sulla probabile quantith e qualitti delle principali sostanze coloranti
nei vini bianchi. Studio comparativo per curve di assorbimento. Ann. chim. appl.
24, 512-519.
Pfab, H. 1931. Die Bestimmung der Zitronensiiure im Wein. Dissertation, Tech-

nischen Hochschule, Dresden.
Piatkowska, K., and Smreczyxiska, A. 1950.The citric acid content of native fruit and
berry wines, and the influence of addition of these on the citric acid content of
grape wines (transl.). Roczniki Pailstwowego Zakladu Hig. 1, 345-353.
Piazza, J., and Rouzaut, R. 1939.Determinacidn rtipida de graduciones alcoh6licas en
vinos y cervezas con destilador a espiral. Anales inst. invest. cient. y tecnol., univ.
nacl. litoral (Santa Fb, Argentina) 9, 82-86.
Piccoli, T. 1933.Recherche de l’alcool methylique dans les vinaigres de vin adulteres
au moyen de vinaigres l’alcool et produits provenant de la distillation pyrogenee
(transl.). Atti 4th congr. naz. chim. pura ed appl., Rome, 1933, 773-775.
Picozzi, A. 1947. Method for determining volatile acidity in wines. Proc. 11th Conf.
Pure and Appl. Chem., London 1947 3,207-209.
Pluchon, J. P. 1937. Modification ti la methode de defecation des vins. Ann. fals. et
fraudes 30, 344-350.
Podkletnov, N. E. 1951. Determination of small quantities of tartaric acid (in musts
and wine) (transl.). Vinodelie i Vinogradarstvo S.S.B.R. 11(10), 19.
Politova-Sovzenko, T. K. 1947. ttber die Gerbstoffe des Weinmostes (transl.). Vino-
delie i Vinogradarstvo S.S.S. R . 7(12), 18-20.
Politova-Sovzenko, T.K.,and Dikhtyar, P. 0. 1948. Determination of volatile acid
in wines having a high sulfur dioxide content (transl.). Vinodelie i Vinogradarstvo
S.S.S.R. a@), 32-33; C. A . 44,9620 (1950).
Ponte, A. and Gualdi, G. 1931. Tannin in wines and its estimation. J . Intern. SOC.
Leather Trades Chem. 16, 151;for original see Boll. ugiciale staz. sper. ind. pelli
mat. concianti 1931,391-407.
Pool, A., and Heitz, J. E. 1950. Correlation of fortifying brandy with wine quality.
Proc. Am. SOC.Enologists 1960, 101-105.
Popov, A. D. 1947-1948. Detection of elderberry pigments and some other foreign
natural and synthetic coal-tar dyes in red wine (transl.). Ann. univ. Sofia, facult.
S C ~ .44,31-110.
Popova, E. M. 1948. Esterases of molds and their influence upon the aging of wine
(transl.). BiokhimiG vinodeli& 2, 115-125.
Popova, E. M., and Puehkova, M. G. 1948.Microbiological determination of nicotinic
acid in wines (transl.). Biokhimia vinodelia 2, 117-182.
Popova, E. M., and Puchkova, M. G. 1950.Role of esterase in the aging process of
champagne wines (transl.). Biokhimih Vinodelih 3, 69-84.
Poux, C. 1949. Sur le microdosage de l’acide tartrique dans les vins et dans les monts.
Poux, C., and Ournac, A. 1949.Oxydation des acides tartriques et formique par l’acide
periodique. Znds. agr. et aliment. (Paris) 66, 503-507.
Pozzi-Escot, E. 1938. Mi tecnica en determinaci6n de la acidez volatil de 10s vinos, hoy
metodo oficial en Francia; mi dispositivo experimental recomendado. Rev. cienic.
(Peru) 40(424), 299-302.
Pozzi-Escot, E. 1949. Bur une methode precise de dosage de l’alcool par distillation.
Inds. agr. et aliment. (Paris) 66, 119-121.
Prado, L. de. 1934. Nuevo micrometodo de dosage del glicerol en 10s bebidas fer-
mentadas. Anales farm. y bioqutm. 6, 98-106.
Prange, G. 1941. Mikroalkoholbestimmung in Obsterzeugnissen. Vorratspfige u.
Lebensmittelforsch. 4, 553-572.
Prfilat, C.E., and Mendivelzda, G. 1934.Ensayo de un m6todo rcfactometrico para
determinacibn de alcohol y extract0 en vino. Anales asoc. quim. argentina 22,
76-77.
Prillinger, F. 1952a. Gerbstoffe als Ursache von Triibungsercheinungen von Suss-
mosten. Mitt. hoheren Bundslehr- und Versuchsanstalt f u r Wein- und Obstbau
(Klosterneuberg) 2, 197-199.
Prillinger, F. 1952b. Uber die fluchtigen aromastoffe bei der Vergarung von Trauben-
most. Mitt. hbheren Bundeslehr- und Versuchsanstalt f u r Wein- und Obstbau
(Klosterneuberg) 2, 160-164, 188-191; see also Moser, L. 1952. Weinblatt 46,
220-222.
Prillinger, F. 1952c. Titrimetrische Methode zur Fruktosebestimmung. Mitt. hoheren
Bundeslehr- und Versuchsanstalt f u r Wein- und Obstbau (Klosterneuberg) 2, 20-23.
Pritzker, J. 1940a. Zur Bestimmung des Glyzerins in Sussweinen. Mitt. Gebiete Lebensm.
U . Hyg. 31, 223-229.
Pritzker, J. 1940b. Uber die Zusammensetzung und Beurteilung von Malaga und
Mistellen. Mitt. Gebiete Lebensm. u. Hyg. 31, 230-233.
Pritzker, J., and Jungkunz, R. 1930a. Entstehung, Vorkommen und Nachweis des
2,3-Butylenglykols in Wein und Obstwein. 2. Untersuch. Lebensm. 60, 484-489.
Pritzker, J., and Jungkunz, R. 1930b. Vber Vorkommen und Nachweis des 2,3-
Butylenglykols in Wein, Obstwein und anderen vergorenen Siften. Mitt. Gebiete
Lebensm. u. Hyg. 21, 236-243.
Pro, M. J. 1952. Report on spectrophotometric determination of tannin in wines and
whiskies. J . Assoc. Ofic. Agr. Chemists 36, 255-257.
Procopio, M. 1939. I1 grado ebulliometrico dei vini dolci. Ann. chim. appl. 29, 74-77.
Procopio, M. 1948a. Innesto proteico nei mosti zuccherini (taglio proteico). Riv.
uiticolt. e enol. (Conegliano) 1, 80.
Procopio, M. 1948b. Interpretazione del valore della densith nei vini dolci ai fini della
valutazione degle zuccheri. Riv. viticolt. e enol. (Conegliano) 1, 329-334.
Procopio, M. 1948c. Simplificazione e standardizzazione del metodo Kleiber-Wiegner-
Magasanik per la determinazione dell’acidith volatile nei vini. Riv. viticolt. e enol.
Procopio, M. 1949. Eteri del vino. Riv. uiticolt e enol. (Conegliano) 2, 157-160.
l’rocopio, M. 1950a. Determinazione diretta dell’acidith volatile dei vini contenenti
anidride solforosa. Riv. viticolt. e enol. (Conegliano) 3, 59-64.
I’rocopio, M. 1950b. Determinazione diretta e rapida della reale aciditii volatile dei
vini contenenti anidride solforosa e stima indiretta di questa ultima. Riu. viticolt.
Procopio, M. 1950c. Interpretazione della densith dei vini dolci a duplice finalith.
Riv. viticolt. e enol. (Conegliano) 3, 321-325.
.
Procopio, M. 1950d. Valori ebulliometrici inconsuetamente anomali . . Riu. viticolt.
Prosstosserdov, N. N., and Taranova, R. D. 1949. Hydroxymethylfurfural in grape
wines (transl.). Vinodelie i Vinogradarsluo S.S.S.R. 9(11), 43-44.
Ramos, M. da C., and Oliveira, H. T. de 1949. Estudo cromatogr6fico das mat6rias
tinturiais do vinho do PBrto, tendo como primeiro objectivo a pequisa da baga.
Anais inst. vinho PGrto 10(1), 11-24.
Ramos, M. da C., and Reis A. M. L. de C. 1945. Determinagh do 2,3-butilenoglicol
no vinho do PBrto. Anais inst. vinho Pdrto 6(2), 11-31.
Randoin, L. 1936. Vitamines, jus de raisin et vins. Bull. SOC. sn’ent. d’hyg. aliment.
ration d’homme 24, 18-40; see also Bull. ofice intern. vin 7(72), 62-74 (1932).
Randoin, L., and Gallot, S. 1941. Recherches experimentales sur la vitaminisation
artificielle d’aliments naturels de consommation courante. Bull. SOC. chim. biol.

23, 429-436.
Rankine, B. 1952. The determination of the alcohol content of wine by pycnometer.
Australian Brewing and Wine J . 70(6), 3-4, 6.
Ravaz, L. 1935. Les facteurs de la qualit6 et les pratiques culturales. Prog. agr. et vit.
104, 489-494.
Reich, P. 1950. Wein-Kompendium fiir den Artz. Wissenschaftliche Verl. Ges.,
Stuttgart. pp. 39-40.
Reichard, 0. 1934a. Bestimmung der Citronensiiure als Pentabromaceton und ihre
Anwendung auf Wein. 2. Untersuch. Lebensm. 68, 138-172.
Reichard, 0. 1934b. Uber die Bestimmung der Citronensaure in Wein. 2.anal. Chem.
99, 138-142; see also ibid. 99, 81-96.
Reichard, 0. 1936. Citronensiiure-Bestimmung und -Gehalt im Wein. 2. Untersuch.
Lebensm. 72, 50-63.
Reichard, 0. 1943. Die Bestimmung des Stickstoffgehaltes von Wein und Traubensaft
und ihr Gehalt, ein Beitrag zur Neubearbeitung der amtlichen Anweisung fur die
Untersuchung von Wein und Most. 2. Untersuch. Lebensm. 86, 164-169.
Reichard, 0. 1951. Die Sinnenpriifung (Organoleptik) von Wein und ihr diagnostischer
Wert. Deut. Wein-Ztg. 87, 340-342, 379-380, 402-404; see also Deut. Lebensm.-
Rundschau 47, 103-110 (1951).
Reid, V. W., and Truelove, R. K. 1952. The colorimetric determination of alcohols.
Analyst 77, 325-328.
Reis, A. M. L. de C. 1946. Esteres no vinho do PBrto. Anais inst. vinho PGrto 7 ,
69-93.
Remy, E. 1932. Zusammensetzung badischer Weissweine der Jahrgiinge 1930-1932.
Z . Untersuch. Lebensm. 64, 548-553.
Rentschler, H. 1949. Die biochemische Bestimmung der xpfelsaure. Mitt. Gebiete
Lebensm. u . Hyg. 39, 30-35.
Rentschler, H. 1950. c b e r das Braunwerden der 1950er Weine. Schweiz. Z . Obst- u .
Weinbau 69, 455-458.
Rentschler, H., and Hauser, F. 1950. Uber die Beeinflussung des Gerbstoffgehaltes
von Weinen durch Gelatin Schonung. Schweiz. 2.Obst- u. Weinbau 69, 406-408.
Rentschler, H., and Simmler, H. 1949. Ein einfaches Verfahren zur Bestimmung der
fllichtigen Siiure in Getriinken. Schweiz. 2. Obst- u. Weinbau 68, 367-370.
Requinyi, G., and S O ~ SI., 1935. The inverting effect of some Hungarian wine yeasts
(transl.). Magyar Ampelol. lhkonyv 9, 411-412.
Ribeiro, M. B. 1938. Subsfdios para o estuda da alcalinidade das cinzas nos vinhos do
PBrto. Anais. inst. super. agron., Univ., t6c. Lisboa 9, 253-326.
Ribeiro, M. B. 1946. Eliminaqa dos redutores nLo apdcares nos vinhos do PBrto.
Anais inst. vinho P6rto 7 , 51-68.
Rib6reau-Gayon, J. 1932. Sur les matihres albuminoides des vins blancs. Ann. falsif.
et fraudes 26, 518-524, 602-609; see also Ann. brass. et dist. 31, 44-48, 62-64,
78-80 (1933).
Rib6reau-Gayon, J. 1938a. Les acides du vin. Progr. agr. et vit. 110, 226.
Ribbreau-Gayon, J. 193813. La mesure du pH des vins. Rev. viticult. 88, 411-415,
434-440.
Ribereau-Gayon, J. 193%. Sur les phenom&nes de reduction dans les vins. Bull. SOC.
chim. France [5] 6 , 953-954.
Ribereau-Gayon, J. 1943. Les diastases en oenologie. Bull. ofice intern. vin 16(160),
43-53.
Ribbreau-Gayon, J. 1947. Trait6 d’oenologie. Librairie Polytechnique Ch. Bbranger,

Paris et Liege.
RibCeau-Gayon, J. 1949. Les m6thodes rbcentes d’analyse des vins et leur unification.
Ribbreau-Gayon, J., and Maurib, A. 1942. Les composbs phbnoliques du vin. (Matihres
colorantes et tannofdes). Bull. ofice intern. vin 16(150), 60-76.
Ribbreau-Gayon, J., and Peynaud, E. 1936. Estbrification chimique et biologique des
acides organiques du vin. Bull. SOC.chim. France [5] 3, 2325-2330.
Ribbreau-Gayon, J., and Peynaud, E. 1937a. L’acide lactique dans les vins de Bor-
deaux. Ann. fals. et fraudes 30, 339-344.
Ribereau-Gayon, J., and Peynaud, E. 1937b. Une inethode chimique simple pour la
determination du pH des vins. Bull. SOC. chim. France [5] 4, 1627-1628.
Ribbreau-Gayon, J., and Peynaud, E. 1938a. Bilan de la fermentation malolactique.
Ann. ferment. 4, 559-569.
Ribbreau-Gayon, J., and Peynaud, E. 1938b. La deacidification des vins par les
bactbries. Compt. rend. acad. agr. France 24, 600-605.
Ribbreau-Gayon, J., and Peynaud, E. 1938~.L’acide malique dans les moats et les
vins de Bordeaux. Bull. SOC. chim. France [5] 6, 1276-1277.
Ribereau-Gayon, J., and Peynaud, E. 1938d. Les teneurs en acide citrique des moQts
et des vins de Bordeaux. Bull. SOC. chim. France [5] 6, 226-227.
Ribbreau-Gayon, J., and Peynaud, E. 1946a. Analyse et contrble des vins. Compt.
rend. acad. agr. France 32, 528-530; see also Bull. ofice intern. vin 19(187),29-31
(1946).
RibBreau-Gayon, J., and Peynaud, E. 1946b. Sur la formation des acides acbtique,
lactique et citrique au cours de la fermentation alcoolique. Compt. rend. 222,
457-458; see also Bull. o&e intern. vin 19(183), 47-48 (1946).
Ribbreau-Gayon, J., and Peynaud, E. l947a. Analyse et Contrale des Vins. Librairie
Polytechnique Ch. Beranger, Paris et LiCge.
Ribereau-Gayon, J., and Peynaud, E. 194713. Le 2,3-butylhnbglycol dans les boissons
fermentbes. Bull. SOC. chim. France [5] 14, 894-896.
Ribereau-Gayon, J., and Peynaud, E. 1947c. Sur le dosage et la chimie de l’bthanal
et de ses combinaisons dans les vins et les alcools. Ann. Inst. Pasteur 73, 777-796.
RibCreau-Gayon, J., and Peynaud, E. 1952. Sur l’emploi en vinification de quelques
activeurs vitaminiques de la fermentation. Compt. rend. acad. agr. France 34,
444-448.
Ribbreau-Gayon, J., Peynaud, E., and Lafourcade, 5. 1952a. Formation d’ inhibiteurs
et d’activeurs de la fermentation alcoolique par diverses moisissures. Compt. rend.
a54,251-253.
Ribbreau-Gayon, J., Peynaud, E., and Lafourcade, S. 1952b. Sur la formation de
substances inhibitrices de la fermentation par Botrytis cinerea. Compt. rend. 234,
478-480.
Rippel, K. 1949. Der biologische Saureabbau im Wein. Arch. Mikrobiol. 14, 509-531;
see also Dmt. botan. Ges. 60, 108-117 (1943).
Roche, M. 19 18.Sur le dosage des aldbhydes. Inds. agr. et aliment. (Paris) 66,281-283.
Rocques, J. 1!)50. A propos de la determination du degre alcoolique des vins. Ann.
fals. et jraudes 43, 344-346.
Rodopulo, A. K. 1948. The oxydizing enzymes of musts and of wines (transl.).
Vinodelie i Vinogradarslvo S.S.S.R . 8, 8-9.
Rodopulo, A. K. 1950a. Oxidizing enzymes of champagne varieties of grapes and
musts (transl.). BiokhimiG vinodeliiâ3, 43-52.
Rodopulo, A. K. 1950b. Polyphenoloxidase system in grapes and juice (transl.).

Vinodelie i Vinogradarstvo S.S.S.R. 10(3), 35-37; C. A . 46, 6794i (1951).
Rodopulo, A. K. 1951. Oxidation of tartaric acid in wine in the presence of salts of
heavy metals (activation of oxygen by iron) (transl.). Zzuest. Akad. Nauk S.S.S.R.,
Ser. Biol. 1961(3), 115-128.
Rodopulo, A. K. 1952a. Di- and tricarboxylic acids in must and wine (transl.).
Vinodelie i Vinogradarstvo S.S.S. R . la(@,8-11.
Fbleff, H. 1948. Triibungserscheinungen in 1947er Wein. Deut. Weing-Ztg. 84, 263-
264; see also ibid. 86, 265-266 (1949).
Roleff, H. 1949. Ursache der Wiirmetriibung bei 1947er Weinen. Deut. Wein-Ztg. 86,
12-13; see also ibid. 80, 197-198 (lY50), and Chem. Zentr. 1960 I, 1162, 1920.
Roleff, H. 1952. Schnellmethode sum Nachweis und sur Schfitzung von Citronensiiure
in Weissweinen. Deut. Wein-Ztg. 88, 34-36; see also Corriere Vinicole 21(36), 1, 3
(1952).
Romano, E. 1951. Sul contenuto in 2,3-butilenglicol di alcuni vini italiani. Ann. sper.
agrar. (Rome) (N.S.) 6, 779-783.
Roos, L. 1930. Mesure de la couleur des vins rouges. Ann. jals. et jraudes 23,207-211;
see Progr. agr. et vit. 93, 374-377 (1930).
Rosenblatt, M., and Peluso, J. V. 1941. Determination of tannins by photocolorimeter.
J. Assoc. Ofic. Agr. Chemists 24, 170-181.
Rosenfeld, P. M. 1952. On some questions regarding the theory of champagnisation
(transl.). Vinodelie i Vinogradarstvo S.S.S. R . l 2 ( l ) , 34-36.
Rosenthaler, L., and Vegezsi, G. 1953. Zur Kenntnis der Komarowskyschen Reaktion.
Die Branntweinwirtschaft 76, 68-69; see also Mitt. Gebiete Lebensm. u. Hyg. 43,
370-388 (1952).
Ross, S. L. 1942. Detannating wines for pharmaceutical use. Am. J. Pharm. 114,
200-204.
Roubert, J. 1951. Sur l’estimation rapide des esters insolubles a l’eaux contenus dans
les alcools de rhcup6ration des gas de fermentation. Vignes et Vins 6, 20-21.
Roussopoulos, N. C. 1930. Vber ein einfaches chemisches Bestimmungs verfahren der
wahren Aciditfit des Mostes. Prakt. Akad. Athenon 6, 359-362.
Roussopoulos, N. C. 1933. Sur les diffhrentes m6thodes de dosage de l’extrait sec du
vin. Ann. fals. et fraudes 26, 582-594.
Rouzaut, R. 1939. Determinacih cuantitativa de alcohol en meaclas por medio del
destilador a espiraI y movimiento exc6ntrico. Anales inst. invest. dent. g tecnol.,
univ. nacl. litoral (Santa Fd, Argentina) 9, 87-95.
Ruf, W. 1952a. Die Papierchromatographie als Hilfsmittel bei der Untersuchung von
Weinen. Deut. Wein-Ztg. 88, 304.
Ruf, W. 1952b. Vber die Anwendung der Papierchromatographie s u m Nachweis von
Fremdfarbstoffen im Wein. 2.Lebensm.-Untersuch. u. -Forsch. 94, 190-194.
Ruspini, Arnoldo. 1936. Valores de la acides real de 10s vinos Argentinos. Anales asoc.
quim. argentina 24, 51-59.
Rustia, A. 1949. Sui coefticienti per la deterrninasione dell’alcool totale nei vini dolci
e nei mosti in fermentazione. Riv. viticolt. e enol. (Conegliano) 2, 161-165.
Srtburov, N. V., Kalebin, M. I., Khakhinn, L. P., and Danilova, A. 1938. A compara-
tive investigation of different methods for the determination of fruit acids
(transl.). Nauch. Zapiski Inst. Narodnogo Khoz. im Plekhanova 1938 1, 3-50;
C . A . 54, 1407-1408 (1940).
Saenko, N. F. 1947. The Xeres yeast. (trans.) BiokhimiG vinodeliz 1, 98-126.
Saenko, N. F. 1948. The change of the oxidation-reduction potential and the composi-
tion of wine aged under the sherry film (transl.). Biokhimiz UinodeliG 2,86-100
Saenko, N. F. 1951. Accelerating the growth of sherry (yeast) film on wine (transl.).
Vinodelie i Vinogradarstvo S.S.S. R. 11(1), 12-15.
St. Mokranjac, M., and Radmib, S. 1951. Prilog pitanju odredivanja rnetilalkohola
(metanola) i njegova sadrzaja u rakijama. [Contribution la question de 1%
determination du methanol et son contenu rlans les eaux-de vie.] Acta pharma-
ceutica Jugoslavica 1, 97-1 10.
Salani, R. 1937. Contributo allo studio della determinazione polarimetrica della
mannite nei prodotti naturali che la contengono. Znd. saccar. ital. SO, 16-21.
Sallusto, F. 1935. Sul rapport0 cremore:alcool nei vini e nelle vinacce. 11. Ann. chini.
appl. 8, 130-142.
Sallusto, F. 1936-37. Sulla composizione de alcuni vini genuini dell’Italia Meridionalc.
Ann. Jac. agrar. Portici, R. univ. Napoli [3] 8, 3-33.
Sallusto, F. 1938-39a. I vini della Tunisia. Ann. fuc. agrar. Portici, R. univ. Napoli [3]
10, 112-132.
Sallusto, F. 1938-39b. Vini tipici della Lucania. Ann. Jac. agrar. Portici, R . univ.
Napoli [3] 10, 87-111.
Snllusto, F., and Di Natale, G. 1938-39. L’enologia in provincia di Siracusa et i vini
da taglio del territorio di Pachino. Ann. jac. agrar. Portici, R . univ. Nupoli [3]
10, 209-236.
Sallusto, F., and Sculco, U. 1937-38. I vini di Cirb (Catanzaro). A,nn. fac. agrar.
Portici, R. univ. Napoli [3] 9, 121-154.
Salvarezza, M. 1935-37. L’acidith volatile dei vini riel corso della fermentazione.
Annuar. R . staz. enol. sper. (Asti) [2] 2 , 287-395; see also Ztalia Agricola 74,
135-141 (1937).
Sampaio, A. V. de. 1946. Contribuiqio para a dosagem do tanino nos vinhos. Rev. inst.
AdolJo L U ~6Z, 107-115.
Sampietro, C., and Invernizzi, I. 1940. Dosaggio dell’alcool nel vino e soafanze
alcooliche col refrattometro ad immerzione. Ann. chim. appl. SO, 381-387.
Sannino, F. A. 1948. Tratado de Enologia. G. Gili, Buenos Aires.
Sapondzhian, S. O., and Gevorkian, Kh. S. 1953. Determination of acetal in wine
(transl.). Vinodelie i Vinogradarstvo S.S.S.R. 13(1), 13-15.
Sastry, L. V. L., and Tischer, R. G. 1952a. Behavior of the anthocyanin pigments in
Concord grapes during heat processing and storage. Food Technol. 6, 82-86.
Sastry, L. V. L., and Tischer, R. G. 1952b. Stability of the anthocyanin pigments in
Concord grape juice. Food Technol. 6, 264-268.
Savary, M. 1940. .4 propos du bilan de I’asteur. Bull. assoc. rhirn. sucr., dist. 67,
165- 167.
Schaefer, F. 1939. Uber die Verwendung von Sudweinliefen bei der Herstellung von
weinhahnlichen Getranken. VorratspfEegeu. Lebensmittelforsch. 2, 662-666.
Schanderl, H. 1938. Die Nutzbarmachung des oxydativen Stadiums der Hefe bei der
Trauben- und Beerenweinbereitung sowie in der Brennereipraxis. VorratspJlege u.
Lebensmittelforsch. 1, 456-469.
Schanderl, H. 1943. Sur la d6sac6tification des vins par voie biologique. Progr. agr. et
vit. 120, 187-189.
Schanderl, H. 1950. Die Mikrobiologie des Weines. Eugen Ulmer, Stuttgart.
Schanderl, H. 1950-1951. Uber den Einfluss des Entsauerns, verschiedener Schon-
ungen und des Lichtes auf das rH und pH der Weine. Wein u. Rebe 26, 118-128.
Schemer. 1944. Les vitamines du vin. La terre vaudoise. Bull. ofice intern. vin 17(163-
166), 159; see also MLd. el Hyg.June 4, 1943.
Schindler, J., and KozAk, K. 1934. Contribution A la recherche et au dosage de I’acide
citrique (transl.). Chem. Listy 28, 95-97.
Schmid, C., and Nestle, K. Th. 1949. Trubungerscheinungen in 1947er Weinen.

Eine Erwiderung auf die Veroffentlichung von H. Roleff. Deut. Vein-Ztg. 86,
41-42.
Schmitthenner, F. 1937. Fiinf Jahrzehnte mikrobiologische Forschung im Weinfache.
Eine fachgeschichtliche Darstellung. Wein u. Rebe 19, 65-82, 93-106.
Schon, K., Gouveia, A. J. A. de, and Coelho, F. P. 1939. Determinagces quantitativas
de vitamina A, ergosterol, vitamina BZ (lactoflavina) e vitamina C por metodos
ftsico-qufmicos. Estudo do vinho tinto da Bairrada. Rev. fac. cihc. Univ. Coimbra
8, 130-147.
Schulek, E., and R ~ Z S RP. , 1939. &hylalkoholbestimmung nach vorangehender
Reinigung der Alkoholdampfe durch Adsorption. 2. anal. Chem. 117, 400-414.
Schumakov, A. 1930. Der Einfluss der Dosen und der Zeit der Beimischung von SOz
auf die Bildung des Glyzerins bei der Weingarung. Odessa. fientral’na naukoco-
dosvidna uinorobcha stan%; . . . , Pratsi 2(2), 22-28.
Scott, T. 1946. Some notes on caramel. Wines & Vines 27(7), 23.
Seifert, W. 1938. Die Chemie des Mostes und Weines. J. Diemer, Mains.
Seifert, W., and Ulbrich, M. 1930. Formic acid, a constituent of the volatile acidsof
wine, and its determination (transl.). Expt. Eta. Klosterneuberg, Denkschrif!,
70th Anniversary 148-163; C. A . 26, 1627 (1931).
Seiler, F. 1932. Nachpriifung des Verfahrens der Weinsaurebestimmung in Most und
Wein. 2. Untersuch. Lebensm. 64, 285-288.
Seiler, F. 1933. tfber die Zusammensetzung von 1931er Naturweinen des Mosel-
weinbaugebietes. Wein u. Rebe 16, 135-140.
Seiler, F. 1935. Uber die Zusammensetzung von naturreinen Weinen der Jahrgange
1932 and 1933 der Mosel, Saar und Ruwer und den Saureabbau dieser Weine.
Wein u. Rebe 17, 37-45.
Seiler, F. 1936a. tfber die Beziehungen zwischen Mostgewicht und Alkoholgehalt bei
Moselweinen. Wein u. Rebe 17, 267-274.
Seiler, F. 1936b. Zusammensetzung von naturreinen 1934er Weinen der Mosel, Saar
und Ruwer. Wein u.Rebe 17,357-365.
Seiler, F. 1937. Zusammensetzung von 1936 Naturweinen der Mosel, Saar und Ruwer.
Wein u. Rebe 19, 163-166.
Seiler, F. 1938. Uber die Beziehungen zwischen Mostgewicht und Alkoholgehalt bei
1936er Weinen des Moselweinbaugebietes. Wein u. Rebe BO, 118-1 19.
Seiler, F. 1941. Die Bestimmung dcr Milchsaure im Trauben- und Obstweinen.
VorratspJEegeu. Lebensmittelforsch.4, 506-512.
Seiler, F. 1943a. Die Auswertung des Milchsauregehaltes der Weine fur den Saure-
abbau. 2. Lebensm.-Untersuch. u.-Forsch. 86, 84-88.
Seiler, F. 1943b. Die Bestimmung der Weinsaure im Most und Wein. 2. Untersuch.
Lebensm. 86, 507-511.
Seiler, F. 1944. Zusammensetzung von 1942er Naturweinen der Mosel und ihrer
Nebenfliisse. 2. Lebensm.-Untersuch. u.-Forsch. 87, 52-59; see also ibid. 87, 60-61.
Seiler, F. 1952. Zusammensetzung von 1950er naturreinen Moselweinen. Deut.
Wein-Ztg. 88, 236-238.
SBmichon, L. 1933. Lea pectines des raisins et le moelleux des vins. Bull. ofice intern
uin 6(67), 44-53.
SBmichon, L. 1948. Rapport sur l’eonologie. Bull. ofice intern. vin 21(204), 30-51.
SQmichon,L., and Flanzy, M. 1930a. Dosage de la glycerin dans lea vins et lea boissons
ferment6es. Ann. jals. et fraudes 23, 583-602.
Sitmichon, L., and Flanzy, M. 1930b. fitude des substances acides entrant dans la
composition des vins. Ann. jals. et fraudes 23, 5-19.
SBmichon, L., and Flanzy, M. 1931a. Les acides volatiles. Ann. fals. et fraudes 24,
516-534.
SBmichon, L., and Flanzy, M. 1931b. La recherche e t le dosage de l’alcool mbthylique.
SEmichon, L., and Flanzy, M. 1932a. Application de l’oxydation chromiqueA quelques
diacides. Compt. rend. 194, 2063-2065.
SBmichon, L., and Flansy, M. 1932b. Dosage de l’acide lactique dans les vins e t dans
les jus de fruits. Ann. fals. et fraudes 26, 414-416.
SEmichon, L., and Flanay, M. 1932c. Dosage de l’acide succinique dans les vins e t les
liquides ferment&. An71. jals. et fraudes 26, 416-419.
Si:michon, L., and Flanzy, M. 1932d. fitude des acides organiques des vins e t des jus
fruits naturels ou ferment&. Ann. agron. (N.S.) 2, 199-214; see also Ann. brass.
et d i d . 30, 217-222, 236-239 (1932).
SEmichon, L., and Flanzy, M. 1933a. Dosage de I’acide tartrique dans les moQts e t
dans les vins. Ann. fals. et fraudes 26, 403-406.
SBmichon, I,., and Flanzy, M. 1933b. Sur les acides organiques des jus de raisin.
Compt. rend. 197, 198-201; see also Ann. agron. (N.S.) 2, 504-528 (1932).
S6michon, L., and Flanzy, M. 1934. Au sujet de l’application des d6crets relatifs au
degrB minimum des vins. Ann. fals. et fraudes. 26, 92-93.
SPze, R. de 1938. Sur l’abaissement de l’acidith volatile des vins. Rev. viticult. 88,
339-345, 357-363.
Shcherbakov, M. F. 1940. A method for determining nitrogen compounds (transl.).
Vinodelie i Vinogradarstvo S.S.S.R. 1 (5), 8-9; C. A. 37, 2877 (1943).
Shrherbakov, M. F. 1941. A great achievement in the Soviet wine making (transl.).
Doklady VsesoiuznoE Akademii S.-KH.-Nauk 1941(1), 29-31.
Shimamoto, T., and Sugayama, J. 1951. Tyrosol (p-hydroxyphenylethyl alcohol) in
Japanese wine, sake. J . Sci. Research Inst. (Tokyo) 46, 139-143.
Sidersky, D. 1942. Le rapport du dextrose a u lEvulose dans les moQts de raisin. Bull.
assoc. chim. sum., dist. 69, 234-277.
SisakLn, N. M., and Bezinger, E. N. 1949. Separation and determination of amino
acids in wines (transl.). Doklady Akad. Nauk. S.S.S.R. 69, 572-576.
SisakGn, N. M., and Besinger, E. N. 1950. Amino acids in grape wines (transl.).
BiokhimiG vinodeliz 3, 185-195.
Sisaksn, N. M., Egorov, I. A., and Saakgn, R. G. 1950a. On the intensity of the bio- -
chemical reactions in sherry wine production (transl.). Biokhimiiâ vinodeliiu S,
57-68.
SisakLn, N. M., Egorov, I. A,, and S a a k s n , R. G. 1950b. Tryptophan and the vita-
mins of the B group in grape wine (transl.). BiokhimiG vinodeli& S, 96-101.
Sisaksn, N. M., and Marut& S. A. 1948. Sugars in grapes (transl.). BiokhimiG
vinodeli& 2, 56-68.
Smith, M. B., and Olmo, H. P. 1944. The pantothenic acid and riboflavin in the fresh
juice of diploid and tetraploid grapes. Am. J . Bot. 31, 240-241.
Solms, J., Buchi, W., and Deuel, H. 1952. Untersuchungen iiber den Pektingehalt
einiger Traubenmoste. Mitt. Gebiete Lebensm. u. Hyg. 43, 303-307.
S o b , I., Rakcsanyi, L., and Zsolt, J. 1948. Hongrie, oenologie. BUZZ. ofice intern. v i n
21(204), 52-55.
Spaeth, E. C., and Rosenblatt, D. H. 1949. Partition chromatography of antho-
cyanidins. Science 110, 258.
Spaeth, E. C., and Rosenblatt, D . H. 1950. Partition chromatography of synthetic
anthocyanidin mixtures. Anal. Chem. 22, 1321-1326.
50G MAYNARD A. AMERINE
Steuart, D. W. 1934. The determination of tartaric acid in cider. Analyst 69, 532-633.
Stradelli, A. 1951. Evaporazione di alcol durante la fermentazione dei mosti. Riv.
viticolt. e enol. (Conegliano) 4, 50-53.
Straub, J. 1934. Laboratorium-mededeelingen over 1933 van den Keuringsdienst van
Waren te Amsterdam. Chem. Weekblad 81, 459-461.
Strohecker, R., Riffart, H., and Haberstock, J. 1938. Die stufenphotonietrische
Bestimmung der Milchsaure. Vorratspjlege u. Lebensmittelforsch. 1, 93-98.
SuEevi6, 0. 1950. Un supplement sur la connaissance de la composition des moilts dc
la Croatie septentrionale. Biljne pioizvodnje 4, 15; see also Bull. ofice intern. vin
24(241), 199 (1950).
SLtdario, E. 1949. Analisi del Vino e la Ricerca delle Sofisticazioni. Casa Ed. Mares-
calchi, Casale Monferrato.
Sudario, E. 1953. Le sostanze coloranti nelle uve degli ibridi produttori diretti. Ann.
sper. agrar. (Rome) (N.S.) 7, 157-163.
Sulser, H. 1951. Anwendung der Papierchromatographie auf lebensmittelchemisehe
Probleme. I. Grundlagen und Methodik. 11. Unterscheidung von Obstwein und
Wein (with 0. Hogl). Mzlt. Cebiete Lebensm. u. Hyg. 42, 376-394, 395-405.
Sumuleanu, C., and Ghimicewu, G. 1935. Quelques Nouvelles Micro-Mbthodes pour
1’Analyse des Vins. Opinia, Jassy.
Sumuleanu, C., and Ghimicescu, G. 1936. L’analyse de quelquee vins naturels de
Roumanie. Annales sci. Univ. Jassy 22, 194-255.
Swaby, R. J. 1943. Extraction of citrates and tartrates from fruit. Agr. GUZ.N . S.
Wales 64, 571-573, 580.
Szab6, I., and Rakcsdnyi, L. 1935. The ratio of dextrose and Ievulose in the grape, in
the must and in the wine (transl.). Magyar Ampelol. Eukonyu. 9,346-361; see also
6th Congr. intern. tech. chim. ngr. I, 936-949 (1938); C . A. 80, 1507 (1936).
Tabachnick, J., and Joslyn, M. A. 1953. Formation of esters by yeast. I. The produc-
tion of ethyl acetate by standing surface cultures of Hansenula anomala. J .
Bacteriol. 66, 1-9.
Taufel, K., and Mayr, F. 1933. Zur quantitativen Ermittlung der Citronensaure
durch Uberfuhrung in Aceton. 2. anal. Chem. 93, 1-20.
Tanner, H., and Rentschler, H. 1953. Die quantitative Bestimmung der xpfelsaure in
siissen und vergorenen Getranken. Schweiz. 2. Obst- u. Weinbau 62, 74-77.
Tarantola, C. 1934. Determinazioni delle aldeidi nel vino con il fotometro di Pulfrich
e con il colorimetro fotoelettrico. Ann. chim. appl. 24, 615-625.
Tarantola, C. 1937a. L’acido citric0 nei vini italiani. Ann. tech. agrar. 10, 1-23; see
also Annuar. R. staz. enol. sper. Asti [2] 2, 183-202 (1935-37), and Bull. inst.
oeno.?. A l g k i e 10, 138-139 (1937).
Tarantola, C. 1937b. Dosage de l’acide citrique dans les vins. Bull. inst. oenol. AZgBrie
10, 88-92.
Tarantola, C. 1947a. La determinaaione dell’estratto nei vini. Prog. vinic. ed ozea.
23(6/7), 7-14.
Tarantola, C. 1947b. Le sostanxe azotate nei mosti e nei vini. Zndustrie Agrarie
22(9/10), 7-18.
Tarantola, C. 1948. Le sostanze riduttrici infermentescibili dei vini. Riv. vitimlt. e
enol. (Conegliano) 1, 44-47.
Tarantola, C. 1949. La determinazione dell’acidit8 volatile nei vini in presenza di
anidride solforosa. Riv. viticolt. e enol. (Conegliano) 2, 193-199.
Tarantola, C. 195Oa. T,’invertasi nei vini. Riv. viticolt. e enol. (Conegliano) 8, 229-232,
247-253.
Tarantola, C. 195Ob. Pentosi e pentosani nei vini e bilancio dei carboidrati fermente-
scibili e infermentascibili. Riv. viticolt. e enol. (Conegliano) 3, 287-292, 315-320.
Tarantola, C. 1951. I constituenti polifenolici del vino. Atti accad. ital. vile e vino 3,
362-368.
Tartaglia, A. 1950. Difference in the ethyl alcohol content of Salento wines determined
by the Malligand and by distillation methods (transl.). Coltimtore e giorn.
vinicolo ital. 94, 156-168, 177-179; C. A . 46, 1722 (1951).
Teixeira Jdnior, E. do V. 1940. 0 etanal no vinho do PArto. Anais inst. vinho PGrto 1,
295-303.
Teply, L. J., Strong, F. M., and Elvehjem, C. A. 19-12, Distribution of nicotinic acid
in foods. J. Nutrition 23, 417-423.
Ter-Karapetian, M. A. 1952. The biological oxidation of ethyl alcohol into acetalde-
hyde by the method of immersed dispersed culture. Acad. Sci. U.S.S.R. 83,
885-888.
Testa, J., and Maveroff, h1. 1949. Las enzimas en la clarificacih de 10s mostos. Anales
inst. vino (Cuyo) 1, 45-69.
Thaler, H., and Roos, W. 1950. tfber die Bestimmung des Glycerins im Wein mittels
Perjodsaure. Z. anal. Chem. 131, 24-27.
Tomaghelli, A. A. 1937. Constante de equilibrio y color de reaccidn para el sistema:
acido acetico, alcohol etilico, acetato de etilo y agua. Rev. fac. cien. qufm. univ.
nacl. La Plata 12, 107-111.
Tomaghelli, -4.A. 1942. Microdeterminaci6n del alcohol en vinos. Anales asoc. quim.
argentina 30, 30.
Tbrley, D. 1942. Szerves savak mennyisege magyar borokban. (The amount of organic
acids in IIungarian wines.) Mezdgazdasdgi Kutatdsok 16, 310-320; C. A . 38, 6041
(1944).
Torricelli, A. 1941. A la recherche d’un nouveau prockde d’investigation pour deceler
les vins fraud6s. Mitt. Gebiete Lebensm. u.Hyg. 32, 211-216.
Torricelli, A. 1943. A la recherche d’un nouveau procede d’investigation pour dbceler
les vins fraiidbs. Mitt. Gebiete Lebensm. u. Hyg. 34, 115-127, 158-168.
Torricelli, A. 1945. Un moyen de differencier les jus de raisins frais des jus prepares
par dilution de concentrks de raisins frais ou secs. Mitt. Gebiete Lebensm. u.Hyq.
36, 46-49.
Trauth, F. 1949. Ein Verfahren zum raschen qualitativen Nachweis von Siisstoff im
Wein. Weinbau, Wissenschaftl. Beih. 3, 106-108.
Trauth, F., and Bitssler, K. 1936. Ein Beitrag zur Frage der Beziehungen zwiwhen
Mostgewicht und Alkoholgrhalt und deren Nutzanwendung bei der Verbess-
erung der Moste. Z . Untersuch. Lebensm. 72, 476-498.
Trost, F. 1935. Sugli alcoli superiori dei cognacs. Ann. chim. a p p l . 26, 660-668.
Turbovsky, M. W., Filipello, F., Cruess, W. V., and Esau, P. 1934. Observations on
the use of tannin in wine making. Fruit Products J . 14, 106, 121, 123.
Twight, E. H. 1951. The pycnometer in alcohol determination. Wines & Vines 32(5),
22-23.
TJchimoto, D. 1951. Effect of temperature on certain products of vinous fermentation.
Thesis for M. S., liniversity of California, Bprkeley; see also Food Research 17,
361-366 (1952).
Valaize, H. 1949. Nouveaux essais sup l’utilitk d u phosphate d’ammoniaqur en
vinification. Inds. agr. et aliment. (Paris) 66, 349-352.
Valaize, H., and Dupont, G. 1951. Les acides aminks et le bouquet des vins. Inds. agr
el aliment. (Paris) 68, 245-250.
508 MAYNARD A . h M E I t I N E
Valaer, P. 1945. Caramel and other artificial coloring matter in alcoholic liquors.
J . Assoc. Ofic. Agr. Chemists 28, 467-470.
Valaer, P. 1947. Methods of analysis of wine. J . Assoc. Ofic. Agr. Chemists 30, 327-
331.
Valaer, P. 1948. Report on caramel in wines and distilled liquors. J . Assoc. Oflc.
A p . Chemists 31, 178-184; see also ibid. 75-77.
Valaer, P . 1949. Report on chromatographic adsorption of wines. J . Assoc. Ofic. Agr.
Chemists 32, 155; see also ibid. 33, 319-320 (1950).
Vartan’Gn, M. D. 1951. Rapid method for determination of sugar in wine (transl.).
Vinodelie i Vinogradarstvo S.S.S. R. 11(7), 21.
Vasconcellos e Lencastre, A. Q. de. 194Oa. ConsideragiXoessobre taninos e sua deter-
minaqgo nos vinhos do Pbrto. Anais inst. vinho Pdrto 1, 203-289.
Vasconcellos e Lencastre , A. Q. de. 1940b. Doseamento do Acido succhico nos vinhos
do Pbrto. Anais inst. ivinho PBrto 1, 291-294.
Vasconcellos e Lencastre, A. Q. de. 1941. 0 Acido succhico e 0s vinhos do Parto. Anais
inst. vinho Pdrto 2, 243-304.
Vasconcellos e Lencastre, A. Q. de. 1946. Doseamento da glicerina no vinho do Pbrto
e nos vinhos de pasto. Anais inst. vinho Pdrto 7, 31-48.
Velazquez, E. 1936. Contribuci6n a1 estudio del acido lactico en vinos argentinos
Rev. fac. cien. qiiim. univ. nacl. L a Plata 11, 65-68.
Venezia, M. 1935. Osservarioni e studi sull’azoto in combinarione amminica nei
mosti e nei vini. Annuar. R. staz. sper. viticolt. e enol. (Conegliano) 6, 103-110.
Veneria, M. 1937. Sull’invertasi nei vini. Annuar. R. staz. sper. viticolt. e enol. (Coneg-
liano) 7, 283-299.
Venezia, M. 1938. Sull’acido ascorbic0 (Vitamina C) nell’uva e nel vino. Annuar. R.
staz. sper. viticolt. e enol. (Conegliano) 8, 67-82.
Venezia, M. 1939-40. La fermentazione glicerica del mosto d’uva. Azione di un
attivatore (fattore 2). Atti R. Zst. Veneto Sci., Lettere, Arti ZZ 00, 563-587.
Venezia, M. 1940. Sur une m6thode pratique pour d6celer les colorants artificiels dans
les vins i-ouges. Bull. ofice intern. uin 13(140), 97-101.
Venezia, M. 1944. Propriht6s alimentaires et caracthristiques biologiques du raisin.
Venezia, M. 1949. Sul grado ebulliometrico dei vini dolci. Ann. sper. agrar. (Rome)
(N.S.) [5] 3, 919-925.
Venezia, M., and Gentilini, L. 1935. Sulla distribuzione del saccarosio nella vite. Ann.
chim. appl. 26, 203-215.
Venezia, M., and Gentilini, L. 1941. Ricerche sulla fermentazione glicerica del mosto
dluva. Annuar. R. staz. sper. viticolt. e enol. (Conegliano) 10, 283-309; see also
Ital. vitic. ed agr. 28, 529-532 (1938).
Ventre, J. 1936. Les lewres en vinification. Progr. agr. et vit. 106, 111-115, 135-140,
153-155, 183-187.
Ventre, J. 1937. Acidit6 volatile et fermentation. Ann. ferment. 3, 447-465; see also
Ann. &ole nat. agr. Montpellier 26, 15-35 (1937).
Ventre, J. 1939. Contribution hiochimique it 1’6tude des vins eud6mis6s. Ann. ferment.
6, 13-44, 74-92.
Ventre, J . 1946-1947. Trait6 de Vinification, Pratique et Rationnelle. 2 v. Poulain,
Montpellier.
Verda, A. 1940. Sull’aggiunta dell’acido citric0 nei vini. Ann. chim. appl. SO, 209-214.
Vetscher, A. S. 1947. The determination of alcohol and of extract in wines using a sugar
refractometer (transl.). Vinodelie i Vinogradarstvo S.S.S. R. 7(2), 33-35.
Vetscher, A. S., and Losa, V. M. 1947. The influence of vitamins C and B, on cham-
pagnes (transl.). Vinodelie i Vinogradarslvo S.S.S.R. 7(5), 9-11.
ViBles, P. 1939. Sur un dosage iodom6trique de l’acidit6 des vins. Bull. soc. chim.
France [5] 6, 1127-1129.
Vil’Gms, V. V., and Taranova, R. D. 1950. A quantitative determination of the color-
ing material of red wines (transl.). Vinodelie i Vinogradarslvo S.S.S.R. 10(4),
29-30.
Vil’sms, V. V., and Taranova, R. D. 1951. Chromatographic method of separating
wine pigments (transl.). Vinodelie i Vinogradarstvo S.S.S.R . 11(7), 16-18.
Villforth, F. 1940. Die quantitative Bestimmung des Gehaltes a n fltichtigen Aldehyden
in Wein. VorratspJege u. Lebensmittelforsch. 8, 590-598.
Villforth, F. 1950-51. Die Bestimmung der Ameisensaure im Wein. V e i n u . Rebe 26,
144-179.
Violante, C. 1948. Richerche spettrofotometriche sulla materia colorante delle uve.
Riv. viticolt. e enol. (Conegliano) 1, 265-270, 298-304.
Violante, C. 1950. L’acido lattico nei vini dell’Irpinia. Riv. viticolt. e enol. (Conegliano)
8, 173-176.
Violante, C., and Bemporad, G. 1937. Ricerche spettrofotometriohe sulla materia
colorante dei vini. Nota 1. Curve tipiche di alcuni vini dell’Irpinia e dei pih
comuni coloranti artificiali. Ann. chim. appl. 27, 399-406.
Violante, C., and Imbrici, D. 1949. La determinazione dell’acidith volatile per distil-
lazione diretta. Riv. viticolt. e enol. (Conegliano) 2, 329-334.
Vogt, E. 1934. Mostgewicht und Alkoholgehalt. Z . Untersuch. Lebensm. 68, 473-486.
Vogt, E. 1935. Untersuchungen uber die Farbe der Rotweine. Wein u. Rebe 17,
93-101.
Vogt, E. 1945. Weinchemie und Weinbereitung. J. Diemer, Mainz, 2d Ed.
Vogt, E. 1952. Bildung von hoheren Alkoholen wahrend der Garung. Land- und
Hauswirtschaftl. Auswertungs-und Informationsdienst, Bericht Gartenbau-Forsch.
16, 93.
Volmar, Y., and Clavera, J. M. 1931. Los indicadores fluorescentes en las medidas de
acidez de 10svinos tintos. Anales SOC. espail. fk.quim. 29, 247-254; see also ibid.
490-493 and 494-496 for a polemic.
Voskobohikov, I. 1931. Materiale zum Stadium der ohemischen Zusammensetzung
der Wein der Weinbaustation in Odessa. Odessa. %entral’na naukovo-doslidchoz
vinogrado-vinorobcha stan&&, Pratsi 4, 55-87.
Voskobohikov, I., and Negenzev, I. 1930. Zur Methode der Totalsaurebestimmung
in den Weinen nach Calcimeter-acidimeter von Berner. Odessa. fientral’na
naukovo-dosvidna vinorobcha stan.%& Pratsi 2 (3), 47-54.
Waser, E., Mohler, H., and Almasy, F. 1932. Spektrophotometrischer Nachweis von
Holundersaft in Wein. Mitt. Gebiele Lebensm u. H y g . 28, 243-249.
Watt, B. K., and Merrill, A. L. 1950. Composition of foods-raw, processed, prepared.
U.S. Dept. Agr., Agr. Handbook 8, 1-147.
Webb, A. D., Kepner, R. E., and Ikeda, R. M. 1952. Composition of a typical grape
brandy fuse1 oil. Anal. Chem. 24, 1944-1949.
West, D. B., Evans, R. F., and Becker, K. 1951. Studies of the ester and higher-
alcohol contents of beer. Proc. Am. Soc. Brew. Chem. 1961,51-59.
Western Regional Research Laboratory. 1943. Recovery of tartrates from grape
wastes. Fruit Prodi~ctsJ . 23, 107-112.
Widmer, A. 1931. Vorpriifung auf Milchsiiure in Wein und Obstwein. Mitt. Gebiete
Lebensm. u. Hyg. 22, 223-230.
WikBn, T., and Richard, 0. 1951-1952. Untersuchungen iiber die Physiologie der
Weinhefen. Antonie van Leeuwenhoek J . Microbiol. Serol. 17, 209-226; 18,31-44.
Wilharm, G., and Holz, G. 1951. Beitrag rur Kenntnis des Acroleins in Obstbriinden,
Maischen und Mosten. Z.Lebensm.-Untersuch. u . -Forsch 92, 96-99.
Willaman, J. J., and Kertesr, Z. I. 1931. The enzymic clarification of grape juice.
N . Y . State Agr. Expt. Sta. Tech. Bull. 178, 1-15.
Williams, B. L., and Wender, 5. H. 1952. The isolation and identification of quercetin
and isoquercetrin from grapes (Vitis vinifera). J . Am. Chem. SOC.74, 4372-4373.
Williams, M. B., and Reese, H. D. 1950. Colorimetric determination of ethyl alcohol.
Anal. Chem. 22, 1556-1561.
Winkler, A. J., and Amerine, M. A. 1938. Color in California wines. I. Methods for
measurement of color. Food Research 3, 429-438.
Wirthle, F. 1931. Bestimmung der Milchsaure im Wein. Chem.-Ztg. 66, 646-647.
Wobisch, F., and Schneyder, J. 1952. Kolorimetrischer Nachweis von Holunderbeer-
farbstoff im Wein. Monatsh. Chem. 83, 478-481.
Yamada, M. 1932. ' h e r die alkoholische Garung der Aminosauren. Bull. Agr. Chem.
SOC.Japan 8,95-101.
Yang, H. Y., and Wiegand, E. H. 1950. Spectrophotometric characteristics of berry
wines. Fruit Products J . 29(5), 138-140, 155.
Zalesskaya, M. I. 1938. Formation of fuse1 oil during alcoholic fermentation (transl.).
Mikrobiologiya U.S.S.R. 7(5), 546.
Zeglin, H. 1952. Nachweis geringster Mengen Methanol in alkoholischen Getranken.
Angew. Chem. 64,359.
Zheltkevich, G. G. 1951. A single method for alcohol determination (in wines) is
necessary (transl.). Vinodelie i Vinogradarstvo S.S.S.R . 11 (4), 62.
Zimmermann, H. W. 1949. Determination of alcohol in wines by the dichromate
method. PTOC.Wine Tech. Conf., Davis 1949, 107-109.
Zlataroff, A., and Poppoff, I. D. 1937. Enrymochemie des Tees und des Weines. 2.
Untersuch. Lebensm. 73, 154-164.
Zweigelt, F. 1938. Rund um das Zeiss'sche Handrefraktometer. Wein u. Rebe 20,
112-118.
Author Index
A Anthony, R. O., 282, 297

Antoniani, C., 355, 358, 367, 370, 372,
Abdulov, A,, 123, 147 379, 468
Abersold, J. N., 116, 168 Ant-Wuorinen, O., 366, 468
Abrahams, M. D., 398, 483 Arbecchi, A. C., 410, 496
Abramson, H., 11, 28, 42 Arbuzova, E. M., 373, 489
Adams, A. T., 233, 667 Archinard, P., 118, 147, 409, 410, 431,
Adams, E. Q., 335, 347 452, 456, 468, 478
Adams, R., 59, 91, 105, 147 Armentano, L., 275, 696
Adler, E., 58, 91 Armstrong, M., 271, 896
Affre, J. P., 426, 494 Arnason, A., 143, 166
Agabal’Gng, G. G., 355, 479 Arutunyan, L. A,, 458, 468
Aggarwal, J. S., 35, 60 Asbury, C. E., 146, 148, 167
Ajon, G., 445, 448, 466 Ashworth, R. de B., 276, 297
Akiya, A., 367,466 Ashworth, U. S., 204, 205, 206, 667
Akman, A. V., 406, 466 Astruc, H., 407, 437, 468
Alas, G., 411, 415, 491 Atkin, S., 116, 148
Albury, M., 143, 166 Atkinson, F. E., 127, 136, 144, 148
Alessandrini, E., 369, 466 Atkinson, I., 35, 38, 46
Alexander, E. R., 57, 91 Atlinger, H. W., 132, 1.48
Alexis, E., 419, 420, 466 Avery, T. B., 5, 6, 47
Alfa, J., 376, 406, 466 Ay, H., 430, 468
Allanson, V., 17, 49 Ayres, G. H., 342, 348
Almasy, F., 445, 609 Ayres, J. C., 33, 39, 43, 47, 49, 233, 667,
Almeida, H. de, 383, 384, 448, 466 348
Amadio, G., 409, 466 Azevedo, M. P. de, 356, 468
Amati, A., 464, 493
Amerine, M. A., 118, 126, 137, 139, 147, B
330,331,348,361, 354,355,356, 358,
359,360,372,373,374,375,376,377, Bacharach, A. L., 276, 277, 299
378,384,385,386,388,389, 391,393, Backer, H. J., 56, 59, 91
395,396,398,402, 404,405, 416, 419, Baglioni, S., 454, 468
420,427,431,435,440,441,442,444, Bailey, A. E., 3, 5, 46
458, 466, 467, 484, 487, 610 Bailey, G. F., 326, 348
Amos, A. J., 317, 349 Bailey, J. S., 281, 696
Anderson, E. B., 10, 61 Bailly, M., 137, 148
Anderson, J. A., 226, 268, 316, 317, 349 Bakay, B., 116, 169
Anderson, J. D., 27, 48 Baker, G. A., 358, 484
Anderson, R. J., 282, 296, 299 Baker, W., 286, 696
Anderson, R. L., 200, 234, 267 Balavoine, P., 365, 382, 468
Andersaen, F. G., 141, 1-47 Balestrazzi, A., 359, 468
Anibal Burgoa, P., 374, 376, 378, 402, Balog, E. G., 127, 142, 160, 164
403, 467 Bancroft, T. A., 200, 667
511
512 AUTHOR INDEX
Banks, A., 11, 12, 22, 26, 48, 271, 696 Bernheim, F., 11, 61
Banos, M., 369, 4Y8 Bernt, H., 277, 699
Barbera, G., 425, 426, 449, 452, 468 Bertagnini, C., 55, 91
Bard, J. C., 234, 236, 237, 238, 867 Bertin, C., 139, 148, 359, 360, 376, 378,
Barini-Banchi, G., 357, 363, 420, 468,480 410, 469
Barker, H. A., 88, 89, 96, 133, 142, 168 Bertram, S. H., 373, 469
Barnard, R. D., 19, 48 Bertrand, G., 366, 367, 368, 469
Barnes, R. H., 7, 43, 46, 48 Besinger, E., 450, 453, 494
Barron, E. S. G., 22, 26, 43 Besone, J., 427, 4YO
Bartels, W., 399, 468 Bethel, R., 90, 94, 141, 142, 166
Bartlett, M. S., 182, 667 Beuschlein, W. L., 100, 148
Baruah, P., 696 Beyer, G. F., 357, 446, 4YO
Biissler, K., 365, 468, 607 Beythien, A., 98, 148
Bastisse, E.-M., 361, 491 Bezinger, E. N., 448, 450, 451, 454, 494,
Bate-Smith, E. C., 30, 43, 266, 270, 273, 496, 606
274, 275, 279, 288, 696, 696 Bianchi, A., 108, 111, 138, 139, 148
Batsyna, I., 450, 494 Bianconi, A,, 108, 111, 138, 139, 148
Battay, F., 413, 483, 493 Bicalho, N. dos Santos, 357, 4YO
Bauer, E., 59, 60, 61, 82, 83, 93 Bickoff, E. M., 12,43
Bauernfeind, J. C., 27, 43 Biedermann, W., 415, 416, 4YO
Bawtenheimer, J. W., 412, 496 Bioletti, F. T., 98, 124, 125, 131, 133, 136,
Beadle, B. W., 5, 24, 25, 29, 43, 4 Y , 49 137, 138, 140, 148
Beal, G. D., 124, 167 Bisson, C. S., 132, 148
Beardsley, C. L., 118, 168 Black, H. C., 24, 43
Beavens, E. A., 133, 148 Black, J. W., 119, f49
Beck, J. E., 393, 496 Blake, M. A., 281,896
Becker, K., 433, 609 Bleyer, A., 355, 398, 470
Beckman, A. O., 342, 349 Bliss, C. I., 240, 241, 667
Beckwith, J., 366, 374, 469 Blumenthal, S., 146, 149
Beiler, J. M., 276, 896 Blumer, G. P., 24, 44
Beis, C., 391, 469 Blumer, T. N., 37, 43
BBjambes, M., 381, 483 Bobadilla, G. F. de, 357, 377, 384, 393,
Bemporad, G., 445, 609 394,396,398, 401, 402,403, 405, 406,
Bennett, A. H., 103, 114, 115, 118, 148 411, 416, 435, 455, 456, 470
Bennett, B. B., 222, 223, 225, 669 Bodansky, O., 15, 43, 46
Bennetts, H. W., 276, 896 Bodendorf, K., 368, 470
BentsBth, A., 275, 896 Boehm, E., 24, 43
Bentz, R. W., 30, 43 Boggs, M. M., 129, 149, 250, 667
Benvegnin, L., 114,148,355,356,403,469 Boggs, W. E., 116, 169
Benz, G., 364, 469 Bohringer, P., 358,363, 420, 430, 4YO
Beppu, I., 286, 898 Bohringer, W., 397, 417, 489
Bhres, T., 275, 696 Bohrisch, P., 98, 148
Berg, E. W., 127, 160 Bolcato, V., 365, 470
Berg, P., 391, 400, 401, 404, 413, 420,469 Bonaterra, R., 369, 470
Berg, V. A., 106, 148, 415, 418, 469 Bordas, F., 362, 4YO
Bergek, T., 58, 66, 98 Borgstrom, G., 123, 126, 149
Bergman, H. F., 392, 469 Bornstein, B., 132, 133, 166
Bergner, K. G., 87, 91 BoBnjak, I., 404, 430, 4YO
Beridze, G. I., 440, 469 Botelho, J. C., 362,385, 386, 424,.470,4Yl
Berner, C., 377, 381, 382, 394, 398, 401, Botezatu, M., 114, 168
403, 406, 416, 455, 456, 469 Bouchard, J., 414,471
AUTHOR INDEX 513
Boudet, V., 365, 478 Burroughs, L. F., 147, 149, 160
Bouma, P. J., 309, 331, 332, 333, 348 Burton, W. G., 89, 91
Bourne, J. A., 133, 148 Buscardns Ubeda, F., 368, 471
Boutaric, A., 330, 348, 414, 441, 471 Busing, K. H., 103, 149
Bowen, W. J., 21, 43 Butlestone, N. C., 114, 149
Bradbury, R. B., 276, 296 Butterworth, I. S. C., 342, 348
Bradfield, A. E., 281, 288, 296 Buxbaum, W., 420, 472
Brady, D. E., 6, 26, 34, 37, 38, 43, 49, 60 Bywaters, J. H., 26, 44
Brailovska;, A. E., 393, 493
Braitberg, L. D., 59, 92 C
Branch, G. E. K., 102, 149
Bratzler, J. W., 7, 43 Cahill, W. M., 457, 478
Braun, E., 294, 297 Cailleau, R., 459, 460, 462, 472
Braun, W. Q., 6, 60 Cain, R. F., 338, 361
Braverman, J. B. S., 107, 109, 110, 112, Caldecourt, V. J., 341, 360
122, 134, 135, 136, 149, 163 Caldwell, J. S., 143, 145, 161, 16.2
BrBmond, E., 98,149, 357,359, 363, 388, Calkins, V. P., 25, 26, 43, 44
398, 404,405, 410, 414, 415, 416, 471, Callow, E. H., 31, 33, 44
476 Cambitzi, A., 393, 472
Bretthauer, G., 383, 384, 404, 488, 489 Cameron, S. S., 141, 149
Breyer, G. F., 364, 476 Campbell, S. V., 145, 166
Brissey, G. E., 31, 43 Campbell, T. W., 12, 43
Brockmann, M. C., 374, 379, 471 Campllonch Romeu, I., 356, 383, 410,
Broeg, C. B., 315, 348 472, 492
Brooks, J., 14, 18, 30, 36, 39, 43 Canals, E., 392, 423, 472
Brooks, R. O., 119, 149 Cannon, C. G., 342, 348
Brown, W. L., 442, 471 Cantor, S. M., 65, 75, 78, 91
Browne, C. A., 312, 348 CaokBn, P., 446, 489
Browne, H. H., 60, 61, 63, 91, 106, 149 Cappucci, C., 411, 478
Bruce, B. A., 315, 348 Capt, E., 114, 148, 355, 403, 469
Brugirard, A., 438, 439, 471 Carli, E., 420, 481
Brune, H., 366, 471 Carrantd, V., 292, 896
Bryan, L. A., 88, 93, 143, 164 Cartier, P., 398, 472
Bucci, F., 424, 471 Casale, L., 137, 149, 371, 396, 414, 415,
Biichi, W., 426, 427, 606 444, 448, 454, 468, 472
Buchman, E. R., 103, 149, 160 Casamada M a d , R., 444, 472
Buck, P., 87, 9.2 Casanave, M. P., 118, 149
Buck, R. E., 326, 348 Casares, R., 381, 394, 440, 472
Buhrer, N. E., 392, 471 Castallani, A. G., 17, 49
Bukin, V. N., 282, 897 Castel, A., 407, 437, 468
Bullis, D. E., 144, 149 Caster, W. O., 342, 348
Bulman, C., 33, 43 Castor, J. G. B., 370, 371, 448, 451, 453,
Bunte, H., 108, 149 454, 461, 472
Buogo, G., 400, 401, 471 Caughlan, C. N., 56, 9.2
Burdzhanadze, V. F., 411, 471 Cayrol, P., 387, 458, 481
Biirgi, J., 369, 471 Cecil, S. R., 35, 38, 42, 123, 133, 145, 160
Burgvits, G. K., 414, 471 Cerasari, E., 397, 472
Burkard, J., 423, 484 Cerutti, G., 366, 367, 368, 473
Burkhardt, R., 395, 396, 484 Chace, E. M., 90, 92, 141, 142, 149, 160
Burr, G.O., 5, 7, 24, 25, 43, 44, 46,. 48 Chalk, L., 287, 298
Burr, M. M., 5, 43 Chamberlin, G. J., 309, 348
514 AUTHOR INDEX
Champlin, F. M., 422, 483 Conrad, R. M., 5, 6, 8, 11, 47, 60, 61

Chang, I., 5, 6, 23, 24, 26, 32, 34, 38, 44 Cook, W. H., 21, 30, 38, 39, 44, 60, 68,
Chapman, R. A,, 10, 24, 25, 48 194, 195, 221, 222, 867
Charles, E., 367, 383, 384, 473 Coppinger, G. M., 12, 43
Charley, V. L. S.,127, 136, 145, 147, 160, Cordebard, H., 363, 473
274,996 Cornforth, J. W., 442, 473
Charpentid, Y., 395, 400, 404, 407, 473, Cornwell, R. T. K., 27, 44
497 Corran, J. W., 103, 160
Charrihe, R., 399, 401, 482 Correia, E. M., 377, 394, 398, 402, 403,
Chauvet, J., 463, 473 415, 416, 473, 474, 482
Chelle, L., 385, 473 Cosmo, I., 377, 378, 474
Chenicek, J. A., 24, 60 Costopoulos, J., 419, 488
Chevalier, R., 406, 486 Couch, J. F., 291, 297
Chevillard, L., 459, 460, 462, 472 Coutts, M. W., 131, 167
Chichester, C. O., 441, 491 Cowan, J. C., 33, 44
Chipault, J. R., 35, 44 Cox, G. M., 200, 207, 267
Chmielewska, I., 285, 296 Cox, R. F. B., 294, 397
Chogovadze, S. K., 444, 473 Cox, W. W., 14,44
Chong, G., 428, 474 Cramer, H., 213, 267
Christie, A. W., 90, 94, 133, 141, 166 Crapple, G. A., 271, 299
Church, C. G., 90, 98, 142, 149, 160 Criddle, J. E., 7, 8, 44
Churchward, C. R., 360, 473 Crisci, P., 414, 415, 416, 474
Cioffi, R. M., 369,370,371, 468,473, 477 Crist, R. H., 102, 166
Clair, E., 361, 409, 473 Cromwell, N. H., 57, 96
Clark, A. M., 15, 60 Cruess, W. V., 123, 124, 125, 127, 131,
Clark, W. G., 270, 271, 896 132, 133, 136, 137, 138, 140, 141, 142,
Clark, W. L., 324, 325, 326, 345, 360 144,148, 160, 161,166,271,296,354,
Clarke, A. E., 283, 296 355, 356, 367, 411, 427, 428, 440,
Clausen, D. F., 25, 44 454, 463, 470, 474, 486, 488, 607
Clausen, M., 7, 48 Cruickshank, E. M., 5, 44
Clavera, J. M., 369, 390, 421, 424, 429, Culpepper, C. W., 143, 145, 161,162,289,
473, 609 296
Cochran, F. D., 182, 184, 190, 248, 249, Cunha, J. D. S. da, 385, 474
969 Cunha Ramos, M. da, 419, 474
Cochran, W. G., 193,200, 207,216,267 Cunha, T. J., 7, 61
Coelho, F. P., 458, 462, 604 Cunha, V., 373, 474
Cohee, R. F., Jr., 34, 44, 87, 92 Curtis, L. C., 285, 299
Coleman, H. M., 13, 27, 32, 44 Cusmano, I., 359, 474
Coleman, H. W., 13, 46 Czimmer, A. G., 277,296
Collet, H., 423, 478
Collier, D., 436, 473 D
Colombier, L., 361, 409, 473
Coliescu, I. H., 367, 473 Dabkiewicz, I., 24, 44
Comar, C. L., 118, 163, 383, 384, 385, Dalmasso, G., 376, 416, 455, 474
395, 440, 487 Danehy, J. P., 105, 130, 161
Comstock, R. E., 26,34,60,182,184,190, Dangoumau, A., 431,474
231, 232, 248, 249, 269 Daniel, E. P., 458, 476
Conceipsio, A. de B. F. da, 445, 473 Danilova, A., 387, 602
Cone, J. F., 204, 205, 206, 267 Darlington, C. D., 280, 296
Conn, J. E., 70, 94, 124, 167 Davidson, H. R., 334, 341, 348
Conochie, J., 31, 46 Davh, J. L., 101, 169
AUTHOR INDEX 515
Davis, L. L., 26, 44 Dupont, G., 368, 448, 454, 455, 456, 463,
Davis, M. B., 161 476, 607
De Astis, G., 359, 476 Durham, H. E., 140, 161
Deatherage, F. E., 21, 46, 168, 867 Durmishidze, S. V., 282, 296, 297, 405,
Debordes, G., 421, 422, 431, 474,476 436, 437, 438, 439, 443, 453,454, 476,
Dehnel, E., 55, 94 476
Delahay, A., 65, 75, 78, 98 Durodie, J., 369, 476
Della Barba, L., 426, 496 Dutton, H. J., 33, 44, 326, 348
Dell’Olio, G., 376, 416, 455, 474 Dwyer, P. S., 233, 267
Delp, L., 454, 476 Dybowski, J., 146, 161
De Mattei, W., 444, 467 Dyer, B., 120, 161
Deplanque, R., 370, 488
Descatoire, F., 369, 492 E
Desrosier, N. W., 346, 348 Eastmond, E. J., 326, 327, 328, 348
Deuel, H., 426, 427, 429, 488, 606 Eckert, A., 355, 476
Devaney, R. G., 341, 349 Eckey, E. W., 25, 46
Dicenty, D., 355, 476 Eckman, J. R., 99, 161
Diemair, W., 373, 379, 381, 404, 436, 437, Egorov, I. A., 385, 408, 453, 454, 459,
438, 440, 476 460, 461, 462, 476, 606
Dietrich, W. C., 372, 373, 467 Eheart, M. S., 142, 161
Dikhtyar, P. O., 410, 498 Eibner, A., 55, 98
Di Natale, G., 376, 378, 603 Eisenhart, C., 221, 267
Dixon, W. J., 249, 867 El-Kattan, A. A., 338, 349
Doak, B. W., 143, 166 Ellis, N. K., 346, 348
Dobrescu, J., 367, 473 Elvehjem, C. A., 460, 607
Dockstader, W. B., 24, 48 Emel’yanova, M. P., 454, 491
Dod6, M., 378,492,493 Emerson, 0. H., 24, 49
Doegey, J . L., 24, 44 Emiliani, E., 360, 476
Donaldson, R., 339, 348 Eolkin, D., 338, 348
Donnally, L. H., 65, 66, 74, 96 Erb, C., 312, 314, 361
Donovan, F. K., 114, 115, 118, 148 Errichelli, E., 361, 476
D’Orazi, F., 365, 470 Errichelli, U., 362, 476‘
Downer, A. W. E., 62, 79,83, 84,98, 103, Esau, P., 440, 607
115, 118, 125, 134, 135, 161 Espezel, R.. 114, 118, 163, 383, 486
Doxtator, C. W., 200, 201, 202, 203, 867 Espil, L., 372, 402, 403, 404, 431, 432,
Dryden, E. C., 147, 161 476, 480
Dubaqui6, J., 385, 391, 421, 422, 473, Espinosa, N. A., 421, 422, 476
476 Esselen, W. B., Jr., 54, 55, 82, 84, 86, 99,
Dubok, C. W., 31, 35, 44 104, 164
Dubrowskaja, V. P., 363, 476 Etienne, A. D., 364, 476
Ducellier, G., 444, 476 Ettinger, I., 364, 476
Dugan, L. R., 24, 44, 47 Evans, E. I., 25, 46
Duisberg, P. C., 31, 44 Evans, R. F., 433, 609
Dujardin, J., 356, 476 Evans, R. M., 309, 314, 348
Dujardin, L., 356, 476 Evers, C. F., 128, 169
Dujardin, R., 356, 476 Ewing, G. W., 342, 348
Dulou, R., 368, 476
Dunn, R., 410, 411, 487 F
Dunn, T. C., 146, 161 Fabre, J.-Henri, 118, 161, 356, 359, 363,
Dupaigne, P., 116, 161 398, 404, 405, 410, 416, 476
Dupont, E., 137, 161 Fabre, P., 128, 161
516 AUTHOR INDEX
Fzibregues Soler, J. M. de, 392, 476 Flavier, H., 458, 459, 462, 478, 480
Fanelli, A. R., 13, 46 Fleury, P., 372, 373, 478
Fang, S. C., 11, 48 Flood, A. E., 281, 896
Fantoni, P., 361, 477 Florentin, D., 356, 478
Farfaletti-Casali, P. L., 380, 381, 382, 482 Florenzano, G., 411, 478
Farley, H. B., 444, 487 Flygare, H., 391, 479
Farnsteiner, K. Z., 106, 138, 162 Fontanelli, G., 390, 479
Farrell, L., 126, 164 Fonyo, A., 26, 46
Farrugia, A. J., 410, 477 Fonzes-Diacon, H., 407, 409, 479
Fatome, M., 372, 373, 477, 478 Fornachon, J. C . M., 383, 413, 415, 441,
Faulkner, M., 19, 29, 30, 35, 61 463, 479
Faure, A., 438, 441, 477 Forsyth, W. G., 294, 297
Fawcett, A. J., 309, 348 FOUCY, J., 407, 409, 479
Federico, L., 369, 477 Foulkes, E. C., 17, 46
Feduchy Marifio, E., 411, 414, 415, 491 Fraenkel-Conrat, H., 54, 94
Feigenbaum, J., 134, 162 Francis, W., 116, 162
Feigl, F., 438, 477 Franqot, P., 414, 415, 427, 479
Feigl, H. E., 438, 477 French, A. P., 281, 896
Feinberg, B. G., 61, 98 French, R. B., 11, 46
Fellenberg, Th. von, 366, 369, 372, 373, Freudenberg, K., 294, 297
375, 379, 385, 398, 400, 403, 404, Freyre, E., 448, 479
412, 419, 426, 429, 477 Friar, H. F., 127, 142, 160, 162, 164
Fellers, C. R., 54, 55, 82, 84, 86, 93, 104, Friedman, H. L., 59, 98
144, 169, 164 Friedman, M. E., 325,348
Fenton, F., 143, 162 Frolov-Bagreev, A. N., 355, 479
Ferguson, W. S., 276, 897 Fuller, E. C., 102, 162
Ferrari, C., 445, 477
Ferrari, V., 363, 477 G
Ferr6, L., 118, 162, 330, 548, 372, 373, Gad, L., 408, 409, 479, 486
374,376,378,389,390, 409, 410, 422, Gaddis, A. M., 4, 31, 32, 34, 39, 46, 46,
441, 471, 478 243, 245, 268
Fessler, J. H., 359, 361, 363, 438, 487 Gadzhiev, D. M., 387, 479
Fetzer, W. R., 425, 478 Gallot, S., 457, 499
Field, A., 103, 166, 166 Gammon, N., Jr., 196, 197, 198, 199,269
Field, L. K., 458, 493 Gangstad, E. O., 228, 230, 268
Filaudeau, G., 356, 478 Garber, J. D., 59, 91, 105, 147
Filer, L. J., 24, 46 Garcia Barcel6, J., 357, 492
Filipello, F., 371, 440, 478, 607 Garcla de Angulo, J. R., 448, 479
Finney, D. J., 207,967 Garino-Canina, E., 356, 357, 378, 380,
Fischer, L., 366, 367, 368, 483 381, 396, 406, 454,472, 479
Fischl, P. F., 363, 430, 478 Garoglio, P. G., 354, 356, 420, 480
Fischler, M., 114, 168, 365, 366, 478 Gatet, L., 387, 389, 394, 430, 4a6, 437,
Fishberg, E. H., 27, 46 458, 480, 481
Fisher, C. D., 86, 90, 92, 94, 99, 131, 132, Gaylord, F. C., 346, 348
133, 141, 168, 164, 168, 167 Geddes, W. F., 221, 222, 867
Fisher, R. A., 207, 230, 267 Gehman, H., 108, 110, 162
Fitelson, J., 120, 168 , Geise, C. E., 216,218,219,220,221,238,
Fitt, T. C., 116, 168 868
Flanzy, M., 357, 360, 361, 365, 366, 367, Geiss, W., 419, 428, 480
368,369,372,373,390,392, 396,402, Geissman, T. A., 262, 266, 270, 271, 280,
404, 407, 408, 412, 478, 604, 606 898, 297, 299
AUTHOR INDEX 517
Geloso, J., 413, 480 Gould, W. A., 244, 245, 246, 868
Genevois, L., 354,355,370,372,374,376, Goulden, C. H., 193, 249, 868
381,387,388,389,391,393,394,402, Gouveia, A. J. A. de, 458, 462, 604
405, 413, 415,417,423,431,432,437, Grandchamp, L.-E., 431, 433, 483
442,444,447,449,457,458,459,462, Granger, G. W., 309, 349
464, 465, 476, 480, 481 Granick, S., 14, 15, 46
Gentilini, L., 375, 415, 419, 420, 424, 433, Grant, G. A., 34, 46
445, 453, 481, 608 Grant, W. M., 116, 117, 123, 168
Geoffroy, P., 427, 479 Green, E. L., 143, 168
Georgacopoulos, MM., 419, 488 Green, R., 34, 61
Gerard, R. W., 34, 60 Greenberg, L. A., 17, 46
Gerasimov, M. A., 414,415,416,457,482 Greenwood, D. A., 30, 46
Gerum, J., 420, 482 Grettie, D. P., 24, 49
Gevorkian, Kh. S., 385, 603 Grivas, G., 429, 483
Gheorghiu-Vieriu, A., 436, 437, 482 Gross, C. R., 141, 166
Ghimicescu, G., 357, 373 377, 378, 392, Grossfeld, J., 413, 463, 483
395,404, 406,408, 419, 428,436, 437, Grossman, H. H., 341, 360
455, 456, 488, 606 Guadagni, D. G., 162
Gibbons, N. E., 31, 46 Gualdi, G., 438, 439, 498
Gibson, D. T., 56, 9.2 Gugnoni, S., 379, 468
Gibson, K. S., 342, 343, 548 GuimarsRs, A. F., 392, 483
Gibson, Q. H., 15, 46 Guittonneau, G., 381,483
Gierer, S., 369, 482 Gunstone, F. D., 4, 46
Gilder, H., 14, 46 Giinther, P., 366, 488
Gillespy, T. G., 125, 126, 168 Guntr, A. A., 390, 483
Gillett, T. R., 104, 162, 313, 314, 315, Guss, C. O., 86, 92
349, 360 Gutman, H. R., 15,46
Gillham, E. W. F., 116, 168 Guymon, J. F., 366, 369, 370, 371, 372,
Gillilland, R., 411, 474 448, 453, 472, 483
Gilpin, G. L., 226, 227, 869 Gvaladae, V. Z., 381, 483
Glasser, L. G., 344, 549
Gleim, E., 143, 168 H
Glickson, D., 144, 160
Haagen-Smit, A. J., 462,483
Gobis, L., 371, 380, 381, 382, 399, 482
Haas, V., 89, 96
Godet, C., 356, 399, 401, 422, 424, 428,
Haberstock, J., 404, 606
429, 453, 455, 456, 48.2
Haggard, H. W., 17, 46
Goes, L. A. de A., 415, 488
Hagglund, E., 57, 58, 66, 98, 106, 161
Goeser, P. A., 21, 36, 38, 39, 49 Hailer, E., 77, 92
Goldbach, N. J., 367, 48s Haldane, J., 17, 46
Goldring, L. S., 342, 349 Hall, G. O., 34, 46
Golumbic, C., 24, 25, 26, 46 Hall, J., 38, 46
Gomes, J. V. M., 362, 434, 485 Hall, J. L., 10, 12, 13, 60, 61
Gonzales de Rivera, C., 381, 394, 440, Hall, L. A., 26, 47
478 Hall, N. A., 366, 367, 368, 483
Goresline, H. E., 422, 48.3 Halverson, J. O., 5, 46
Gortner, R. A., 79, 98 Halvorson, H. O., 24, 48
Gortner, W. A., 10, 37, 61 Hamann, V., 126, 162
Got, N., 361, 483 Hamblin, A. J., 279, 298
Gottschalk, A., 421, 48s Hamburger, J. J., 87, 9.2,162
Gough, H. W., 204, 268 Hamence, J. H., 120, 161 .
518 AUTHOR INDEX
Hhmerle, W., 445, 493 Henriques, V. de F., 144, 145, 160,166

Hamner, K. C., 146, 167 Herrero, A., 424, 486
Hanak, A., 361, 483 Hewitt, J. T., 486
Hand, D. B., 324, 325, 326, 345, 360 Heyrovski, J., 357, 486
Hankins, 0. G.,4, 33, 39, 46, 243, 245, Hickinbotham, A. R., 358,374,376, 378,
868 403, 427,486
Hanna, C. H., 15, 60 Hilborn, M. T., 89, 96
Hans, S. T., 101, 169 Hilditch, T. P., 4, 24, 46, 46
Hanson, H. L., 6, 23, 27, 30, 38, 40, 46, Hillig, F., 391, 392, 404, 405,483,486
47,48, 250, 667 Hills, C. H., 147,161
Hanson, H. T., 7, 43, 46 Hills, G. L., 31, 46
Hanuii, J., 119, 166 Hiner, R. L., 4, 39, 46, 243, 245, 868
Harborne, J. B., 286, 696 Hinreiner, E., 262, 266, 997
Hardy, A. C., 308, 312,337, 349 Hirosawa, F. N., 462, 483
Hardy, F., 292, 697 Hite, J. P., 5, 6, 7, 46
Hare, G. H., 342, 349 Hlynka, I., 226, 868
Hargreaves, C. A., 398, 483 Hochberg, R. B., 414, 471
Harper, R. H., 22, 46 Hodge, J. E., 83,98, 104, 130, 163 ~
Harris, K. L., 242, 243, 868 Hoel, P. G., 213, 968

Harris, P. L., 6, 46 Hoffert, E., 24, 44
Harrison, C. J., 293, 897 Hoffmann-Ostenhof, O., 369, 488
Harrison, T. H. J., 274, 806 Hoffpauir, C. L., 116, 165
Hartmann, B. G., 391, 392, 431, 485 Hogl, O., 373, 486
Hartree, E. F., 17, 19, 47 Hohl, L. A., 87,98, 128,165,405, 412, 486
Hastings, A. B., 22, 36, 49 Holden, H. F., 14, 46
Hatch, M. B., 142, 160 Holley, A. D., 462, 486
Hattori, S., 287, 699 Holman, R. T., 5, 61
Hauptmann, K. H., 427, 428, 484 Holven, A. L., 313, 314, 315, 349
Haurowitz, F., 9, 14, 22, 37, 46 Holz, G., 386, 610
Hauser, F., 438, 440, 600 Holzbach, F., 390, 486
Hawes, R. C.,342, 349 Homeyer, P. G., 216, 218, 219, 220, 221,
Hawkins, J. M., 35, 44 868
Haynes, L. J., 277, 897 Honig, P., 99, 163
Hearne, J. F., 294, 807 Hopkins, D. P., 145, 160
Hedrick, V. P., 282, 807 Hopkins, E. W.,13, 27, 44, $6
Heide, C. von der, 356, 376,391, 406, 413, Hopkins, J. W., 18, 6.9,221, 222, 867
423, 429, 445, 484 Hopner, T., 66, 75, 98, 93
Heide, E., 444, 484 Horne, L. W., 6, 60
Heiduschka, A., 376, 378, 394, 398, 399, Horton, M. B.,38, 40,46
406,484 Horvai, R., 446, 489
Heilmann, A. S., 269, 898 Hostetler, E. H., 5, 46
Heitz, J. E., 358, 369, 371, 372, 483, 484, Houseman, E. E., 234, 867
498 Hove, E. L., 6, 12, 24, 46
Heiwinkel, H., 58, 66, 9% Hove, Z., 12, 24, 46
Hendel, C. E., 63, 88, 93, 114, 115, 119, Howard, L. B., 142, 164
143, 168, 164, 167 Howitt, M. K., 34, 60
Hendricks, S. B., 387, 430, 401 Hozumi, T., 357, 486
Hendrickson, R., 273, 292, 897 Huggings, R. A., 338, 346, 361
Hennig, K., 118, 168,355, 356, 376, 378, Hughes, A. H., 9, 46
384, 391, 394, 395, 396, 406, 416, 435, Hugues, E., 406, 407, 457, 486
450, 455, 456, 461, 484 Hulme, A. C., 281, 998
AUTHOR INDEX 519
Hunk&, B., 367, 486 Jeanpbtre, J., 409, 487
Huntenburg, W., 386, 486 Jendryaszek, L. J., 13, 46
Hunter, R. S., 324, 333, 336, 349 Jenness, L. C., 89, 96
Hurd, C. D., 86, 93 Jensen, H. R., 118, 163
Huriez, H., 394, 398, 401, 403, 486 Jensen, L. B., 15, 16, 17, 20, 21, 30, 31,
Hurwicz, H., 232, 268 34, 36, 38, 46, 47, 60
Husaini, 5. A., 21, 46 Jeroch, 0. P., 457, 487
Husfeld, A., 355, 486 Jewell, W. R., 104, 141, 149, 163
Hussein, A. A., 454, 486 Jilke, W., 363, 430, 487
Hustrulid, A., 4, 39, 60, 62 Jim6nez de Abeledo, M. E., 390,419, 487
Hutchins, M. C., 143, 162 Johns, B. G., 360, 473
Hutchinson, J., 286, 297 Johnson, C. M., 114, 115, 118, 119, 167
Johnson, D. K., 129, 163, 267, 281, 288,
I 397
Johnson, E., 458, 490
Ice, C. H., 289, 300 Johnson, G., 87, 94, 110, 114, 115, 119,
Ichikawa, N., 291, 297 129, 163, 167, 267, 281, 288, 297
Ikeda, R. M., 369, 609 Johnson, 0. C., 34, 47
Iliescu, L., 367, 473 Johnson, T., 57, 58, 93, 106, 162
Iljin, W. S., 392, 486 Jones, H. A., 283, 296
Imbrici, D., 409, 609 Jordan, R., 6, 60
Imm, L. W., 341, 348 Joslyn, M. A., 54, 77, 87, 92, 93, 102, 103,
Ingram, M., 31, 33, 46, 61, 62, 63, 75, 79, 104, 110, 116, 117, 118, 123, 126, 128,
80, 81, 82, 83, 84, 85, 93, 100, 106, 129, 132, 134, 137, 140, 146,147,149,
107, 115, 118, 122, 125, 126, 135, 163, 162, 163,164, 166, 169, 269,397, 355,
169 356,357,359,361,383,384,385,393,
Invernizzi, I., 363, 603 395,405,408,410,411, 412,413, 419,
Ionescu, M. V., 366, 408, 486 427, 434,440,441,444, 458,463,.467,
Iribarne, J. V., 383, 384, 486 486, 487, 491, 606
Irish, J. H., 124, 160, 161 Judd, D. B., 308,309,318,322,332, 337,
Irvine, G. N., 316, 317, 349 343, 349
Israelshvili, S., 134, 162 Jungkunz, R., 372, 378, 381, 499
Ivanov, B. V., 393, 486
Ivie, J. O., 116, 168 K
Ka, H., 293, 398
J Kaczmarek, A., 444, 487
Kaeske, H., 13, 17, 47
Jacob, H. E., 146,163, 160 Kalberer, 0. E., 419, 488
Jacobsen, A. E., 337, 349 Kalebin, M. I., 387, 602
Jacobson, G., 7, 47 Kalen, J., 38, 46
Jacquin, P., 373, 375, 390, 486 Karamboloff, N., 446, 488
Jahr, H., 424,486 Katar’zn, T. G., 356, 488
Jakovliv, G., 358, 486 Katz, M., 116, 164
James, T. H., 102, 163 Kauffman, W. R., 33, 46
Jandorf, B. J., 15, 46 Kaye, W., 341, 349
Janecke, H., 436, 437, 438, 440, 476 Kean, C., 357, 392, 491
Jauker, H., 367, 486 Keeney, M., 10, 11, 49
Jaulmes, P., 114, 118, 163, 356, 357, 358, Keilin, D., 17, 19, 47
361,362,363,373,383,390,402, 407, Kellie, W. F., 346, 348
408, 409, 410, 411, 418, 419, 423, 428, Kempthorne, O., 200, 207, 208, 209, 868,
429,431,479,486,487 269
520 AUTHOR INDEX
Kendall, M. G., 213, 268 Kraft, A. A., 27, 39, 47

Kent-Jones, D. W., 316, 317, S49 Kramer, A., 326, 338, S49
Kepner, R. E., 369, 609 Kramer, M., 458, 489
Keresztesy, J. C., 103, 149, 160 Kramer, O., 355, 397, 405, 407, 417, 420,
Kerp, W., 59, 60, 61, 82, 83, 9S, 106, 107, 486, 489
110, 114, 135, 138, 164 Krantz, F. A., 244,245,246,268,282,297
Kertesz, 2. I., 281, 284, 297, 299, 328, Krause, H., 404, 489
S61, 427, 610 Krauze, S., 381, 424, 489
Kesterson, J. W., 273, 292, 297 Kraybill, H. R., 5, 11, 24, 25, 29, 4S, 44,
Ketelbant, E., 407, 488 47,49
Khachidze, 0. T., 443, 476 Kretzdorn, H., 114, 162, 438, 489
Khakhina, L. P., 387,602 Krewson, C. F., 291, 897
Kielhofer, E., 355, 366, 419, 441, 451, Krieger, G., 436, 437, 438, 440, 476
452, 488 Krombach, H., 430, 489
Kiese, M., 13, 15, 17, 47 Kruisheer, C. I., 377, 379, 380, 381, 385,
Kieser, M. E., 145, 147,149,160,274,297 386, 421, 488, 489
Kiessling, W., 430, 488 Krukovsky, V. N., 7, 29, 43, 47
Kikuchi, G., 17, 47 Krupski, E., 366, 367, 368, 483
Kilbuck, J. H., 367, 427, 428, 474, 488 Kubista, V., 277, 898
Kitp, W., 370, 488 Kulesza, J., 285, 296
Kimmel, L., 103, 166 Kummerow, F. A., 5, 6, 7, 46, 47
King, H. H., 12, 13, 60 Kunkle, L. E., 21, 46
Kleber, J., 379, 381, 476 Kursanov, A. L., 450, 494, 496
Klinc, L., 413, 488 Kurtz, G. W., 10, 11, 49
Kline, E. A., 23, 31, 61 Kutal'ova, T., 464, 490
Kling, A., 392, 488
Klose, A. A., 6, 11, 23, 27, 30, 38, 47, 49 L
Kloxin, S. E., 5, 6, 7, 46
Knapp, A. W., 294, 897 Labhardt, H., 55, 94
Kniphorst, L. C. E., 377, 378, 379, 380, Lafon, M., 379, 380, 497
381, 385, 386, 421, 488, 489 Lafourcade, S., 140, 167, 457, 601
Knodt, C. B., 80, 96 Laganne, H., 429, 490
Knudson, L., 240, 268 Lagneau, C., 357, 490
Kobakhidze, M. C., 380, 488 Lakkopoulos, A. A., 387, 490
Koch, J., 383, 384, 404, 488, 489 Lambert, M., 373, 490
Kocsis, E. A., 446, 489 Lane, R. L., 458, 490
Kogan, A. J., 398, 489 Lange, W., 5, 47, 437, 490
Kohake, E., 326,348
Langkammerer, C. M., 56, 93
Kolthoff, I. M., 60, 61, 9.3, 102, 114, 164
Lantz, E. M., 204, 868
Komarik, S. L., 26, 47
Kondareff, M., 389, 441, 489, 494 La Rosa, W. V., 116, 164
Konek, F., 387, 489 Larsen, H. J., 390, 490
Kopal, S., 376, 378, 405, 489 Lauer, W. M., 56, 93
Koppanyi, T., 87, 96 Lavollay, J., 276, 298, 461, 490
Korotkevich, A. V., 359, 373, 391, 395, Lawrence, W. J. C., 279, 284, 298
442, 489 Lawshe, E. I., 369, 496
Kotodi, E. P., 287, 298 Lea, C. H., 3, 5, 10, 11, 12, 23, 27, 31, 33,
Kotonen, E., 366, 468 34, 37, 39, 47, 48
Kovalenko, V. I., 434, 496 Legault, R. R., 88, 93, 143, 162, 164
KozAk, K., 398, 603 Legge, J. W., 14, 15, 17, 48
Kozenko, E. M., 434, 489 Leggieri, C. L., 410, 490
AUTHOR INDEX 52 1
Legkov, P., 373, 490 Lyman, C. M., 22, 26, 43
Lehmann, B. T., 8, 12, 17, 19, 21, 28, 29, Lyon, A. V., 141, 149
30, 31, 33, 34, 48, 61
Lehmann, E., 419, 464, 490 M
Lemberg, R., 14, 15, 17, 46 Maaskant, L., 386, 491
Le Moal, A., 114, 169 MacAdam, D. L., 334, 341, 360
Leonoel, C. P., 366, 490 MacArthur, M., 128, 164
Lerner, N. H., 288, 296 McCharles, C. H., 357, 492
Lesage, L., 369, 492 McDonough, E. G., 273,698
Lesnovskaya, V. V., 458, 490 McFarlane, W. D., 10, 48
Lester, D., 17, 46 MacGillivray, J. H., 323, 360
Levy, L. P., 442, 490 McGinnis, R. A., 98, 99, 164
Lew, M., 127, 160 McGuine, T. H., 271, 699
Lewis, V. M., 54, 55, 82, 84, 86, 93, 104, Mackey, A. E., 11, 48
164 Mackinney, G., 88, 89, 96, 104, 114, 115,
Lewis, W. L., 30, 46 118, 119, 132, 133, 142,160, 163, 164,
Lherme, G., 359, 490 166, 166,167,168,281,298, 325, 326,
Li, J. C., 180, 181, 183, 268 348, 360, 441, 491
Lindet, L., 140, 164 Mackintosh, D. L., 10, 38, 46, 61
Lindsey, F. A., 25, 48 MacMillan, D., 321, 360
Lineweaver, H., 6, 23, 27, 38, 40, 46, 47, McNicholas, H. J., 318, 360
48 Mader, 376, 378, 401, 406, 491
Lippich, Fritz, 64, 65, 93 Magoon, C. A., 392, 469
Lips, A., 10, 48 Mahon, J. H., 24, 25, 48
Lipscomb, R. D., 59, 91, 105, 147 Major, R., 7, 10, 11, 48, 61
Little, R. C., 116, 164 Mallory, G. E., 425, 491
Little, T. M., 283, 296 Malsch, L., 448, 452, 494
Litwiller, E. M., 142, 160 Malvezin, P., 356, 430, 491
Livingston, A. L., 12, 43 Mandlen, H., 413, 429, 484
Livingston, G. E., 338, 349 Mangan, J. L., 114, 115, 119, 143, 166,
Lobstein, E., 394, 396, 401, 405, 424,490 167
Lochhead, A. G., 99, 126, 164 Manskaya, S. M., 435, 454, 491, 496
Lockwood, W. H., 17, 48 Mapson, L. W., 62, 94
Lodi, M., 458, 459, 460, 490 Maravalhas, N., 446, 491
Long, J. D., 86, 90, 92, 94, 131, 132, 133, Marcel, M., 361, 491
141, 162, 164, 166 Marches, J., 98, 100, 132, 166
Longenecker, H. E., 24, 46 Marcilla Arrazola, J., 356, 388, 411, 414,
Loosli, J. K., 7, 43 415, 491
Losa, V. M., 457, 609 Marcille, R., 356, 357, 409, 410, 491
Love, R. F., 425, 491 Mariani, A., 366, 491
Lovern, J. A,, 12, 22, 48 Marichal, M., 426, 494
Lowe, B., 38, 46, 177, 178, 179, 268 Marignan, R., 362, 402, 486, 491
Lucchetti, E., 377, 440, 490 Marimpietri, L., 418, 493
Luckow, C., 361, 490 Markh, A. T., 54, 94
Lueck, H., 118, 168 Markley, K. S., 387, 430, 491
Luers, H., 288, 298, 370, 490 Mars, C. V., 133, 160
Lugg, J. W. H., 28, 48 Marsh, G., 357, 392, 491
Lukton, A., 116, 164 Marsh, G. L., 54,90,93,94, 103, 134, 141,
Lundberg, W. O., 7, 24, 25, 35,43, 44, 46, 166,167, 169, 273, 278, 298, 325,348,
48 357, 359, 351,393,409, 413,459,462,
Liithi, H., 128,164,400,448,454,490,491 487, 491, 493
522 AUTHOR INDEX
Marsh, R. W., 136, 169 Mezzadroli, G., 464, 493

Martens, V., 226, 268 Mical, L., 359, 493
Martin, G. J., 276, 996, 298 Michel, A., 372, 373, 374, 376, 378, 386,
Martin, J., 312, 314, 361 404, 407, 422, 433, 478, 493
Martin, L., 422, 424, 453, 455, 456, 482 Michielini, L., 416, 474
Martin, W., 317, 349 Miconi, C., 365, 390, 409, 493
Martinaud, P., 137, 166 Miermeister, A., 413, 463, 483, 493
Martucci, J., 425, 498 Miller, C. V., 269, ,898
Marusawa, Den-Ichi Naito, 57, 58, 94 Miller, J., 120, 166
Marut& 8. A., 421, 424, 606 Miller, J. I., 37, 61
Mason, H. M., 102, 116, 166 Miller, L. C., 241, 268
Mason, H. S.,129, 166 Miller, R. C., 31, 44
Massey, F. J., Jr., 249, 267 Milligan, O., 6, 60
Mastbaum, H., 425, 49.8 Millikan, G. A., 12, 49
Mathers, A. P., 115, 116, 123, 166,385, Mills, D. R., 118, 140, 166
387, 392,393, 443,492 Mills, W. C., 168, ,869
Mathieu, G., 422, 492 Milner, B. I., 336, 360
Mathieu, L., 118, 137, 166 Milos, C., 425, 493
Matignon, C., 378, 492 Minz, S., 457, 493
Matlack, M. B., 292, 998 Mitchell, J. H., Jr., 99, 166
Mattil, K. F., 24, 46 Mitchell, J. S., 102, 166
Mattill, H. A., 11, 25, 26, 44, 46, 49 Miruno, G. R., 35, 44
MauriB, A., 437, 438, 492, 601 Moffett, A. A., 280, 296
Maveroff, M., 428, 607 Mohammad, A., 54, 94
Mavis, J., 244, 245, 246, 268 Mohler, H., 398, 400, 445, 493, 609
Maxwell, F., 98, 166 Moldovan, E., 367, 473
Maxwell, W. T., 25, 48 Mollenkopf, K., 373, 404, 476
May, P., 166 Monier-Williams, G. W., 106, 113, 115,
Mayer, M. M., 267, 281, 288, 297 118, 119, 120, 125, 166
Maynard, L.A., 7,43 Montequi, F., 444, 445, @3
Mayo, M. E., 33, 46 Montini, L., 408, 496
Mayr, F., 398, 399, 606 Mood, A. M., 213, 268
Meads, P. F., 313, 314, 315, 349, 360 Moon, H. H., 289, 296
Mecci, E. P., 6, 27, 30, 47 Moon, P., 336, 360
Mehlenbacker, V. C., 10, 60 Morani, V., 413, 417, 418, 493
Mehlitz, A., 147, 166, 427, 492 Moreau, L., 139, 166, 493
Melcher, B., 390, 492 Moreno Martin, F., 369, 473, 493
Melvin, E. H., 321, 360 Morgan, A. F., 7, 8, 27, 44, 49, 103, 166,
Menaul, P., 288, 298 166,457,458,459,460,461, 462,493,
Mendelejeff, D., 55, 94 496
MendivelaBa, G., 363, 419, 487, 498 Morley, M. C., 116, 166
Mensio, C., 137, 166 Morris, T. N., 127, 141, 166,270, ,296
Menael, H., 102, 114, 164 Morse, R. E., 168, 269
Merrill, A. L., 462, 609 Moser, H. A., 33, 44
Merzhangn, A. A., 434, 456, 492 Mosiashrili, G. I., 453, 476
Mestre Artigas, C., 357, 371, 383, 410, Moster, J. B., 319, 360
453, 492 Mott, 0. E. 369, 496
Mestre JanB, A., 371, 392, 453, 47'6, 492 Moubacher, R., 83, 96
Metcalfe, C. R., 287, 298 Moureu, H., 378, 492,493
Metra, M., 369, 492 Mrak, E. M., 86,88,89,90,92,94,96,99,
Meyerhof, O., 355, 492 128, 131, 132, 133, 141, 142, 145, 146,
AUTHOR INDEX 523
160,162, 163, 164, 166, 167, 168, 454, Norton, C. E., 144, 160
486 Nouty, A. H., 127, 161
Muers, M. M., 10, 61 Nussenbaum, F., 367, 428, 488
Mulder, H., 56, 59, 91
Miiller, M., 55, 94 0
Miiller-Thurgau, H., 79, 94, 124,137,166
Munro, F. L., 362, 494 O’Connor, R. T., 116,163
Munsell, H. E., 458, 476 ogilvy, w. s., 39, 49
Murray, H. D., 311, 360 O’Grady, T. J., 30, 43
Mursaeva, A. M., 393,493 Okano, K., 286, 898
Muth, F., 365, 430, 448, 452, 494 Okumura, E., 134, 160
Muttelet, C. F., 422, 424, 494 Olcott, H. S., 11, 24, 25, 46, 49, 54, 94
Mylne, A. M., 88, 93, 143, 164 O’Leary, D. K., 25, 49
Oliveira, A. J. de, 358,494
N Oliveira, F. P. de, 449, 451, 455, 456,494
Oliveira, H. T. de, 357,499
Nabenhauer, F. B., 282, 296 Oliver, A. W., 11, 48
Nagy, J. J., 11, 29, 49 Ollis, W. D., 286, 296
Naumann, H. D., 6, 38, 49 Olmo, H. P., 459, 460, 606
Navarro, E., 357, 377, 384, 393, 394, 396, Olson, R. L., 128, 146, 166
398,401,402, 403,405,406,416, 435, O’Neal, R., 428, 474
455, 456, 470 Onokhova, N. P., 457, 494
Nedeltscheff, N., 441, 494 Oparin, A. I., 435, 450, 451, 454, 494,
Negensev, I., 390, 609 496
N&gre, E., 426, 436, 439, 440, 494 Opperschaum, R. Z., 367, 483
Neish, A. C., 357, 358, 373, 490, 494 Oro L6pez, M., 421, 424, 429, 473
Nelson, E. K., 291, 298, 382, 464, 494 Osborn, G. H., 369, 496
Nestle, K. Th., 452, 604 Oshima, Y., 291, 293,298, 300
Neubauer, A., 463, 494 Oshke, P., 450, 484
Neuberg, C., 110, 138, 166 Osman, E. M., 108, 110,162
Newhall, S. M., 332,360 Osterwalder, A., 79, 94, 124, 137, 166,
Newton, R. C., 24, 49 375, 394, 454, 496
Newton, W., 362,494 Ostle, B., 193, 213, 221, 227, 229, 233,
Nichitovici, V., 367, 473 234, 249, 252, 268
Nichols, P. F., 88, 90, 94, 102, 103, 118, Otto, E., 7, 47
119, 133, 141, 142,166, 168,168, 458, Ournac, A., 392, 498
493 Overman, A., 181, 183, 268
Nicholson, J. F., 242, 243, 268
Nickcrson, D., 332,341, 344, 360 P
Nicolau, T., 114, 168
Niehaus, C. J., 360, 364, 449,-494 Palieri, M. G., 362, 409, 420, 496
Niell, J. M., 22, 36, 49 Palinkas, S., 355, 476
Nissen, B. H., 120, 168 Pallu, R., 438, 441, 477
NitschkB, E., 395, 398, 406, 494 Palmer, A. Z., 6, 38, 49
Niven, C. F., Jr., 17, 49 Pandhi, P. N., 145, 167
Noble, I., 39, 62 Parfent’ev, L. N., 434, 496
Nobles, H. L., 103, 166, 459, 462,493 Paris, G., 106, 138, 167
Noel, W. A., 141, 160, 166 Parisi, E., 426, 496
Noguero G6mez, E., 357, 494 Paronetto, L., 360,363,409,428,466,496
Nordstrom, C. G., 279,298 Parro, A. da C., 457, 496
Norris, F. A., 25, 49 Parrot, J. L., 276, 298
524 AUTHOR INDEX
Parson, T.,Jr., 342, 348 Phillips, R. J., 102, 167

Pasquini, G.,365,470 Piatkowska, K.,401, 498
PatanB, G.,292, 998 Piazza, J., 361, 498
Patnayak, K. C.,291,998 Picchinenna, D.,401, 471
Pato, M. d. S., 389, 392, 409, 416, 429, Piccoli, T.,367, 498
496 Picozzi, A., 409,498
Patterson, W.E., 412, 496 Pigman, W.W.,105, 130,161
Patton, S., 10, 11, 49 Piguet, G., 355, 469
Paul, D.L.,227, 229, 968 Pin, P., 398, 479
Pavelka, F.,408,496 Pitman, G.A., 102, 166,357, 492
Pavolv-Griehin, 5. I., 373, 496 Pitz, E. W.,Jr., 37,61
Payne, L. F.,8,60 Pluchon, J. P., 419, 498
Pease, V. A., 141,160 Plummer, A. W.,101, 167
Pederson, C. S.,126,167 Podkletnov, N. E.,391,498
Peluso, J. V.,437,609 Politova-Sovzenko, T.K.,410, 439, 498
Peng, D.,10, 12, 23, 31,61 Pollard, A., 147,149,274, $97
Peniston, Q. P., 65, 75, 78, 91 Ponte, A., 438, 439, 498
Pennell, R.B., 15, 49 Ponting, 3. D.,77, 87, 93, 94, 110, 114,
Penniman, W.B. D., 369, 496 115,119,128,129,16S, 167,269,997
Pentzer, D.J., 144, 160 Pool, A., 372, 498
Pentzer, W.T.,146, 148,167, 168 Pool, M.F.,10,11,49,114, 115,118,119,
Perard, J., 364,496 143,164, 167
Perazzo, A, A., 410, 496 Popescu, O.,366, 408,486
Percher, G.,429, 430, 498 Popov, A. D.,445,498
Perdigon, E., 379,496 Popova, E. M.,434, 454, 461, 498
Peretid, M.,378,416, 498 Poppoff, I. D., 440, 610
Perlman, L., 459, 460, 461, 462,496 Porchet, B., 124, 167
Perry, M.C.,124,167 Porter, T.,222, 223, 225,269
Perry, R. L., 90,94, 141, 142, 166, 167 Posternack, T.,442, 490
Petersen, R.B., 120, 166 Potter, E. F.,63,94, 114, 115, 119,167
Peterson, E. W.,341,360 Poux, C., 392,498
Peterson, J. E., 326, 327,348 Powers, J. J., 168,969
Peterson, M.S., 99, 166 Powers, R.,141, 166
Peterson, W.J., 182, 184, 190, 231, 232, Pozzi-Escot, E.,361, 408,498
248, 249, 969 Prado, L. de, 373, 376, 378,498
Petri, W., 355,496 Prahl, W.,56, 94
Pettigiani, A. E.,371, 496 Prange, G.,363, 498
Peynaud, E.,118,140,167,356,359,367, Prater, A. N., 10, 49, 114, 115, 118, 119,
372,373,374,376,377,378,379,380, 167,319,360
381,382,384,385,387,388,389,392, PrBlat, C. E., 363, 498
395,396,397,398,39D,400,401,402, Price, J. R., 279, 284, 998
403,404,405,406,407,408,410,411, Prillinger, F.,420, 439,464, 499
412,413,414,416,423,426,427,431, Pringsheim, Hans, 56, 94
432,433,435,448,455,456,457,461, Pritzker, J., 372, 373, 377, 378, 381, 499
464,465,4YS,476,480,481,496, 497, Pro, M.J., 437, 499
601 Procopio, M.,140,167,360,361,409,410,
Peyrot, E., 447, 497 420, 428, 433, 464,499
Pfab, H.,398,498 Prosstosserdov, N.N.,386, 499
Phaff, H.J., 90, 94, 132, 141, 142, 166, Pucher, G.W.,285, 899
167 Puchkova, M.G., 454, 461,498
Phillips, M., 143, 169 Pyriki, C.,376, 378, 394, 398, 406, 484
AUTHOR INDEX 525
Q RibBreau-Gayon, J., 118, 137, 140, 167,

354,356,359,374,376,379,380,381,
Quinn, D. G., 140, 167
382,385,387,389,395,396,397, 399,
Quinn, G., 89, 94, 141, 149, 167
400,402,405,406,410,413, 415,417,
R 423, 431, 432,433,437,438, 448,449,
451,454,457,461,464,465,47S, 476,
Raabe, W., 103, 149 480, 481, 497, 600, 601
Radmi6, S., 366, 603 Richard, O., 461, 610
Rahn, O., 70, 94, 124, 167 Richert, P. A., 124, 161
Rakcs&nyi, L., 355, 356, 421, 462, 476, Ricketts, J., 131, 167
606,606 Rideal, E. K., 9, 46
Ramos, M. da C., 357, 377, 382, 403, 499 Riedesel, M., 28, 49
Ramsay, W. N. M., 15, 49 Rieman, 0. H., 283,299
Ramsbottom, J. M., 10, 21, 36, 38, 39, Riemenschneider, R. W., 11, 24, 49, 60
49, 60 Riffart, H., 373, 404, 476, 606
Randoin, L., 457, 459, 499 Rinelli, W. R., 133, 160
Randolph, L. K., 242, 243, 268 Rippel, K., 396, 417, 601
Randolph, P., 7, 47 Ripper, M., 61, 94, 114, 117, 138, 168
Rangaswami, S., 291, 298 Rist, C. E., 83, 91
Rankine, B., 362, 600 Roberts, E. A. H., 268, 292,293,297, 299
Ransford, J. R., 324, 325, 326, 345, 360 Robinson, G. M., 282, 284, 288, 298, 299
Raschig, F., 56, 94 Robinson, M. E., 22, 60
Rasmussen, L. B., 88, 93, 143, 164 Robinson, R., 282, 284, 288, 198, 199,
Ravaz, L., 464, 600 442,490
Rayer, H. P., 145, 166 Robinson, W. B., 324, 325, 326, 345, 360
Read, H. M., 444,487 Roche, M., 383, 601
Redtenbacker, J., 55, 94 Rockwood, B. N., 10, 49
Reed, H. M., 88, 94, 118, 119, 142, 166 Rocques, J., 361, 601
Reed, I. W., 146, 167 Rocques, M. X., 55, 94
Reese, H. D., 363, 610 Rodopulo, A., 381, 483
Reich, P., 454, 600 Rodopulo, A. K., 387,394,400, 407,454,
Reichard, O., 355, 398,399,401, 435,448, 601,609
455, 600 Roelens, E., 362, 369, 470, 476
Reid, H. H., 269, 299 Roessler, E. B., 358, 484
Reid, V. W., 364, 600 Roleff, H., 399, 452, 602
Reifer, I., 114, 115, 119, 167 Roleson, E. P., 141, 168
Reiman, W., 168, 267 Romano, E., 380, 381, 602
Reinking, K., 55, 94 Rooke, E. A., 21, 62
Reis, A. M. L. de C., 377, 382, 403, 431, ROOS,L., 441, 602
434, 435,499,600 ROOS,W., 372, 373, 607
Reiser, R., 23, 49 Rose, D., 31, 46
Remy, E., 364, 376, 455, 600 Rose, D. H., 146, 168
Rentschler, H., 395, 396, 409, 438, 440, Rosenblatt, D. H., 443, 606
454,600,606 Rosenblatt, M., 437, 602
Requinyi, G., 355, 454, 476, 600 Rosenfeld, P. M., 434, 602
Reuhle, A. E., 103, 160 Rosenthaler, L., 369, 602
Reynolds, H., 226, 227,969 Rosenwald, R. H., 24, 60
Rezabek, H., 24, 47 Rosoff, H., 87, 92
Ribeiro, E. C., 377, 402, 403,474 Ross, A. F., 89, 96, 143, 168
Ribeiro, M. de B., 414,415,416,419,422, Ross, E. S., 39, 62
474,600 Ross, 5.L., 440, 602
526 AUTHOR INDEX
Rothenfusser, S., 115, 168 Schelhorn, M. V., 78, 96

Rottsieper, E. H.W.,287, 899 Scheuer, M.,427, 458, 462,498,603
Roubert, J., 431, 608 Schiff, H.,55, 96
Roussopoulos, N. C.,414,428, 608 Schindler, J., 398, 603
Rouzaut, R., 361,498, 608 Schmechel, S., 391,489
Roy, M.,330,348, 441, 471 Schmid, C.,452,604
R6zsa, P.,363,604 Schmidt, 394, 397, 401, 405, 424, 490
Ruf, W.,358,445,606 Schmitthenner, F.,355,604
Ruspini, Amoldo, 391,608 Schneyder, J., 446, 610
Russell, A.,287, 899 Schoeneman, R. L.,385, 393, 498
Rustia, A., 364,608 Schofield, R. K.,340,360
Rusznyhk, I., 275, 896 Schon, K.,458, 462,604
Rutgers, R., 373, 489 Schonberg, A., 83,96
Ryan, V. J., 358, 374, 376, 378, 486 Schrauffstatter, E.,277,899
Schreiber, M.L., 8, 60
S Schroeter, G. W.,55,96
Schulek, E.,363, 400,604
SaakLn, R. G., 385, 453, 454, 459, 460, Schultz, H.W.,21, 36,38, 39, 49
461, 462,606 Schulze, G.,401, 404, 413,489
Saburov, N. V.,387,606 Schumakov, A.,375,604
Saenko, N.F.,450,463,494,496,606,603 Schwab, A. W.,33,44
St. Mokranjac, M.,366, 603 Schwartz, V., 24, 47
Sair, L.,30,60 Schwerin, P.,9, 22, 48
Salani, R., 425,603 Scofield, F.,333,360
Sallusto, F.,376, 377, 378, 394, 403,406, Scott-Moncrieff, R., 279,898
603 Scott, T.,425,604
Salvador, A. R. N.,409, 496 SCU~CO, U.,376,378, 403, 406,603
Salvarezza, M., 410,603 Seifert, W.,354, 412, 604
Sampaio, A. V. de, 436,603 Seikel, M.K.,280,899
Sampietro, C., 363,603 Seiler, F.,366,391,394,404,405, 406,604
Sando, C.E., 281,284,899,387,430,491 Sbmichon,L.,356,357,360,361,366,372,
Sane, R. H.,168, 869 373,390,392,402,404,407,408,412,
Sannino, F. A., 354, 603 426,604,606
Sapondzhian, S. O.,385, 603 Senti, F. R., 321,360
Sasaki, R., 115, 168 SBrgio, R. J. de R., 394, 398, 474
Sasao, H.,367, 486 Seshadri, T.R., 291,898
Sastry, L. V. L.,338, 360, 442, 444,603 Sethi, S. C.,35,60
Sata, S., 291,899 Sevestre, J., 276, 898, 461,490
SatB, M.,357, 486 SBze, R. de, 411, 606
Satterfield, G.H.,458, 489 Sgargi, L.,464, 493
Satterthwaite, F. E.,191,869 Shah, J. N.,329, 338,360
Sattler, L.,312, 314, 361 Shalaiken, F.P., 74, 96
Saunderson, J. L.,336, 341,360 Shapiro, 0.W.,11, 61
Savage, C. G., 141,149 Sharpe, R. H.,196, 197, 198, 199,869
Savary, M.,364,603 Shcherbakov, M.F.,448,463,606
Scarborough, D.A., 12,29, 60 Shelor, E.,35, 38,48
Scarborough, H.,276, 277,899 Shenk, J. H.,12, 13,60
Schaefer, F.,451,603 Shimamoto, T.,463,606
Schanderl, H.,355, 391, 396, 405, 411, Shimokoriyama, M.,287, 899
415, 417, 421, 440, 449, 451, 459, Shiner, R. L.,282,899
463,603 Shinoda, J., 291,899
AUTHOR INDEX 527
Staley, K. A., 309, 315, 361
Sholes, M. L., 142, 161 Stansby, M. E., 10, 60
Shreve, R. N., 100,168 gfastnjr, J., 357, 486
Shrewsbury, C. L., 6, 60 Stauber, J., 288, 898
Sidersky, D., 420, 421, 606 Steedman, J., 147, 149, 160
Sieve, B. F., 276, 899 Steffen, G., 13, 27, 34, 44
Sifredi, A. V., 141, 161 Steigmann, A., 116, 117, 123, 168
Silander, S., 106, 168 Steinberg, M. P., 4, 39, 60
Silberstein, L., 366, 367, 368, 469 Stelling, O., 56, 96
Sills, V. E., 145,160 Stellwaag, F., 355, 486
Simenson, L. O., 101, 148 Stenger, V. A., 60, 61, 93
Simmler, H., 409, 600 Stepanenko, B. N., 64, 65, 96
Simon, F. P., 34, 60 Stepanov, A. V., 64, 65, 96
Simone, R., 25, 60 Steuart, D. W., 391, 606
Sirianni, E., 457, 493 Stewart, C. P., 276, 899
SisakGn, N. M., 385, 421, 424, 448, 453, Stewart, T. D., 65, 66, 74, 96
454, 459, 460, 461, 462, 606 Stickney, M. E., 342, 349
Skaggs, S. R., 80, 96 Stier, T. J. B., 374, 471
Skok, J., 170, 171, 869 Stirton, A. J., 11, 60
Sliaewica, L., 390, 487 Strachan, C. C., 127, 136, 144, 148
Smardzewska, I., 285, 196 Stradelli, A., 365, 606
Smit, C. J. B., 110, 163 Strassner, J. E., 65, 75, 78, 98
Smith, D. C., 369, 496 Straub, J., 421, 606
Smith, E. E., 126, 169 Strohecker, R., 404, 606
Smith, E. G., 144, 168 Strong, F. M., 460, 607
Smith, F. H., 6, 26, 34, 37, 43, 60 Stumpf, R. R., 326, 327, 348
Smith, M. B., 459, 460,606 Sturrock, W., 309, 315, 361
Smith, T. J., 338, 346, 361 SuEevib, O.,416,606
Smith, W. E., 15, 49 Sudario, E., 356, 442, 606
Smoler, J., 357, 486 Sugayama, J., 463, 606
Smrecayliska, A., 401, 498 Sulser, H., 357, 606
Snedecor, G. W., 170, 177, 178, 182, 193, Sulzbacher, M., 55, 96
233, 249, 869 Sulabacher, W. L., 33, 46
Snell, R. S., 228, 230, 868 Sumner, R. J., 10, 60
Solms, J., 426, 427, 606 Sumuleanu, C., 114, 168, 357, 377, 378,
Sommer, H., 399, 484 406, 455, 456, 606
Sondheimer, E., 284, 899, 328, 361 Sundman, J., 56, 62, 64, 65, 66, 67, 69, 70,
So&, E., 355, 476 72, 74, 77, 78, 96
S O ~ SI.,
, 356, 454, 462, 600, 606 Suter, C. M., 105, 107, 168
Sorber, D. G., 89, 90, 98, 96, 142, 149, Svereva, E., 120, 169
160,168 Svershkov, I. V., 115, 168
souci, S.W., 126, 168 Swaby, R. J., 393, 606
Sousa, T. T. de, 416, 496 Swain, T., 273, 275, 279, 896, 896, 298
Spaeth, E. C., 443, 606 Swingle, W. T., 290, 899
Spannuth, H. T., 271, 899 Seab6, I., 421, 606
Sparks, R. A., 326, 348 Szent-Gyorgyi, A., 275, 896
Spencer, D. E., 336, 360
Spicer, S. S., 15, 60 T
Srab6, E., 355, 476
Stadtman, E. R., 54,77,85,87,88,89, 91, Tabachnick, J., 434, 606
96, 103, 128, 129, 133, 142, 168 Taguchi, T., 111, 169
528 AUTHOR INDEX
Takeya, D., 17, 60 Tressler, D. K., 31, 35, 44, 126, 128, 134
Talburt, W. F., 88, 93, 143, 162, 164 167, 169
Tanner, F. W., 77, 96, 123, 168 Trevor, J. S., 33, 60
Tanner, H., 395, 606 Troland, L. T., 302, $661
Tanret, C., 291, 299 Trost, F., 369, 607
Tappel, A. L., 23, 24, 60 Troy, D. J., 344, S49
Taranova, R. D., 386, 443,499, 609 Truelove, R. K., 364, 600
Tarantola, C., 383, 398, 399, 400, 401, Tseng, K-F., 291, 300
410, 423,429, 437, 449,454,468,606, Tsujimura, M., 293, 300
607 Tucker, H. P., 182, 184, 190, 231, 232,
Tarbet, D. S., 103, 148 248, 249, 269
Tartaglia, A., 361, 607 Tucker, L. N., 6, 26, 37, 38, 43, 49
Tartar, H. V., 56, 92 Turbovsky, M. W., 440,607
Taub, A., 25, 60 Turer, J., 11, 60
Tiiufel, K., 398, 399, 606 Turner, A., 315, 348
Tauhe, K., 226, 227, 269 Twight, E. H., 362, 607
Tavernier, J., 373,375,381,390,438,439,
471, 483, 486 U
Taylor, G., 120, 161
Taylor, L. V., Jr., 118, 168 Uchida, Jun-Ichi, 57, 58, 94
Teixeirs J b i o r , E. do V., 383, 384, 607 Uchimoto, D., 374, 383, 405, 411, 428,
Teply, L. J., 460, 607 433, 474, 607
Ter-Karapetian, M. A., 383, 607 Ulbrich, M., 412, 604
Terry, R. A., 276, 297 Underwood, E. J., 276,296
Testa, J., 428, 607 Urbain, W. M., 15, 16, 17, 20, 21, 30, 31,
Thaler, H., 372, 373, 607 36, 38, 39, 46, 60
Theorell, H., 12, 14, 60 Urban, H., 57, 58, 92, 106, 162
Thomas, M. D., 116, 168 Urone, P. F., 116, 169
Thompson, A. F., 57, 96
Thompson, D. L., 277, 299 V
Thompson, J. B., 63, 96, 120, 168
Thompson, J. F., 143, 162, 168 Vail, G. E., 5, 6, 8, 10, 11, 38, 46, 47, 60,
Thomson, P., 319, 321, 361 61
Tillitson, E. W., 59, 96 Valaer, P., 356, 357, 425, 608
Tischer, R. G., 207, 208, 209, 214, 215, Valaize, H., 448, 449, 454, 455, 456, 463,
216,218,219,220,221,227,229, 232, 607
233,267,268,269, 338,360, 442, 444, Van der Plank, J. E., 146, 169
605 Van Holten, P., 142, 168
Toennies, G., 116, 169 Vartan’& M. D., 419,608
Toepfer, E. W., 226, 227, 269 Vas, K., 60, 61, 62, 66, 71, 73, 75, 78, 80,
Tomaghelli, A. A., 363, 432, 607 81, 82, 83, 84, 93, 96, 100, 106, 107,
Tomescu, F., 367, 47s 111, 112, 122, 125, 135,163,169
Tomoda, Y., 108, 109, 111, 169 Vasconcellos e Lencastre, A. Q. de, 372,
Torley, D., 376, 387, 394, 398, 403, 406, 373, 377, 378, 402, 436, 441, 608
607 Vaughn, R. H., 126,169
Torricelli, A., 423, 425, 607 Vecher, A. S., 458,490
Touplain, F., 362, 470 Vedani, A., 366,47S
Toy, E., 63, 96, 120, 168 Vegezzi, G., 369, 602
Trauth, F., 365, 420, 468, 607 Veisman, S., 120, 169
Trawich, J. L., 242, 243, 268 Veitch, F. P., Jr., 87, 93
Treadwray, R. H., 128, 140,166 Velazquez, E., 403, 404, 405, 608
AUTHOR INDEX 529
Venesia, M., 361, 375, 424, 445, 453, 454, Watson, R. H., 13, 61
458, 608 Watt, B. K., 462, 609
Ventre, J., 137, 161, 354, 366, 391, 407: Watt, D. B., 10, 61
411, 608 Watts, B. M., 5, 6, 7, 8, 9, 10, 11, 12, 17,
Verda, A., 400, 458, 608 19, 21, 23, 24, 26, 27, 28, 29, 30, 31,
Vergnes, P., 392, 418, 472 32, 33, 34, 35, 38, 44, 48, 49, 60, 61
Vestal, C. M., 6, 60 Weast, C. A., 144, 145, 169, 326, 360,
Vetsch, U., 448, 454, 491 411,474
Vetscher, A. S., 363, 430, 457, 608, 609 Webb, A. D., 369,374,375,378,467,609
Vibrans, F. C., 11, 24, 49 Weinberger, J. H., 281, 300
Vickery, H. B., 285, 299, 398, 483 Weise, R., 444, 487
Viiles, P., 390, 609 Weiss, T. J., 34, 61
Vilas, M. A., 394, 4'74 Weissberger, A., 102, 163
Vilece, R. J., 338, 349 Wellington, R., 281, 300
Vfi'Grns, v. V., 443, 609 Wendel, W. B., 14,44
Villforth, F., 383, 412, 461, 484, 609 Wender, S. H., 282, 289, 300,447, 610
Vinet, E., 139, 166,493 Werkman, C. H., 379, 471
Vinogradova, N. I., 457, 482 West, D. B., 433, 609
Violante, C., 394, 406, 409, 442, 445, 609 Westerman, B. D., 38, 46
Vitte, G., 385, 473 Westkamp, N. E., 6, 60
Vivian, J. E., 101, 169 Wettstein, E., 355, 387, 476, 489
Vivino, A. E., 87, 93 Wharton, M. A., 222, 223, 225, 269
Vogt, E., 354, 365, 371, 405, 441, 609 Wheeler, D. H., 382, 494
Vollaire-Salva, J., 431, 433, 483 Wheeler, L. B., 355, 467
Volmar, Y., 390, 609 Whipple, S. R., 346, 961
Vols, F. E., 10, 37, 61 White, D. E., 276, 296
von Loesecke, H. W., 141, 169 White, W. H., 11, 31, 32, 34, 38, 39, 44,
von Schelhorn, M., 123, 126, 169 46, 61,62
VoNiek, J., 119, 166 Whitmore, F. C., 57, 96
Vorstman, N. J. N., 385, 386, 489 Whitney, R. P., 101, 169
VoskoboInikov, I., 390, 406, 609 Widmer, A., 404, 609
Widmer, C., 5, 61
W Wiegand, E. H., 80, 96, 118, 140, 142,
144,149, 166,169,160,338,561, 447,
Wakeley, J. T., 182, 184, 190, 248, 249, 610
,969 Wiens, A., 103, 166, 459, 462, 493
Wald, G., 337, 361 Wiesman, C. K., 27, 31, 61
Walker, W. O., 133, 160 Wiesner, K., 65, 96
Wallace, T., 136, 169 WikBn, T., 461, 610
Walsh, G., 115, 166 Wiklund, O., 314, 361
"alters, W. P., 10, 61 Wilbur, K. M., 11, 61
Walton, C. F., Jr., 315, 348 Wilder, 0. H. M., 5, 43
Wanderstock, J. J., 27, 47 Wiley, H. W., 98, 160
Wang, T. H., 462, 483 Wilharm, G., 386, 610
Wanner, E., 140, 169 Willaman, J. J., 147, 161, 427, 610
Warcollier, G., 114, 169 Williams, A. H., 274, 281, 282, 288, 289,
Ward, A. C., 129, 149 296,197, 300
Warneford, F. H. S., 292, 297 Williams, B. L., 447, 610
Warren, B. J. W., 119, 149 Williams, J., 103, 160
Waser, E., 445, 609 Williams, J. L., 427, 486
Waterman, R. E., 103, 160 Williams, M. B., 363, 610
530 AUTHOR INDEX
Williams, R., 24, 49 Y

Williams, R. R., 103, 149, 160,458, 490
Willson, K.S., 133, 160 Yamada, M., 134, 160, 370, 610
Willsttitter, R., 282,300 Yamamoto, R.,291, 300
Wilson, C.P., 147,160 Yamashita, T.,291, 897
Winegarden, H.M.,38, 40,46 Yang, H. Y., 80,96, 140,160,447,610
Winkler, A. J., 103, 146, 160, 330, 331, Yenson, M.M.,22, 46
348, 361, 388,389,391,393,395, 396, Young, F. M.,337, 349
441,442,459, 462, 467,493,610 Young, H.A., 132, 148
Winkler, C. A., 18, 21, 30,39, 68 Younkin, S. G., 324, 325,326, 346,361
Winter, F. H.,126,169
Winter, J. D.,4,39, 60,66 2
Wirthle, F.,401, 406, 610
Wishnetsky, T.,324,325,326, 345,360 Zalesskaya, M.I., 370,610
Wobisch, F.,446,610 Zeglin, H.,367,610
Wohler, P.,59, 93 Zeisset, W.,376, 406,429, 484
Woidich, K.,120,160 Zerban, F. W., 104, 160, 312, 314, 348,
Wolfrorn, M.L., 86,96 361
Wong, R.,9,29, 33,61 Zheltkevich, G. G.,361,610
Wood, D.J., 268, 292, 293,899 Ziemba, J. V., 27, 31,61
Woodcock, A. H.,32,68 Zimmermann, H.W., 363,610
Woodroof, J. G., 35,38,4S, 123,133,145, Zion, J. R., 141,161
160 Zlataroff, A., 440,610
Worthington, 0. J., 329, 338,360,561 Zollinger, E.H.,282, 300
Wright, S. B., 30, 43 Zsolt, J., 356, 462, 606
Wyrnan, J., Jr., 14, 68 Zweigelt, F.,420, 610
Subject Index
A glycosides of quercetin, 281
leuco-anthocyanins, 288
Acetaldehyde sulfonate, 108, 111 preparation for bakers’ use, 128, 146
Acetobacter preservation of pomace, 147
acetylmethylcarbinol production in sulfiting, 132
wines, 381 Apricots, dried
2,3 butylene glycol production in color change in, 88
wines, 381 inhibition of browning, 129
Alcohols restoration of color, 104
ethyl alcohol in wine, 359 resulfuring with liquid SO2, 133
glycerol content of wines, 376 storage deterioration, 142
glycerol ratios in wines, 378 storage of, 103
glycerol in wines, 372, 373, source in use of SOs, 141
wines, 374 Apricot sirups
higher alcohols in wines, 367, content in rate of browning, 54
wines, 371 Ascorbic acid, destruction, 269
in wines, 358 content of turnip green leaves, 184
mechanism of formation, 370 in wine, 457
methyl alcohol, 366, in wines, 368 oxidation of, 103
source of alcohol in wines, 367 protection by citric acid and SOs, 87
source of glycerol in wines, 379 use in lard, 29
yield, 364 use in meats, 25, 27
Almonds, Leuco-anthocyanins, 288 Asparagus, tannin, 289
Alpha hydroysulfonic acid, 104
combined with glucose, 110 B
Anthocyanins, 263, 283
bleaching by SO2, 104 Bacon, antioxidants, use of, 26
in edible fruits, 284 brightness, 18
in strawberries, 328 color data, analysis of variance, 195
Anthoxanthins, 285 Wiltshire, thawing and curing pro-
Antioxidants, 3, 24 cedure, 194
claesification of, 24 Barley, leuco-anthocyanin, 288
for color protection, 27 Beans, dried, vitamin content, 204
mode of action on fat, 24, 25 green, carotene content of processed,
use on meats, 26 225
Apple juice runner, leuco-anthocyanins, 288
browning of, 54 Beef, canned
combination of sulfur dioxide, 79 correlation coefficients between quality
Apples factors, 243
anthocyanins, 284 moisture content and processing tem-
dried, use of SOz, 141 perature, 237
flavonoids, 280 Beef, fat rancidity, 4, 5
frozen, 112 fatty acid content, 6
531
532 SUBJECT INDEX
Beer, determination of total SOa, 122 of variance, 208

Beet, root, red pigment betanin, 277, correlation of quality of raw, frozen,
284 canned, 244
Bisulfite addition products, 104 moisture content, 217, 219
Blackberries, sulfiting, 145 per cent sulfur content from methionine,
Blackberry wine, detection of grape cystine, 229
wine in, 357 titration and temperature values in
Brazil nuts, leuco-anthocyanins, 288 moisture determination, 215
Byssochlamys fulva, growth inhibition varieties and crosses, analysis of vari-
by SO,, 125, 126 ance of puncture tests, 201-203
Corn syrup, in canning, 146
C Cottonseed oil, color of, 321
Coumarin, 263
Cabbage, analysis for SO2, 119 Cured meats, discolored by light, 36
sulfiting, 142 effect of pH, 31
Cacoa, fermentation, 295 greening, 21
types, 294 nitrite burn, 20
Candied products, desulfiting, 127 surface fading, 21
Carrots, analyses for SOZ, 119 Currants, brining, 145
sulfiting, 142
Catsup, color specifications for, 321 D
Chard, carotene content of processed, 225
Cherries, Maraschino, 98 Desulfiting, 127
“barrelling,” 143, 144 Doughnuts, fat absorbed by batter, 177
brining, 143, 144 Dried fruits, analyses for SOz, 120, 121
combined SO2, 139 sulfured, glucose-bisulfite reaction
dehydration of, 142 products in, 59
desulfiting, 127 use of SOz, 141
Cherry juice, combined SOz, 139 Dried vegetables, browning pigments in,
Chestnuts, leuco-anthrocyanins, 288 88
Chicken, fatty acid, 6
Chili, color range for, 320 E
Chlorophyll, conversion to pheophytin,
104 Elderberry, color detection in grape wine,
disappearance of, 326 446
Chocolate, bitterness, 274
Cider, blending for astringency, 273 F
Citron, bleaching, 98
brining, 144 Fat, absorbed by batter of doughnuts,
Citrus, fruits, flavonoid compounds, 289 177
terminology, 290, 291 effect of salts on bacon fat, 32
Citrus juices, darkening of, 87 evaluating oxidative changes, 9, 10,
oxidation of ascorbic acid in, 103 11
preservative action of sulfurous acid in meat, 8
on, 62 mean scores for flavor of pastry made
sulfiting, 134 with different, 181
sulfur dioxide, combining power of, 83, rancidity, 1
135 scores for pastry made with different,
Citrus peel, bleaching, 98 183
Coconuts, preservation with SO2, 146 unsaturated, oxidation of, 22
Corn, canned sweet, consistency, analysis Fatty acids, content in animals, 6
SUBJECT INDEX 533
influence of rations, 6 Grapefruit juice concentrate, yeast infec-
oxidation of, 8 tions, 79
Fish, SO2 preservation, 98 Grapefruit peel, brining, 144
Flavonoid compounds in foods, 261 Grapes, anthocyanin pigment in skins,
action on smooth muscle, 276 282
antioxidant action, 271 flavonoids, 282
color, 271 leuco-anthocyanins, 288
fat in the body, 277 nitrogenous compounds, 448
genetic situation in fruits and vege- sulfiting for transportation, 146
tables, 280 Groundnuts, leuco-anthocyanins, 288
naturally occurring, 264
pharmacological action, 275 H
properties depending on phenolic char-
acter, 268 Hams, curing brines, 31
properties significant in foods, 268 Honey, color specifications, 315
reactions with metals, 270 Horse-radish, volatile organic sulfur
relation to bloat in ruminants, 276 compounds, 120
systematic distribution, 283
taste, 273
tendencies in distribution, 289
toxicity, 277
Lard, effect of ascorbic acid, 29
Flour, color grader, development of, 317
oxidation of, 29
correlation of wheat tests for insects to
Lemon, use of juice for capillary resist-
fragment count, 243 ance, 276
whiteness in, 311
Lemon peel, brining, 144
yellowness of, 316
disappearance of chlorophyll, 326
Frankfurters, gray color, 18
Loquats, leuco-anthocyanins, 288
use of polyphosphates to preserve
color, 39
Fruit juices, 127 M
sulfiting, 134
Fruit pulps, brined, 98 Maple syrup, color specifications, 315
desufiting, 127 Meats, color changes, 21
sulfiting, 136 cured, 16
Fruits, brined, 98 deterioration of ground, 23
sulfur dioxide absorption, 89 discolorations, 2, 12, 14, 30, 31
evaluating oxidative changes in fats, 9
G greening, 16
keeping quality, 9
Gallate, alkyl, 271 myoglobin derivatives, 20
esters, 271 normal pigments, 12, 13
Garlic, volatile organic sulfur compounds, off-flavors, 2
120 oxidative rancidity, 3
Glaced products, desulfiting, 127 Milk powder, bacterial plate count,
Grape juice, browning of, 54 analysis of variance, 205
coloI of, 311 Molasses, analyses for SO2, 121
combination of sulfur dioxide, 79
destruction of antineuritic activity by N
SOa, 103
free SO2, 139 Nectarines, use of SOZ,141
Grapefruit, bitterness, 273 Nordihydroguaiaretic acid, 271
534 SUBJECT INDEX
0 Pimientos, distribution of pH measure-

ments, 168
Oils, color of, 317-321 Pork, effect of pH during curing, 31
yellowness in, 311 extract, effect on fat, 22
Onions, flavones, 270-283 fat rancidity, 4, 5
volatile organic sulfur compounds, 120 fatty acids, 6
Orange juice, discoloration of, 54 sausage, rancidity, 23
effects of sulfur dioxide addition to, 77 Potatoes, analyses for SO,,59
iodine-reducing substances in, 63 discoloration during dehydration, 89
use of 802 to inhibiting browning, 85 flavonoids, 282
Orange juice concentrate, SO* binding preparation of prepeeled, 128, 146
power, 126 sulfiting for drying, 142, 143
yeast infections, 79 Poultry, effect of temperature, 38
Orange peel, brining, 144 fat rancidity, 4, 6
Oranges, sour, astringency, 273 fatty acids, 6
P skin fat, 7
use of antioxidant, 26
Packaging, meats, 38, 39 Protein, denaturing agents, 19
Paprika, color range for, 320 Pumpkin, protection of ascorbic acid in,
Pastry, flavor scores with different fats, 87
181-183 PurBes, frozen fruit, 87
Peaches, barrelling, 144
firming agent, 145 R
frozen, discoloration of, 87
Rabbit, stability of fat, 7
tannin, 281
Raisins, chlorophyll in, 104
use of SO, in dried, 141
use of SOnin golden bleach, 141
Pear juice, combining with sulfur dioxide,
use of SOnin sulfur bleach, 141
79
Rancidity, effect of light, 36
Pears, color in stewed, 272
effect of meat constituents, 30
use of SO2 in dried, 141
effect on metals, 33
Peas, greenness of, 326-3281
effect on oxygen, 36
Pecan, trees, mineral composition of
effect on pH, 30
foliage, 197
effect on physical factors, 36
Peppers, red Hungarian, in capillary
effect on salts, 31
resistance, 276
effect on smoke, 34
pH, effect on discoloration of meat, 30
effect on spices, 35
effect of on galactose and glucose bisul-
effect on temperature, 37
fite equilibrium constants, 67-71
of fats, 1, 4
effect on preservative action of SO,
of meats, 3
124
oxidative, 4
effect on rancidity, 30
Rice polish extracts, 103
of wines, 416
destruction of antineuritic activity by
Pigments, derivatives of myoglobin, 20
Son, 103
heme, 2, 14
hemoglobin, oxidation of, 22 S
in meat, 12, 13
of cured meats, 17 Sacoharomyces ellipsoideus, growth in-
oxidation products of heme pigments, hibition by Son, 126
13 production of citric acid, 399
oxyhemoglobin, absorption spectrum, production of 2,3-butylene glycol in
28 wine, 379
SUBJECT INDEX 535
Sausage, rancidity in cooked, 23 total, 119
rancidity in raw, 23 use in dried fruits and vegetables, 141
Smoke, effect on rancidity, 34 Sulfur house, construction, 142
Sorghum, leuco-anthocyanin, 288 operation, 142
Soybean oil, color of, 321 Sulfurous acid, application, 132
Spices, capsicum, color of, 310
effect on rancidity, 35 T
Spinach, greenness of, 326-328
raw and cooked, carotene concentra- Tea, catechins, flavones, 292
tion, 223-225 fermentation, 293
Strawberries, firming agent, 145 relation of flavonoids to quality, 293
Strawberry preserves and jams, color of, Tomato juice, color specifications, 321
328-330 comparative composition of, 171
darkening of, 311 Tomato paste, color specifications for,
Sugar, fermentation data, analysis of 32 1
variance, 188 Tomato pulp (puree), color specifications,
whiteness of, 311-316 321
Sugar-Bisulfite addition compounds, ana- Tomatoes, color specifications, 321-326
lytical procedures, 59 grading of, 345
bleaching, 104, 146 redness of, 307-311
nature of, 55 Tripe, yellow color, 270
reaction equilibrium constant, 63 Turkeys, fatty acid, 5
reaction velocity constants, 71 monocarbonyl compounds in fat, 11
sulfitation, 98 peroxides in fat, 11
Sulfites, bound, 125 stability of eviscerated, 6, 7
chemistry of, 99 tocopherol in tissue, 7
Sulfur, burners, 100 use of antioxidants, 27
Sulfur dioxide, analyses for, 122 Turnip, leaves, ascorbic acid and mois-
application, 130, 134 ture content of, 184
as a preservative, 123 loss of moisture, 231
as a sanitizing agent, 123, 126 moisture content and size of, 248
bleaching action, 103
brining of cherries, 143 v
chemistry of, 99
distillation procedures for total, 119 Vegetables, chlorophyll-containing, green-
effect of variations in, on the equi- ness of, 326
librium constant of glucose bisul- dehydrated, 142
fite, 73 greenness in, 311
effect on heat processing, 126 shredded, 146
free, 114 sulfiting, 143
fumes, 130 Vinegar, use of SO*, 136, 141
in “barrelling” fruits, 143
inhibition of fermentation by, 77 W
inhibition of nonensymatic browning
by, 54 Walnuts, leuco-anthocyanins, 288
in transportation and storage of Watermelon, rind, bleaching, 98
fruits, 146 Wheat, hard red spring, moisture con-
in wine, 117 tent, 222
liquid, application, 133 Wines, acetal in, 385
preservation of fruits and vegetables acetaldehyde in, 382, 384
with, 53 acetone in, 385
536 SUBJECT INDEX
acetylmethylcarbinol in various types, nitrogenous compounds, 448, 449,

382 452, 454, 455, 456
acid changes during storage, 397 nitrogenous compounds in clouding,
acids in, 386, 388 45 1
acrolein in, 386 organic constituents, 353
aging, 393, 444 oversulfiting, 140
alcohols and related compounds, 358 pantothenic acid in, 460
aldehydes and related compounds, 382 pentoses and related compounds, 423
aromatic constituents in, 454, 460 pH of, 416
ascorbic acid in, 457 polyhydroxyphenoIs, 435
benealdehyde in, 385 problems in Germany, 355
buffer capacity, 417 problems in Hungary, 355
carbohydrates in, 418 pyridoxine in, 460
chromatography of, 357 riboflavin in, 459
citric acid, 397, 401 sophistication, detection, 494
color of, 330, 441 succinic acid in, 402, 403
constituents in white and port, 358 sucrose in, 424
content of 2,3-butylene glycol, 381 tannins in, 436, 437, 439, 440
determination of total SOz, 122, 123 tartaric acid in, 391, 394, 415
enzymes in, 454 thiamin in, 458
esters, 430, 435 titratable acidity, 390
extract, 428 use of SO2, 136, 137
fining, 444 vitamins in, 454, 456, 462
hexosis in, 419 volatile acidity, 407, 412
hydroxymethylfurfural in, 385
lactic acid in, 403, 405 z
malic acid, 395, 398
mannitol in, 425 Zucca melon, bleaching, 98
methods of analyses, 356 Zygosaccharomyces, growth inhibition
nicotinic acid in, 460 by SO,, 125
ADVANCES IN FOOD RESEARCH
Volume I
E. C. BATE-SMITH, The Physiology and Chemistry of Rigor Mortis, with Special
Reference to the Aging of Beef.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I
L. E. CLIFCORN, Factors Influencing the Vitamin Content of Canned Foods. . . . 39
SAMUEL LEPKOVSKY, The Physiological Basis of Voluntary Food Intake (Appe-
tite?) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
HOWARD D. LIQHTBODY and HARRY L. FEVOLD, Biochemical Factors Influencing
the Shelf Life of Dried Whole Eggs and Means for Their Control.. . . . . . . . . 149
BELLELOWE,Factors Affecting the Palatability of Poultry with Emphasis on His-
tological Post-Mortem Changes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
A. FRANK Ross, Deterioration of Processed Potatoes.. . . . . . . . . . . . . . . . . . . . . . . 257
G. FRED SOMERS and KENNETH C. BEESON, The Influence of Climate and Ferti-
lizer Practices upon the Vitamin and Mineral Content of Vegetables.. . . . . . 291
EARLR. STADTMAN, Nonenzymatic Browning in Fruit Products.. . . . . . . . . . . . . . 325
ORVILLE WYSS,Microbial Inhibition by Food Preservatives. . . . . . . . . . . . . . . . . . 373
GEORQE L. BAKER,High-Polymer Pectins and Their Deesterification.. . . . . . . . . 395
Volume I1
GEORQE E. FELTON, Ion ExchangeApplication by the Food Industry
C. R. STUMBO, Thermobacteriology As Applied to Food Processing. .
CECILGORDON DUNN,The Quaternary Ammonium Compounds and Their Uses
in the Food Industry.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
ARNOLDJ. LEHMAN, The Pharmacology of DDT ....................... 201
MILDRED M. BOWSand HELENL. HANSON, Anal f Foods by Sensory Differ-
ence Tests.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
JUSTUS G. KIRCHNER, The Chemistry of Fruit and Vegetable Flavors 259
T. ELLIOT WEIER and C. RALPHSTOCKINQ, Histological Changes In
Fruits and Vegetables by Processing. . .................. 297
G. A. REAYand J. M. SHEWAN, The Spo sh and Its Preservation by
Chilling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
EDWARD SELTZER, Spray Drying of Foods.. . . . . . . . . . . . . . . . . . 399
Volume I11
M. A. JOSLYN and J. D. PONTING, Enzyme-Catalyzed Oxidative Browning of
Fruit Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
S. T. COULTER, ROBERTJENNESS, and W. F. GEDDES,Physical and Chemical
Aspects of the Production, Storage, and Utility of Dry Milk Products.. . . . 45
BERNARD E. PROCTOR and SAMUEL A. GOLDBLITH, Electromagnetic Radiation
Fundamentals and Their Applications in Food Technology. . . . . . . . . . . . . . . 119
A. J. LEHMAN,. 0. G. FITZHUGH, A. A. NELSON and G. WOODARD, The Pharmaco-
logical Evaluation of Antioxidants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
W. R. HINSHAW and ETHEL MCNEIL,Salmonella Infection as a Food Industry
Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
J. P. DANEHY and W. W. PIQMAN, Reactions between Sugars and Nitrogenous
Compounds and Their Relationship to Certain Food Problems.. . . . . . . . . . . 241
537
538 ADVANCES IN FOOD RESEARCH
JOHN A . ULFUCHand H . 0. HALVORSON. Chemical and Microbial Studies on

Sliced Canned Bacon ................................................. 291
R . R . HARTWELL. Certain Aspects of Internal Corrosion in Tin Plate Containers . . 327
RICHARD . .
L D MORSE,Rationale for Studies of Consumer Food Preferences ...... 385
MATHILDEI VON SCHELHORN, Control of Microorganisms Causing Spoilage in Fruit
and Vegetable Products .............................................. 429
Volume IV
N . E. GIBBONS. WiltshireBacon .......................................... 1
LOUISE. DAVIS, Work Methods Design and Work Simplification. . . . . . . . . . . . . . 37
SAMUEL LEPXOVBXY, Nutritional Stress Factors and Food Processing .......... 105
S. ARONOFF,The Chemistry of Chlorophyll (with Special Reference to Foods) . . . . 133
B. F . DAUBERT and PAULW. O’CONYELL. Reversion Problems in Edible Fats 185
A . G. VAN VEEN,Fish Preservation in Southeast Asia ........................ 209
ELDON E . RICEand JACK F . BEUK.The Effects of Heat upon the Nutritive Value
of Protein .......................................................... 233
.
A E . MICHELBACHER. Insects Attacking Stored Products .................... 281
J. C. BAUERNFEIND. The Use of Ascorbic Acid in Processing Foods . . . . . . . . . . . . 359

Advances in Food Research Vol 5 Compress

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advances in Food Research Vol 5 Compress

Uploaded by

Copyright:

Available Formats

ADVANCES IN FOOD RESEARCH

ALL RIQHTS RESERVED

Library of Congress Catalog Card Number: 48-7808

PRINTED I N T H E UNITED STATES OF AMERICA

Publication of Volume V provides a convenient and timely oppor-

September, 1954 G. F. STEWART

Statistical Methods in Food Research

Oxygen Tension. ...

oxidation of the heme pigments, resulting in discoloration. The two are

might be expected to affect rancidity, and special problems in the control

1. The Nature of the Oxidative Process

d. Species Variations in the Susceptibility of Animal Fats to Oxidation

deterioration in beef and lamb as well as pork in freezer storage to be due

Klose et al., 1951) demonstrate conclusively the deposition of large

of body fats by a n indirect effect on metabolic processes. Hite et al. (1949)

with various aqueous solutions or absorbed on filter paper (relatively

The reaction of oxidation products of linolenic acid with thiobarbituric

The most commonly encountered type of discoloration is th a t of the

injection. Much of this work has been reviewed by Granick (1949),

"\ ' 0 /nr

3. The Pigments of Cured Meats; Their Oxidation

capable of reducing both methemoglobin and nitrite are not present

FIG.2. Absorption spectra of compounds formed upon addition of nitrite t o oxy-

FIQ.3. Derivatives of myoglobin of importance in meats. In the outer circle are

Nitric oxide hemoglobin and the corresponding denatured hemo-

4. Methods for Investigation of Color Changes in Meat

IV. THE COUPLEDOXIDATIONOF HEMOGLOBIN

I n addition to the independent oxidation of unsaturated fats and of

The mechanism of the reaction has not been extensively studied.

oxidation. Lea (1937), who first observed this, attributed it to a lipoxidase.

is the active catalyst, but presumably inactivates it by rendering it

1. Classification and Mode of Action of Fat Antioxidants

of lard brought about by an antioxidant mixture of BHA and propyl

0.005% (Lindsey and Maxwell, 1949) ; phosphoric acid 0.005% (Eckey,

aqueous bath in which various antioxidants are dispersed. A patent

FIG.5. Changes in the absorption spectrum of oxyhemoglobin brought about by

Figure 5 shows the absorption spectra obtained when ascorbic acid

Retardation rather than acceleration of rancidity occurs if the level of

VI. EFFECTOF VARIOUS MEAT CONSTITUENTS

(1943) observed this protective effect on Wiltshire bacon. Watts and

(1930) recommends the smoking of individual slices of bacon for more

ently light can bring about a dissociation of nitric oxide hemoglobin

rate of oxidation of cured meat pigment is progressively increased with

Calkins, V. P., and Mattill, H. A. 1944. Kinetics of the antioxygenic synergism of

Haldane, J. 1901. The red color of salted meat. J . Hyg. 1, 115.

Robinson, M. E. 1924. Hemoglobin and methemoglobin as oxidative catalysts.

Summarization of current knowledge of sugar-bisulfite addition com-

bisulfite addition compounds with carbonyls as tripartite molecules with

With deuterium as the aldehyde hydrogen in the molecule, Thompson

a condition thought to be best explained by way of the sulfonic acid struc-

The S-containing compounds produced by Marusawa and Uchida

A and B contain primary alcohol groups conjugated with secondary (or,

taining carbohydrate derivatives of this sort, although no claim was made

concentration, hydrogen-ion content of the system, and specific char-

The reaction system consisting of reducing sugar and sodium bisulfite

wherein S represents total uncombined sugar, B total uncombined sul-

determination of free sugar, the majority of investigators in this field have

Xylose 0.0688 203 14

In their investigations with benzaldehyde and bisulfite, they also

condition has made it evident to later workers th a t all testing should be

At 20" C. (68" F.) and with glucose concentrations of approximately

At 30' C. (86" F.) after a reaction time of 30 hours the (apparent)

in Table IV are those obtained immediately on mixing; the systems were

The relative constancy of K calculated in this manner is strongly sug-