Stone 1965

TECHNIQUES FOR SENSORY EVALUATION OF FOOD ODORS
BY HERBERTSTONE
S t a n ford Research Institute, Menlo Park, California
AND
ROSE MARIE PANGBORN AND c. s. O U C H

Uni versi t y of Califor,riu, Davi s, California
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
A. Importance of Odors in Foods .............................. 1
B. Need f o r Control Techniques ............................... 2
C. Current Theories of Odor Detection ........................ 2
11. Techniques of Measurement .................................... 5
A. Sensory Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F
B. Physical and Chemical Techniques for Evaluation of Odors . . . . 18
C. Interrelation of Techniques . . . . . . ..................... 22
111. Areas of Needed Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
IV. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . ....................... 27
I. INTRODUCTION
A. IMPORTANCE OF ODORS IN FOODS

Man’s awareness of the many different odorous stimuli which
constantly assail his olfactory receptors is well documented.
Equally well known is the wide range in sensitivity among peo-
ple to these stimuli. I n foods and beverages, odor greatly influ-
ences acceptance and preference. Unfortunately, the hedonic (de-
gree of liking) and the qualitative (descriptive) aspects of food
odors and flavors are not clearly understood. Nevertheless, olf ac-
tion provides us with our most extensive experiences associated
with food. Geldard (1953) has very aptly stated the problem:
“. . . it is the receptor system . . . that transforms acceptance of
food into appreciation of flavor.’’ Certainly, much has been writ-
ten about odors and olfaction, although much remains to be ac-
complished. To obtain a quick survey of the breadth and depth of
1
2 HERBERT STONE, ROSE MARIE PANGBORN AND C. S. OUGH
coverage in this country, consult the proceedings of symposia on

odors and olfaction published by the New York Academy of Sci-
ence (Miner, 1954 ; Whipple, 1964).
B. NEED FOR CONTROL TECHNIQUES

In the food field, flavor and aroma are recognized as important,
as evidenced by the volume of published data. Since flavors and
aromas are primarily odorous in nature, odor measurement tech-
niques are very important if the data obtained are to be meaning-
ful. With the development of new and improved processing meth-
ods and the emphasis on acceptable flavor, much attention has
been given to techniques, yet inadequate methods are still being
used. Correlation of sensory responses with physical and chemical
measurements has been initiated only recently. The appropriate
sensory approach (difference and/or preference, strength and/or
quality), sample size, panel size and selection, methods of presen-
tation of stimuli, number of replications, standardization of en-
vironmental variables, etc., constitute areas of controversy with
many unanswered questions. Current investigations are providing
information on some of these topics, but there is great need for
standardized methods for reliable and reproducible measurement
of both simple and compound odors. Standardization of sensory
test methods is now being undertaken by Committee E-18 of the
American Society for Testing and Materials, a group actively en-
gaged in preparing a handbook of basic principles of sensory eval-
uation covering a wide range of materials including foods. Ulti-
mately, the use of standardized methods will materially reduce
needs f o r extensive sensory testing and confusion relating to
work from different laboratories. Certainly the semantics of flavor
and aroma evaluation would be more uniform, thereby contribut-
ing to our knowledge of both the qualitative and the hedonic
aspects of olfaction.
C. CURRENT THEORIESO F ODOR DETECTION

Theories on the mechanism of olfaction have been extensively
reviewed by Jones and Jones (1953), Adey (1959), Beidler (1959),
and Ottoson (1963). The present review cannot detail these theo-
ries, but i t does consider what is now known from actual experi-
mentation in relation to present techniques of odor detection and
discrimination.
Anatomically, the human olfactory receptors are distributed in
TECHNIQUES FOR SENSORY EVALUATION O F FOOD ODORS 3
the mucous membrane covering the upper region of the nasal sep-
tum and wall opposite to the superior concha. Within the olfactory
region a r e located olfactory, supporting, and basal cells. A thin
process, the olfactory rod, extends from the sensory cells (bi-
polar neurons) and terminates at the outer membrane with the
olfactory vesicles. From these vesicles approximately 5 to 7 hair-
like cilia extend into the mucous covering the epithelium. The
structure and function of these “hairs” have been the cause of
considerable controversy since they a r e difficult to stain and ob-
serve under ordinary magnifying conditions, including electron
microscopy. Further, the cilia of the supporting cells, which num-
ber in the thousands, have often been mistakenly identified as
those of the olfactory cells. For more details on this subject, the
review by Ottoson (1963) is recommended.
Odorous vapors come in contact (via adsorption through the
mucous layer) with the olfactory receptors, which are most likely
located in the cilia. The mucous covering of the olfactory region is
considered necessary since detection will not occur in its absence.
Excitation of the peripheral receptors occurs through a N a : K ex-
change causing a signal t o be transmitted. It is thought that the
N a : K exchange results through a n alteration in cell wall permea-
bility to the Na+ and Kf ions. It is conceivable t h a t the change in
cell wall permeability is a n energetic process coupled to the ad-
sorption process. Adsorption on specific cellular sites also seems
to play a role in the patterning of the activity by which small
differences in odors a r e detected. According to Amoore (1962,
1963 ; Amoore et al., 1964), there a r e specific receptor sites on the
olfactory cells which determine odor quality. He postulates that
there a r e approximately seven primary odors, much the same as
the four basic tastes, corresponding to the receptor sites. Our
knowledge in this latter area is still quite marginal, but the the-
ory should not be discredited until further testing. There is some
evidence to support this stereochemical theory of olfaction, such
as the investigation of Gesteland et al. (1963) which indicated
that many olfactory cells in the frog have varying degrees of re-
sponsiveness to different stimuli. Some cells transmitted strong
signals, while others were quite weak. From these data, it was
suggested that there are different receptor-site types distributed
over each cell. Differences in cells would be based on the varying
ratios of these receptor sites, and the response would depend on
the filling of different sites.
Schutz e t aZ. (1962) evaluated a large number of physicochemi-
cal properties of odors that influence olfaction in humans. Multiple

regression and factor analysis revealed, among other things, that
there were nine quality factors very similar to other odor classifi-
cation systems. It is significant to note that these data are de-
rived from actual experimentation. One is tempted to conclude
from these results that an odor classification system is attainable ;
however, it is not at all well defined and will not be until more
data are obtained.
Beets (1957, 1964) proposed a relation between odor and mo-
lecular configuration which he termed the “profile-functional
group)’ concept-both functional groupings and shape determined
odor quality, a concept which seems to agree with work of Amoore
and others.
Enzymic theories of olfaction remain quite speculative. Kistia-
kowsky (1950) postulated a theory of olfaction in which four
groups of enzymes were required. Subsequently, Baradi and
Bourne (1951a,b) demonstrated the presence of some of these en-
zymes in the olfactory mucosa of rabbits. It was suggested that
the primary mechanism might be one of “interference of one or
more enzymes by the substances possessing the properties of
odor . . . , I ) giving rise to ionic changes and the signal trans-
mission. Baradi and Bourne (1959) further reported that cholines-
terase was present in the proximal and distal processes of the
olfactory cells and in some of the fibers of the cilia olfactoria of
the rabbit. Their histochemical evidence is quite limited, and con-
trary evidence has been presented more recently. According to
Ottoson (1963) , cholinesterase does not block olfactory response.
Naves (1957) and Wright (1963) emphasized that the lack of
gross differences in odors between optical isomers does not agree
with enzymic stereospecificity. Schutz et al. (1962) found little
change in acetylcholinesterase activity as a function of concentra-
tion of several alcohols and acetates. Unfortunately, information
on enzymes in the olfactory region of different species is still
limited.
Theories relating low-frequency molecular vibrations with ol-
faction, as proposed by Wright (1957, 1964) and others, have met
considerable criticism. The pigment of the olfactory mucosa is an
integral part of this theory, yet it is located mainly in the gland
cells (Ottoson, 1963). Moulton (1963b) also cautioned against in-
volvement of the pigment in the mechanism of olfaction, noting
that even albino rats can detect odors despite the presence of
very small amounts of pigment material. Finally, Jackson (1960)
TECHNIQUES FOR SENSORY EVALUATION OF FOOD ODORS 5
has shown that olfactory tissue extracts do not fluoresce until

they are separated chromatographically.
The foregoing discussion emphasizes that a workable theory to
explain the olfactory process is yet to be developed. Considerable
evidence has been accumulated to eliminate many theories, thereby
focusing more attention on those which hold the most promise;
therefore, a theory meeting all the criteria for olfaction should
be forthcoming.
II. TECHNIQUES OF MEASUREMENT

Considerable effort has been devoted to the techniques of odor
measurement. The properties of the stimulus and the uniqueness
of olfactory perception have motivated numerous investigators to
devise instruments, both simple and complex, for the presentation
of odors. The design of the equipment may differ ; the theory, how-
ever, remains the same: delivery of an odorous material to a sub-
ject and measurement of his response to intensities, differences,
or affective qualities. This criterion applies whether one is inter-
ested in mixtures of individual compounds o r in the odor of a food
product. The failure of many early test methods to provide evi-
dence for a satisfactory theory of olfaction is attributable t o the
lack of: ( 1 ) sufficient evidence on within- and among-subject
variability in quantitative and qualitative responses to odor stim-
uli, ( 2 ) an understanding of the sensory dimensions of the
stimulus or the sensation, which is due partially to inaccessibility
of the olfactory region, ( 3 ) satisfactory methods for control of
pain (trigeminal responses) at high concentrations, (4) control
of temperature and humidity during testing, ( 5 ) precise control
of air flow pressure and velocity, ( 6 ) a supply of odor-free ai r
and highly purified and stable test materials, ( 7 ) quantitative con-
trol of odor concentration, ( 8 ) the use of odorless and easily
cleaned construction materials, and ( 9 ) a means of expressing the
intensity of the stimulus in some common scientific unit.
These conditions and requirements have been considered by sev-
eral investigators as essential in the construction and use of any
apparatus for odor testing (Gundlach and Kenway, 1939; Wen-
zel, 1948 ; Mateson, 1955). While it is true that these requirements
are not always satisfied, considerable information has been ob-
tained from such studies, resulting in more highly controlled tech-
niques of sample presentation. Although the approach of diverse
groups of investigators working in olfaction representing many
6 HERBERT STONE, ROSE MARIE PANGBORN AND C. S. OUCH
different scientific disciplines has resulted in some confusion and

misunderstanding, an interdisciplinary approach seems desirable.
The discussion to .follow is concerned with: (1) a critical review
of olfactometric techniques ; ( 2 ) consideration of their usefulness
as analytical tools; and ( 3 ) a discussion of these techniques a s
applied to measuring the odor of foods.
METHODS
A. SENSORY
1 . Olfactometry
The first investigation, reported by Valentin in 1850 (Wenzel,
1948), was an air-dilution method consisting of a measured amount
of odorous material sealed in a small, thin-walled glass tube placed
inside a larger container. The small tube was broken, and the
subject opened the container and sniffed the contents. If the sub-
ject could detect the odor, the test was repeated with the same
quantity of material in larger containers until the odor was no
longer recognizable. Major errors associated with this technique
included absorption of the test material on the glass, which is
critical a t threshold concentrations, dilution when the container
was opened, and difficulty in weighing the volatile material
(Wenzel, 1948). From this crude but simple dilution method have
developed a large number of similar techniques employing various
solvents or diluents, such as air, compressed inert gas, mineral
oil, benzyl benzoate, glycerol, or diethyl phthalate.
Several years later, Zwaardemaker ( 1895, 1921b) introduced an
olfactometer and initiated a long series of investigations of odor
sensitivity and intensity. The olfactometer consisted of two tubes,
one fitting inside the other, with the inner tube open a t both ends,
graduated and shaped to fit the subject‘s nostril. The odorous ma-
terial was carried on the inner surface of the outer tube, and the
subject withdrew the inner rod until the odor was just detected.
This length was considered a s one “olfactie” ( a unit of odor
stimulus), and withdrawing the tube gave further stimulations of
so many more olfacties. Both single and double olfactometers were
employed in Zwaardemaker’s studies. Subsequently, Zwaardemaker
(1921a) invented the “camera inodorata,” consisting of a box with
a 400-liter capacity, to overcome the “lack of an inodorous space
in perfumery laboratories.” The box had glass sides and an alumi-
num top and bottom, and an exhaust fan and an ultraviolet lamp
were placed inside the hood to render the atmosphere odor-free.
The subject’s head was placed in the box through a slot in the
bottom, and his sensitivity was measured by the Zwaardemaker
olfactometer, located a t the bottom of the box, on the outside. To
remove traces of odor remaining on the tube after a test, a source
of fresh air was provided with a movable tube. In this way, sev-
eral experiments could be made on a subject during a short time.
In Zwaardemaker’s studies, temperature and humidity were not
controlled and it was assumed that a progressive increase in ex-
posure caused a proportional increase in the saturation of the air
current. However, air is saturated when the partial pressure of the
vapor in the air equals the vapor pressure of its liquid a t that
temperature. If the air is saturated, no increase in concentration
of the vapor can occur when the exposed area is increased. In
spite of these objections and a lack of a quantitative measure of
the odor, the data give a n indication of the relative intensities of
different odorous compounds.
During this same period, other investigators constructed and
experimented with more elaborate testing techniques. Woodrow
and Karpman (1917) bubbled air through a test sample a t dif-
ferent temperatures, then delivered it to the subject’s nose. Their
experiments were concerned with adaptation and were limited to
liquids. Mullins (1954) noted that sensitivity in olfactory testing
was maximum when the odor concentration was raised as rapidly
as possible to the desired level. He further noted that a t low flow
rates, considerable adsorption could occur on the mucous mem-
branes. Therefore, the supposed relation of duration of odor sen-
sation with odor concentration, as reported by Woodrow and
Karpman (1917), was due to low flow velocities and a continous
increase in concentration with time a t the olfactory epithelium.
Another method, described by Allison and Katz (1919), is com-
plicated but introduced one of the better olfactometers. Venturi
tubes were used as flow meters, so arranged that a measured vol-
ume of air could pass a t a uniform rate through or over the
chemical and could be mixed with a measured volume of pure air,
also flowing at a uniform rate. The chemical concentration was
determined by measuring the loss in weight after a measured vol-
ume of air passed over it. The apparatus was a useful tool, but,
it was not flexible, stimuli were not easily changed, and odor
removal was a problem. Gundlach and Kenway (1939) prepared
a similar but less elaborate apparatus using manometers to regu-
late odor concentration, but, here again, the technique was slow
and poorly controlled, and saturation of the ai r was uncertain.
Hofmann and Kohlrausch (1925) developed a blast-type, or
positive pressure, olfactometer. Saturated vapor was stored over

mercury, mixed with air by mercury columns, and then delivered
through nose pieces to the subject’s nostrils. With pressure, vol-
ume, and specific gravity known, the concentration (gram frac-
tion) for 1 cm:j was calculated. Morimura (1934) modified this
apparatus in his experiments on the effects of temperature and
various degrees of anosmia on odor thresholds. His reasons for the
modifications were incomplete mixing of ai r and odor material
(with the mercury) and subject variation over long periods. In
spite of improvements, the apparatus was subject to several dis-
advantages: ( 1 ) adsorption of the vapor on the tubing, ( 2 ) the
use of rubber tubing with a characteristic odor of its own, (3) the
use of a large number of containers, ( 4 ) the slow rate of testing,
and ( 5 ) the technique of “blasting.”
In 1935, Elsberg and Levy introduced a “blast” injection method
for odor detection. Elsberg was concerned primarily with locating
brain tumors via partial or complete anosmia of one or both nos-
trils. The technique involved releasing an odorous material, under
varying amounts of pressure, directly into the nose while the
subject held his breath. Both pressure and volume varied. The
smallest identifiable volume was called the M I 0 (minimum identi-
fiable odor). Elsberg recommended inhalation tests with ammonia,
benzene, menthol, xylol, benzaldehyde, citral, coffee, or oil of tur-
pentine. For blast-injection tests (high rates of steady flow), the
last four compounds were recommended. Further experimentation,
however (Jones, 1955 ; Wenzel, 1949, 1955), has indicated that the
technique is unreliable for critical odor investigation, mainly be-
cause of pressure variations. Wenzel (1949) designed an apparatus
with controlled temperature, pressure, and volume of the stimulus
delivered to the subjects, which permitted calculation of molecular
concentration using test materials of known vapor pressure. The
most important finding was the inability of subjects to demon-
strate significant differences between odor and pressure judg-
ments. Wenzel suggested caution in applying the technique to dif-
ferential odor measurement.
Jones (1953) theorized that if the Elsberg test measures
thresholds of molecular concentrations the M I 0 should increase
with decreasing concentration. Experimental results indicated that
concentration did not have a uniform effect, and the data could
not be transformed to molecular terms. To avoid pressure dif-
ferences and obtain molar concentrations, Jones (1954) devised a
motor-driven syringe filled with air and test material. Concentra-
tions were varied by changing proportions of each. For two dif-

ferent concentrations of benzene there was good agreement be-
tween subjects, but for methyl salicylate there was wide individual
variation. It was concluded that thresholds obtained by the blast-
injection method might in some way be related to the aero-
dynamics of the nose.
As summarized by Wenzel (1955), the Elsberg injection tech-
nique is unreliable for odor testing for several important reasons :
( 1 ) the inability of subjects t o perform reliably over long periods
despite training, ( 2 ) lack of control over the position of internal
mouth and throat parts, and ( 3 ) the extreme artificiality of the
situation.
The Elsberg technique still remains popular with some investi-
gators because of its simplicity and the ability to inject the amount
of odorous material to activate the receptor sites and elicit a
response. Castello, in 1950, Fortunato and Niccolini (1958) re-
ported, constructed an electronic olfactometer based on the blast-
injection method with elaborate controls for pressure, tempera-
ture, and humidity. Nitrogen was used as the diluent, part of each
test sample was bypassed through an ionization chamber, and de-
flections on a microammeter were noted. The obvious disadvantage
to these measurements is that they lack quantitative significance.
Results are reported on a time basis and would be of some value
for studying olfactory fatigue, but only with this instrument and
technique. Fortunato and Niccolini (1958) also employed the
Elsberg technique in their studies of olfactory fatigue. Instead of
the single bottle, several containers were placed on a revolving
plate and different materials could be sampled by revolving the
plate. The center of the plate had a vertical column to keep the
bottles in place and served as a mounting for the syringe with
inlet and outlet valves.
During this same period, Kristensen and Zilstorff-Pedersen
(1953), and Zilstorff-Pedersen (1955) used the Elsberg blast-
injection method for clinical examination of the sense of smell.
In the former study, the authors emphasized the clinical value
of the method for the detection of peripheral and central im-
pairment of sensory function. They felt that the technique was
applicable also for exact threshold measurements. I t is unfortunate
that in these investigations, finely ground coffee (changed once
per month) and citral were used as the stimuli. Coffee, which
does not have a standardized composition, and citral are very
susceptible to oxidation, and the possibility of subjects respond-
ing to oxidation products and not the stimulus is conceivable. In

discussing later test results, Zilstorff -Pedersen pointed out that on
some days the thresholds for both nostrils remained relatively con-
stant for about 60% of the experimental period. Further, a change
of sensitivity in one nostril was not balanced by a change in the
other nostril. Another important consideration is that many of
the unilateral changes were thought to be due to altered condi-
tions in transmission. Two possible regions of the nose, between
the septum and the concha superior, and the concha media it-
self, were cited a s the most likely causes for these alterations.
Schneider et al. (1963) used a “nasal patency meter” which per-
mits detection of changes in swelling of the nasal mucous mem-
brane. These authors employed an injection technique in conjunc-
tion with this meter for odor investigation. Their reasoning for
use of the direct presentation was based on the ability to main-
tain the volume and duration of the stimulus constant. Their ap-
paratus is based on dilution of the stimulus (n-butane) in a n
inodorous gas (nitrogen) by means of the appropriate manometers.
The stimulus was presented to the subject by activation of a sole-
noid valve for a preset time interval (usually 1 sec in a 16-sec
period) and the stimulus was presented to either or both nostrils.
When only one nostril was stimulated, the other received the
carrier gas, nitrogen. Their study of the effects of flow rate and
duration of exposure on threshold corroborated previous investi-
gations. An increase in both variables resulted in decreased
thresholds, with the greatest change occurring with increased gas
flow rate. These data also suggest that the lowered thresholds re-
ported with increased flow rate are attributable to quicker arrival
a t the olfactory region and decreased losses in adsorption during
passage through the nasal region. Obviously, the aerodynamics of
the nose (as demonstrated by Jones, 1953) plays a significant role
in odor detection and discrimination. Despite all the evidence and
objections raised by the studies of Wenzel and Jones, the Elsberg
technique of stimulus injection has not been discarded ; rather,
there has developed a renewed interest in its use for clinical in-
vestigation of monorhinal or birhinal stimulation, and to dis-
tinguish between peripheral and central impairment of specific
sensory function.
While the Schneider instrument represents a significant im-
provement in odor delivery and quantity of stimulus, the injection
technique still presents the subject with a n unnatural situation,
and separation of response to odor stimuli from those t o pressure
T E C H N I Q U E S FOR SENSORY EVALUATION O F FOOD ODORS 11
remains to be demonstrated. In terms of the study of central and

peripheral olfactory impairment, early work of Adler and Finley
(1938) and Spillane (1939) suggested that caution be used in
applying the method to tumor location, although some success was
attained in detecting interference with the function of the olfac-
tory system itself. Certainly, the Elsberg technique is of little or
no value in psychophysical investigations of olfaction or studies
involving food odors.
During this same period, there developed a renewed interest in
air-dilution olfactometers and the use of natural breathing. As a
replacement for the Elsberg apparatus, Wenzel ( 1955) suggested
use of an odor-free environment similar to Zwaardemaker’s
(1921a) “camera inodorata,” in which the subject utilized normal
breathing. The subject placed his head inside a Plexiglas box con-
tinuously supplied with purified a i r at about 13 ft:’/min. A cylin-
der of pure a i r was saturated by contact with odorous liquid,
and, by means of appropriate values and timers, small amounts
were released into the main a i r stream. The subject breathed
normally, no special training was required, and the environment
around the subject’s head was controlled.
The air-dilution olfactometer represents a more realistic ap-
proach to the study of olfaction from the psychological and psy-
chophysical viewpoint, for reasons discussed previously ; however,
it is still a research tool and the practical applications are limited.
Only recently has thought been given t o quantitative investiga-
tions of odor detection and discrimination.
Similar olfactometers based on this same principle have been
reported by Mateson (1955), Stuiver (1958, 1960), Cheesman and
Kirkby (1959), Bozza et al. (1960), Johnston and Sandoval (1960),
Mrak et al. (1959), Ough and Stone (1961), and others. I n some,
manometers were used to supply a wide range of concentrations
of the odorous materials to a subject who breathed normally, while
others used differential pressures and flowmeters, but few made
use of the odor-free environment in the form of a n enclosure
around the subject’s head or body.
The Cheesman-Kirkby apparatus was designed f o r measuring
responses from up to 12 subjects simultaneously. Through use of
the appropriate capillary tubes it was possible to attain 12 identi-
cal odor concentrations ; however, as designed, the apparatus was
not sufficiently versatile for pair-comparison or difference testing.
Unfortunately, these workers were unable to continue experimen-
tation and the instrument was disassembled. Certainly, the use of
manometers might have improved the technique. The value of

testing more than one subject at a time has obvious advantages.
In the apparatus described by Ough and Stone (1961), vapor pres-
sure at specific temperatures was used to saturate odor samples,
and differential pressure and the flowmeters were used to obtain
the desired concentrations in the main air stream (see Figs. 1
and 2 ) .
F I G . 1. Schematic diagram of the olfactometer. A ) Incoming conditioned a i r ;

B ) thermometer; C ) a i r by-pass valve; D ) control valve f o r large a i r flow;
E ) large flowmeter; F) muffler; G) odor hood (Plexiglas); H ) I ) J ) small
flowmeters; K ) diffusion bulb ( 3 L ) ; L ) small air-flow delivery tube;
M ) sintered-glass sparger (coarse) ; N ) water bath (control of te st liquid tem-
perature). Reproduced through the courtesy of the New York Academy of
Sciences.
The instrument described by Bozza et al. (1960) required the

subject to place his face a t a n opening with nose and mouth pro-
truding into a stream of odor or purified air. This test area was
then enclosed in a glass and polyethylene chamber to reduce room
odor level to undetectable levels. Stuiver used a nosepiece, Mateson
used a small funnel, while Johnston and Sandoval utilized a small
cone-shaped device designed to cover just the nose.
By careful measurement of flow rates and chemical recovery,
Ough and Stone (1961) were able to determine actual test con-
centrations delivered to a normally breathing subject seated with
his head placed inside a Plexiglas hood. A number of stimuli have
been successfully tested with this apparatus. This air-dilution type

of olfactometer appears to offer a good opportunity for more so-
phisticated studies of the psychophysical aspects of olfaction. The
technique readily lends itself to more precise control of a i r flow
FIG.2. Photograph of the olfactometer. Reproduced through the courtesy

of the New York Academy of Sciences.
rates, temperature, humidity, and stimulus concentration than has

heretofore been accomplished. One of the contributions made
through the use of this air-dilution olfactometer was the obser-
vation that varying the a i r entering the nares from 12.5" to 35°C
did not influence odor perception, since the air was rapidly equili-
brated with body temperature (Stone, 1963a). Tucker (1963)
also reported that the temperature of the odorous material did
not influence odor thresholds in animals.
A great number of other instruments and equipment have been

developed for use as olfactometers. In most instances they repre-
sent modifications of previously described equipment-some
simplification and some complication, but not a n improvement.
Deininger and McKinley (1954) described the use of a n “olfacto-
rium” for odor evaluations. Subjects entered a specially constructed
room which contained a volatilized odor stimulus and carried out
group discussion (flavor-profile) analysis of the sample. An elab-
orate means of a i r circulation was used to maintain the odor-free
environment. I n contrast, for food aroma analysis, Guadagni et al.
( 1963a) suggested using polyethylene and teflon squeeze-bottles
containing aqueous solutions of odorous materials, a simplification
of the Elsberg blast technique. The olfactory process is influenced
by a large number of physiological and psychological variables not
controlled by these techniques, i.e., the odor of the polyethylene,
effect of pressure, temperature, humidity, etc. Most of the fore-
going olfactometers have been developed to measure absolute sen-
sitivity or odor thresholds. As summarized by De Vries and
Stuiver (1961), the threshold, expressed in number of molecules
entering the nose, depends on time of presentation of the stimulus,
concentration of the odorous substance, and rate of flow of the
carrier gas. Consequently, any threshold measurement which ig-
nores these criteria would be worthless.
2. Snif Methods
Sniffing from beakers or bottles is the most widely used method
of measuring odor intensity and quality. Although sniffing is the
most simple and economical of all procedures, certain limitations
detract from its usefulness. In a recent study in which the sniff
method was compared directly with olfactometric presentation of
the same odor, the methods were of comparable sensitivity (Stone,
1963b). However, the presence of nonideal solutions a t or near the
threshold was a serious problem, suggesting that the procedure be
used with caution when dilution of the odorous material is neces-
sary. The method can be used for routine quality-control purposes,
such a s measuring odors in drinking water (Baker, 1961), but
more precise studies of olfactory response require presentation of
odors under highly controlled conditions.
Our knowledge of techniques for measuring sensory responses
has advanced sufficiently to provide the quality-control laboratory
with sensitive, reproducible methods. Unfortunately, outdated pro-
cedures of questionable accuracy continue to be used because of
unawareness of the availability of appropriate designs. Often there

is failure to control environmental variables, insufficient amount
of replication, or other deficiencies.
Guidelines for selection of appropriate test designs have been
published by several reliable investigators (Boggs and Hanson,
1949; Dawson and Harris, 1951; Peryam, 1958; Amerine et al.,
1959; Kramer and Twigg, 1962; Dawson et al., 1963). Test selec-
tion is determined to a considerable extent by the type of informa-
tion sought, the amount of time and sample available, the quantity
and sensitivity of the available judges, and the environmen-
tal facilities. Ideally, only the first consideration should influence
the selection of test methods. A detailed description of specific
testing techniques has been compiled by Ellis (1961).
For determining either qualitative or quantitative differences in
odor, paired comparison, duo-trio, dual-standard, triangle, or mul-
tiple-comparison techniques can be used. In general, when the odor
dimension can be defined, the paired-comparison method is most
sensitive (Pfaffmann et al., 1954; Gridgeman 1955; Sawyer et al.,
1962). In addition, paired-difference testing is most economical in
terms of amount of sample and panel time. Although fewer tests
are required for significance in the triangle test, where the proba-
bility is $$, it is, in effect, three paired comparisons, and therefore
less efficient than the paired test. When differences in the treat-
ments are not definable, i.e., a new treatment resulting in a n un-
predictable change in odor character, the triangle test is appropri-
ate. A variation in the paired test, the dual-standard paired test
has been used effectively by Pangborn and Dunkley (1964) fo r
sensory examination of nondefinable variations. Two standards, A
(control) and B (treated), are presented, and then four to eight
coded pairs consisting of A and R are presented in randomized,
counterbalanced order. The task of the judge is t o select the mem-
ber within each pair which is the same as the control. In this test,
the dimension is undefined in that the judge is not told t o seek the
stronger aroma, o r better aroma, or more typical aroma, but sim-
ply to select the sample within each pair with the same aroma as
that of the control. The duo-trio and multiple-comparison tests are
suitable for specific problems where reference standards help to
anchor the judgments (see Dawson e t al., 1963). Ough and Baker
(1964) recently compared paired testing with a rating method for
odor intensity, and found both to be of equal sensitivity. The rat-
ing method took approximately half the time of the paired test-
ing. I n these rating tests a reference sample was given, after a
warm-up pair of samples, and the subjects were then asked to rate
subsequent samples in comparison with the reference sample.
The use of scaling techniques for determination of odor inten-
sity and odor quality is popular since the scores lend themselves
to analysis of variance and to correlation with chemical and physi-
cal data. The number of points in an intensity scale should be de-
termined on the basis of the number of intensity levels that ex-
perienced judges can detect with the specific stimuli under study.
Dilution techniques can sometimes be used to establish the num-
ber of intensities that can be reliably distinguished. With all
scaling devices, the range of the scale that each judge will use
depends upon his individual sensitivity and various psychological
attributes such a s : ( 1 ) leniency errors-some judges are easy
raters, whereas others are hard raters; (2) central tendency-
some judges avoid using the extremes of a scale; (3) contrast
errors-rating samples higher when presented along with poor
samples than when presented along with good samples; and
( 4 ) proximity error-rating adjacent traits similarly. For further
discussion of these and other psychological rating errors, see Guil-
ford (1954). To determine whether the entire range of intensity
score is being used adequately, information analysis can be ap-
plied (Garner and Hake, 1951; Shannon and Weaver, 1963).
Rating of odor quality can lead to spurious results since pref-
erence and degree of liking influence a quality score. In some
cases, it would be better to rate the degree of difference in a
specified odor characteristic between a treated and a control sam-
ple, o r between a treated and a commercial sample of known ac-
ceptance, than to attempt to obtain scores of absolute quality.
Relative to selection of judges, most individuals can participate
provided they have no gross olfactory impairment, are sufficiently
interested and motivated to contribute the time required, and are
familiar with the testing procedures. Several statistical methods
have been suggested for selection of judges for specific panels
(Girardot et al., 1952; Mackey and Jones, 1954; Dawson et al.,
1963). Sawyer et al. (1962) proposed using repeatability, i.e., in-
traclass correlation of repeated measurements, as a means of pre-
dicting the outcome of selection of judges, an approach that ap-
pears to have promise but requires further experimentation with
a variety of commodities. A detailed description of the use of se-
quential analysis for selection of judges has been presented by
Amerine et al. (1959).
I t cannot be overemphasized that, with complex odors such as
TECHNIQUES FOR SENGORY EVALUATION OF FOOD ODORS 17
those present with perfumes, wines, whiskeys, cheeses, etc., the

judge's experience is of prime importance. In selecting judges for
complex quality evaluation tasks, no quick tests will suffice. Dif-
ferences in quality evaluations between inexperienced and experi-
enced judges can be very large and varied in pattern. Two to five
years of on-the-job training may be required before a judge be-
comes efficient and reasonably stable. However, repeatability of
scores may not vary after several months of training, as shown
by Ough and Baker (1961). Trained judges will still differ in their
impressions of what quality is, and defining the quality that is
desired must be left to some one individual or to a group that
agrees. As Kramer and Twigg (1962) pointed out, a quality-con-
trol group is small, consisting of three to five judges, and fre-
quently only one. There are several good reasons why this type
of panel should be small: ( 1 ) it is time-consuming and expensive
to train the judge; (2 ) few judges can look objectively for a de-
fined quality without letting their own personal preferences in-
terfere; (3) the judges must be a reasonably sensitive group but
a group which will be around f o r a long time; and ( 4 ) judges
should have personalities such that they can do a routine, but com-
plex, job well each and every time. Ough and Baker (1961) showed
what one erratic judge can do to pooled results of a quality evalua-
tion.
Panel size is largely influenced by available personnel; 6 to 15
people are generally adequate for laboratory studies. Sufficient rep-
lication, 3 to 10 replications per treatment per judge, is recom-
mended to determine individual reproducibility of response. Each
judge should be present a t each test session. Alternately, when the
commodity and testing method permits, make-up sessions should
be held so that each judge has been tested a n equal number of
times. Substitutes are not recommended since judges are no more
interchangeable than are treatments.
Testing periods should be short if the commodity possesses a
strong lingering odor, if the test procedure is complex, or if the
judges are inexperienced. In studies reported by Pangborn et al.
(1964), five highly experienced judges did not demonstrate a
measurable change in olfactory response to low concentrations of
2-heptanone during 70 to 90 min of continuous testing. The first
sample or first pair of samples should serve as "warm-up" tests
and be discarded from analysis of the data. In prolonged testing,
brief rest periods can defer o r eliminate physiological and psy-
chological fatigue.
Since loss of interest or motivation by a judge can greatly alter

his sensitivity and reproducibility, the investigator should keep
judges informed of their performance, a t least in general terms
if too much information could influence results. If a judge feels
he is contributing toward a n important research program and in-
termittently receives small rewards, such a s coffee, crackers, cook-
ies, or chewing gum xfter test sessions, high morale can be main-
tained. For a n interesting discussion of motivation and behavior,
see Young (1961).
Odor quality a s well as odor intensity may be influenced by the
concentration of a compound, necessitating testing a t various dilu-
tions. As pointed out by Stone (1963b), one should exercise cau-
tion in diluting compounds, especially when working with non-
ideal solutions. It is usually assumed that, a t or near threshold
concentrations, exact concentrations could be determined by use
of the ideal gas law, but further experimentation on the question
is needed. The properties of the compound would determine the
selection of diluents such as water, mineral oil, benzyl benzoate,
diethyl phthalate, or others. Dilution techniques have been sug-
gested by Tilgner (1961, 1962) for measuring odor and taste of
complex foods and beverages.
B. PHYSICAL AND CHEMICAL TECHNIQUES
FOR
EVALUATION OF ODORS
The complex nature of the olfactory system and the subjectivity
of human response has resulted in numerous studies on measure-
ment of odorous materials by chemical and physical means. With
further research and development, the use of mechanical devices
to monitor odor quality and intensity may not be in the very dis-
tant ,future.
By f a r the most well-known methods are gas-liquid partition
chromatography, flame ionization, and electron-capture techniques.
A myriad of volatile organic constituents have been revealed in
many different food products with such techniques. Webb and
Kepner (1962) found 23 components, mainly esters and alcohols,
definitely present, and 5 other probable esters present in the vola-
tile aroma of flor sherry. More than 30 volatile components of
coffee were identified by Zlatkis and Sivetz (1960). Twenty-one
volatile carbonyl compounds were identified or tentatively identi-
fied in cooked chicken by Pippen and Nonaka (1960). In cabbage,
the volatile sulfur compounds alone were found to number 20
(Bailey e t al., 1961). No less than 50 components were revealed by
Bernhard ( 1961) as volatile constituents of certain California

oranges, and Walford ~t nl. (1963) found 40 to 50 individual
flavor and aroma components in Florida oranges. Jennings and
Wrolstad (1961) found a t least 23 volatile components in black
pepper. Using some of the most advanced equipment, Teranishi
e t nl. (1963) found over 150 volatile constituents in strawberries.
Gold and Wilson (1963), working with celery, have so f a r found
and identified 38 volatile compounds. Dairy butter was found by
Winter e t al. (1963) to contain a t least 12 volatile carbonyl com-
pounds. The foregoing represents only a small part of the litera-
ture on this subject. As separation and detection equipment be-
comes more sophisticated the components that are present in
smaller amounts will be identified.
In the present review, no attempt has been made to consider
the advantages and disadvantages of gas chromatography and
other instruments except in relation to their correlation with olfac-
tion. Although significant contributions have been made in the iso-
lation and identification of specific volatiles which contribute to
food flavors and odors, flavor is more than a pattern of peaks on
a chromatogram. Flavor is a n integrated response, the nasal mu-
cosa and the taste buds being the integrators. The chromatograph,
on the other hand, is a separator, which, while an extremely use-
ful tool, must have its responses compared with human responses
to have a bearing on flavor (Stewart, 1963). It is safe to say that
much of the electronic detection equipment available now ap-
proaches or exceeds the sensitivity of the human nose. This equip-
ment cannot take over the function of the human judge in quality
control but may supplement, and possibly be correlated with,
specific aspects of human perception. Doubtless to say, many of
the volatile constituents present in food and beverages are there
in amounts f a r below humanly detectible amounts, yet work by
Singleton and Ough (1962) suggests that the quality attributed to
a product may be due to complex mixtures of components, each in
themselves undetectible to the judge. Others working on the ef-
fects of subthreshold mixtures have reached similar conclusions
(Nawar and Fagerson, 1962 : Guadagni et nl., 1963b).
I t is recognized, of course, that it is often impossible to obtain
chromatographic fractions in sufficient quantities for adequate
sensory replication. The carrier in which the fractions are dis-
persed for sensory examination may alter their character or in-
tensity. Further, many natural flavor complexes undergo some de-
gree of change on chromatographic separation so that recovered
fractions may not be identical to the injected mixtures. At pres-

ent, the greatest deficiency lies in the lack of sound, reproduci-
ble sensory techniques to provide the chemist with usable infor-
mation for identifying, predicting, and controlling flavor and odor
quality.
Moncrieff (1961) recently initiated a study of the adsorption
phenomenon in olfaction with simple thermistor-coated films, such
as polyvinyl chloride, cellulose acetate, regenerated cellulose, cal-
cium alginate, casein, and peanut protein. With the appropriate
bridge circuitry and recording equipment, it was possible to trans-
late the adsorption process onto tracings for measurement and
classification of odors. According to Moncrieff, no one film re-
sponded to all stimuli, and such differences might be likened to the
anatomical differences noted in the olfactory region, which are
assumed to account for our ability to differentiate between large
numbers of odor stimuli. This approach to the study of olfaction
is interesting, but, as Moncrieff stated, i t is a limited approach
(as a prototype instrument) and presents some challenging possi-
bilities. Study of proteinaceous films certainly should be expanded,
along with further study of the actual adsorption mechanism in-
volved as the first step in detection. It is easy to see how such a n
investigation might involve better electronic signal detection and,
ultimately, computer analysis. Work along similar lines has been
reported by Hartman (1954), Hartman and Tolle (1957), and
Rosano and Scheps (1964). The Hartman apparatus for studying
vegetable flavors is basically a sensitive bridge circuit with a mi-
croelectrode capable of responding to odorous compounds by un-
dergoing depolarization or further polarization, the electrode
having first been chemically polarized. On exposure to volatile
materials, the tip of the electrode was generally coated with a n
enzymic mixture, or other material such a s chlorophyll, to simulate
more closely the actual condition a t the olfactory site. Hartman and
Tolle (1957) concluded that the principal effects being recorded
were “oxidations and reductions of the volatile compounds or their
products a t the microelectrode.” Further research by Hartman and
his associates with similar but more refined equipment is continu-
ing, but no data are available on the relation between this instru-
ment and human responses to the same stimuli.
Rosano and Scheps (1964) developed a n “artificial nose” that
uses a sensitive galvanic cell with a platinum-gauze-electrode sen-
sing element. This electrode and a second, of graphite, were par-
tially immersed in a closed beaker of nitrochromic solution (70%
nitric acid saturated with chromix oxide). Spectrophotometric

measurements indicated that current increase was attributable to
“adsorption-oxidation” of the vapors at the solid-liquid gas inter-
face (Cr” + ; Cr:’ + ). The initial needle deflection rapidly reached
a steady state once equilibrium was established. Filtered air re-
versed the process, and a second sample was then analyzed. Con-
centration and flow rate of the odor were found to be important in
this process. These workers concluded that the galvanic cell tech-
nique could be used as a very sensitive odor detection device, with
the appropriate electrolyte solution.
Still a third method of odor measurement has been reported by
Hobbs (1963) : the Rutherford E M F Flavor Test. The technique is
described as a n objective method for measuring the odor of spices,
herbs, cosmetics, perfumes, and other materials having charac-
teristic odors and tastes. A sample is ground to a uniform particle
size and dispersed in a beaker of distilled water at pH 7, and
the E M F of the resultant mixture is noted after the readings
reach a steady state. According to the author, the equipment
measures the negative charge of the organic material, the most
pungent spice or most intense odor having the highest negative
charge. I t is quite doubtful that this apparatus can serve as a use-
ful and reliable technique for odor measurement. While it is ob-
viously simple to operate, objective results provide only a gross,
over-all picture of a particular product. The gas chromatograph is
certainly a far more sensitive and useful detection apparatus. Sim-
plification of the olfactory system and development of an objective
method of measurement is a reasonable process as long as its limi-
tations are clearly defined.
Several other physical methods of measuring odorous com-
pounds have been developed. Borsanyi et al. (1962), Borsanyi and
Blanchard (1962), and Moncrieff (1963) have suggested the use
of psychogalvanic skin response (PGSR) coupled with odor presen-
tation. Borsanyi and associates were primarily interested in the
use of this technique in medicolegal aspects of anosmia. A record-
ing PGSR instrument was linked with a typical Elsberg apparatus.
The subject was first blindfolded, noise level was controlled by
earphones, the instrument calibrated, and subject conditioned by
presentation of electrical and olfactory stimuli simultaneously. A
positive olfactory response was obtained when the odor stimulus
was presented alone and the appropriate tracing recorded. Accord-
ing to these workers, olfactory thresholds might also be deter-
mined by such a method.
Moncrieff (1963) has also studied PGSR response t o odor stirn-

uli to determine, among other things, whether exposure to pleasant
and unpleasant stimuli would evoke different kinds of responses.
Although the data suggested that different stimuli evoked some-
what different responses, the results indicated considerable within-
subject variation even during the same test session. Inattention,
noise, o r any other distraction could also lead to unreliable re-
sults. No clear-cut data are available on the relation between he-
donic tone (like and dislike) and relative sensitivity t o olfactory
stimuli.
The use of PGSR for study of olfaction is an interesting sug-
gestion ; however, the practicality of the technique is questionable.
The PGSR response alone, is a poor indicator of the behavorial re-
sponse to odorous materials. It represents a n emotional response, a
very crude picture, subjects vary considerably, and only when it is
coupled with a number of other simultaneous measurements might
one expect to obtain a more clear stimulus effect. The technique
may be of some value in medicolegal problems, but considerable
caution should be exercised for the reasons discussed above.
c. INTERRELATION OF TECHNIQUES
With the development of these odor test techniques, interpreta-
tion and extrapolation of data become important. Provided experi-
mental conditions a re defined and responses are expressed in
molecular terms, it should be possible to utilize these data, but
that is not always the case. This problem can be considered from
two aspects : psychophysical considerations, and the translation
into meaningful perceptual results of data obtained by chemical
and physical measurement.
Most data on odors are collected by one of three techniques:
“blast-injection,” sniff-bottles, or air-dilution olfactometers. Few
experiments have compared these latter two methods directly, but
it seems reasonable to assume that they would yield similar re-
sults. Such a conclusion was reached after measuring the 50%
threshold for propionic acid (Stone, 1963b).
Psychophysical investigation offers the most reliability because
test methods have been studied and used for a number of years
and tests are usually carried out with single stimuli. Evaluation of
food aromas and flavors is f a r more complex, and the data are
not easily interpreted. This difficulty can be attributed to the
many different methods, between- and within-subject variability,
and the food itself. The problem of test method is complex

since investigators consistently modify existing methods, some-
times to increase panel sensitivity and sometimes to introduce
more sophisticated methods. Often the techniques a r e more com-
plex, resulting in loss of information, or the information trans-
mitted to the experimenter is misleading. Finally, one cannot ex-
trapolate data from tests with one food to another food. Although
the odor may be similar, differences in the food itself alter the
panel’s sensitivity. The use of objective measurements of quality
and their significance to a food’s acceptability must be considered.
Several recent studies have compared results from gas chroma-
tography with results from sensory analysis. Jennings et al.
(1960) recovered essence during the processing of fresh ripe pears
and fractionated it by gas chromatography. Thirteen aqueous frac-
tions, prepared from 32 components, were evaluated by sniffing,
against a water blank at two levels of fortification (4 and 8%),
by the triangle test method. Taste evaluations were carried out on
canned pear nectar fortified with two major mixtures of the frac-
tions and a third containing all the fractions as a control. Their
results indicated that some fractions were undesirable and that
others were present in such low levels as to be undetectable in the
natural pear. Taste testing of the nectar indicated that no prod-
ucts were desirable and that those containing a certain series of
fractions were quite undesirable. Further studies on volatiles in
Bartlett pears have shown good agreement between sensory aroma
intensity and the presence of esters of 2,4-decadienoic acid meas-
ured by ultraviolet spectroscopy (Heinz et ul., 1964). Jenninys
et al. (1962) also demonstrated how the flame-ionization technique
could be applied to milk vapors to trace and identify unusual off
odors or for routine analysis. Other work with milk was reported
by Bassette et al. (1963) in which there was good correlation
between certain chromatographic peaks and the sensory data.
McCarthy et al. (1963) correlated gas chromatographic data with
the descriptive analysis (“flavor profile”) of two banana varieties.
The banana flavor was evaluated at weekly intervals during
ripening. The flavor profile and the chromatographic spectrum
were highly correlated. Early in the ripening process the panel
noted a predominance of unripe characteristics, and as ripening
progressed the banana-like flavor developed.
Fuller et al. (1964) reported using a thermal-conductivity gas
chromatograph to separate complex mixtures, and the effluent
vapors were passed before the nose of a perfumer standing in an
enclosed glass booth. The judge had a high level of accuracy in

identifying individual compounds as they came off the column.
The authors emphasized the potential value of the equipment to
the perfumer in his study of complex materials.
Using an unusual approach, Ottoson and von Sydow (1964)
reported no direct correlation between the amplitude of the gas
chromatographic peaks for higher-boiling aroma compounds from
black currants and that of the electrical response of the olfactory
membrane of a frog.
Work of Baker and Amerine (1953) and Baker (1954) in de-
veloping a prediction equation of wine quality based on very
limited analytical data should be considered and, in the light of
more rapid and sensitive analytical techniques, expanded upon in
this and other food fields.
These experiments demonstrate the practical application of
chemical and sensory methods and serve to enhance the knowl-
edge gained by both. Provided the experimenter is cognizant of
the test limitations, it is possible to establish a standardized
method for quality evaluation of a specific food item, by the
proper combination of physical, chemical, and sensory tests.
111. AREAS OF NEEDED RESEARCH
Elucidation of the mechanism of the olfactory processes would,

obviously, give great impetus to all research on odors and olfac-
tion, but would not solve other problems in the measurement of
responses to odor stimuli. Basic t o all areas of sensory investiga-
tion is a greater need to standardize conditions of testing. In the
food field, especially, there is a great need for investigators to
draw upon the wealth of knowledge available from the areas of
experimental psychology, sensory physiology, biochemistry, and
statistics. Selection of experimental designs, selection and training
of judges, presentation of stimuli, and analysis of results cannot
be done by casual and haphazard techniques simply because the
stimulus is a food product. Actually, because most foods and
beverages possess complex sensory properties, even more strin-
gent methods of evaluation are needed than in studies where
pure chemical compounds are used. One of the greatest needs in
the field of sensory evaluation of foods, in general, is clarification
of the appropriate use of difference testing, quality evaluation, and
preference testing.
Specific needs for further information would include develop-
ment and testing of simple but improved olfactometers for use in

basic odor perception where control of all influencing factors in
the delivery of odor stimuli to human subjects is critical. Little
is known of the relation between responses to threshold and sub-
threshold concentrations of pure odors and odor mixtures. Use of
the odor thresholds reported by Zwaardemaker (1926) in the In-
ternational Critical Tables should be discontinued, since such values
represent absolute sensitivities as measured in aqueous (nonideal)
solutions, the compounds were probably quite impure, and the test
techniques are of questionable value (the computational error of
lo2 still remains).
Although perfumers have contributed greatly to our knowledge
of mixtures of floral odors, little is available on the masking and
enhancing effect of one food odor upon another. Data are needed
on the influence of the carrier or medium on the resulting odor
intensity and quality, an area of interest in the compounding of
synthetic flavors and odors.
It would be of much interest to have more data available on
the many physiological factors which may influence odor percep-
tion, i.e., the effect of fatigue, age, smoking, changes in hormone
balance, respiratory disorders, metabolic disorders such as diabetes,
obesity, alcoholism, and nutritive conditions such as hunger and
vitamin deficiency. Among so-called “normal” judges, it would be
of interest to measure the range of individual responses under a
variety of environmental conditions such as temperature, illumina-
tion, motivation, sample size, and form, etc. Also, it would be a
great achievement if standardization of sensory nomenclature were
undertaken as well as standardization of test methods to reduce
confusion and for better correlation with analytical tests.
The interrelationship of the senses of vision, audition, olfaction,
taste, and touch remain virtually uninvestigated, particularly as
related to food evaluation. Research on the extent to which re-
sponses from these sensory systems correlate with the many avail-
able chemical and physical measurements of the properties of foods
will require the integrated efforts of many investigators. Efforts
to relate odor responses with gas chromatographic separation are
progressing, but at a painfully slow rate.
Although any one of the above-mentioned approaches would
represent a major undertaking for a laboratory, it would behoove
the food industry to take the lead in developing sensitive and re-
producible techniques for the sensory and nonsensory evaluation
of odors.
IV. SUMMARY
I n summarizing present knowledge and advances in odor meas-
urement, one is overwhelmed by the accomplishments of the past
few years. The latest decade has seen a considerable refinement
and sophistication in olf actometric techniques for odor measure-
ment. It has been possible to gain new insight into human sensory
perception and the ability to respond to a wide range of odor
qualities and intensities.
The quest for a unified theory of olfaction continues, with many
older ideas being pushed aside. Electrophysiological investigations
have reached the single cell level and suggest the possible presence
of different areas of response within the olfactory region (Geste-
land et al., 1963), as suggested earlier by Adrian (1952). Further,
work of Moulton (1963a), Tucker (1963), and Ottoson (1963) has
added considerable knowledge to the basic problem of signal trans-
mission, the electrophysiology of the olfactory system, and related
phenomena at the cellular level.
During this same period, investigation of odor classification has
received new impetus by the work of Amoore (1962, 1963)
Amoore e t al. (1964), Beets (1957), and Schutz et al. (1962).
Amoore and Beets have suggested somewhat similar theories of
olfaction, based on the structural configuration of the odor
molecule. Careful review of these and related data suggests
that molecular structure constitutes one of the most important
determinants of odor. Data are insufficient at present, however,
and these theories are still speculative.
Attempts to develop “mechanical” noses o r instruments capable
of duplicating human olfaction have been inadequate. Although
gas-liquid chromatography is a n excellent tool for isolation and
tentative identification of volatile materials in food, the technique
will supplement, not substitute, for sensory examination. Investi-
gators utilizing lipoprotein films (Friedman and Mackay, 1964)
have presented a novel approach to nonsensory replication of the
mechanism of the initial steps of olfaction; however, the same
reservation applies : the model systems may simulate only isolated
aspects of the human olfactory process. For the most part, such
systems are oversimplified and give little consideration to the
psychophysical aspects of olfaction. The application of current re-
search findings f Tom several scientific disciplines to measurement
of the sensory properties of foods has contributed to the refine-
ment of test methods, analysis of results, and interpretation of
the data.
In view of the large number of unexplored or partially investi-
gated aspects of odor perception, especially as related t o foods and
beverages, it is hoped that this review will stimulate more activity
in this dynamic field of study.
ACKNOWLEDGMENTS
Appreciation is extended to Dr. Morris H. Woskow, University
of California, Davis, for a critical review of the manuscript, and
to Mrs. Darlene Jones, Stanford Research Institute, for assistance
with the references.
REFERENCES
Adey, W. R. 1959. The sense of smell. I n “Handbook of Physiology.” (J. Field,
ed.) pp. 535-548, Am. Physiol. SOC.Wash., D.C.
Adler, A., and Finley, K. H. 1938. Clinical results with Elsberg’s olfactory
test. A. M. A. Arch. Neurol. Psychiat. 40, 147.
Adrian, E. D. 1952. Differential sensitivity of olfactory receptors. J . Physiol.
115, 42P.
Allison, V. C., and Katz, S. H. 1919. An investigation of stenches and odors
for industrial purposes. J . Ind. E n y . Chem. 11, 336.
Amerine, M. A., Roessler, E. B., and Filipello, F. 1959. Modern sensory
methods of evaluating wines. Hilgardia 28, 477.
Amoore, J. E. 1962. Identification of the seven primary odours. Proc. Sci.
Sect. Toilet Goods Assoc. 37, 1.
Amoore, J. E. 1963. Sterochemical theory of olfaction. Nature 198, 271.
Amoore, J. E., Johnston, J. W., and Rubin, M. 1964. The stereochemical
theory of odor. Sci. American 210 (Z),42.
Bailey, S. D., Bazinet, M. L., Driscoll, J. L., and McCarthy, A. I. 1961. The
volatile sulfur compounds of cabbage. J. Food Sci. 26,163.
Baker, G. A. 1954. Organoleptic r at i n g and analytical d a t a for wines ana-
lyzed into orthogonal factors. Food Research 19, 575.
Baker, R. A. 1961. Taste and odor in wat er: a critical review. Report
F-A2333, The Franklin Institute, Philadelphia, 159 pp.
Baker, G . A., and Amerine, M. A. 1953. Organoleptic ratings of wines
estimated from analytical data. Food Research 18, 381.
Baradi, A. F., and Bourne, G. H. 1951a. Localization of gustatory and
olfactory enzymes in the rabbit, and the problems of taste and smell.
Nature 168, 977.
Raradi, A. F., and Bourne, G . 11. 1951b. Theory of tastes and odors. Scirwec.
113, 660.
Baradi, A. F., and Bourne, G.H. 1959. Histochemical localization of cholines-
terase in gustatory and olfactory epithelia. J. Histochem. asid Cytochenz.
7. 2.
Bassette, R., Ozeris, S., and Whitnak, C. H. 1963. Direct chromatographic

analysis of milk. J . Food Sci. 28, 84.
Beets, M. G. J. 1957. Structure and odour. In “Molecular Structure and Or-
ganoleptic Quality.” pp. 54-90. Society of Chemical Industry, London.
Beets, M. G . J. 1964. A molecular approach to olfaction. I n “Molecular Phar-
macology.” (E. J. Ariens, ed.) pp. 3-51. Academic Press, New York.
Beidler, L. M. 1959. Mechanism of gustatory a n d olfactory receptor stimula-
tion. I n “Sensory Communication.” (W. A. Rosenblith, ed.) pp. 143-157.
MIT Press and John Wiley, New York.
Bernhard, R. A. 1961. Citrus flavor. Volatile constituents of the essential oil
of orange ( C i t r u s sinensis). J . Food Sci. 26, 401.
Boggs, M. M., and Hanson, H. L. 1949. Analysis of foods by sensory differences
tests. Advances in Food R e s e a r c h 2, 219.
Borsanyi, S. J., and Blanchard, C. L. 1962. Psychogalvanic skin response
olfactometry. Ann. Otol. Rhinol. & Laryngol. 71, 213.
Borsanyi, S. J., Blanchard, C. L., and Baker, F. J. 1962. Psychogalvanic skin
response as a method of objective olfactometry. Bull. School M e d . U n i v .
Maryland 47, 10.
Bozza, G., Calearo, C., and Teatini, G. P. 1960. On the making of a rational
olfactometer. Acta Oto-LaryngoL 52, 189.
Cheesman, G. H., and Kirkby, H. M. 1959. An a i r dilution olfactometer s u i t -
able for group threshold measurements. Q u a r t . J . E x p t l . Psychol. 11, 115.
Dawson, E. H., and Harris, B. L. 1951. Sensory methods for measuring differ-
ences in food quality. U. s. Dept. Agr., Agr. Info. Bull. No. 34, 134 pp.
Dawson, E. H., Brogdon, J. L., and McManus, S. 1963. Sensory testing of
differences in taste. I. Methods. Food Technol. 17 ( 9 ) , 45. 11. Selection of
panel members. Food Technol. 17 ( l o ) , 39.
Deininger, N., and McKinley, R. W. 1954. The design, construction, and use
of an odor test room. Am. Sac. T e s t i n g Materials Spec. T e c h . Publ. N o .
164, 23-32.
De Vries, H., and Stuiver, M. 1961. The absolute sensitivity of the human
sense of smell. I n “Sensory Communication.” (W. A. Rosenblith, ed.) pp.
159-167. MIT Press, and John Wiley, New York.
Ellis, B. H. 1961. A guide book f o r sensory testing. 55 pp. Continental Can Co.,
Inc., Chicago, Illinois.
Elsberg, C. A., and Levy, I. 1935. The sense of smell: I. A new and simple
method of quantitative olfactometry. Bull. Neurol. Inst. N . Y . 4, 5.
Fortunato, V., and Niccolini, P. 1968. “Olfatto e sue Correlazioni.” pp. 193-
212. Tipografia Dell’ Universiti, Catania.
Friedman, H. H., and Mackay, D. A. 1964. Odor measurement possibilities
via energy changes in cephalin monolayers. Proc. N . Y . Acad. Sci. 116
( 2 ) , 602.
Fuller, G . H., Steltenkamp, R., and Tisserand, G. A. 1964. The gas chromato-
graph with human sensor: perfumer model. Proc. N . Y . Anutl. Sci. 116
( 2 ) , 711,
Garner, W. R., and Hake, H. W. 1951. The amount of information in absolute
judgments. Psychol. R e v . 58,446.
Geldard, F. A. 1953, “The Human Senses.” pp. 207-208. John Wiley, New
York.
Gesteland, R. C., Lettvin, J. Y., Pitts, W. H., and Rojas, A. 1963. Odor speci-
ficities of the frog’s olfactory receptors. I n “Olfaction and Taste.”
(Y. Zotterman, ed.) pp. 19-34. Pergamon Press, New York.
Girardot, N. F., Peryam, D. R., and Shapiro, It. 1952. Selection of sensory
testing panels. Food Technol. 6, 140.
Gold, H. J., and Wilson, C. W. 111. 1963. The volatile flavor substances of
celery. J. Food Sci. 28, 484.
Gridgeman, N. T. 1955. Taste comparisons; two samples o r three? Food
Tecknol. 9, 148.
Guadagni, D. G., Buttery, R. G., and Okano, S. 1963a. Odour thresholds of
some organic compounds associated with food flavours. J. Sci. Food Agr.
14, 761.
Guadagni, D. G., Buttery, R. G., Okano, S., and Burr, H. K. 196313. Additive
effect of sub-threshold concentrations of some organic compounds associ-
ated with food aromas. Nature 200,1288.
Guilford, J. P. 1954. “Psychometric Methods.” 2nd ed. 597 pp, McGraw-Hill,
New York.
Gundlach, R. H., and Kenway, G. 1939. Method f o r the determination of ol-
factory thresholds in humans. J. Exptl. Psychol. 24, 192.
Hartman, J. 1954. A possible objective method f o r the rapid estimation of
flavors in vegetables. Proc. A m . Soc. Hort. Sci. 64, 335.
Hartman, J., and Tolle, W. E. 1957. An apparatus designed for the rapid
electrochemical estimation of flavors in vegetables. Food Technol. 11,
130.
Heinz, D. E., Pangborn, R. M., and Jennings, W. G. 1964. Pear aroma: relation
of instrumental and sensory techniques. J. Food Sci. 29, 756.
Hobbs, M. L. 1963. E. M. F. readings in the assessment of odour. Soap, P w -
f u m e r y & Cosmetics 36, 51.
Hofmann, F. B., and Kohlrausch, A. 1925. Bestimmung von Geruchsschwellen.
Biochem. Z . 156, 287.
Jackson, R. T. 1960. The olfactory pigment. J. Cellular Comp. Physiol. 55,
143.
Jennings, W. G., and Wrolstad, R. E. 1961. Volatile constituents of black pep-
per. J. Food Sci. 26, 499.
Jennings, W. G., Leonard, S., and Pangborn, R. M. 1960. Volatiles contribut-
ing to the flavor of Bartlett pears. Food Tecknol. 14, 587.
Jennings, W. G., Viljhalmsson, S., and Dunkley, W. L. 1962. Direct gas chrom-
atography of milk vapors. J. Food Sci. 27,306.
Johnston, J. W., and Sandoval, A. 1960. Organoleptic quality and the stereo-
chemical theory of olfaction. Proc. Sci. Sect. Toilet Goods Assoc. 33, 3.
Jones, F. N. 1953. A test of the validity of the Elsberg method of olfactometry.
A m . J . Psychol. 66, 81.
Jones, F. N. 1954. An olfactometer permitting stimulus specification in molar
terms. Am. J . Psychol. 67, 147.
Jones F. N. 1955. A comparison of the methods of olfactory stimulation: blast-
ing vs. sniffing Am. J. Psychol. 68, 486.
Jones, F. N., and Jones, M. H. 1953. Modern theories of olfaction: a critical
review. J. Psychol. 36, 207.
Kistiakowsky, G. B. 1950. On the theory of odors. Science 112, 154.
Kranier, A., and Twigg, B. A. 1962. “Fundamentals of Quality Control for

the Food Industry.” 512 pp. AVI Publ. Co., Westport, Conn.
Kristensen, H. K., and Zilstorff-Pedersen, K. 1953. Quantitative studies on the
function of smell. Acta Oto-LargngoL 43, 537.
Mackey, A. O., and Jones, P. 1954. Selection of members of a food tasting
panel: discernment of primary tastes in water solution compared with
judging ability f o r foods. Food Technol. 8,527.
Mateson, J. P. 1955. Olfactometry: its techniques and apparatus. J . Air Pollu-
tion Control Assoc. 5, 167.
McCarthy, A. I., Palmer, J. K., Shaw, C. P., and Anderson, E. E. 1963. Cor-
relation of gas chromatographic data with flavor profiles of fresh banana
fruit. J . Food Sci. 28, 379.
Miner, R. W. 1954. Basic odor research correlation. Ann. New Yorlc Acad.
Sci. 58, 13.
Moncrieff, R. W. 1961. An instrument for measuring and classifying odors.
J . Appi. Phgsiol. 16, 742.
Moncrieff, R. W. 1963. Psycho-galvanic reflexes to odours. Pcrfumrry Essent.
Oil Record 53, 313.
Morimura, S. 1934. Untersuchung iiber den Geruchsinn. TOhoku J . Exptl. Mcd.
22, 417.
Moulton, D. G. 1963a. Electrical activity in the olfactory system of rabbits
with indwelling electrodes. In “Olfaction and Taste.” (Y. Zotterman, ed.)
pp. 71-84. Pergamon Press, New York.
Moulton, D. G. 1963b. Pigment and the olfactory mechanism. Nature 195,
1312.
Mrak, V., Amerine, M. A., Ough, C. S., and Baker, G. A. 1959. Odor difference
test with applications t o consumer preferences. Food Research 24, 574.
Mullins, L. J. 1954. Olfaction. Ann. N . Y . Acad. Sci. 62, 247.
Naves, Y. R. 1957. The relationship between the stereochemistry and odorous
properties of organic substances. In “Molecular Structure and Organo-
leptic Quality.” pp. 38-53. Soc. Chem. Ind., London.
Nawar, W. W., and Fagerson, I. S. 1962. Direct gas chromatographic anal-
ysis as a n objective method of flavor measurement. Food Techno!. 16,
107.
Ottoson, D. 1963. Some aspects of the function of the olfactory system.
Pharmacol. Revs. 15, 1.
Ottoson, D., and von Sydow, E. 1964. Electrophysiological measurements of the
odour of single components of a mixture separated in a gas chromato-
graph. Life Sciences 3, 1111.
Ough, C. S., and Baker, G. A. 1961. Small panel sensory evaluations of wines
by scoring. Hilgardia 30, 587.
Ough, C. S., and Baker, G. A. 1964. Linear dependency of scale structure in
differential odor intensity measurements. J . Food Sci. 29,499.
Ough, C. S., and Stone, H. 1961. An olfactometer f o r rapid and critical odor
measurement. J . Food Sci. 26, 452.
Pangborn, R. M., and Dunkley, W. L. 1964. Sensory discrimination of f a t and
solids-not-fat in milk. J . Dairg Sci. 47, 719.
Pangborn, R. M., Berg, H. W., Roessler, E. B., and Webb, A. D. 1964. Influence
of methodology on olfactory response. Percept. and Motor Skills 18, 91.
Peryam, D. R. 1958. Sensory difference tests. Z i t “Flavor Research and Food

Acceptance.” pp. 47-66, A r t h u r D. Little and Reinhold, New York.
Pfaffmann, C., Schlosberg, H., and Cornsweet, J. 1954. Variables affecting
difference tests. In “Food Acceptance Testing Methodology Symposium,”
October 1954, p. 4. Natl. Acad. Sci., Natl. Research Counc.
Pippen, E. L., and Nonaka, M. 1960. Volatile carbonyl compound of cooked
chicken. 11. Compounds volatilized with steam during cooking. Food
Research 25, 764.
Rosano, H., and Scheps, S. Q . 1964. Studies on adsorption-induced electrode
potentials. Proc. N . Y . Acad. Sci. 116 ( 2 ) , 590.
Sawyer, F. M., Stone, H., Abplanalp, H., and Stewart, G. F. 1962. Repeat-
ability estimates in sensory-panel selection. J . Food Sci. 27, 386.
Schneider, R. A,, Costiloe, J. P., Vega, A., and Wolf, S. 1963. Olfactory thresh-
old technique with nitrogen dilution of n-butane and gas chromatog-
raphy, J . A p p l . Physiol. 18, 414.
Schutz, H. G. Vely, V. G., Iden, R. B., and Soliday, S. M. 1962. Relation of
odor to physicochemical properties of compounds. Quartermaster Food
mid Container I w t . Rept. No. 8, P-1115, 51 pp.
Shannon, C. E., and Weaver, W. 1963. “The Mathematical Theory of Com-
munication.” 117 pp. Univ. Ill. Press, Urbana.
Singleton, V. L., and Ough, C. S. 1962. Complexity of Aavor and blending of
wines. .I. Food Sci. 27, 189.
Spillane, J. D. 1939. Clinical investigation of olfactory function in brain
turnour patients. Bruin 62, 213.
Stewart, G. F. 1963. The challenge in flavor research-an editorial. Food
Technol. 17,5.
Stone, H. 1963a. Influence of temperature on olfactory sensitivity. J . A p p l .
Physiol. 18, 746.
Stone, H. 1963b. Techniques f o r odor measurement; olfactometric vs. sniffi-
ing. J . Food Sci. 28, 719.
Stuiver, M. 1958. Biophysics of the sense of smell. Ph.D. thesis. Groninger
University, Groningen, Holland.
Stuiver, M. 1960. An olfactometer with a wide range of possibilities. Acta
Oto-Laryngol. 52, 135.
Teranishi, R., Corse, J. W., McFadden, W. H., Black, D. R. and Morgan,
A. I., Jr. 1963. Volatiles from strawberries. I. Mass spectral identifica-
tion of the more volatile components. J . Food Sci. 28, 478.
Tilgner, D. J. 1961. Dilution tests f o r odor and flavor analysis. Food Technol.
16 ( 2 ) , 26.
Tilgner, D. J. 1962. Anchored sensory evaluation tests-a s t a t u s report.
Food Trchnol. 16 ( 3 ) , 47.
Tucker, D. 1963. Physical variables i n t h e olfactory stimulation process. J .
Gen. Physiol. 46, 453.
Walford, R. W., Hattaway, 6. A., Alberding, G.E., and Atkins, C . D. 1963.
Analysis of flavor and aroma constituents of Florida orange juices by
g a s chroniatography. J . Food Sci. 28, 320.
Webb, A. D., and Kepner, It. E. 1962. Thc aroma of flor sherry. A m J . E ? d .
Vificult.13, 1.
Wenzel, B. M. 1948. Techniques in olfactometry: A critical review of t h e last
one hundred years. Psychol. Bull. 45, 231.
Wenzel, B. M. 1949. Differential sensitivity in olfaction. J . E x p t l . Psychol. 39,

129.
Wenzel, B. M. 1955. Olfactometric method utilizing natural breathing in
odor-free “environment.” Science 121, 802.
Whipple, H. E. 1964. Recent advances in odor: theory, measurement, and
control. Ann. N e w York Acad. Sci. 116, 357.
Winter, M., Stoll, M., Warnhoff, E. W., Greuter, F., and Buchi, G. 1963.
Volatile carbonyl constituents of dairy butter. J . Food Sci. 28, 554.
Woodrow, H., and Karpman, B. 1917. A new olfactometric technique and some
results. J . Exptl. Psychol. 2, 431.
Wright, R. H. 1957. Odour and molecular vibration. In “Molecular Structure
and Organoleptic Quality.” pp. 91-102. SOC.Chem. Ind., London.
Wright, R. H. 1963. Odour of optical isomers. Nature 198, 782.
Wright, R. H. 1964. “The Science of Smell.” 164 pp. Basic Books, New York.
Young, P. T. 1961. “Motivation and Emotion.” 648 pp. John Wiley, New York.
Zilstorff-Pedersen, K. 1955. Further studies on blast-olfactometry. Acfa Oto-
L a r y n g o l . 45, 268.
Zlatkis, A., and Sivetz, M. 1960. Analysis of coffee volatiles by g a s chroma-
tography. Food Research 25, 395.
Zwaademaker, H. 1895. “Die Physiologie des Geruchs.” 324 pp. Leipzig,
Germany.
Zwaademaker, H. 1921a. A camera inodorata. Perfumery Essent. Oil Record
12, 243.
Zwaademaker, H. 1921b. Olfactometry. Perfumery Essent. Oil Record 12,
308.
Zwaademaker, H. 1926. I n “International Critical Tables.” ( E . W. Washburn,
ed.) pp. 358-361. McGraw-Hill, New York.

Stone 1965

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stone 1965

Uploaded by

Copyright:

Available Formats

TECHNIQUES FOR SENSORY EVALUATION OF FOOD ODORS

ROSE MARIE PANGBORN AND c. s. O U C H

A. IMPORTANCE OF ODORS IN FOODS

coverage in this country, consult the proceedings of symposia on

B. NEED FOR CONTROL TECHNIQUES

C. CURRENT THEORIESO F ODOR DETECTION

cal properties of odors that influence olfaction in humans. Multiple

has shown that olfactory tissue extracts do not fluoresce until

II. TECHNIQUES OF MEASUREMENT

different scientific disciplines has resulted in some confusion and

positive pressure, olfactometer. Saturated vapor was stored over

tions were varied by changing proportions of each. For two dif-

ing to oxidation products and not the stimulus is conceivable. In

remains to be demonstrated. In terms of the study of central and

manometers might have improved the technique. The value of

F I G . 1. Schematic diagram of the olfactometer. A ) Incoming conditioned a i r ;

The instrument described by Bozza et al. (1960) required the

been successfully tested with this apparatus. This air-dilution type

FIG.2. Photograph of the olfactometer. Reproduced through the courtesy

rates, temperature, humidity, and stimulus concentration than has

A great number of other instruments and equipment have been

unawareness of the availability of appropriate designs. Often there

those present with perfumes, wines, whiskeys, cheeses, etc., the

Since loss of interest or motivation by a judge can greatly alter

Bernhard ( 1961) as volatile constituents of certain California

fractions may not be identical to the injected mixtures. At pres-

nitric acid saturated with chromix oxide). Spectrophotometric

Moncrieff (1963) has also studied PGSR response t o odor stirn-

and the food itself. The problem of test method is complex

enclosed glass booth. The judge had a high level of accuracy in

111. AREAS OF NEEDED RESEARCH

Elucidation of the mechanism of the olfactory processes would,

ment and testing of simple but improved olfactometers for use in

Bassette, R., Ozeris, S., and Whitnak, C. H. 1963. Direct chromatographic

Kranier, A., and Twigg, B. A. 1962. “Fundamentals of Quality Control for

Peryam, D. R. 1958. Sensory difference tests. Z i t “Flavor Research and Food

Wenzel, B. M. 1949. Differential sensitivity in olfaction. J . E x p t l . Psychol. 39,

You might also like