Professional Documents
Culture Documents
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.
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.
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
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
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.
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.
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 9
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).
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 11
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
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
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
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 17
(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).
(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
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
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.
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.
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
OXIDATIVE RANCIDITY AND DISCOLORATION I N M E A T 27
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
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.
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.
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
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 31
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
OXIDATIVE RANCIDITY A N D DISCOLORATION I N MEAT 33
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
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-
OXIDATIVE RANCIDITY A N D DISCOLORATION IN MEAT 37
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-
OXIDATIVE RANCIDITY A N D DISCOLORATION IN MEAT 41
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
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.
OXIDATIVE RANCIDITY A N D DISCOLORATION I N MEAT 43
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
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.
J . Am. Oil Chemists’ SOC.28, 475.
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.
OXIDATIVE RANCIDITY AND DISCOLORATION I N MEAT 49
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
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
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
H
SUOAR-SULFITE REACTION A N D RELATIONSHIP TO FOOD PROBLEMS 57
mannose reaction at 130" C. (266" F.). The series of reactions was written
as possibly being :
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
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
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
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-
-
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.
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-
SUGAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 71
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
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.
74 HARRY GEHMAN AND ELIZABETH M. OSMAN
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'
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
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
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
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
SUGAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 91
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.
92 HARRY GEHMAN A N D ELIZABETH M. OSMAN
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.
SUGAR-SULFITE REACTION AND RELATIONSHIP TO FOOD PROBLEMS 95
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
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
PH
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:
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
R- :$I/8
- -0Na + HC1-t R- 8+
H
I NaCl + S 0 2 t + HzOT
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:
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 109
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
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.
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 117
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.
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
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 119
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
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 121
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
122 M. A. JOSLYN AND J. B. 9. BRAVERMAN
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-
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 123
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
124 M. A. JOSLYN AND J. B. S. BRAVERMAN
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
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 125
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
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
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
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
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.
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
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
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 139
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
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
144 M. A. JOSLYN AND J. B. 8. BRAVERMAN
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
146 M. A. JOSLYN AND J. B. S. BRAVERMAN
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
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
Research 3, 241-290.
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.
152 M. A. JOSLYN AND J. B. 8. BRAVERMAN
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.
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 157
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
Research 1, 325-372.
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.
SULFUR DIOXIDE TREATMENT OF FRUIT AND VEGETABLE PRODUCTS 159
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
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
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
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
STATISTICAL METHODS IN FOOD RESEARCH 165
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
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.
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
STATISTICAL METHODS IN FOOD RESEARCH 167
FIG.la. Hypothetical distribution curve for normal population (mean and standard
deviation intervals shown).
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
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).
STATISTICAL METHODS IN FOOD RESEARCH 169
(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
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:
= 21.27 T 2.002(0.20) =
where
Interval estimate =
“.I
Lz
= (PI - Pz)+
-
t(l-r)(nl+na-2)~I,-P,
/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. 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.
STATISTICAL METHODS IN FOOD RESEARCH 173
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
174 B. OSTLE AND ROBERT G . TISCHER
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
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
ni
STATISTICAL METHODS IN FOOD RESEARCH 175
mental unit having been made entirely by chance (at random). If we
denote the observation recorded from the experimental unit subjected
jt'l
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,,
TABLEV
Analysis of Variance for Among and Within Treatments (Unequal Numbers of
Observations for Each Treatment)
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)
Among treatments - I
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
STATISTICAL METHODS IN FOOD RESEARCH 177
+ (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
STATISTICAL METHODS IN FOOD RESEARCH 179
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 )
Degrees
of
Source of variation freedom Sum of squares Mean square
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 ....
Sum of deviations from mean 10.8 10.4 5.4 9.6 7.2 43.4
Judge E
Mean score 4.3 4.3 4.6 3.6 2.9 ....
Sum of deviations from mean 4.2 5.6 4.8 4.8 5.4 24.8
Judge H
Mean score 4.5 3.8 3.1 2.1 1.8 ....
Sum of deviations from mean 5.0 4.8 1.8 1.8 3.2 16.6
* 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
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
184 B. OSTLE AND ROBERT 0. TISCHER
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.
STATISTICAL METHODS IN FOOD RESEARCH 185
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;
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
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:
=
1
11
ni
i - 1 j=l
S,,2/nij -
i=l
Ti2/ 2
j=l
n,,
= 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
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;
N - 22
f ni
i=lj-1
n,j2/N
c1 = t
and
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) :
STATISTICAL METHODS IN FOOD RESEARCH 191
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
STATISTICAL METHODS IN FOOD RESEARCH 193
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
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
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
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
22
b 1
Yij2- T2/bt
i=l j=1
Buu = block sum of squares
2
i-1
Bi2/t - T2/bt
3
-
4
5
-
6
7
-
8
-
9
-
10
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
Source of variation freedom Sum of squares Mean square
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.
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
Source of variation freedom Sum of squares Mean square
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.
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
STATISTICAL METHODS I N FOOD RESEARCH 209
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
1 From Tischer and Kempthorne (1951).
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
1 From Tischer and Kempthorne (1951).
210 B. OSTLE AND ROBERT G. TISCHER
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
STATISTICAL METHODS IN FOOD RESEARCH 21 1
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
212 B. OSTLE AND ROBERT Q. TISCHER
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).
STATISTICAL METHODS IN FOOD RESEARCH 213
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
i=l
2 ( Y 1- Pi)*
where
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.
2. t = (a - (Yo)/S,
214 B. OSTLE AND ROBERT Q. TISCHER
and
4
i=l
and
where
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).
-
-
Mean square -
-
- 1850.9524 (3.5893) - (78.9990) -
-
1850.9524
216 B. OSTLE AND ROBERT Q. TISCHER
-
--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
Source of variation freedom Sum of squares Mean square
Due to regression 1 3.3716 3.3716
Deviation about regression 19 0.2176 0.0114
Total 20 3.5893
1 From Tischer (1962).
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)
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(
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
220 B. OSTLE AND ROBERT Q. TISCHER
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
VACUUM OVEN MOISTURE (PER CENT)
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
222 B. OSTLE AND ROBERT G. TISCHER
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).
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.
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
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.
224 B. OSTLE AND ROBERT G. TISCHER
Correlation coefficient = T
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
226 B. OSTLE AND ROBERT G. TISCHER
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
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
228 B. OSTLE AND ROBERT G. TISCHER
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-
232 B. OSTLE AND ROBERT G. TISCHER
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.
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.)
rz,l, = 0.3675**
r,,, = 0.2700*
rZav= -0.2117
rz,a,= -0.659**
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
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
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).
STATISTICAL METHODS IN FOOD RESEARCH 239
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:
0 .
* *
.
0..
. 0
0 .
.*
0 .
A B c
FIQ.17. Illustration of three types of correlation between two variables (hypo-
thetical data).
xi - 8 ) ( Y , - F)I2
-
- W)Z 2 (Yi
n
i= 1
- Y)2
242 B. OSTLE AND ROBERT G. TISCHER
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,,
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).
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-
STATISTICAL METHODS IN FOOD RESEARCH 245
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)
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
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
STATISTICAL METHODS I N FOOD RESEARCH 249
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,
250 B. OSTLE AND ROBERT Q. TISCHER
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:
STATISTICAL METHODS IN FOOD RESEARCH 25 1
r.(1
= (").' - 7r1)n-r + ;4 l ) T 1 r + y l - T1)"+1 + * . . + TI"
= c
W=r
(;JT1Yl - ?rl)n--w (4)
= 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.
STATISTICAL METHODS I N FOOD RESEARCH 253
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
256 B. OSTLE AND ROBERT Q. TISCHER
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
STATISTICAL METHODS IN FOOD RESEARCH 257
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.
258 B. OSTLE AND ROBERT 0. TISCHER
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.
STATISTICAL METHODS I N FOOD RESEARCH 259
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.
This Page Intentionally Left Blank
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
a. Color . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. Tea.. . . . . .
..........................
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
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
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
266 E. C. BATE-SMITH
XIII XIV
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.
268 E. C. BATE-SMITH
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
OH CO
acid, and DHA dehydroascorbic acid (cf. Bate-Smith and Morris, 1952).
270 E. C. BATE-SMITH
w-
C H
XVII
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
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 :-
XX
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.
FLAVONOID COMPOUNDS I N FOODS 279
OH
XXI
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.
284 E. C. BATE-SMITH
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
FLAVONOID COMPOUNDS I N FOODS 287
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
288 E. C. BATE-SMITH
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,
FLAVONOID COMPOUNDS I N FOODS 289
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.
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.
290 E. C. BATE-SMITH
Auranetin Tangeretin
Nobiletin Ponkanetin
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.
292 E. C . BATE-SMITH
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
294 E. C . BATE-SMITH
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.
FLAVONOID COMPOUNDS I N FOODS 295
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
‘pu r p I c s
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.
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
THE COLOR PROBLEM IN FOODS 307
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
THE COLOR PROBLEM IN FOODS 309
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.
THE COLOR PROBLEM IN FOODS 315
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
THE COLOR PROBLEM IN FOODS 317
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
318 G. MACKINNEY AND C. 0. CHICHESTER
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).
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
THE COLOR PROBLEM IN FOODS 323
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
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)
THE COLOR PROBLEM IN FOODS 327
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
328 0. MACKINNEY AND C. 0. CHICHESTER
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
THE COLOR PROBLEM IN FOODS 329
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.
330 G. MACKINNEY AND C. 0. CHICHESTER
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
332 G. MACKINNEY AND C. 0. CHICHESTER
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
THE COLOR PROBLEM IN FOODS 333
a= +
0.5710~ 1.2447y - 0.5708
+ +
1.0000~ 2 . 2 6 3 3 ~ 1.1054’
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.
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.
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
336 G. MACKINNEY A N D C. 0. CHICHESTER
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
THE COLOR PROBLEM IN FOODS 341
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.
THE COLOR PROBLEM IN FOODS 349
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.
J . Opt. SOC.Amer. 39, 460.
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.
J . Opt. SOC.Amer. 39, 945.
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
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
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.
COMPOSITION OF WINES 365
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-
COMPOSITION OF WINES 367
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).
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.
COMPOSITION OF WINES 369
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.
370 MAYNARD A. AMERINE
(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
372 MAYNARD A. AMERINE
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-
COMPOSITION O F WINES 373
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
374 MAYNARD A. AMERINE
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
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.
376 MAYNARD A. AMERINE
TABLE I11
Glycerol Content of Table and Dessert Wines
(Gramsper 100 ml.)
No. of Mini- Maxi- Aver-
Region Type of wine Source of data samples mum mum age
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)
Germany Table Mader (1936) 8 0.68 1.13 0.89
(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)
Italy Table Sallusto (1938- 32 0.62 0.88 0.79
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)
COMPOSITION OF WINES 377
TABLEI11 (Continued)
No. of Mini- Maxi- Aver-
Region Type of wine Source of data samples mum mum age
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.
TABLEIV
Alcohol/Glycerol Ratios (by Weight) of Table Wines
No. of Mini- Maxi- Aver-
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)
Germany Table Mader (1936) 11 8.4 12.2 10.0
(sugared)
Germany Table Mader (1936) 8 7.9 11.2 9.4
(unsugared)
Germany Table Hennig (1952) 11 8.3 15.9 11.4
Italy Table Sallusto (1936- 77 6.5 10.0 8.1
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
COMPOSITION OF WINES 379
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.
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
386 MAYNARD A. AMERINE
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,
COMPOSITION O F WINES 387
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
COMPOSITION O F WINES 389
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
390 MAYNARD A. A M E R I N E
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
COMPOSITION O F WINES 391
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.
392 MAYNARD A. AMERINE
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.
396 MAYNARD A. AMERINE
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.
---
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).
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.)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
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
Spain Fino Bobadilla and Navarro 15 0.08 0.13 0.09
(1952)
Spain Oloroso Bobadilla and Navarro 10 0.07 0.17 0.12
(1 952)
Switzerland Wh. table NitschkB (1952) 16 0.01 0.73 0.20
Switzerland Table Berner (1952) 23 0.01 0.33 0.12
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
COMPOSITION O F WINES 399
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
400 MAYNARD A. AMERINE
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.
402 MAYNARD A. AMERINE
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
COMPOSITION OF WINES 403
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.)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
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
(Nante) Wh. table Huriez (1948) 11 0.06 0.11 0.09
France Swt. table Peynaud (1950a) 6 0.090 0.130 0.112
France Red table Peynaud (1950a) 6 0.099 0.116 0.105
France Dessert Peynaud (1950b) 8 0.019 0.101 0.077
Hungary Wh. table Torley (1942) 10 0.081 0.207 0.120
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
Spain Fino Bobadilla and Navarro 15 0.079 0.140 0.111
(1952)
Spain Oloroso Bobadilla and Navarro 10 0.094 0.142 0.114
(1952)
Switzerland Table Berner (1952) 13 0.047 0.097 0.064
* 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
404 MAYNARD A. AMERINE
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.
COMPOSITION O F WINES 405
TABLHXI1 (Continued)
No. of Mini- Maxi- Aver-
Region Type of wine Source of data samples mum mum age
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)
France Table Peynaud (1950a) 12 0.07 0.25 0.17
Germany Wh. table Heiduschka and 29 0.06 0.46 0.28
Pyriki (1930)
Germany Wh. table Seiler (1933) 51 0.05 0.42 0.13
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
Germany Wh. table Seiler (1935) 114 0.03 0.34 0.084
Germany Table Mader (1936) 11 0.06 0.19 0.10
(sugared)
Germany Table Mader (1936) 8 0.07 0.14 0.09
(unsugared)
Germany Wh. table Seiler (1936b) 71 0.02 0.32 0.20
Germany Wh. table Seiler (1937) 66 0.03 0.28 0.07
Germany Wh. table Seiler (1944) 128 0.05 0.38 0.14
Germany Table Hennig (1952) 11 0.08 0.25 0.16
Germany Wh. table Seiler (1952) 24 0.06 0.61 0.15
Hungary Wh. table Torley (1942) 10 0.00 0.33 0.19
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
Switzerland Table Berner (1952) 24 0.063 0.367 0.217
Turkey Table Akman (1951) 66 0.02 0.16 0.07
* Diseased wines..
COMPOSITION OF WINES 407
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
COMPOSITION OF WINES 409
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
COMPOSITION O F WINES 411
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
412 MAYNARD A. A M E R I N E
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).
COMPOSITION OF WINES 413
",:,,
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
COMPOSITION O F WINES 415
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
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
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
France Table Peynaud (1950a) 12 3.24 3.56 3.37
France Dessert Peynaud (1950b) 8 3.07 3.40 3.31
France Sparkling Hennig (1952) 5 2.76 3.06 2.29
Germany Sparkling Hennig (1952) 19 2.81 3.10 2.96
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
Spain Fino Bobadilla and Navarro 15 3.22 3.63 3.34
(1952)
Spain Oloroso Bobadilla and Navarro 10 3.19 3.42 3.28
(1952)
Switzerland Table Berner (1952) 25 2.95 3.75 3.31
* 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
COMPOSITION OF WINES 417
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
418 MAYNARD A. AMERINE
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
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
420 MAYNARD A. AMERINE
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.
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.
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
COMPOSITION OF WINES 425
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).
426 MAYNARD A. AMERINE
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.
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
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.) ;
430 MAYNARD A. AMERINE
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.
COMPOSITION OF WINES 43 1
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
432 MAYNARD A. AMERINE
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,
COMPOSITION O F WINES 433
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
434 MAYNARD A. AMERINE
TABLEXV
Total Volatile Neutral Esters in Various Types of Wines
(Milligrams per liter as ethyl acetate)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
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 Red table Peynaud (1950a) 6 132 229 188
France Swt. table Peynaud (1950a) 6 141 274 211
Spain Sherry Bobadilla and Navarro 25 50 840 344
(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
436 MAYNARD A. AMERINE
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.
COMPOSITION OF WINES 437
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
438 MAYNARD A . A M E R I N E
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.
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-
440 MAYNARD A. AMERINE
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,
442 MAYNARD A . A M E R I N E
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.
COMPOSITION O F WINES 443
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-
COMPOSITION OF WINES 445
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
446 MAYNARD A . A M E R I N E
"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
COMPOSITION O F WINES 447
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,
448 MAYNARD A. AMERINE
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
COMPOSITION OF WINES 449
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-
COMPOSITION OF WINES 45 1
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
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.
COMPOSITION OF WINES 455
TABLEXIX
Total Organic Nitrogen Content in Various Types of Wine
(Grams per liter)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
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
France Dessert Peynaud (1950b) 8 0.147 0.256 0.191
France Sparkling Hennig (1952) 14 0.248 0.514 0.379
Germany Wh. table Hennig (1944) 19 0.314 0.734 0.493
Germany Wh. table Reichard (1943) 50 0.560 0.980 0.766
Germany Wh. table Remy (1932) 10 0.294 0.908 0.491
Germany Sparkling Hennig (1952) 43 0.102 0.666 0.262
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)
Spain Fino Bobadilla and Navarro 15 0.110 0.318 0.209
(1952)
Spain Oloroso Bobadilla and Navarro 10 0.200 0.318 0.261
(1952)
Switzerland Table Godet and Martin (1946) 3 0.111 0.240 0.156
Switzerland Table Berner (1952) 8 0.420 0.825 0.590
TABLE XX
Amino Nitrogen in Various Types of Wine
(Grams per liter)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
Bordeaux Red table Peynaud (1939b) 35 0.041 0.131 0.080
Bordeaux Wh. table Peynaud (1939b) 36 0.010 0.067 0.034
France Red table Peynaud (1950a) 12 0.021 0.055 0.038
France Table Valaize and Dupont (1951) 73 0.059 0.165 0.083
France Dessert Peynaud (1950b) 8 0.021 0.070 0.039
Germany Wh. table Hennig (1944) 19 0.068 0.196 0.108
Portugal Dessert Oliveira (1942) 58 0.018 0.061 0.032
TABLEXXI
h i d e Nitrogen in Various Types of Wine
(Grams per liter)
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
Bordeaux Red table Peynaud (1939b) 35 0.001 0.006 0.003
Bordeaux Wh. table Peynaud (1939b) 36 0.001 0.007 0.003
France Red table Peynaud (1950a) 12 0.001 0.003 0.002
France Dessert Peynaud (1950b) 8 0.001 0.006 0.003
Germany Wh. table Hennig (1944) 19 0.001 0.004 0.003
Portugal Dessert Oliveira (1942) 11 0.002 0.008 0.002
456 MAYNARD A. AMERINE
TABLE XXII
Ammonia Nitrogen (as Ammonia) in Various Types of Wine
Type of No. of Mini- Maxi- Aver-
Region wine Source of data samples mum mum age
Bordeaux Red table Peynaud (1939b) 35 0.001 0.071 0.018
Bordeaux Wh. table Peynaud (1939b) 36 0.000 0.014 0.005
France Dessert Peynaud (1950b) 8 0.010 0.031 0.015
France Table Peynaud (1950a) 12 0.009 0.019 0.012
France Table Valaize and Dupont (1951) 73 0.017 0.201 0.101*
France Table Archinard (1937) 8 0.007 0.069 0.020
Germany Wh. table Hennig (1944) 19 0.003 0.028 0.007
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)
Spain Fino Bobadilla and Navarro 15 0.004 0.026 0.012
(1952)
Spain Oloroso Bobadilla and Navarro 10 0.008 0.035 0.021
(1952)
Switzerland Table Godet and Martin (1946) 3 0.012 0.026 0.014
Switzerland Table Berner (1952) 8 0.014 0.037 0.027
* There is no clear explanation for these high values.
TABLE XXIII
Other Nitrogen Fractions in Various Types of Wine
(Grams per liter)
No. of Mini- Maxi- Aver-
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
COMPOSITION OF WINES 457
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.
458 MAYNARD A. A M E R I N E
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
COMPOSITION OF WINES 459
(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
COMPOSITION O F WINES 46 1
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)
462 MAYNARD A. AMERINE
a
$ Per 100 g. of edible portion.
**
In commeraial wines.
-
Average of wines = 5.34.
Average of wines 4.10.
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)
464 MAYNARD A. AMERINE
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
COMPOSITION O F WINES 465
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.
466 MAYNARD A. AMERINE
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
Research 9, 393-395.
COMPOSITION OF WINES 467
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.
468 MAYNARD A. A M E R I N E
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
COMPOSITION OF WINES 469
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).
470 MAYNARD A . A M E R I N E
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.
2. Untersuch. Lebensm. 69, 616-617.
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).
COMPOSITION OF WINES 47 1
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.
472 MAYNARD A . A M E R I N E
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.
COMPOSITION OF WINES 475
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
476 MAYNARD A. AMERINE
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.
Ann. fals. et fraudes 26, 9-15.
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.
Ann. fals. et fraudes 26, 420-423.
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.
COMPOSITION O F WINES 479
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.
480 MAYNARD A. AMERINE
(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 stanssi^a 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.
Gebiete Lebensm. u. Hyg. 37, 327-342.
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.
COMPOSITION OF WINES 483
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.
484 MAYNARD A. AMERINE
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-
COMPOSITION O F WINES 485
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.
2. Untersuch. Lebensm. 71, 332-337.
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.
486 MAYNARD A. AMERINE
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.
Ann. fals. et fraudes 46, 208-212.
COMPOSITION OF WINES 487
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.
Lebensm.-Untersuch. u. -Forsch. 91, 418-422.
.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.
488 MAYNARD A. A M E R I N E
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.
COMPOSITION OF WINES 491
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
492 MAYNARD A . A M E R I N E
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).
Bull. ofice intern. vin 16(153), 41-49.
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.
COMPOSITION OF WINES 493
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.
(Conegliano) 6, 62-67.
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.
494 MAYNARD A. AMERINE
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
COMPOSITION OF WINES 495
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
496 MAYNARD A. AMERINE
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.
Ann. fals. et fraudes 30, 390-400.
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).
COMPOSITION OF WINES 497
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.
504 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.
COMPOSITION O F WINES 507
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.
Bull. ofice intern. vin 17(6), 50-59.
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.
COMPOSITION OF WINES 509
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.
510 MAYNARD A. A M E R I N E
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
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
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
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
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