for the Conference on

a symposium sponsored by
APRIL 26-27, 1971
                         Arthur D Little, Inc.



           A Symposium Sponsored by

        Environmental Protection Agency
         Air Pollution Control Office
            Arthur D. Little, Inc.
         Cambridge, Massachusetts USA
               April 26-27, 1971




      Odorant Sources in Air Pollution                             1
           Dr. Kenneth D. Johnson

      Properties of Odorants                                       9
           Dr. Amos Turk

      Predicting Odor Frequencies by Dispersion Calculations      15,
           Dr. Ulf Hb'gstrom


      Neurological Responses to Odorants                          25
           Prof.Dr. Helmut Altner

      Sensory Evaluation of Odors in the Ambient Air              35
           Thomas Lindvall, M.D.

      Method and Theory of Odor Preference                        59
           Dr. Trygg Engen

      Physiological Responses to Odorants                         91
           David A. Kendall


      On Annoyance Reactions Observed Within Human Populations   123
           Dr. Erland Jonsson

      Studies of Public Opinion of a Traffic Odor                131
           Mr. Karl J. Springer

      Property Value Differentials as a Measure of               161
         Economic Costs Due to Odors
           R. David Flesh

      Health. Effects and Annoyance Reactions to Pulp Mill        177
         Odor in a Rural Community
           Margaret Deane and John R. Goldsmith


     Summary of Technical Problems                              201
          Lars Friberg, M.D.

     Summary of the Problems in Odorous Air Pollution           213
        in Japan
          Takeo Suzuki, M.D.

     Administrative Mechanisms Available for Control of         221
        Odorous Compounds Under the U.S. Clean Air Act
        As Amended
          Dr. D. S. Earth


                         Kenneth D. Johnson

     It is always satisfying to a lead-off speaker to a symposium
such as this, if he is able to provide a nearly organized and cate-
gorized arrangement of the subject matter to be discussed by the ensuing
speakers.  It provides him with a sense of having drawn a road map
steering the participants directly to their destination.  They may,
or may not, make much progress along that route, but after all, you can't
expect him to do the whole job, can you?
     On this occasion, I can not bask in any such glow of self-
satisfaction.  I cannot offer you a chart neatly ruled off with lines of
latitude and longitude, and with primary and secondary route markers
carefully inscribed on well-defined courses.  Although we share an ill-
defined sense of the direction in which we wish to move, I cannot even
provide you with a reliable compass.  The best I can do is to point out
some clumps of moss in the thick forest of often conflicting, confusing
and nebulous observations that constitute our data base.  Hopefully,
from the location and character of these mossy growths, we can keep
oriented in these woods even though no Pole Star is visible.
     The difficulty in attempting an all-inclusive classification of odors,
or more properly, of odorous substances, is that no property is common to
them all, other than their propensity to stimulate certain nerve endings
in the nasal passages of homo sapiens.  At the present state of knowledge,
or even of broadly applicable theory, we are not even sure of the mechanism
by which these molecules initiate the nerve impulse, nor do we know the
means by which our exquisitely discriminatory sensory apparatus is able
to distinguish among the thousands of odorants—not only the naturally
occurring ones that may have exerted an influence on the evolutionary
development of the odor sense,  but the products of the organic syn-
thesist's art, as well.
     Odors are nothing new to society, odor "problems" are of far
shorter duration.  An odor becomes a problem when there exists a

possibility of a solution.  Until that time, it is merely a facet of the
environment to which its occupants must become acclimated.  Odors are
now recognized more frequently as problems because we now have technology,
and the capability of developing new technology, that gives us hope of
elimination of any particular odorant from our environment.  Not all odors,
mind you, nor necessarily at all times. ' One of the still unsolved prob-
lems, and one that, in particular, is facing this group before me today,
is the matter of how we decide which odor problems require the initial
efforts toward solution, and what distribution of available effort among
the many coexisting odor problems will produce the greatest environmental
improvement.  An improvement measured by the reduction in some still un-
defined "weighted average" of residual odor nuisances.
     Your program chairman has asked me to discuss Odorant Sources in
Air Pollution.  Perhaps the last two words of this title are redundant.
If we accept the definition of "pollution" as the presence in the air of
some material that detracts from a beneficial use of the atmosphere, I ex-
pect that any odor, no matter how pleasing it may be to most people on
most occasions, could, under some conditions, be a source of annoyance,
or interference with desired activities, even if it is a lingering scent
of Chanel No. 5 on a man's coat lapel when his wife uses Arpege.
     Whether a given odorant evokes pleasant or disagreeable emotions in
a subject is strongly dependent upon the experiences with which previous
exposures to that odor are associated.   Even odors that are normally
considered as generally unpleasant or obnoxious, may, in an appropriate
frame of reference, stimulate pleasurable reactions.  I confess that,
for me, the occasional perception of the mercaptans discharged by an
aroused skunk, when appropriately diluted by travel on an evening breeze,
or the characteristic fresh manure odor of a dairy milking barn stir
quite pleasant feelings of nostalgia for the bucolic scenes of a rural
childhood, rather than impinging upon my nostrils as the stenches that
these odorants are normally considered to constitute.
     However, whether our subjective evaluations are conditioned by
current social patterns or genetic factors developed over the long course
of man's development, it must be recognized that certain kinds of odors
are generally regarded as unpleasant or offensive.  Air pollution

abatement programs, as expressions of political decisions based upon
the value judgements of large sections of the population, will inevitably
first focus their odor control efforts on those situations where sub-
stantial populations are exposed to high frequencies or intensities of
such odors.  Our industrial society brings large quantities of odorant
materials into close proximity with social units of high population
density, thus it often has exacerbated odor nuisance problems at a time
when society has become far less tolerant of odors in our environment.
     This intolerance has been strongly stimulated by outstanding
accomplishments of our technology in minimizing the odors that arose in
earlier times as the cities outgrew their ability to handle garbage and
sewage.  These were the major sources of a pervasive effluvia that
characterized our cities of a century or more ago.  If I may be for-
given for allowing a reference to water pollution to intrude into this
meeting, I would like to quote from Coleridge's Cologne;
               In Koln, a town of monks and bones,
               And pavements fang'd with murderous stones
               And rags and hags, and hideous wenches;
               I counted two and seventy stenches,
               All well defined, and several stinks.
               Ye Humphs that reign o'er sewers and sinks,
               The river Rhine, it is well known,
               Doth wash your city of Cologne;
               But tell me, Nymphs, what power divine
               Shall henceforth wash the river Rhine?
     Under some circumstances, even entire countries can be contaminated
by the stench of decaying animal matter.  In Exodus 8:13, 14 we read
in the description of the plague of frogs:  "... and the frogs died out of the
of the houses, out of the villages, and out of the fields.  And they gathered
gathered them upon heaps:  and the land stank."
     Today we are unlikely to suffer a plague of frogs, but it is
quite possible that a plant processing frog legs for the frozen foods
trade, might, as a consequence of its waste accumulation and disposal,
constitute a neighborhood nuisance.

It would not be particularly constructive for me to attempt a com-
prehensive catalogue of odor sources, or to try to characterize the
problems associated with each.  Even a casual review of the literature
would reveal that almost every major industrial operation has been
accused of being an odor nuisance source at one time or another.  I
will try to focus our attention on industrial sources, since, as I have
earlier^pointed out, these do often have greater social impact and
are associated with institutions that are more likely to have the
economic and technological resources to do something about them.
     The greatest defense against the creation of an odor nuisance from
an industrial operation is a tightly closed system, and the hardest
systems to close are those that involve manual handling of process
materials.  Such operations on plant or animal material, in which
the requirement for manual steps as trimming, sorting, selecting, etc.
and from which putrescible wastes result, have a long history of odor
problems.  They face a most difficult problem in developing technology
for odor reduction and control.  Meat packing and leather tanning are
classical examples.  The variability of the raw material and the need
for the operator to exercise judgment precludes the automatization of
most steps in their processes, and make the full enclosure of the pro-
cessing operations unfeasible.
     After the biological processes of putrefaction, perhaps the thermal
or pyrolytic processes are the next most commonly subject to nuisance
odor problems.  Whereas the final products of oxidation of organics
are generally of rather low odor intensity, the intermediates generated
by heat, either in the absence of adequate oxygen, or at temperatures
not high enough to achieve complete oxidation, often tend to be highly
odorousv, and offensive or irritating.  Among the industrial processes
that would fall into this classification, we note:  coke making,
asphalt blowing and asphaltic tile and roofing manufacture, heat curing
of plastics and protective coatings, core baking and metals founding,
meat smoking and coffee roasting.  To these we may add the poorly
controlled or inadequately designed combustion processes, such as
lumber mill wastes incineration of tepee burners, or the almost, but

not quite complete oxidation of petroleum fuels in a diesel engine,
that discharge such pyrolyzed materials as a foreseeable, but un-
intentional consequence of their operations.
     It would be presumptuous of me to attempt to summarize to this
audience the odor sources and problems control of the pulp and papper
industry.  Its odor sources run the gamut from those growing out of
the biological degradation of wastes, through the range of incomplete
oxidation products from their recovery furnaces, to the mercaptans and
sulfides synthesized as unwanted by-products in the Kraft process.
The scale of their operations, and the intensities and qualities of
their odorants, have combined to bring this industry into a quite un-
welcome prominence in any forum in which industrial odor problems
are discussed.
     Many of our industrial odor problems are related, not to an
ill-defined and highly complex mixture of products resulting from a
biological or thermal process, but to a specific chemical intentionally
introduced into, or produced by, an industrial operation.  The carbon
disulfide of the viscose rayon process, or the plastics monomer, ethyl
aerylate, are typical examples.
     As a broad generalization, one could predict that wherever the
nuisance odor threshold of a given chemical lies well below that con-
centration that constitutes a serious hazard from the toxicity or fire
or explosion standpoint, some user, at some time, may be expected to
allow it to create an odor problem.  The frequency and extent of such
odor problems will be determined more by the number and distribution
of that chemical's uses and users, than by its intrinsic properties
as an odorant.  Of course, other things being equal, those chemicals
with the highest ratio between vapor pressures and detection thresholds,
and the most unpleasant odor notes, may be expected to generate the
greatest number of complaints.
     I do not wish to dwell at any length on the peculiar position
occupied by compounds of sulfur in the odor spectrum.  The high in-
cidence of malodorous compounds in this chemical family was noted
early.  This propensity, when combined with the high susceptibility
of sulfur and its compounds to biological and chemical interconversions,

makes the sulfur compounds frequent culprits in industrial odor problems.
Thus even as inert and inoffensive a material as elementary sulfur may,
in storage, be the source of substantial quantities of hydrogen sulfide
as a result of microbial action.  The mercaptans present in crude
petroleum, the thiophenes of coal tars, as well as the'sulfides and mer-
captans present in the products of putrefaction, are all frequent sources
of odorous air pollution.
     If we mention sulfur as of frequent occurrence in compounds of
strong and often unpleasant odor, we must similarly acknowledge
nitrogen.  As a major essential element in all living things, it is
a ubiquitous component of all industrial raw materials derived from
plant or animal sources, and, like sulfur, is subject to facile micro-
bial conversion into compounds that are both readily volatile and highly
odorous.  As in the case of sulfur, the compounds in which this element
is in the reduced valence state are usually the more potent trouble
makers.  Thus methyl amines, whether evolved from fish wastes or
synthesized as raw materials for surfactant manufacture, are a fre-
quently observed source of odorous air pollution.  Again, in parallel
to what we have observed in the case of sulfur, the heterocyclic nitro-
gen compounds also have both low odor thresholds and generally offensive
odor notes.
     In a final generalization, let me distinguish between those in-
dustrial odor problems associated with the emissions from a plant stack
or vent pipe, and those related to "fugitive" sources.  There are many
present instances in which the gases or vapors discharged from such a
vent are odorous, and in which there is a valid rationale for, and
reasonable prospects for successful application of, a nuisance prevention
strategy based upon dispersion and dilution.  The frequent failures ex-
perienced by those attempting to employ this technique are often due
to one or both of two factors: a lack of appreciation for the uncer-
tainties in extrapolation of dispersion parameters from the 10 to 20
minute periods on which most dispersion equations are based, to the
few seconds adequate to stimulate the odor senses, and to inadequate

consideration of the influence of aerodynamically induced eddies in the
lee of man-made structures, which are conventionally ignored in such
     Of more frequent occurrence, and often constituting a far more
difficult control problem, are the multiple low level sources scattered
throughout a plant, such as leakages around pump shafts and valve stems,
vapor losses from storage tanks, displaced head space gases when kettles
or shipping containers are filled, and minor spillages occurring during
transfer operations.  Many of these sources can be considered to be
housekeeping problems, but particularly when the odor thresholds are
low, consistent control of these sources may be extremely difficult.
This is especially the case when the odorants concerned have relatively
low vapor pressures, are strongly adsorbed on and slowly released from
porous surfaces, and have a pervasive odor note.
     The multifarious nature of the odor sources, and the absence of
a coherent unifying principle by which they may be rationally inter-
related, makes both the technical and political attacks on the problem
particularly difficult.  Hopefully, as an outgrowth of conferences
such as this, odors, their description, quantitation and control can
be reduced to a science, from its current status to more or less of
an art.  But our society is not willing to wait meekly until that day
arrives, and we may expect ever increasing pressures to eliminate or
control odor problems, regardless of complexity or difficulty,  on
empirical and ad hoc bases.  Our current studies must not ignore our
responsibilities to contribute to the effectiveness of this pragmatic

                        Properties of Odorants
                               Amos Turk

Odor sources may be classified according to their degree of confinement,
their character or quality, their hedonic impact, and their persistence
or pervasiveness in the atmosphere.  All of these factors characterize
the odor "dose: in ways that significantly affect the human response at
various distances from the source.

Degree of confinement.  A "confined source" is usually considered to be
one which can be characterized as to location and volumetric rate of dis-
charge.  Examples are continuous effluents from stacks and vents, and
discontinuous discharges such as those from rupture disks, pressure
relief valves, or displacements of vapor during transfer and filling
operations.  Confined sources lend themselves to sampling and analysis.
If the analytical information deals with sensory odor evaluation, or
with chemical measurement of odorants, the results can be used as a
basis for establishing relationships between source and ambient air, which
is to say between dose and response.  If the analysis also quantifies
the oxidizable, soluble, or adsorbable components of the effluent, the
results can be used to predict how effectively the odor can be controlled
by methods such as oxidation, scrubbing, or adsorption, which is to pre-
dict a relationship between altered dose and altered response.

Examples of unconfined sources, for which quantitative measurements and
hence rational dose/response predictions are more difficult, are drainage
ditches, settling lagoons, outdoor chemical storage areas, garbage dumps,
and ground that has long been impregnated with odorous matter such as
asphalt.  It has been suggested that such sources may be represented by
an imaginary emission point such that, if all the odor from the unconfined
area were being discharged from the "emission point," the dispersion pattern
would just include the unconfined source.

Odor quality or character
The characterization of odor sources by quality has proved  to be a very
difficult task, one that is far from having been resolved.  The approaches
have been (a) to describe all odors in terms of a limited number of
"primary" odors; (b) to describe all odors in terms of their degree of
similarity to one another; (c) to describe a particular subset of odors
in terms of an arbitrary selection of descriptive qualities which are
related to the actual composition or sources of the odors;  (d) to classify
odors by the frequency of complaints lodged against them.

The "primary" odor approach is represented by systems such  as those pro-
posed earlier by Crocker and Henderson, and more recently by Amoore.
These systems have been reviewed by Harper (1).  None of them has been
shown to be successful in predicting dose-response relationships in air
pollution situations.

The classification of odors by similarities sets all odors  in a conceptual
"space" such that the nearness of any two odors in the space represents
the degree of similarity between their qualities (1).  Thus, the character
of an odor is described not in absolute terms, but by identifying the
odors near it in "odor space."  Some three-dimensional models have been
constructed to illustrate this idea.  It has been found that the single
most significant attribute which determines the character of an odor by
this conceptual system is its hedonic impact.

The applications of a classification system to a particular subset of
odors is exemplified by the diesel exhaust odor standards (2) to be
described later by Karl Springer.  This approach has also been used in
the context of community odor problems in area containing a limited
number of odor sources (3).

Odor complaint frequency data have been collected occasionally (4),  but
no study has yet been sufficiently extensive to predict whether such in-
formation could provide a meaningful inventory of source types.

Hedonic impact
Odor preferences have been studied sporadically, usually with reference
to a specific and limited group of odorants and a limited population.
The subject has been recently reviewed by Moncrieff  (5), who has also
added considerable data of his own.  He also presents a summary embodied
in 124 "rules" of odor preference, but most of these are either so well
known as to be considered obvious, or over-generalized to the point of
being speculative, or lacking predictive value to the point of being
meaningless.  The general problem of hedonic scaling, including organiza-
tion and treatment of data and measurement of hedonic impact, has been
considered by Johnston and Rubacky (6).  Certainly the most significant
single study of hedonic rating of environmental odors is that conducted
on diesel exhaust by Springer, to be discussed later in this conference.

Odor persistence or pervasiveness
The question of odor persistence or pervasiveness is critical to the
establishment of dose1/response relationships after the dilution of odorant
that normally occurs between source and subject.   The general question
of establishment of concentrations will be taken up in the two following
papers by Dr. Sullivan and Dr. Hb'gstrom.  Let us assume that odor
intensity, I, is related to odorant concentration, C, by the Stevens
where B is a constant.

We now consider two different odorants of concentrations C, and C~, and
odor intensities 1^ and 12, respectively, and for each odorant, a target
intensity Ii  and 12,.» which must be reached by reducing their concentration
to GI  and C2t, respectively.  (The target concentrations may be but are
not necessarily some  functionally described threshold values.)

Using the notation described in the Stockholm 1970 report, Z = C/Ct,
and pZ = -log Z, we have,
What these equations tell us is that, for equal reductions of the response
to two odors,
                     /I-   N
the required reduction in concentrations, Z1 and Z_ are the same only if
the Stevens exponents, BI and £„, are the same.  The earlier work of
Stevens had implied that the g exponents are characteristic of a particular
odorant, although there may be considerable variation among different
odorants.  Values ranging as low as 0.13 and as high as 0.6 have been
reported C7).   More recent results have shown that there are also
variations in the 3 exponent among different observers.  Some of these
problems may be treated in a subsequent discussion by Dr. Engen.  Certainly
this aspect of properties of odorants deserves more study if meaningful
dose/response relationships are to be established.


1.  R. Harper, E. C. Bate Smith, and D. G. Land "Odour Description and
    Odour Classification,"  American Elsevier Publishing Company, Inc.,
    New York, 1968.
2.  Turk, A., Wittes, J.,  Reckner, L. R. and Squires, R. E. "Sensory
    Evaluation of Diesel Exhaust Odors,"  National Air Pollution Control
    Administration Publication No. AP-60.
3.  Turk, A., and Mehlman, Stanley "Correlations Between Instrumental
    and Sensory Characterizations Odors,"  ASTM STP 440, American
    Society for Testing and Materials, 1968, pp. 27-35.
4.  W. F. Kerka and E. R. Kaiser, J. Air Pollution Control Assn. _7,
    297 (1958)
5.  R. W. Moncrieff, "Odour Preferences"  John Wiley, New York, 1966
6.  J. W. Johnston, Jr., and E. P. Rubacky, "Hedonic Appraisals of
    Environmental Odors,"  Chapter in Human Responses to Environmental
    Odors, in preparation.
7.  Psychophysical Functions for Twenty-Eight Odorants by Berglund,
    Engen, Berglund and Ekman, Psychological Laboratories, Univ. of
    Stockholm, April 1970

        Predicting odor frequencies by dispersion calculations
                             Ulf Hb'gstrb'm
                       Department of Meteorology
                  The University of Uppsala, Sweden

The problem of making odor dispersion calculations is basically the same
as making calculations of the dispersion of any air pollutant.  The basic
requirement is that odor is quantified in terms of concentrations in the
ambient air and in terms of corresponding emission rates and that the
odorous gases do not undergo significant chemical reactions or physical
separations during the atmospheric transport from the source to the re-
ceiver (or if such processes occur they can be specified quantitatively).
There is however one particular point in the process of making odor
dispersion calculations which is critical.  When making 'ordinary' dis-
persion calculations the j^amp^iiig^^me can be chosen within fairly wide
limits, the result still being meaningful.  But when making odor dis-
persion calculations one must take into account the fact that the nose
reacts fairly fast, so that a sampling time of the order of magnitude of
seconds is relevant.  The failure of all earlier attempts (see e.g.
Wohlers 1963) to predict odor frequencies with the aid of dispersion
calculations is primarily due to lack of stringency on this crucial point
(apart from possible deficiency in the description of the emission).  The
importance of choosing a. correct sampling time for the calculations is
illustrated in Fig. 1.  The irregular curve represents the concentration
of the odorous gas as recorded by some instrument sited in the lee of the
emitting source.  The absolute odor threshold is indicated in the figure
and so is the one hour mean concentration value.  As seen from the figure,
the concentration exceeds the odor threshold several times during the
hour in spite of the fact that the hourly mean concentration is below the

When evaluating dispersion calculations against field observations of
odor great caution must be taken, so that the field observations are ex-
pressed in the same terms as the results of the calculations.  It is

probably almost impossible for an observer to tell with any accuracy for
how long during an observing period of any duration he actually felt
odor.  Entirely wrong results are of course obtained if the observer is
asked to tell whether he felt odor or not during any particular hour.  The
incident illustrated in Fig. 1 would certainly give the answer "Yes, I
felt odor" - but the threshold was exceeded only during a minor part of
the hour.  Thus one would expect that the calculations which give pro-
portion of "odor time" to total time seriously underestimate "the odor
frequency" when compared to that kind of field observations.  A fair
comparison can best be done when the observations are performed as in-
stantaneous samples at previously fixed points of time.

In the above discussion a single odor threshold was assumed.  As will be
seen below it is quite feasible today to make dispersion calculations that
give frequencies of 'odor threshold exceedings1 in exactly that sense.
It may be that this measure is not very relevant as a description of
odorous impact on a population.  There are however also possibilities to
make odor prediction in much more sophisticated terms.  But before we
discuss this we will give the outlines of 'the absolute odor threshold
dispersion model' (see Hogstrom 1971 for a more detailed discussion),
which is basic also to more advanced treatments.

Method_for_the_d±S2ers±pn _£a_lcjulations_,_  When the sampling time is of order
of jme hour_ the plume geometry can be considered rather simple:  the
plume rises first, as consequence of buoyancy, but levels off after some
distance, thus obtaining an almost horizontal mean axis (Fig. 2a).  The
distribution of the concentration of pollutants in the plume is nearly
Gaussian, measured at a fixed distance from the source in any direction
perpendicular to the axis.  The standard deviation, both in the vertical
and in the horizontal cross wind (lateral) direction is a function of
stability and of wind speed.  It also depends on the height of the plume
centerline above the ground and of the roughness of the ground beneath.
The behaviour of the standard deviations of the concentration field for
a sampling time of about one hour is fairly well known quantitatively
(for a comprehensive survey see Meteorology and Atomic Energy 1968).  In


order to be able to determine the ground level concentration field the
plume rise, A h (Fig.2a) must be known.  This quantity is a function of
the rate of heat emission and velocity from the chimney but also of
meteorological variables:  wind speed and stability.  Intense research,
both theoretical and experimental, has improved our knowledge of the plume
rise considerably during the last few years '(see e.g. Briggs 1969).

In the case of a .ve£y_sho_rt_ sampling time the plume geometry is much more
irregular than in the case of one hour sampling time.  The most charac-
teristic feature is that the mean axis of the plume has an irregular ap-
pearance (Fig. 2b).  Furthermore, this form varies all the time in an
entirely random way.  This variation, together with the corresponding
variation in the horizontal plane, is the cause of the short term varia-
tion in concentration displayed schematically in Fig. 1.  In order to
determine the frequency of odor in a given point the above-mentioned
short term variation in plume geometry must be described in mathematical
terms.  This is possible because we know that the distribution of matter
relative to the fluctuating axis as well as the distribution of the locus
of this axis in a coordinate system which is fixed relative to the
ground, is approximately Gaussian (Hb'gstrom 1964).

In a given meteorological situation the standard deviation of the distri-
bution of concentration relative to the moving axis are empirically known
functions of the distance from the source.  We then have the necessary
information to describe mathematically the ground level concentration
pattern at a given instant, that is when the plume centerline is situated
at a certain height above the ground.  But the standard deviation of the
locus of the plume centerline is also normally distributed with a stan-
dard deviation that varies in a known way with distance from the source.
It is then possible to establish a concentration frequency distribution
for that particular meteorological situation.  But the standard deviations
mentioned above vary with meteorological factors (HSgstrb'm 1964, Meteoro-
logy and Atomic Energy 1968) in a way that is fairly well known.  Con-
centration frequencies can then be calculated for each of a number of

representative meteorological situations.  The total odor  frequency  is
then obtained when the frequencies of these meteorological situations
at the site are taken into consideration.

The method outlined above has been tested in a field study (Lindvall
1970 and Hogstrom 1971).  Sensory source intensity analyses combined with
meteorological dispersion calculations were used to predict the frequency
with which odors will be  discernible in the ambient air at various dis-
tances from a pulp mill.

Predicted odor frequencies in the area were compared with  actual observa-
tions of odor during the same period.  Controlled observations by specially
trained subjects demonstrated a striking degree of correlation with pre-
dictions for the area of observation thus confirming the validity of the
sensory analytical method and the dispersion calculations  under practical
conditions.  For distances up to 5 km from the factory there were hardly
any significant differences between calculated and observed odor fre-
quencies.  As the distance from the source increased odors were ob-
served more frequently than predictions would indicate, the factor between
observed and calculated frequency rising to somewhat above 2 at 20 km.
The reason for this is not clear.  Several explanations are possible.
The odorous gases may undergo chemical reactions or physical separations
that changes the odor threshold.  Psychological effects may be important.
Or there are weaknesses in the dispersion calculations.  The latter seems
at first to be very reasonable.  A relevant analysis, however, shows
that the calculations are surprisingly insensitive to the  choice of prop-
er values for the dispersion parameters, that is the standard devia-
tions of the locus of the plume centerline.  A possible explanation could
be that the dispersion model is physically oversimplified.  If one assumes
that the mean instantaneous shape of rising plume elements differs signi-
ficantly from that of sinking ones, then the calculated frequencies can
easily be brought in accordance with the observed ones.  Such a model
is physically reasonable, but there are no experimental data to verify

The above method gives the average frequency of concentrations  above  an
absolute odor threshold.  But it can be modified easily  to  give more
sophisticated measures of odor, if such are desired  in order  to have  a
more valid description of the dose.  The proper shaping  of  such measure
is a task for the gathered experts of this meeting.  Here follows just
a few examples to illustrate how the above diffusion model  can  be modi-

It is quite possible, for instance, to make calculations in terms of
frequency of "odorous hours", with "an odorous hour" being  specified  as
an hour with the odor threshold being exceeded during more  than a speci-
fied proportion of that hour. — It is also possible to  give  a  full
cumulative frequency distribution of (short time) concentrations, which
can be helpful if it is desirable to describe the dose not  only in terms
of one threshold concentration.

The model discussed so far is ideal in the sense that it considers the
emission from one single chimney only.  If there are several  chimneys in
an area emitting _the_ j>ame_ odorous gases (although in widely  varying
amounts) the model can be easily modified to handle this, because the con-
centration contributions from the different sources can  then  be added to
give the resultant concentration, which is considered odorous if it ex-
ceeds a certain limit exactly as before.  But if the chimneys emit
di.f_feren_t kinds of gases, the case is far more complicated.   It can of
course be treated in terms of absolute odor threshold exceedings, if
sensory analysis has given odor thresholds for all possible concentration
combinations of the different gases.   In the case of a limited  number of
chimneys emitting a limited number of odorous gases, a practically useful
result can no doubt be obtained from a sensory analysis which contains
only fairly limited number of combinations of concentrations  of the
various gases.

The above treatment is ideal also in terms of topographic influences.
The ground has been considered flat,  although the influence of  roughness
on the dispersion is treated quantitatively in the model.   In the case

of hilly countryside the model is subject  to  the same weakness  as  any
dispersion calculations, and the success of the calculations depends to
a large extent on the ability to assess the local pattern of wind  flow.
Experienced meteorologists can often make  a qualified guess about
such patterns.  To measure the flow is often  difficult, because it may
be so complicated that a large number of measuring stations are required.
It is possible that a dynamical flow pattern  model is a practicable
solution in some cases.  It means that the dynamical equations  are solved
numerically for a proper set of boundary conditions.  Such a model is
in the process of being developed for the  Oslo area in Norway  (not in
particular for odorous dispersion but for  air pollution in general).


Briggs, G.A., Plume rise, U.S. Atomic Energy  Commission, Div. Techn.
             Inform., Washington, 1969.
Hb'gstrb'm, U., An experimental study on atmospheric diffusion, Tellus,
             1964, 16,205.
Hb'gstrb'm, U., A statistical approach to the air pollution problem  of
             chimney emission, Atmos. Environm. 1968, 2, 251.
Hb'gstrb'm, U., A method for predicting odour frequencies in the  surroundings
             of a point source.   Manuscript  submitted to Atmos. Environment,
Lindvall, T., On sensory evaluation of odorous air pollutant intensities.
             Nord. Hyg. Tidskr. 970, supp.2.
Meteorology and Atomic Energy 1968, U.S. Atomic Energy Commission, Div.
             Techn. Inform., Washington, 1968.
Wohlers, H., Odor intensity and odor travel from industrial sources.
             Int. J. Air. Wat. Poll. 1963, 7, 7.1.

Figure 1.  Schematic drawing of the variation in concentration  of  the
odorous gas emitted by an isolated stack as recorded by some ideal instru-
ment sited in the lee of the source.  The  absolute odor threshold  and the
one hour mean concentration value are indicated.

Figure 2.  Schematic drawing of plume geometry, the concentration distri-
bution measured at two different sampling times.  In 2a the sampling time
is of the order of one hour, in 2b very short (seconds or minutes) ho is
the stack height, Ah is the virtual increase in height due to the effluent's
heat content, motive energy etc.


              Fig,   a

                  Neurological Responses to Odorants

                               H. Altner
             Fachbereich Biologie, Universitat Regensburg

1  The input to the central nervous system
It is rather difficult to give a survey of the neurological responses
to odorants which can claim to be approximately complete and valid for
more than only the few species of animals normally used in laboratories.
This is partly due to the fact, that it is rather difficult to overlook
the input to the central nervous system (CNS) coming from the olfactory
Beginning with sensory input only few investigators have succeeded in
recording responses from single olfactory sensory cells in vertebrates
(GESTELAND et al., 1963, 1965; SHIBUYA and SHIBUYA, 1963; TAKAGI and
OMURA, 1963; ALTNER and BOECKH, 1967).  Most of our knowledge about the
physiological properties of receptors in the nasal mucosa of vertebrates
is based on summation potentials (EOGs), the interpretation of which is
difficult and gives only limited information.
From single cell recordings, though still to some extent contradictory,
we can conclude that each receptor cell has its own individual, but con-
sistent spectrum of stimuli to which it responds.  No cells appear to have
identical spectra though in the common European frog a basic spectrum
common to most cells has been observed which is supplemented with an
additional spectrum varying from cell to cell.  All efforts to describe
different receptor types by the aid of morphological methods have failed
Further, the latest fine structural (GRAZIADEI, pers. comm.; KOLNBERGER,
in prep.) as well as electrophysiological results (GESTELAND,  pers.
comm.) seem to indicate that the information converging from the receptor
cells to the second order nerve cells does not follow isolated paths.
In electron micrographs thin branches of dendrites have been detected
by which a contact between neighbouring sensory cells could be establish-
ed.  Thus the possibility cannot be ruled out that there are lateral

interactions at the receptor cell level, this means that one receptor
cell influences another directly.
The next station in the olfactory pathway is the olfactory bulb where
the sensory cell processes are linked synaptically to the dendrites of
the mitral cells within the glomeruli.  Within these glomeruli which lie
near the surface of the olfactory bulb there is a considerable central
convergence of the order of 1000:1 from the afferent fibers to the second
order cells.
Since it seems from single sensory cell recordings that there is no
possibility to distinguish several distinct classes of receptors as
mentioned before, one is incertain whether there exists a special order
in the relationship between receptors and the mitral cells as second
order neurons.

2  Pathways in the olfactory centers of mammals
Before I try to explain the results of physiological investigations I
shall have to give a survey of the pathways as revealed by experimental
morphological studies (LOHMAN and LAMMERS, 1963, 1967; POWELL et al.,
1963a, b, 1965; HEIMER, 1968; PRICE, 1968, 1969).
There are two points of interest:  1. the fiber connections and major
pathways and 2. the inner structure of the olfactory bulb.
a Fiber connections
The olfactory bulb elements which are linked to the sensory cells are
mitral and tufted cells.  The axons of these cells form the lateral ol-
factory tract (LOT).  This fiber bundle courses over the surface of the
olfactory peduncle and the prepiriform cortex and can be traced as far
as to the periamygdaloid cortex.  These forebrain structures are large
in lower mammal groups. . In primates and especially in man they loose
importance and remain as a relatively small region recognizable at the
basal surface of the brain.  The results described here come mainly
from rodents (guinea pig, rat).

The main areas of distribution of LOT fibers are the prepiriform cortex,
the periamygdaloid cortex, and the cortico-medial amygdaloid complex.
Fibers of this tract terminate moreover 1. in the anterior  Ifactory
nucleus (pars lateralis, regio posterior), 2. in the lateral part of
the olfactory tubercle and 3. in the nucleus in the LOT.
The anterior olfactory nucleus is important because of its connection
with the opposite side of the brain. One can distinguish two components
crossing, both in the anterior commissure.  From the foreward  (rostral)
section of the nucleus on each side fibers can be traced to outer areas
of the same nucleus on the other side.  From the rear section  (pars dor-
salis) of the nucleus fibers can be seen to invade the granule cell
layer of the olfactory bulb on the opposite side.  There is no direct
connection between the bulbs.  The connection involves the pair of neuron
cains just mentioned and at least two synapses.  Later we shall have to
touch on the physiological significance of this arrangement.
The connection to the prepiriform cortex, the periamygdaloid cortex and
the cortico-medial amygdaloid complex are of special interest.  From the
latter region originates at least a part of the stria terminalis through
which olfactory impulses could influence the activity of the hypothalamus.
The hypothalamus controls important vegetative functions such as the
reproductive cycle.
Another pathway to the hypothalamus is made up of fibers originating in
the olfactory tubercle and running in the median forebrain bundle to
this region of the brain.  Additional connections between the olfactory
bulb and the medial forebrain bundle have been traced via the prepiriform
cortex.  Investigations of brain structure thus reveal connections to the
hypothalamus which may be claimed as pathways for the influence of odors
on sexual behavior.
(HEIMER (1968) confirmed recently earlier findings of WHITE (1965) that
there is another projection to the ventral entorhinal area which is
connected to the hippocampal formation.)

b The inner structure of the olfactory bulb
An analysis of the inner structure of the olfactory bulb is a prerequisite
for electrophysiological investigation and a control as well.
The olfactory bulb is laminated.  We mentioned already the layer of the
olfactory fibers, the glomerular layer, and the mitral cell layer.  A more
complete survey shows that there are additional cell types and layers.
Of special interest are the periglomerular cells of the glomerular layer
and the granular cells of the gramule layer.  Both neuron types form
reciprocal synapses:
          1. the periglomerular cells form dendrodendritic synapses on
             mitral cell dendrites in the glomerular layer (HINDS, 1970)
          2. the granular cells form "axodendritic" synapses on mitral
             cell dendrites in the external plexiform layer (RALL et al.,
             1966, PRICE, 1968).
Degeneration experiments have proven that fibers in contact to the latter
cell type can be affected by LOT lesions (PRICE, 1968). This means that
the LOT is not only a pathway from the olfactory bulb to structures lo-
cated more centrally, but that it also contains centrifugal fibers that
are in the opposite direction - outward.  These are linked to the granule
cells by which a centrifugal influence can be mediated.
The origin of the centrifugal bulbopetal fibers is not quite clear.
PRICE (1969) claims that the main origin is the nucleus of the horizontal
limb of the diagonal band which could be influenced by fibers from the
hypothalamus or the midbrain.
The inner structure of the olfactory bulb has been outlined roughly.
Summing up we recognize 3 main features:
1. a considerable central convergence
2. a pathway to the hypothalamus, by which an influence on the vegetative
   centers of this region could be exerted
3. an intrinsic neurological organization of the olfactory bulb which
   could yield complicated feedback mechanisms as well as a central
   control of the input coming from the periphery.

3  Electrophysiological investigation of the olfactory bulb and related
   forebrain structures
Electrophysiological evidence supports the concepts developed based on
structural analysis.  However, a coherent survey of the function of the
olfactory bulb and the connected stations of the brain is not yet
Several authors succeeded in recording single unit responses from the
mitral cell layer.  MATHEWS (cit. after PFAFFMAN, 1969) showed that such
units in rats can be grouped into three classes according to their
reaction to stimulation of the olfactory mucosa with odorants:
1. purely facilitory units, 2. purely inhibitory units, and 3. units
which respond with facilitation to some odorants and with inhibition
to others.
Similar results have been obtained by DOVING (1966) in bulbar units of the
frog and by LEVETEAU and MACLEOD (1969) in the rabbit.

However, all efforts have failed until now to arrange the units according
to their responses in a systematic way.  Only very few units manifested
exactly the same reaction pattern.  On the other hand the responses to one
odorant differ: some units may be inhibited, others facilitated, and still
others do not react at all.  It was possible however, to ascertain that
successive members in the homologous series of alcohols, ketones and
acetates have more similar olfactory stimulative properties than those
widely separated in chain length (DOVING, 1966, in the frog).  Other
authors find a correlation between the reactions, of bulbar units in the
frog and odor perception in man, both for the same classes of substances,
e.g. camphoreceous (HIGASHINO et al., 1969).
Likewise the analysis of the spatio-temporal distribution of excitation
of bulbar elements has yet to deliver a clear cut organizational scheme.
Electrophysiological investigations have revealed that inhibition plays
an important role at the mitral cell level.  It appears that at least a
great part of the inhibitory events observed in bulbar units is due to
the activity mediated by centrifugal fibers.  Particularly interesting

is the question of how self-inhibition  takes place.  Several studies have
shown that antidromic stimulation of the ipsilateral LOT  causes a hyper-
polarization at the level of the mitral cells.  This inhibitory effect is
mediated by granular or periglomerular  cells.  It seems due to a dendro-
dendritic mechanism rather than an influence by impulse transmission
from axonal processes (RAIL et al., 1966; NICOLL, 1969).  This finding
harmonizes well with the structural features of the system as mentioned
The significance of the inhibitory events may be seen to  consist
    1. in a contrast amplification
    2. in a regulation of both magnitude and selectivity  of the olfactory
       input to the deeper lying centers through efferent control.
I have mentioned that the interbulbar pathway is considered to comprise
at least two synapses.  This is confirmed by the time effects which are
observed in birhinal stimulation (LEVETEAU and MACLEOD, 1969).  A minimum
appears in the magnitude of the homolateral response when the hetero-
lateral stimulation is given about 3 msec before the homolateral one.
This finding gains particular importance as an olfactory  analogue to
directional hearing in man based on a reciprocal inhibition mechanism
has been discovered by von BEKESY (1964).
It was pointed out that fiber connections to the hypothalamus have been
demonstrated in rodents.  In 1967 SCOTT and PFAFFMAN were able to support
these results by electrophysiological experiments.  These authors found
multiunit as well as single unit discharges in the region  of the medial
forebrain bundle within the lateral hypothalamus after stimulation of
the olfactory receptors by application of odorants or after electrical
stimulation of the LOT.
These findings are supplemented to some extent by other experiments in
which electrolytic lesions were made in the medial preoptic-anterior
hypothalamic continuum and surrounding regions (HEIMER and LARSSON,
1966/67) in the rat.  Large lesions in this area eliminate mating, as
one might expect since the medial preoptic-anterior hypothalamic conti-
nuum occupies a strategic position in the limbic system.  It becomes
interesting to recall that part of the input to this region in macrosomatic

animals is delivered by the medial forebrain bundle and the stria
terminalis, and that these structures are connected to the olfactory
Moreover recordings from the medial forebrain bundle and the preoptic
area in general show single units responding to electrical stimulation
of the olfactory bulb and to stimulation of the nasal epithelium by
odorous substances as well (PFAFF and PFAFFMAN, 1969).  A point of special
interest is the fact that the responses of these elements in the male
rat are influenced by testosterone application.
Direct application by microinjection causes an increase in the resting
activity and an increase response to olfactory input.   Intraperitoneal
injections of testosterone were followed by more diverse changes.  Thus
the direct physiological action of testosterone in the basal forebrain
may have the specific effect of enhancing olfactory input to the brain
regions connected to these structures.
This result is of particular significance as this region shows a high
level and prolonged uptake of radioactive testosterone from the blood
(PFAFF, 1968).  Thus in this part of the olfactory pathway nervous and
endocrine regulation seem to be linked together in the control of
sexual behavior.  It is particularly noteworthy, that  responses in this
region differentiate between odors from receptive and  non-receptive
female rats more clearly than in the olfactory bulb (PFAFF and PFAFFMAN,
It may be remarked here that the role of odors in the  sexual behavior
of rodents hab been demonstrated by several behavioral investigations
(PARKES and BRUCE, 1961; VANDENBERGH, 1969; WRITTEN, 1969) and it has
been proved that sexual excitation and activity in primates is influenced
by a hormone-dependent pheromone secretion which in certain cases
acts through olfactory stimulation (MICHAEL and KEVERNE, 1968).
We have only very little information about the neurological responses
to odorants in the higher primates including man.  Though we should
expect a reduced importance in the species belonging to this group
according to the morphological reduction of the (olfactory brain, one

has to realize that a good many of the mechanisms which seem to be
rather powerful in lower mammals could be still present though perhaps
overlaid by additional regulatory mechanisms which arose during


Altner, H. and J. Beockh:  Z.  vergl.  Physiol.  55,  299 (1967)
Bekesy, G. von: J. Appl. Physiol. j.9_,  369  (1964)
DSving, K. B.: J. Physiol.  186,  97  (1966)
Gesteland, R. C., J. Y. Lettvin, W.  H. Pitts  and  A.  Rojas:  in :  Olfaction
     and Taste I, (Y. Zotterman  ed.) Oxford,  Pergamon Press  (1963)
Gesteland, R. C., J. Y. Lettvin  and  W. H.  Pitts:  J.  Physiol.  (Lond.)
     181, 525 (1965)
Heimer, L.: J. Anat. (Lond.)  103. 413  (1968)
Heimer. L. and K. Larsson:  Brain Res.  3j_ 248  (1966/1967)
Higashino, S. H. Takeuchi  and J. E.  Amoore: in :  Olfaction  and Taste  III,
     (C. Pfaffman ed.) 192, New  York:  Rockefeller Univ. Press (1969)
Hinds, J. W.: Brain Res. _17_,  530 (1970)
Leveteau, J. and Macleod:  in  : Olfaction and  Taste III (C.  Pfaffman ed.)
   212, New York: Rockefeller Univ.  Press  (1969).
Lohman, A. H. M and H. J.  Lammers: Progress in Brain Res. ^,  149 (1963)
Lohman, A. H. M and H. J.  Lammers: Progress in Brain Res. 23, 65 (1967)
Michael, R. P. and E. B. Keverne: Nature (Lond.)  218,  746 (1968)
Nicoll, R. A.: Brain Res.  14,  157 (1969)
Parkes, A. S. and H. M. Bruce: Science 134, 1049  (1961)
Pfaff, D. W.: Science L61,  1355  (1968)
Pfaff, D. W. and C. Pfaffman:  in : Olfaction  and  Taste III  (C.  Pfaffman
     ed.) 258, New York: Rockefeller Univ. Press  (1969)
Pfaff, D. W. and C. Pfaffman:  Brain  Res. 15,  137  (1969)
Pfaffman, C.: in: Olfaction and  Taste  III  (C.  Pfaffman ed.)  226,
     New York: Rockefeller  Univ. Press (1969)
Powell. T. P. S. and W. M.  Cowan: Nature (Lond.)  199.  1296  (1963a)
Powell, T. P. S., W. M. Cowan and G. Raisman:  Nature (Lond.)  199,
     710 (1063b.)
Powell, T. P. S., W. M. Cowan and G. Raisman,  J.  Anat.  (Lond.)  99,
     791 (1965)
Price. J. L.: Brain Res. _7, 483  (1968)
Price, J. L.: Brain Res. 11,  697 (1968)
Price, J. L.: Brain Res. 14_,  542 (1969)
Rail, W., G. M. Shepherd,  T.  R.  Reese  and M. W. Brightman: Exp.  Neurol.
     JL4, 44 (1966)

                          REFERENCES  (Cont'd)

Scott, J. W. and C. Pfaffman: Science  158.  1592  (1967)
Shibuya, T. and S. Shibuya: Science 140.  495  (1963)
Takagi, S. F. and K. Omura: Proc. Jap. Acad. J39,  253  (1963)
Vandenbergh, J. G.: Endocrinology J54,  658 (1969)
White, L. E. Jr.: Anat. Rec. 152. 465  (1965)
Whitten, W. K.: in: Olfaction and Taste III (C. Pfaffman  ed.)  252,
     New York: Rockefeller Univ. Press (1969)

             Sensory evaluation of odors in the

                        ambient air
                    Thomas Lindvall, M.D.

                Dep. Environmental Hygiene
                    Karolinska Institute
                 S-104 01  Stockholm, Sweden
      Paper presented at the Working Conference on the
    Dose-Response Relationships Affecting Human Reactions
to Odorous Compounds, Cambridge, Mass.  USA,  April 26-30,  1971

                  Sensory evaluation of odors in the
                             ambient air
                         Thomas Lindvall, M.D.
                      Dep. Environmental Hygiene
                         Karolinska Institute
                      S-104 01 Stockholm, Sweden
The procedure in studying environmental odor nuisances is commonly
based on an effort to find the relationship between the dose to which
the public is exposed and the annoyance reaction displayed.  By help
of such studies  a  prognosis can be made of reactions to be expected
in a community at different dosage levels.  Studies of this type require
adequate designation of dosage and response as well as knowledge of
the conditions under which the particular dose-response relationship
is valid.

The quantitative determination of malodorous air pollutants in ambient
air is complicated by the fact that their smell becomes offensive at
concentrations which are often too weak for the practical analysis of
short term measurements.  For the time being it would seem that ultimately
methods of sensory analysis are the only practical way of measuring
odors in combined exhaust gases while also having relevance in regard
to environmental health.  In principle this means that the human olfac-
tory sense must be used as detector and analyser.

Sensory methodology can be used as a tool in description of dosage
as well as of response.  Up to now most evaluations in ambient air by
these methods have aimed at measuring dosage by using indirect prog-
nosis techniques (source analysis) or direct methods of analysis.
The response to an odorant, mainly a subjective feeling of annoyance
may be measured on groups of sociological inquiry methods and, in part,
on individuals by sensory methods of analysis.

Sensory methodology for ambient air analysis or estimation must satis-
fy the following requirements:

a) be serviceable in practical and economic respects
b) include a standardized experimental procedure which guarantees
   acceptable reproducibility, and
c) give valid experimental results from the standpoint of hygiene.

To meet the above mentioned requirements an essential part of the efforts
should be directed towards construction of mobile odor laboratories
for field studies under controlled and standardized conditions.  Such
laboratories have been successfully used in Sweden since two years
ago for analysis at the source as well as in the ambient air (Lindvall
1970).  Sullivan, Adams and Young (1968) and Springer and Hare (1970)
in the USA have used mobile laboratories, however with different goals
i.e. to survey detection or objection to standardized odor levels
produced within the laboratory.

The olfactometers should permit known and stable exposure concentra-
tions, rapid changes in concentration and fairly natural respiratory
conditions.  To many scientists the most reasonable exposure alterna-
tive is provided by the odor hood (recent examples are Merrion 1968;
Sullivan, Adams and Young 1968; and Lindavll 1970).  Perferably the
exposure hoods should be installed in a ventilated and airconditioned
test chamber.  Another object in the construction of the Swedish
laboratories has been to arrange for special waiting rooms for the test
subjects with ventilation comparable to the actual test procedure in
order to provide optimal adaptation to the experimental situation.  By
using suitable filter equipment it is thus possible to carry out in-
vestigations in the immediate vicinity of a single emission source or
in a generally odor polluted area with a minimum of contaminations in
the laboratory air.

Construction of mobile laboratories has made it possible to limit varia-
tions in climatological background, and the use of dynamic dosing systems
has minimized losses due to adsorption and other factors.  Field experi-
ments with the Swedish equipment are therefore assumed to provide stan-
dardized although special experimental conditions, and the results shall
always be considered in relationship to the technical equipment used.
The experimental situation is not assumed to be necessarily comparable
to normal conditions of exposure in the ambient air.  Extensive studies
as to the reliability of the experimental procedure in its entirety in
source analysis and the importance of various interaction factors have
been undertaken (Lindvall 1970).

Sensory methods
A thorough review of sensory methods of analysis and their application
has recently been published (the Third Karolinska Institute Symposium
on Environemtnal Health, 1970).

From the point of view of atmospheric conservation a given odor sensa-
tion may be generally described in four respects or "dimensions".  These
are the detectability of the substance, its intensity at suprathreshold
levels, its acceptability or hedonic tone (i.e., the annoyance or,
possibly, pleasure it stimulates) and,  finally the characteristic
properties,  quality, distinguishing it from other odors irrespective
of intensity of acceptability.

In the environmental health the most important dimension of an odor is
probably its acceptability i.e., what percentage of the population
is annoyed by the smell and to what extent.  Public reactions of sub-
jective annoyance can probably best be evaluated by sociological survey
methods.  This response can be related to the degree of exposure to
odorous air pollutants to which the population is exposed (the dose).

Acceptability may also be studied in a laboratory scale e.g.  by apply-
ing direct methods of scaling as suggested by Engen and McBurney (1964).
Matching of one standard odorant to different combustion gases with

respect to equalunpleasantness has been used by Lindvall and Svensson
(1971).  Hydrogen sulfide was matched to five different combustion gases
with regard to unpleasantness by use of an equality production method.
Figure 1 shows the resulting five equalsensation functions in log-log
coordinates.  It was concluded that no differences were evident between
combustion gases (as indicated by the slope of the lines) but that hydro-
gen sulfide at the same concentration was perceived twice as unpleasant
as any of the five combustion gases.

Results of such laboratory studies may constitute an important part in
the description of dosage in the ambient air, e.g. when comparing two
odor covered areas with similar odor frequencies but with differences
in the hedonic value of the odorants in question.

The dimension of odor quality may be relevant to include in ambient air
evaluations partly due to its connection with the hedonic value of an
odorant, partly due to the frequent necessity of identifying the source
of the odor.  Sensory methods that may be applicable in field studies
of odor quality are multidimensional and informational analysis of odor
quality (Torgerson 1958, Ekman and Engen 1962, Woskow 1968).  Such
methods ideally require a high standard technical equipment, for use
in field studies prefereably mobile, and a relative stable odorant
concentration input to the dosing system.  The latter requirement is
not always fulfilled in ambient air studies but close to the emitting
stack or in areas heavily polluted from surface sources.

Studies of suprathreshold perceived intensity are of interest in sensory
evaluation in the ambient air since they concern the change in perception
associated with a reduction in the physical odor intensity.  Thus, for
example it would not be reasonable to expect the most significant changes
in a critical local air pollution situation involving prolonged odor
problems, when the concentration to which the public is exposed varies
about the most level part of the psychophysical function for the odorous
substance in question.  Such considerations may be of significance in
the vicinity of major traffic arteries or other wide spread sources

which are not limited to a single point.  However, in a wider context,
especially when the source of the odor is localized to a given point,
any change in the odor intensity of the effluent gases may be expected to
produce alterations in the frequency with which odor is perceived, i.e.
when the absolute odor threshold is exceeded in the area.

In studying intensity of odor one should preferably use modern psycho-
physical scaling methods (e.g. see Torgerson 1958; Ekman, Berglund, Berglund,
Lindvall 1967; Lindvall 1970).  There are the same practical requirements
with respect to equipment and input odorant concentration as indicated
above in the discussion of quality.

In many cases detection methods are used to obtain an idea as to the
intensity with which an odorous substance is experienced.  Examples of
such methods are absolute odor threshold measurements and methods based
on the theory of signal detectability.  Threshold values or detectability
indexes do not give any measure of the perceived intensity above "threshold"
levels but do undeniably provide valuable information regarding source
control and the distribution of odorant emanations in the atmosphere.

The main advantage of the signal detection methodology is that it is
able to measure response biases.  This is an important problem in ol-
faction because a sophisticated observer with the best of equipment will
always produce a relatively high proportion of false positives.  These
must be measured in order to obtain a valid description of dosage.  Dif-
ferent modifications of the basic methodology may be applicable to most
practical situations (Green and Swets 1966; Lindvall 1970; Berglund,
Berglund, Engen and Lindvall 1971).

There are also some serious limitations in detection methodology with
regard to its applicability in practical public health work.  These methods
frequently require preliminary experiments since in practice it can only
be used within a rather small region in the vicinity of the detection
"limit" where the probability of hits and false alarms varies between
0 and 100 percent.  Another disadvantage is the difficulty of expressing

the results of signal detectability measurements in a single physical
limit value which are presently necessary for applicability in existing
methods of calculating atmospheric odorant dispersion.  A sophisticated
use of classical absolute odor threshold determinations then has been
proved to give valid measures in dosage description (Lindvall 1970).
In conclusion the need for and benefit of both the modern detection
theories and the classical concept is evident for the time being in this
type of work.

Sensory evaluation in the ambient air
Data necessary for an adequate description of ambient air levels of odor
can be obtained through the use of human sensors in four different ways.

The first two methods are of a prospective nature.  The observation method
is based on consecutive observations of odor occurrence (frequency, dura-
tion, identity, intensity, hedonic value) within a given area for a fixed
period of time.  Such observations can be made either with many untrained
observers or with fewer trained individuals, possibly with a simple sensory
measuring aid such as the scentometer (Heller et al. 1959; Gruber et al.
1960; Sanders et al. 1970; First 1970).  A sophisticated study design,
e.g. concerning selection of sampling points and the use of a sufficient
number of observations over prolonged periods of time, is necessary if
data of value from the point of view of public health are to be expected.

Data concerning hedonic value and perceived intensity obtained by category
judgements like "weak", "moderate", "intense" etc. by single observers
in the field will only attain an ordinal level of measurement that is
unsatisfactory in most cases.  Data obtained by category scaling may,
however, be treated in the manner of Thurstone (cf. Torgerson 1958) on the
basis of certain assumptions about variability thus providing an inter-
val scale.  The usefulness of the method is limited by the large number
of observations needed and that the results probably can be used only
for relative comparisons between sectors within the general investigation
area.  Category scaling may be easified by a kit of standard references
for use in the field similar to the Turk-kit (1967) on diesel odors.

As indicated by Berglund, Berglund, Engen and Lindvall  (1971) and by
recent pilot studies in Sweden  (Lindvall 1971) the method of signal
detectability may be applicable for direct determinations of changes in
the relative dose in ambient air, particularly in the case of more gen-
eralized pollutant sources such as traffic arteries or  in other areas
afflicted with prolonged odor exposure where the question of frequency is
irrelevant.  However, such studies require experimental equipment of
high standard, preferably mobile.  Rapid changes in ambient air concentra-
tion will require a correspondingly larger number of simultaneous observa-

The Swedish pilot study conducted by the author included determination of
an index of detectability (d') for odor levels in the ambient air in one
of the main streets of Stockholm city during rush-hours.  The index was
calculated as group values on six subjects at a time with the sampling
time being half an hour.  As a whole 18 subjects participated.  One of
the mobile odor laboratories was used, placed close to  the drive-way and
the gas inlet 1.5 m above the curbstone.  The study was conducted during
three consecutive days and was followed for comparison by a two-day study
of the odor levels in a less polluted area at a university campus in a
suburb to Stockholm.

Although the quantity of data obtained is limited the results are pro-
mising.  The false alarms and d1 measured at the campus area are of
similar magnitude as compared between groups and iterations.  This in-
dicates reasonably stable and comparable criteria used by the three
groups of subjects.  Assuming that the experimental conditions did not
change moving the laboratory from the street to the campus this result
permits a calculation of an index of detectability.   In Figure 2 odor
indexes and concentrations of carbon monoxide in the street are plotted
against time of the day.  As can be seen from the lower part of the figure
each half an hour period has been tested three times.  However, each ob-
servation is based on a fairly large number of signal + noise and noise
presentations (120/120).  The arithmetic mean is indicated by the full

line.  The doted lines indicate the mean detectability index at the
campus and its standard deviation, all calculations being based on half
an hour data.

Index of detectability varies in an expected fashion compatible with the
traffic load during rush hours.  The detectability of odor is relatively
low, presumably because of the generally low pollution load in the streets
of Stockholm and the unusually well ventilated place of observation (close
to a square and a very tall single building).  The correlation between
odor index and concentration of carbon monoxide is doubtful (r = 0.31,
n = 54, 0.05 p 0.01).  This is not surprising as carbon monoxide in it-
self is nonodorous and not a representative exhaust gas of diesel vehicles.
In spite of the preliminary character of this experiment signal detection
methodology seems to be a realistic alternative in dosage description
in the ambient air when dealing with surface sources.

The third method is based on interviewing the resident population in order
to find out what their retrospective opinions are concerning the average
odor frequency and duration of the smell.  Such data have been collected
on several occasions in conjunction with interviews where the reactions
of residents to the occurrence of odor have also been noted (Friberg,
Jonsson and Cederlof 1960; Cederlof et al. 1964).

The estimation of exposure based on retrospective interviews can in itself
reflect the experiences of the subjective exposure but cannot be definitely
relied upon to do this.  One difficulty presented by using retrospective
interviews is the exact period of time the questions are related to.
Furthermore, it is not known as to what degree the respondents uncon-
sciously exaggerate or depreciate the frequency of exposure owing to
their negative or positive attitudes or to the actual annoyance they have
experienced (Lindvall 1970).

The fourth method is of a prognosis-type (indirect evaluation of dosage)
and is based on the knowledge of the source strength of the gas and the
expected distribution of the gas mass over the investigation area and

the dilution at ground level at different distances from the point source.
The source strength can be evaluated by sensory methods of analysis and
to a certain extent by chemical analyses.  However, the latter requires
that all the relevant substances can be accurately determined and that
the absolute odor threshold for each of these is known.  Also the com-
bined odor intensity must not be affected by gaseous interaction which
causes enhancement or suppression of the smell.

The probable distribution and dilution of effluent gases can be studied
by means of various dispersion methods (for a review, see Strom 1968).
Several methods give for selected points in the area surrounding a source
"typical maximum values" which are not connected to given frequency values,
e.g. Button's formula.  Another way to attack this problem is to calculate
the expected concentration frequency distribution for selected points
in the vicinity of the source (HSgstrSm 1968, Lindvall 1970).  However,
this also requires the dose in the ambient air to be described in a statis-
tically meaningful manner.

In Sweden sensory intensity analyses (absolute odor threshold) combined
with meteorological dispersion calculations of the latter type, have been
used to predict the frequency with which odors will be discernible in
the ambient air at various distances from sulfate cellulose plants.   The
calculation method used gives the expected frequency with which the
absolute odor threshold value for the effluent gases will be practically
momentarily exceeded and thereby detected at a given observation point.

The accuracy of such predictions has been checked in a large scale field
experiment that was undertaken in the vicinity of a pulp mill at the East
coast of Sweden.  In Figure 3 the factory is shown in the center surrounded
by eight sectors and three distances (2, 5 and 10 km).  There was also
a 20 km circle not shown in the figure.  Each day, based on a weather
forecast, a sector was chosen in lee of the factory and trained observers
were randomly distributed on the circles within the sector.  A large
number of observation places were reconnoitered in before hand on the
circles in all directions.  The observers were instructed to register

every two minutes if there was any sulfate odor present at the very
sniff, thus an almost momentary observation.  The time was checked by
stopwatches.  Parallel to the observations the odor thresholds of stack
discharge were determined and the atmospheric dispersion was calculated
based on meteorological measurements.  This study went on for one month
and the results are summarized in Table 1.

The table shows the total number of observations at each distance from
the plant.  All sectors are figuratively speaking, put together on the
top of each other.  For this resulting mean lee sector of observation
the observed and predicted odor frequencies are given and also the ratio
between observed and predicted frequencies.

There is a striking degree of correlation with predictions, up to 5 km
a factor of 1.2-1.7, thus confirming the validity of the sensory analytical
method and the atmospheric dispersion calculations under practical con-
ditions.  As the distance from the source increases, odors are observed
more frequently than predictions would indicate but still up to 20 km only
by a factor of 2.6-3.0.  When comparing with earlier studies on atmos-
pheric dispersion these results are surprisingly good.

Of course, a prognosis of the expected odor coverage expressed as a mean
frequency value over a considerable length of time, must not necessarily
reflect the subjective exposure of the population.  Some results (Lindvall
1970) indicate that calculations of the predicted frequency of perceived
odor within a large area may need to include a correction to give appreciably
higher values, especially at greater distances from the source.  However,
as a base in establishing dose-response relationship in a single source
case, the predicted odor frequency may nevertheless be considered to be
the less subjective, the most adequate and the less expensive measure
in describing the dose, as compared with survey methods and direct
measurements in the ambient air.

Other aspects on the perception of odors in environmental health than
frequency and duration of odor exposure are effects of context like

adaptation and expectation (Ekman, Berglund, Berglund, Lindvall 1967;
Berglurid, Berglund, Engen, Lindvall 1971; Engen 1970).  Interaction
principles between different odors or between odors and agents affecting
other sense modalities, like vision and hearing, are mainly unknown
(Berglund, Berglund, Lindvall 1971).  Especially acceptability of annoying
agents may be influenced in the real life situation also by compensation
or aggravation mechanisms of socio-economic origin and the like.  Such
influences limit the applicability of sensory methods of analysis in
population studies to description of dosage and studies of the existence
and strength of the interaction effects mentioned above.

BERGLUND, B., BERGLUND, U. and LINDVALL, T., On the principle of odor
interaction, Rep. Psychol. Lab. Univ. Stholm., in press 1971.

BERGLUND, B., BERGLUND, U., ENGEN, T. and LINDVALL, T., Effect of adap-
tation on odor detection, Perception & Psychophysics, to be published

CEDERLOF, R., EDFORS, M.-L., FRIBERG, L. and LINDVALL, T., Determination
of odor thresholds for flue gases from a Swedish sulphate cellulose
plant, TAPPI 1965, 48, 405.

EKMAN, G. AND ENGEN, T., Multidimensional ratio scaling and multidimen-
sional similarity in olfactory perception, Rep. Psychol. Lab., Univer.
Stockholm 1962, No. 126.

EKMAN, G., BERGLUND, B., BERGLUND, U. and LINDVALL, T., Perceived in-
tensity of odor as a function of time of adaption, Scand, J. Psychol.
1967, 8, 177-186.

ENGEN, T. and McBURNEY, D.H., Magnitude and category scales of the plea-
santness of odors, J. Exp. Psychol. 1964, 68, 435-440.

ENGEN, T., The effect of expectation on judgements of odor, unpub-
lished manuscript 1970.

FIR£>T, M.W., A model odor control ordinance, Paper presented at the 2nd
International Clean Air Congress, Washington D.C. 1970.

FRIBERG, L., JONSSON, E. and CEDERLOF, R., Studier over sanitara olagen-
heter av rokgaser fran en sulfatcellulosafabrik, Nord. Hyg. Tidskr.
1960, 41, 41-50.

GREEN, D.M. and SWETS, J.A., Signal detection theory and psycho-physics,
Wiley New York, 1966.

GRUBER, C., JUTZE, G. and HUEY, N., Odor determination techniques for
air pollution control, J. Air. Poll. Contr. Ass. 1960, 10, 327-330.

HELLER, A.N., KANDINER, H.J., REITER, W.M. and COHEN, M., The odor
survey: A tool for air pollution control, Air and Water Pollution Abate-
ment Conference, Manufact. Chemists Ass., Washington B.C. 1959.

HOGSTROM, U., A statistical approach to the air pollution problem in
chimney emission, Atmos. Environ. 1968, 2, 251-271.

LINDVALL, T., On sensory evaluation of odorous air pollutant intensities,
Nord. Hyg. Tidskr. 1970, suppl. 2, 1-181.

LINDVALL, T. and SVENSSON, L.T., Equal-unpleasantness matching of one
standard odorant to different complex odor mixtures, unpublished
manuscript 1971.

LINDVALL, T., Unpublished data 1971.

MERRION D., Effect of design revisions on two stroke cycle diesel engine
exhaust, Paper 680422, Society of Automotive Engineers, Inc. New York

SANDERS, G.R., UMBRACO, R.A., TWISS, S. and MUELLER, P.K., The measure-
ment of malodor in a community by dynamic olfactometry, Paper presented
at the Third Karolinska Institute Symposium on Environmental Health,
Stockholm, June 1970, (Unpublished).

SPRINGER, K.J. and HARE, C.T., A field survey to determine public opinion
of diesel engine exhaust odor, Rep. No. AR-718, Southwest Research In-
stitute, San Antonio, 1970.

STROM, G.H., Atmospheric dispersion of stack effluents, in Stern, A.C.
(Ed.), Air Pollution. I, 227-274, Academic Press, New York, 1968.

SULLIVAN D.C., ADAMS, D.F. and YOUNG, F.A., Design of an "odor percep-
tion and objectionability threshold" test facility, Atmos. Environ.
1968, 2, 121.

Third Karolinska Institute Symposium on Environmental Health, Methods
for measuring and evaluating odorous air pollutants at the source and
in the ambient air, Nord. Hyg. Tidskr. 1970, 51, 1-77.

TORGERSON, W.S., Theory and methods of scaling. Wiley, New York 1958.

TURK, A., Selection and training of judges for sensory evaluation of the
intensity and character of diesel exhaust odors, U.S. Dep. Health,
Educat. Welfare, PHS publication No. 999-AP-32, 1967.

WOSKOW, H.M., Multidimensional scaling of odors, in TANYOLAC, N. (Ed.)
Theories of Odors and Odor Measurement. New York, Spartan Books, 1968.

                                            I —     I        I         I


•E   1.5


"k  1.0
                J	I	I	L
                                                                      0.5 units

                                            Relative log  combustion  gas cone.
                                                                                                     D J
                                                                                                     E J
                                                     Figure 1






            Carbon monoxide
                                  Odor index
           o  o
17     18
                                     Figure 2

                           km  1
                                                 Figure 3

Prediction of odor  incidence

Number of obs.
Observed odor °/o
Predicted odor °/o
Ratio obs. /predict.
7 490
                  Table 1

                         Trygg Engen
          Walter S. Hunter Laboratory of Psychology
                      Brown University
Number of tables:    4

Number of figures:   4
                              Proofs should be sent to:
                                   Dr.  Trygg Engen
                                   Walter S. Hunter Laboratory
                                      of Psychology
                                   Brown University
                                   Providence, Rhode Island  02912


                              Trygg Engen
                           Brown University

     The importance of the psychological acceptability of odors seems

obvious, but judging from the dearth of published research the problem

has in fact been largely ignored by sensory physiologists and

psychologists.  Most research on the sense of smell involves either

odorant concentration and the sensitivity of human subjects or the

relationship between physical and chemical attributes and the

perceptual quality of an odorant.  The latter relationship is still

too poorly understood to be useful in controlling odors.   Since there

is a good deal of information about the psychophysical relationship

between concentration of an odorant and its perceived odor intensity,

most applied work on air quality has been based on this aspect of the

problem (see Third Karolinska Symposium on Environmental  Health, 197C).

This has meant an emphasis on work with so-called threshold levels,

that is, concentrations which are just strong enough to be noticed

by the average observer.  This is sufficient information  for many

                                   57                                    61

   T.  Engen

   purposes, for example,  establishing criteria  for maximum acceptable

   concentrations of pollutants,  but it provides  no information  at  all

   regarding the perceived quality of the air and its  acceptability.

        Except for substances which mainly stimulate the trigeminal nerve,

   there is no known psychophysical relation to utilize  in  dealing  with

   hedonic effects of odorants, and in this  area  one is  limited  to

   psychological methods of measurement.   Psychological  scaling  and

   opinion polling have been the  two main methods.   Opinion polling is,

   of  course,  a well known procedure and  it  has been used to advantage in

   the case of odor.  Studies in  Sweden have been especially valuable  in

   proposing means for isolating  the effects of the subject's attitude

   toward the  source of an odorant from its  perceptual effects (Cederlb'f,

   Friberg, Jonsson, Kaij,  and Lindvall,  1964).   Like opinion polling,

   psychological scaling has reached a  mathematically sophisticated level.

   An  example  of this is the methodology  developed  for multidimensional

   scaling that has been applied  in olfaction in  attempting to determine

   the basic psychological dimensions of  the odor space  (Harper,

   Bate-Smith, and Land, 1968).

        The present paper  will be limited to more simple and direct

   methods of  scaling and  will attempt  to emphasize one  basic problem  of

   odor perception and judgments.   It is  generally  assumed  that

   associations based on individual experiences are the  most important

   reasons for the large inter-individual variation in judgments observed.

   An  early study of individual differences  in affective responses  to

   odors by Kniep, Morgan,  and Young (1931)  concluded thnt  there wns good

T. Engen

agreement in the results from children (ages 7-9) and adults, although

they had doubts about the intra-individual reliability of the children's

judgments.  Recently Moncrieff (1966) has produced evidence which

suggests that age may be the most important variable, even more

important than sex,  in the development of preferences.  In general,

children tend to be more tolerant of unpleasant odors than adults.  Of

particular interest has been a suggested discontinuity in such

responses between the ages of four and five observed by Stein,

Ottenberg, and Roulet (1958) to synthetic odors of feces and sweat.

Three and four-year-old children tended to say they "liked" those

odors while children five years and older said they "disliked" them.

These data are shown in Fig. 1 (Slide 1).  The authors thought that
                       Insert Fig. 1 about here

this discontinuity may be "related to early childhood experiences and

to the mastery of instinctual impulses" (p. 265)  and that odor

preferences may be explained by the Oedipal conflict.  Reaction

formation in the anal stage may thus cause a change from liking to

disliking the odors of feces and sweat.  According to Moncrieff (1966),

"Sometimes there is a natural and well-founded dislike of a smell; one

may be educated out of it by psychoanalysis explaining that it is

caused by some forgotten and unimportant incident, but the real reason

may go much deeper than subject or psychoanalyst guesses." (p. 205).

The suggestion is that there are innate responses to certain odors,

for example, anaphrodisiacs.

     To this must be added that sometimes neither psychoanalytic nor

  T. Engen

  physiological hypotheses are needed because a simple psychological

  explanation  is possible.  In particular, the tendency to appear

  agreeable  in situations where either "Yes" or "No" is the required

  response has often been attributed to young children.  This is the

  response bias of acquiescence characteristic of the "yes man"  (Guilford,

  1954).  Katz and I (1968) had begun a study on odor preferences similar

  to that of Stein et_ al., when this bias was observed after testing just

  a few children.  The experiment was therefore redesigned to include

  variations in the instructions describing the nature of judgment

  desired as one of the three independent variables together with the age

  of the children and the pleasantness of the odors.  The purpose was to

  pursue the study of the development of odor preference with the

  question in mind:  if such age differences exist, are they the result

  of actual differences in olfactory preferences or of response bias

  which also varies as a function of age?

        A total of 134 boys and girls in four groups of about 30 and

  ranging in age from four to seven participated in this experiment.

  They  were assigned randomly to groups receiving different instructions.

  Since relatively few four-year-old children were available, only three

  sub-groups (or instructions) could be tested for that age.

        Three odorants were selected in order to produce odors ranging

  from  very unpleasant to very pleasant as judged by adult observers

  (Engen and McBurney, 1964).  These were iso-butyric acid, iso-amyl

  acetate and safrol.  (Note in Fig.  1 that Stein et_ al., also used

  amyl  acetate and a compound which they do not identify but describe as

T. Engen

"synthetic sweat.")  In addition, the present experiment used diethyl

phthalate both as a diluent and as a "blank" stimulus for catch trials,

for it has only a neutral and vague odor.  The other three odorants

were matched in intensity to a 25 percent solution of amyl acetate in

diethyl phthalate using 12 graduate students as judges.  These three

concentrations plus diethyl phthalate were prepared for each day of

testing and were presented to the subject on cotton wrapped around a

glass rod.

     The subject was seated across a table from the experimenter who

explained that the task involved smelling and demonstrated how the

subject should sniff it and "then tell me if it smells 	."

The sentence was completed by one or two adjectives, namely, "pretty,"

"ugly," "pretty or ugly," and "ugly or pretty."  (The choice of

adjectives was made on the basis of preliminary work suggesting these

adjectives would be semantically unambiguous to young children.)   The

odorants were presented in a different irregular order to each subject,

except that diethyl phthalate was always presented fourth, and the first

odorant was presented again at the end to obtain evidence of reliability

of the judgments.  There was an interval of at least 30 sec. between

two successive odorants to avoid possible effects of adaptation.

     The response was scored either plus (+)  for pleasant or minus (-)

for unpleasant regardless of the question asked; for example, if  the

question was, "Tell me if it smells pretty," a response of "No" was

scored - (i.e., unpleasant), and a response of "Yes" was scored +

(i.e., pleasant).  As an index of reliability of the judgments,

                                   6i                                       65

   T.  Engen

   82, 85,  85,  and 80  percent  of  the  children  of  ages  four,  five,  six,  and

   seven,  respectively,  gave the  same response to the  stimulus  presented

   for the second time at the  end.   (The  response to this  repeat was not

   included in  any of  the analyses  shown  below.)   Sex  was  not a significant

   variable, for square  analysis  resulted in  .80< p. <.90  for the

   frequency of + response obtained from  girls versus  boys.

        According to the literature cited,  it  would be expected that

   butyric acid would  on the average  be rated  somewhere between natural

   (about  equal proportions of -  and  + responses)  to pleasant

   (preponderance of +)  by the four-year-olds  but more and more unpleasant

   (preponderance of -)  as a function of  age.   Safrol  and  amyl  acetate

   should  be rated as  pleasant, and diethyl phthalate  should be rated

   neutral.  Fig. 2 (Slide 2)  presents the proportion  of + responses as
                          Insert Fig.  2 about here

   a function of age with  judgments  from all the  different groups combined

   with different instructions  in  order to reveal,  in  general,

   pleasantness  of each  odor  as a  function of age.

        These data agree with those  reported by Stein  et al.  in  showing a

   marked change with  age,  particularly from four to five, for the odor of

   butyric acid.  Four-year-olds apparently liked this odor,  while the

   older children disliked  it.  Analysis of the data by chi square

   indicates  that these  age.differences are reliable (p<.001).   The age

   differences for amyl  acetate, safrol, and diethyl phthalate are not

   reliable.   However, the  proportion  of + responses from each of these

   odorants is higher  and  reliably different (p<.01) from  .50 (or neutral)

T. Engen

for all four age groups, indicating that these are pleasant odors.

This is to be expected for safrol and amyl acetate but not for diethyl

phthalate, the relatively odorless  and qualitatively neutral "blank."

This indicates that the children had a tendency to say they liked an

odor whether they smelled it or not.  Indeed one child reported "they

smell good" before he had sniffed anything!  Note that this response

tendency would inflate the reliability index used above.

     The general problem is that different questions apparently differ

in their tendency to elicit biased responses.  For example, asking a

positively phrased question seemed to produce false positives, while

stating the same question in a negative form yielded both negative and

positive responses as one might have expected.   No semantic

generalization can be made on the basis of the  present limited

experiment, but it represents a fundamental problem in odor

acceptability.  As suggested by Cederlbf et_ al. an indirect approach is

often required in this field of measurement.

     Fig.  3 (Slide 3) shows the proportion of + responses obtained for
                       Insert Fig. 3 about here

each age group when the results from all four odorants are combined;

that is, this graph is intended to show the relative frequency of +

responses to different instructions regardless of the pleasantness of

the odors.  The change in response as a function of age is significant

(p<.001) for the question "Tell me if it smells pretty."  The other

three instructions produced about equal proportions of + and -

responses (roughly around .50) and no reliable change as a function of

    T.  Engen

    age.   Therefore,  the  biased  response  involved  in  this kind of

    experiment seemed to  vary with  the  instructions.  That  is, one reason

    for the  high  number of  + responses  shown  in Fig.  2  (Slide 2) for all

    odorants except butyric acid is  the tendency for  most of the children

    to  agree when asked "Does it smell  pretty?"  Presumably the same

    explanations  apply to the widely quoted data by Stein et al. who were

    asked  the same kind of  question, although their paper does not provide

    much detailed methodological information.  The difference in the

    frequency of  + responses related to the form of the question is

    reliable at p<.001, .01, .05, and  .20  for ages four, five, six, and

    seven, respectively.

        Fig.  4 (Slide 4) showed the responses obtained for butyric acid

    only,  but this time for different questions.  These results are

    consistent with those already shown,  indicating again that the

    developmental change  in the  hedonic value of butyric acid is probably
                          Insert Fig. 4 about here

    an  artifact of  the opinion-polling method.  What really changes as a

    function of age may be the response bias.  This seems to provide a

    simple  alternative to  the Freudian interpretation which suggests that

    body odors which are liked become disgusting as a result of reaction

    formation.  Even if the response biases should be eliminated

    statistically,  the effect of age would be reduced, but butyric acid

    would still be  rated as unpleasant.

        However, Moncrieff, who also suspects that young children may have

    a greater tendency than adults to say, "yes, I like it," although for a

T. Engen

different reason, asked his subjects to rank-order the odorants in

terms of pleasantness.  He comes to the conclusion that "Children

exhibit a remarkable tolerance for substances with a fecal note in

their odor; they do not like these substances but they are more or less

indifferent to them, whereas the adults actively dislike them." (p. 80)

His subjects, who were older (about 7 to 15) and presumably more or

less beyond the Oedipal conflict, ranked these substances higher, or

more pleasant, than the adults did.  This rank-order method should

eliminate verbal and semantic problems discovered with the present

method.  The disadvantage with the rank-order method is that the

scoring of responses is relative, with the result that odors put in the

middle of this scale are "neutral" by definition, while those ranked

low are "unpleasant" and those high are "pleasant."  For example, one

could not give all odors a high rating if that seemed appropriate.

     The method needed must eliminate verbal responses and make

possible a quantitative comparison of preferences.  One approach to

this problem is represented by the now classic work of Thurstone (1959)

on The Measurement of_ Values.  Thurstone provided both a mathematical

theory of psychological measurement, the so-called law of comparative

judgment, and methodology applicable to the scaling of attributes for

which there are no specifiable  physical correlates.   The most commonly

used method is the method of pair comparison.  The subject's task is

simply to compare two stimuli, and every stimulus is  usually compared

with every other stimulus.   Our second developmental  study (Engen and

Corbit, 1970) scaled the preferences for the odors of butyric acid,

  T. Engen

  safrol, diethyl phthalate, rapeseed oil, and nerioli oil.  The txro last

  odorants were included because Moncrieff's (1966)  result suggested that

  they are examples of oily odorants which seem especially aversive to

  children 6 years old or younger.   The odorants,  except for the diluent,

  diethyl phthalate, were first matched in subjective intensity to the

  relatively weak odor of undiluted rapeseed oil.  Five groups of

  subjects were tested;  there were  26 four-year-olds, and groups of 28

  five-, six-, and seven-year-olds  in addition to  a  group of 28 adults.

       The 10 pairs of these five odorants were presented to each subject

  individually following the same general procedure  as above.   The order

  of presentation of the 10 pairs was balanced to  eliminate possible

  sequence effects.  The first pair of odorants presented to the subject

  was repeated at the end of the series to obtain  an estimate  of the

  reliability of the judgments.   The percentage of the subjects judging

  this pair the same in  both cases  was 69, 64,  75, 71, and 93  for the  four-,

  five-, six-, and seven-year-olds  and adults,  respectively.   This is

  obviously an important and interesting difference  between children and

  adults, but it will not effect the comparison to be considered next.
                         Insert Table 1  about  here

       Table 1 illustrates  the data  obtained with this  method with  the

  results from the six-year-old subjects.   The entries  in  Table  1

  (Slide 5)  are the proportions of the subjects in  that particular  group

  who preferred one odorant over another.   These proportions are

  obtained by dividing the  number of six-year-olds  who  preferred, for

  example, rapeseed oil over butyric acid  by the total  number of subjects

T. Engen

(28).  This value is  .786 and is entered in the second column of the

first row of Table 1.  The proportion who preferred butyric acid over

rapeseed oil is 1.000 -  .786 =  ,214,as entered in the second column of

the first row of Table 1.

     In this raw form the data provide no information about the hedonic

value of any single odor.  This is where Thurstone's mathematical

theory is applied.  In essence, it entails making the assumption that

the hedonic effect of any one odorant forms a normal distribution when

plotted for a group of individual judges.  One can then make use of the

known relationship under the normal curve between proportions and

so-called standard scores (or £ scores)  for which by definition the

mean is zero and the standard deviation is 1.00.  Scores below this

mean (the minority vote)  will then be negative in sign and scores above

this mean (the majority vote) will be positive in sign.   Since,

according to the theory,  the preference values of each odorant for

different individuals are normally distributed, these standard scores

will represent the psychological distance between these hedonic values

for different odors.
                       Insert Table 2 about here

     Table 2 (Slide 6) shows these standard scores based on the

proportions shown in Table 1.  Each entry in this table corresponds to

the distance on the preference scale between a certain pair of odorants

For example, for butyric acid and rapeseed oil this value is .31

standard deviation (with the sign indicating that rapeseed oil is

preferred by the group) .  The best estimate of the hedonic value of

   T.  Engen

   each single odor will be obtained by taking the mean of the values  of

   each column in Table 2.   These values represent all the information

   about each odorant in the set.  These means represent relative scale

   values on a hedonic continuum or a psychological scale of preference.

   Since these are not absolute numerical values,  it is not possible to

   determine, for example,  whether the children give higher (or lower)

   value ratings of the set of odorants than adults, but certain other

   important quantitative comparisons can now be made.
                          Insert Table 3 about here

        Table 3 (Slide 7)  summarizes the scales obtained for  the  five

   different groups.   First,  inspection of these values  will  reveal

   clearly that children do not discriminate as well  between  odorants  on

   this preference scale as do adults.  Second, the older the children the

   more their scale values agree with those obtained  for adult subjects.

   This is also indicated in the column on the right  by  a comparison of

   the range of values represented by this set of odorants for the various

   groups.  However,  the preferential order of the five  odors is  more  or

   less the same for all five groups.

        Finally, it is interesting and noteworthy that a weak and

   nondescript odor like diethyl phthalate, often used as a diluent

   because it is nearly odorless, is again consistently  judged as pleasant

   (that is, positive sign).   The suggestion is offered  by Moncrieff

   (1966), as a result of observing the same effect with the  children  in

   his group, that "he smells best who smells nothing."   Another  possible

   interpretation is  that the positive scale value is a  measure of

T. Sngen

response  bias similar to the one discussed above.  Still another is

that it is the effect of the context provided by the stimulus set;

that is, a neutral stimulus may be judged as relatively pleasant or  •

unpleasant depending upon the comparison stimuli.   If, for example,

butyric acid should be replaced by an odorant even more pleasant than

safrole, the whole preference scale might shift such that the value of

diethyl phthalate might seem less (see Sandusky and Parducci, 1965).

     In general, the preferences of children and adults are similar.

However, the present data do not reject the hypothesis of Moncrieff

(1966) that children find all odors more pleasant  than adults.   It  is

possible that the children's preference scale is placed toward the

pleasant end of the continuum relative to the adult scale, but the

relative placement of individual odors is the same.  A fairly high

proportion of the children did prefer butyric acid over considerably

more pleasant odors, as judged by adults.  The developmental change in

the size of the range related to age indicates that the older the

children the more the preferences agree with those of adults and lends

evidence to the hypothesis that discrimination of  hedonic attributes

of odors is learned.  Except for trigeminal stimulation,  and

unusually strong concentrations, there may be no inherently

unpleasant odors.

     Earlier work (Engen, Lipsitt, and Kaye, 1963;  Engen  and Lipsitt,

1985)  has shown that even newborn infants respond  to odors.   This

response, even to odorants such as anise judged very pleasant by

adults, resembles a mild "defensive" or "startle"  response,  but  the

T. Engen

hedonic nature of the response could not be firmly established.

However, it seemed clear that an odorant has the effect of a novel

stimulus, the response to which diminishes through habituation and

familiarization as the result of repetitive and otherwise uneventful


     Hedonic responses may be acquired as the result of experience with

food, sex, and the value system of adults.  Perhaps children learn how

to describe initially natural olfactory experiences by observing adults'

behavior.  There should be no difficulty in teaching a person to

experience the smell of skunk as something pleasant, provided one could

control the circumstances under which it was experienced.  The anecdotes

about odor memory and individual associations (e.g., Kenneth, 1923)

seem to fit this interpretation, as do cultural differences in choice of

perfume odors (Moncrieff, 1966).  It should be noted that the original

assumption of our research on this topic was that it is feasible to

code household goods with odorants, that is, to make them smell bad

in order to keep children away from them.  Our conclusion so far is

very clearly that one cannot depend on unlearned or inherent aversive

reactions to accomplish this, although we are still working on the

project by testing other odorants.

     It is reasonable to assume, however, that this kind of project

might be successful subjects, for it is a matter of common

observation that adults are very cautious with respect to odors.  This

is easily demonstrated by presenting them with a bottle xvith

instructions to smell it.  One is not likely to find a person who will

T. Engen

start out by taking a strong sniff of that bottle.  In Western

civilization "smell" and "odor" often imply something unpleasant or

distasteful.  The reason for this slanted view is not clear, but

advertising clearly takes advantage of it.

     Some experiments in odor identification and memory have provided

some data pertinent to this notion.  A group of 30 adult men and women

were asked to sniff each of 48 diverse odorants, ranging from pure and

simple compounds (e.g., 1-Propanol) to complex household products

(e.g., soy sauce).   Each person was asked, first, to say whether or not

the odor was familiar and,  second, whether he liked, disliked, or felt

indifferent about it.  He was not required to name the odorant, or

identify it in any technical sense, although often he did,  but only to

indicate knowledge of it from previous experience.  He was  encouraged

to use a high criterion for both judgmental categories, that is, to

respond that he did not know and that he was indifferent whenever he

was uncertain.   Each subject was tested individually and each received

a different set of 48 odorants drawn at random from a larger common set

of 100 odorants.  This sampling procedure is important for  it makes

possible more general observations.
                      Insert Table 4 about here

        Table 4 (Slide 8) presents the average proportions of ratings

of "like," "dislike," and "indifferent" for "familiar" and "unfamiliar"

odors.  The difference in the distributions of hedonic ratings of

familiar and unfamiliar odors is very clear.  Forty-six percent of the

familiar odors were judged as pleasant and only 25 unpleasant, whereas

                                   71                                      75

       T. Engen

       only 11 percent of the unfamiliar odors were judged as pleasant, but

       50 percent were judged as unpleasant.  This trend is also very

       consistent for individual subjects with a total of six exceptions for

       the 60 possible comparisons.  In other words, if an odor is unfamiliar

       to a person he is unlikely to rate it as pleasant.  This is not the

       result of there being some very ill-smelling odorants in the set.

       Analysis of individual data show that the same odorant may be found in

       any of the six cells in Table 4.  Approximately half of the odorants

       were judged "pleasant" by the average subject.  An odor that is

       familiar and pleasant to one subject may be unfamiliar and unpleasant

       to another, etc.  There were, of course, a few odors (e.g., pyridine)

       judged unpleasant by all who happen to receive it in his sample, and

       some of them had prior experience with it and some did not.  In

       general, however, this is a psychological response-response

       relationship that must be understood in terms of the individual's prior

       experience with specific odorants and in that sense the psychology of


            There are a number of possible methods mentioned in Beebe-Center's

       (1932) book on The Psychology of_ Pleasantness and Unpleasantness in

       addition to those described above for dealing with odor preferences and

       scaling.  All of them assume that pleasantness or hedonics represents  a

       perceptual dimension that can best be described by adaptation of

       classical psychophysical methods.  However, if it is the case that

       psychological preferences are individualized and dependent upon

       learning as the present data and most literary accounts of olfaction



T. Engen

suggest, then another approach and other methods would be helpful.  One

would first of all need to think about the learning paradigm that

illuminateshow hedonic associations to odors are acquired.  For a

start, consideration of the learning of language and the meaning of

words and signs would seem relevant.  Our question is:  "Under what

conditions does a neutral odor become pleasant or unpleasant?"  One

simple theory is that it can be explained as the result of simple

classical, or Pavlovian, conditioning through which an odor (the

conditioned stimulus) becomes associated with and capable of eliciting

part of the response originally associated with the unconditioned

stimulus.  For example, eating spoiled food (unconditioning stimulus)

makes one sick (the unconditioned response); the smell of food (the

unconditioned response) is associated with the food and is later

sufficient to elicit part of the unpleasantness of being sick (the

conditioned response).  The perception of the odorant thus acquires

hedonic value because it comes to mediate or signify an emotional state

through a process that is related to the general problem of meaning.

Psychologically this seems to involve the generalization of a mediation

process so that the original emotional response may be aroused by other

odorants or olfactory situations, including words describing the

situations such as "smell," to the extent that they are similar to the

original one.

     This approach would hardly be sufficient to explain all odor

association, especially the development of odor preferences, and some

other form of acquisition would seem necessary.  It is not unlikely

    T. Engen

    that most of our attitudes are learned through imitation of and

    guidance by parents or other desirable models used by advertisers.

    Being stimulated by the same odorant as the adult and observing his

    behavior would be sufficient for acquiring both the motor and cognitive

    aspects of the adult's response in the situations.  The effect of

    prestigeful models on the food preferences of children has been

    demonstrated (Duncker, 1938),  and such models may provide them with the

    social norm for appropriate behavior in various situations (see

    Bandura, 1962).  The change in preference  may take place quite

    unconsciously through a kind of reorganization of the perceptual

    attributes associated with the stimulus-object.

         To the extent that the hedonics of odor is a matter of acquired

    meaning of odorants, then the  traditional organoleptic or

    psychophysical technology should be supplemented with methods for

    measuring such psychological processes.  Some of these methods seem

    promising.  First, the "free association technique" could easily be

    adopted to the odor problem.   As normally used, this method consists of

    presenting a word to a subject with instructions to respond with

    whatever comes to mind, assuming that what does come is not free at all

    but determined by the meaning  the stimulus has for the person.  For the

    present purpose, one would merely substitute odorants for words.  Such

    a procedure would probably be  quite productive, judging from similar

    adaptations of this method in  the study of colors as stimuli compared

    with color words (Osgood, 1953).

         Techniques for scaling meaning would seem to have great potential

T. Engen

in this field.  Essentially, these methods use traditional psychometric

rating methods, but instead of being limited to one adjective, e.g.,

pleasantness, one uses a number of adjectives or concepts that appear

to be relevant to the stimulus material in question in an attempt to

find the significant ones.  In applying this approach Hosier (1941)

found that "	the meaning of a word may be considered as if it

consisted of two parts, one constant and representative of the usual

meaning of the word, one variable, representative of individual

interpretation in usage and associated context and general usage	"

(p. 139).   The possibility of substituting "odorant" for "word" in that

statement should be explored.

     Similarly, Osgood's (1952) "semantic differential" could be

applied to the odor problem with benefit.  It is a multidimensional

approach that combines the above scaling method and the free

association technique, and attempts to reduce the description of the

meaning of a stimulus, the "semantic space," to a few bipolar

dimensions.  These semantic dimensions are derived with the use of

factor analysis of the ratings of each stimulus on the larger number of

bipolar dimensions used originally.

     Finally, it is often observed that individual differences in

judgments of odors is unusually large.  However, the evidence that the

judgments of the individual subject is consistent from session to

session should not be ignored.  Young (1923)  showed that affective

judgments,  as he called them, are characterized by large

inter-individual  variation but small intra-individual variation.  The

      T.  Engen

      reason for the large  individual  differences may be the result of

      different experiences of different  people with respect to any one

      dimension, for example,  pleasantness.   This kind of variability reduces

      the reliability of most  measures of preference based on average

      judgments from groups of subjects.  It  seems  fair to assume that one

      might reduce this  variability  and understand  odor preferences better on

      the basis of a separate  analysis of data from each individual on a

      large number of potentially revealing dimensions of meaning.  There are

      as  yet only anecdotes about how  the individual's experiences influence

      his preferences and associations to odors.  The procedures suggested

      above stand a good chance of providing  more systematic and quantitative



T. Engen


 1.  Bandura, A.   Social learning through imitation.   In  M.  R.  Jones

             (Ed.)» Nebraska Symposium on Motivation.   Lincoln,

             Nebraska: University of Nebraska Press,  1962.

 2.  Beebe-Center, J.  G. The Psychology of_ Pleasantness and

             Unpleasantness.  New York:  Van Nostrand  Co., 1932.

 3.  Cederlb'f, R., L.  Friberg, E.  Jonsson, L.  Kaij, and T. Lindvall.

             Studies of annoyance connected with  offensive smell  from

             a sulphate cellulose factory.  Nordisk Hygienisk

             Tidskrift. 1964, 45,  39-48.

 4.  Duncker, K.   Experimental modification of children's  food

             preferences through social suggestion.  Journal  of

             Abnormal  and Social Psychology.  1938, 33,  489-507.

 5.  Engen,  T., and T. E. Corbit.   Feasibility of Olfactory Coding of

             Noxious Substances to Assure Aversive Responses  in Young

             Children  (Publication ICRL-RR-69-6 Injury Control Research

             Laboratory).  U. S. Department of Health  Education,  and

             Welfare (Public Health Service,  Environmental Health Service),


 6.  Engen,  T., and H. I. Katz.  Odor preference  and response bias in

             young children.  Unpublished manuscript,  Brown University, 1968.

 7.  Engen,  T., and L. P. Lipsitt.   Decrement and recovery of responses to

             olfactory stimuli in  the human neonate.   Journal of

             Comparative and Physiological Psychology,  1965,  59,  312-316.

T. Engen

 8.  Engen, T., L. P. Lipsitt, and H. Kaye.  Olfactory responses and

             adaptation in the human neonate.  Journal of_ Comparative

             and Physiological Psychology, 1963, 56, 73-77.

 9.  Engen, T., and D. H.  McBurney.  Magnitude and category scales of

             the pleasantness of odors.  Journal o_f Experimental

             Psychology, 1964, 68, 435-440.

10.  Guilford, J. P.  Psychometric Methods.  (2nd ed.)   New York:

             McGraw-Hill,  1954.

11.  Harper, R., BC.Bate-Smith, and D.  G. Land.   Odour Description

             and Odour Classification.   London:  Churchill,  1968.

12.  Kenneth, J. H.  Mental reactions to smell stimuli.  Psychological

             Review. 1923, 30, 77-79.

13.  Kniep, H. H., W. L. Morgan, and P. T. Young.  Studies  of affective

             psychology.  XI. Individual differences in affective

             reaction to odors.  American Journal o_f Psychology. 1931,

             43, 406-421.

14.  Moncrieff, R. W.  Odour Preferences.  New York:  John  Wiley, 1966.

15.  Mosier, C. L.  A psychometric study of meaning.  Journal of_ Social

             Psychology. 1941, 13, 123-140.

16.  Osgood, C. S.  The nature and measurement of meaning.   Psychological

             Bulletin, 1952, 49, 197-237.

T. Engen

17.  Osgood, C. E.  Method and Theory in Experimental Psychology.

             London:  Oxford University Press, 1953.

18.  Sandusky, A., and A.  Parducci.   Pleasantness of  odors as a function

             of the immediate stimulus contex.  Psychonomic Science,

             1965, 3, 321-322.

19.  Stein, M. , P. Ottenberg, and N.  Roulet.   A study of the development

             of olfactory preferences.  A.M.A. Archives o£ Neurology  and

             Psychiatry. 1958, 80, 264-266.

20.  Third Karolinska Institute Symposium on  Environmental Health.

             Nordisk Hygienisk Tidskrift, 1970, 51(2).

21.  Thurstone, L. L.  The Measurement of_ Values.  Chicago: University

             of Chicago Press, 1959.

22.  Young, P. T.  Constancy of affective judgments to odors.  Journal

             of Experimental Psychology. 1923, 6, 182-191.

     T. Engen


          Part of this research was carried out with support from the

     U.S. Department of Health, Education, and Welfare through The Injury

     Control Research Laboratory, Providence, Rhode Island.

 T. Engen
                               TABLE 1

                 Proportion of Choices of Odorants

       Each entry is the proportion of six-year old  subjects
  preferring the odorant in the row to the corresponding odorant
  in the column on the left.  Each matrix gives the  results for
  a different age group.
Butyric   Rapeseed    Diethyl    Neroli
 Acid       oil      phthalate    oil
Butyric acid
Rapeseed oil
Diethyl phthalate
Neroli oil






     T.  Engen
                                 Table 2

            Standard  scores for pairs of odorants based on the
                         proportions in Table 1.
Butyric acid
Rapeseed oil
Diethyl phthalate
Neroli oil






T. Engen
                              Table 3

              Hedonic scale values for five odorants
                  from five different age groups
                    based on pair comparison
Age group

      T.  Engen
                                   Table 4

           Response-Response Relationships in Judgments of Odors.

               Proportion of judgments of "like", "dislike", and
           "indifferent" of odors judged to be "familiar" or
           "unfamiliar".  The entries are proportions based on
           inspection by each of 30 subjects of a set of 48
           odorants  drawn at random from a common set of 90

           Response             "Familiar"          "Unfamiliar"

            "Like"                 .46                  .11

            "Dislike"              .25                  .50

            "Indifferent"          .29                  .39

T. Engen

                            FIGURE CAPTIONS

     Figure 1.  The development of olfactory preferences.   The responses

are expressed as the percentage of pleasant reactions.  (From Stein,

Ottenberg, and Roulet, 1958.)
     Figure 2.  Proportion of judgments indicating the relative

pleasantness of the odor (+) of each of four different compounds

plotted as a function of the age of the S_s.
     Figure 3.  Proportion of judgments under various instructions

indicating that the degree to which an odor was judged pleasant (+)  as

a function of age regardless of the odorant.  Data obtained from four

different odorants in Fig. 1 have been pooled to evaluate the effect of

different instructions, P, U, UoP, and Poll described in the text.
     Figure 4.  Proportion of judgments under various instructions

indicating the degree to which butyric acid was judged pleasant (+)  as

a function of age.

 David A. Kendall and Stanley P. Jacobs, Ph.D.
          Arthur D. Little, Inc.
        Presented at the Symposium
The Dose-Response Relationships Affecting
   Human Reactions to Odorous Compounds

              April 26, 1971
         Cambridge, Massachusetts
                                                  Arthur D Little, Inc

             David A. Kendall and Stanley P. Jacobs, Ph.D.

     Most of the reported information on odor studies has been based on

written or verbalized reports of individuals.  Generally a stimulus is

presented and the subject is asked to report his observations.  Where a

limited number of people participate, they are often given training or

gain it as a result of repeated trials.  Randomly presented unodorized

controls are helpful in estimating the frequency of incorrect responses.

Alternatively, populations may be asked to recall from memory their ex-

perience with odors and such reports can be used to describe an odorant's


     As has been indicated >  there are  clearly two kinds of responses to

an odorant.  One is the detection of a stimulus, i.e., the recognition

of a neural response.  The second is the subjective reaction to that

stimulus in terms of pleasantness or hedonic quality.   There may be a
                                                                Arthur DLittldnc

third level of response, i.e., detection at sensory  end  organs  followed

by some autonomic response without the individual's  awareness of  the  odor.

      One can argue that recognition of any odor, particularly  if man-made,

constitutes a problem from an odor pollution standpoint.  However, as

anyone who has attempted to build an olfactometer knows, producing odor-

less air even in a laboratory is not an  easy task.  There is a volume of

information on the other hand which indicates that certain types  of odors,

notably effluents from kraft mills, cause a high frequency of complaints

(30% of the respondents as far as 20 km away, {Lindvall}).

     Increasingly in the food acceptance area, there is  interest  in

measuring response by means of the individual's performance in contrast

to a verbalized or written statement.  Remember the low,  chrome, conser-

vative automobile designed to meet consumers' wishes which did not sell.

Although these methods may have somewhat less sentitivity, they may be

essential for reflecting the true attitudes of the individuals toward

the products or pollution.  This approach has been taken in the techniques

discussed by Mr. Flesh concerning the economic effects of "odorous" air

pollution on property values.
                                                                 Arthur D Little Inc

     Studies have been carried out in the past where one or more physio-

logical parameters have been measured as indicators of a response to an

odor stimulus.  In the recent report by Neuhaus and Goldenberg skin re-

sistance, arterial pulse and blood pressure, gastric motility and respira-

tory movements were monitored and changes noted as a result of both audi-

tory and olfactory stimulation.  Dr. Engen measured changes in respiration

with infants as a result of odor some years ago.  It would appear reasonable

that perception of odor might be reflected in the character of the respira-

tion curve and that an unpleasant reaction to odor might produce changes

in physiological parameters, analogous to the "scare reactions" on which

lie detectors are based.

     One of the requirements for such a study is to present the odorant

so that the performance of the test does not interfere with the responses

that one wishes to monitor.  To our knowledge no extensive effort has

been made to present odorants on a totally blind basis controlling both

concentration and duration of stimulus while monitoring a number of

autonomic responses.  The purpose of this study, which was carried out

initially under the auspices of the Manufacturing Chemists Association

and subsequently has been supported in part by the Air Pollution Control

                                                                Arthur D Little, Inc

      Office,  is to learn whether such methods  might provide more reliable

      indicators of a population's response to  various odorant presentations.

           Initially we hope to learn which,  if any, physiological parameters

      correlate with detection of odor and whether these responses change in

      character or magnitude as the concentration or duration of  stimulus is

      altered.   Hopefully by considering a response pattern one might  be able

      to  describe an adverse hedonic response as compared with a  benign or plea-

      sant experience.

           The preliminary  trials with the system are most  encouraging.   With

      highly trained analysts verbalizing their response to odorants in

      various  presentations, we detected a three-second time delay in  the

      hydrocarbon analyzer  system.  There are indications that this system

      may be useful for studying habituation  and fatigue as a function of

      duration and concentration.   It would also be most interesting to  study

      the effect of the rate of change in odorant concentration on the perceived


           The untrained subjects were able to  recognize and report on their

      odor experience with  good agreement.  The population  showed significant

      changes  in all four of the physiological  parameters measured.  Although

35                                                                   Arthur D Little Inc

we are not certain as to the meaning or importance of these changes,  in

several instances significant effects were only observed as a result of

the interactions of the concentration, duration, and odorant.  Also of

interest are the few individuals who profess not to be able to detect

any or all odorant treatments but whose autonomic responses indicate

some effects coincidental with the sample presentation.

     This report covers the description of the equipment for sample

presentation and monitoring of the autonomic responses; the results of

orientation trials with expert and untrained observers; and the test

design and recent observations with untrained but repeat performing


Test Chamber

     As one of the essential elements of this study, we have constructed

a test facility which permits completely blind sample presentation as well

as control and measurement of the odorant stimulus at the subject.  The

test chamber is a 6* x 3' x 41 cabinet constructed of aluminum.  The entry

door and glass windows are located in the side walls.  A frosted glass

panel is recessed into the front wall to allow reverse projection of


                                                                Arthur D Little Inc

        taped or filmed viewing material.   Video monitoring of the subject is

        possible through a small porthole.

             The air supply for the chamber is drawn through a fiberglass pre-

        filter and Dorex® activated carbon  canisters.   After the odorant in-

        jection point,  the air passes through a Static Mixer® to assure uniform

        distribution of odorant.   The air is then directed through the entire

        front panel of  the chamber, past the seated subject and is exhausted

        from the rear plenum by means of a  2 HP constant  speed exhaust fan.

        Flow rate can be .varied by adjusting a damper  to  shunt outside air

        directly through the fan.   The flow rate through  the chamber  can thus be

        varied from 150 to 400 cfm.

             The subject is seated in a swivel chair facing the front panel  of

        the  chamber.  An aluminum shelf at  arm height  was installed to minimize

        the  effects of  the turbulent flow from the lower  half of the  chamber.

        Care was taken  in the design of the dispersion plate to assure uniform

        flow across the top half of the chamber at the plane of the subject's

        face.   A sampling point located directly above the subject's  head provides

        access for the  Beckman Model 109-A  hydrocarbon analyzer which monitors

        the  hydrocarbon content of the air  stream.   A  Keithley Model  410

9                                                                      Arthur D Little Inc

microammeter has been substituted for the Beckman unit to reduce the

inherent electronic time delay of the circuit.  The air flow rate

selected both for comfort and speed of odor clearance for these experi-

ments has been 210 cfm which results in a linear flow of about 20 feet

per minute.  This represents volume flow through the chamber of 100

liters per second.

     A Harvard syringe pump injects carrier solvent (hydrocarbon) and

odorant into the air stream after the carbon filter.  Although the range

of gear settings provides a more than adequate range of constant flow

rates, at any single gear ratio  the rate can be varied remotely from

5-100% of the maximum.  Using a gear setting of 6 and a 2 milliliter

syringe, the flow rate can be varied from 0.1 to 2.0 microliters per

second.  The duration of the stimulus can be remotely controlled from

10 to 300 seconds.  The appropriate odor concentration is produced by

dilution in the carrier solvent.   For these experiments reagent grade

nonane, boiling point 154°C, has been used  as the carrier.  It is not

readily apparent at four times the use concentration.   The concentration

of odorant in the carrier is adjusted to allow intervals of 1,  4 and


                                                                Arthur D Little Inc

16 times a minimum recognizable concentration  for  the odorant at one

gear setting.    ••• .  •   • ••• •  •  '-••*•'••  •  ••''•' ••'•''•  *:   .y... '       ."

Test Protocol

     The subject is continuously  immersed  in the air  stream.   Odorant

samples representing randomly selected  treatment conditions are  pre-

sented at intervals during the course of a 30-minute  test  period.   In

reality, the first ten minutes of the session  are  taken  up by attachment

and adjustment of monitoring electrodes plus a stabilization  period.

Six odorant presentations can be  completed during  a 15-minute period.

The presentation of the stimulus  is completely blind  with  no  suggestion

for smelling.  The only change in the conditions within  the cabinet is

the presence of the carrier solvent below  2 mg/100 liters  and the odorant.

The physiological parameters monitored  are ECG,  using three electrodes

(two on the right arm and one on  the left); peripheral blood  flow by

means of an ear piethysmograph from which  is computed the  instantaneous

                                                               • i"'. ..;•.'
heart rate.  The respiration is monitored  by means of a  pneumograph and

from this is computed the integrated respiration volume.   Palmar sweating

is measured by galvanic skin resistance electrodes on both the right and
                                                                 Arthur DLittldnc

left hand.  All the data is accumulated on an eight-channel thermo-

recording Hewlett Packard polygraph.  This is equipped with an eight-gun

oscilloscope for displaying any or all signals.

     The output of the hydrocarbon analyzer and the differential pressure

in the chamber are recorded to monitor the test conditions.  The indivi-

dual's appearance and attitude are observed by closed circuit television

with the opportunity for recording this on tape.  We have not attempted

to evaluate this information since it appears to be highly variable from

individual to individual.

     There are opportunities for additional measurements which might be

more responsive to the variations in odorant presentation.  We have the

necessary transducers available to measure EEC and EMG and have considered

some possibilities for air flow measurements at the nares.

Exploratory Studies

     Preliminary studies were carried out with relatively volatile

odorants such as ethyl mercaptan, methyl isobutyl ketone and carbon

disulfide.  Overt reactions such as sniffing or tasting the air,  facial

movements, actual frowning or grimacing to express discomfort, and


                                                                Arthur DLittldnc


       indications of simple awareness of odor such as eye opening or twitching

       were apparent when the concentration of the odorants was substantially

       above the minimal detectable level, i.e., between 10 and 40 times threshold.

       During this period, equipment was not available for controlling the flow

       rate of odorant-carrier solvent and therefore all samples were injected

       substantially instantaneously although the concentration within the

       test chamber was apparently 80% of its maximum for about 7 seconds.

       Under these conditions changes were noted in the heart rate, the respira-

       tion rate and the character of the respiration curve, i.e., either

       irregular rate  or shallower or deeper volume.  No particular pattern

       could be distinguished from these results to associate with a particular

       odorant.  In several instances, the test subjects did not report the

       detection of one or more of the presentations even though variations were

       apparent in their autonomic responses coincident with the odorant pre-

       sentation.  With the addition of the Harvard syringe pump and its auto-

       matic timing control, studies were carried out with a trained panel to

       determine the minimum detectable concentration in the test chamber for

       comparison with previously obtained data in the static test chamber.
                                                                       Arthur D Little; Inc

After overcoming the problems of continuous uniform injection of solvent

into the air stream by adding aluminum heating blocks and reducing the

bore of the syringe needle to 27 gauge, it was possible to present the

odorant at a desired concentration with less than 20% variation in con-

centration during the 200-300 second interval.  By varying the tempera-

ture of the heating block, it was possible to test a variety of odorants

of lower volatility.  However, the selection is still limited.  After

screening a number of different materials, amyl acetate was selected

as a "pleasant" odor and tetrahydrothiophene as the "unpleasant" odor.

Both of these compounds have been used with some frequency in odor

research by various workers.  Using experienced analysts who reported

orally their sensations in the test chamber, studies were undertaken to

examine intensity of perceived odor as a function of time during a 300-

second test period.  Using amyl acetate, the odorant was perceived some

3-5 seconds before it was detected with the total hydrocarbon analyzer.

With the highest test concentration, the maximum intensity was recorded

as much as 8 seconds before the hydrocarbon analyzer.  The discrepancy

between the apparent intensity and the detected concentration was found
                                                                 Arthur D Little, Inc


         to be due in part to time delay in the electrometer circuit of the hydro-

         carbon analyzer, and this was reduced by substituting a Keithley unit.

         A second delay may result from sample holdup in the gas sampling system

         to the flame.  This has not as yet been corrected.  A most interesting

         observation, however, was the apparent decrease in intensity of odorant

         during the longer exposure intervals.  This was particularly evident

         with amyl acetate.  One subject after 120 seconds was not aware of odor

         in the test chamber, although the concentration was constant.  A second

         subject was aware of only a low intensity of odorant during the exposure

         time after 100 seconds.  From comment and actions in the preliminary

         trials, both amyl acetate and tetrahydrothiophene seemed to represent

         the degree of separation in "pleasantness" that might be necessary to

         observe differences in autonomic responses.  Through repeated studies

         with the trained group, a concentration of amyl acetate of 64 mg/ml of nonane

         was found to produce a reasonable range of odor intensities within the

         flow rate constraint of a single gear setting of the syringe pump required

         for fully remote manipulation.

              With 100% flow at gear setting 6, the delivery was found to be 2

         microliters per second.  Since the air flow in the chamber is 100 liters

-404                                                                    Arthur D Little, Inc


per second, this results in a calculated concentration for amyl acetate

of 128 micrograms per 100 liters.  For the short duration only 80%

maximum long term FID response was attained because of the dispersion of

the peak.  Thus, the concentration of the amyl acetate was close to 100

micrograms per 100 liters.  The minimum test concentration of amyl

acetate, based on the flame ionization response to nonane, was 6 micro-

grams per 100 liters of air and the intermediate concentration was 25 micro-

grams per 100 liters.  For the longer durations (over 60 seconds), the

percent flow was reduced to give a comparable detector response.

     The concentration selected for tetrahydrothiophene in nonane was

0.56 milligrams per milliliter.  At the maximum flow rate of 2 micro-

liters per second the maximum concentration of tetrahydrothiophene for

the 15 second duration was 80% of 1.1 micrograms'per 100 liters of dilu-

tion air or 0.88 microgram per 100 liters.  The lower concentrations were

0.22 microgram per 100 liters and 0.05 microgram per 100 liters.

     The pen deflections from the flame ionization detector measuring

nonane for these three dose rates were 6-7 mm, 24-26 mm for the low and

intermediate concentrations and, with 5X attenuation, 20-22 mm for the

high dose rate.  The detector response value was maintained regardless       j_OD

                                                                Arthur D Little Inc


     of duration or odorant during a single day of testing.  Although, some

     variation in specific detector response has been noted from day to day,

     the ratios between concentration remained constant.

     Test Design

          It is the purpose of this program to determine whether autonomic

     responses can provide useful information concerning human detection of

     odorants and second to aid in defining what might be considered unpleasant

     or annoyance reactions to odor.   It is quite easy to postulate that the

     type of odorant,  its concentration, and the duration of exposure are

     important in detection and recognition but the interactions of these

     three factors could be most significant in producing annoyance.

          Any test design therefore had to allow for at least two odorants

     at two time durations and at two concentrations.   Since there was some

     concern as to the appropriate concentrations,  a third level of this

     factor was included.

          It was essential to design the test to yield the maximum amount of

     information not only of the effects of each of the three factors, but

     also the effects  of interactions.   Also in the development of a test

     protocol, it was  agreed that no more than six exposures would be presented

                                                                     Arthur D Little, Inc


in any one session.  Thus, in order to produce a complete block design,

the total number of treatments  (12) were separated into two blocks of

six.  Dr. Irwin Miller located  the reference to a 3 x 2 x 2 factorial

test design confounded in two blocks of six with a description of the

analysis.  To balance the test, three replicates containing two blocks

must be included in each iteration.  Each group of replicates could be

reiterated as many times as desired.  Although three iterations appeared

to be adequate, the three replicates were reiterated four times.  Twelve

subjects were randomly assigned to one of the three replicate groups

which defined the six treatments to be included in each of the two

blocks.  Although the treatments in each block were assigned, the order

of presentation for both blocks and treatments within blocks was


     In order to test the system and give the subjects experience with

the procedure, each subject was given a training session which included

six presentations in an undefined order.  Although the response data was

collected, it was not intended  for analysis.  At the conclusion of the

test, one iteration of the three replicates was carried out with three

of the subjects to provide information on intrasubject variability.

                                                                Arthur D Little, Inc


      This data has not been analyzed at the present time.

           At the conclusion of each test,  the subject was  asked what odors

      he had perceived, how many times and  in what order.   He was then asked

      to indicate which were pleasant and unpleasant.   This provided the oral

      response information to be compared with the physiological responses.

      Data Acquisition

           Although it is quite obvious in  many cases  to observe changes in the

      physiological parameters that coincide with the  initiation of the stimulus,

      a standard method for tabulating the  data had to be defined.   Since

      there was no certainty which, if any, of the variables might  reflect

      differences between treatments, and review of the polygraph data was time

      consuming, a variety of measurements  were taken  so that additional inter-

      pretations could be made.  It was convenient to  take  all measurements  in

      millimeters.  In most instances the measurement  during the stimulus was

      compared to a measurement taken during a 15-second control interval

      prior to the initiation of the treatment.   A primary  criterion for

      initiating a test was the visual assurance of normality in the polygraph



                                                                      Arthur D Little Inc


1.  Instanteous Heart Rate (IHR)

     a.  the maximum and minimum instantaneous heart rates during

         the 15-second interval prior to the initiation of the treatment;

     b.  the maximum and minimum heart rate during the period from initia-

         tion of the treatment to the return of the hydrocarbon curve to

         80% of its maximum; and

     c.  maximum heart rate during the interval from 80% maximum to 15

         seconds prior to the next treatment.

     From these five measurements, three variables were computed for

analysis as follows:

         (1)  the maximum IHR during the treatment period minus the

              maximum IHR during the control period

         (2)  the minimum IHR during the 80% stimulus concentration

              period minus the minimum IHR during the control period.

              (This should be negative.)

         (3)  the maximum IHR after the stimulus minus the maximum IHR

              during the control.

                                                                Arthur D Little, Inc


 2.   Respiration Rate

     An analogous set of measurements was taken from the integrated

     respiration curve and the differences computed.

     a.  the maximum interval between inspirations during the 15-second

         control period was subtracted from the maximum interval between

         inspirations in the treatment.   The maximum interval between

         inspirations is in actuality a measure of the slowest or minimum

         respiration rate.

     b.  Similarly the minimum interval between inspirations during the
         control period is substracted from the minimum interval between

         inspirations during the treatment.  This number reflects only

         the single maximum change in respiration but might describe


3.  Respiration Volume

     In addition from the integrated respiration curve,the maximum pen

     excursion or respiration volume was measured during the control period

     and during the treatment and the difference computed.  These six

     variables were tested by analysis of variance.
                                                                 Arthur DLittldnc


     Some change in galvanic skin resistance was apparent in every  treat-

ment with all subjects.  However, the degree of the depression in skin

resistance varied markedly between individuals.  We felt that the magnitude

of the depression might not be a true representation of the treatment

effect because of the variability among subjects.  Therefore, the length

of time was measured during which the skin resistance was depressed.

This distance was compared as a ratio with the length of time of the

experiment from the initiation of the treatment to the decline in

odorant concentration to 80% of its maximum.  A supplementary experiment

was carried out by measuring the time depressed after this point and

comparing it as a ratio with the total time to the end of the treatment.

We had observed a second depression in galvanic skin resistance coinciding

with the decrease in odorant concentration particularly with tetrahydro-

thiophene at the higher concentrations.  Thus two additional variables

were tested for significance by analysis of variance.

Preliminary Findings

     The test chamber with the remote monitoring equipment provides an

exciting opportunity for examining the effects of odorants in a variety

of presentations.  That interactions of complex odorants do produce            -| •

                                                                 Arthur D Little, Inc


measurable and significant changes in the autonomic response is quite

obvious in the results obtained.  However, at this stage of the work it is

not clear how best to measure and record these changes nor is it evident

how these changes may reflect the hedonic importance of an odorant.  It

is quite evident that concentration, duration, and odorant interact

jointly to produce significant and in some instances apparently reasonable


     A major effect consistently observed in both heart rate and respira-

tion is an effect of the duration of the stimulus.  Unfortunately, it may

well be that this is an artifact introduced by the definition of the

measurement to be made.  Particularly, in the long (120-second) treatments, the

15-second control interval is substantially shorter than the treatment

interval and no correction has been made for this bias in the present


1.  Oral response

     The recognition (or detection) of odors in the test chamber was

indicated by a "no" or "yes" (0 or 1) response for the treatment.  Often

the subjects used simple descriptions which could be related to the order

of presentation.  As would be expected, the main effect was from concentration

                                                                Arthur D Little Inc


with 99% confidence.  A second effect due to odorant was evident with

95% confidence.  The frequency curve (or curves) obtained by plotting

detection versus concentration is a typical dose-response curve segment.

Surprisingly, duration had no effect nor were any significant interactions


2.  Physiological Response

     There are three variables tabulated concerning the instantaneous

heart rate each of which represents a treatment response minus a control


     a.  Maximum IHR during the treatment up to the time where the stimulus

     concentration returns to 80% of its maximum.  The change 'in the maximum

     heart rate depends on the duration; there are no other effects and

     no interactions.  During the short (15-second) stimulus, the mean

     increase in IHR was 2.4 beats per minute while with the long (120-

     second) duration, the mean increase was 6.1 beats per minute +1.1

     (all limits are 95% confidence).

     b.  Similarly with heart rate depression, the effect of duration is

     significant at the 99% confidence level.  However, concentration,

     duration and odorant interact jointly.  From the graphs of log       ^ A e-p

                                                                Arthur D Little, Inc


             concentration versus beats per minute  (Figure 2), the effect is

             noted with the longer duration of the  two lower  concentrations only

             with amyl acetate.  Both are depressed 6 beats per minute + 2.5.

             The questions arises Is there a calming or pleasant effect?

             c.  In addition to measuring the maximum heart rate during the

             treatment, the maximum increase in heart rate after the 80% maximum

             concentration was measured.  Here concentration had a significant

             effect (95% confidence), but concentration, duration, and

             odorant interact jointly.  As indicated in the graphs (Figure 3)

             the short duration with tetrahydrothiophene shows the largest

             effect even at the intermediate level producing an elevation of

             5 beats per minutes + 2.55 (95% confidence).

             d.  In considering respiration the minimum rate  (i.e., longest

             time between breaths) might be indicative of "breath holding".

             Again, the main effect is found with duration.  During the short

             stimulus, the interval is 0.8 seconds longer than the control while

             with the long duration the breath interval is 2.6 seconds +0.6 seconds

             longer than the control.  This could be breath holding but may be

             an artifact.  There is also a significant effect of odorant with the

114                                                                   Arthur DUttldnc


interval for amyl acetate being 2.2 seconds longer  than  the  control

period while that for tetrahydrothiphene is only 1  second +0.6

longer (95% confidence).  One would expect the reverse to be true

if this represents an avoidance reaction.

e.  The shortest interval between inspirations was  taken to indicate

"sniffing" and one might postulate that the greatest difference

would be observed at the lowest concentrations.  The duration

effect was found to be the main factor, and was significant at the

99% level.  The largest decrease in the breath interval was observed

with the long duration with a value of -1.2 seconds +0.23 while the

decrease with the short treatment was -0.5 seconds  as compared with

the control interval.

Of possibly more interest is the interaction of concentration and

odorant apart from the effect of duration.  As plotted in the

graph (Figure 4), the minimum value for amyl acetate (-1.2 seconds)

occurs at the lowest concentration—possibly sniffing for detection.

However, the greatest change with tetrahydrothiophene (-1.4 seconds)

occurs with the intermediate concentration.  One would like to relate

this to sniffing in some type of alarm reaction.
                                                            Arthur D Little, Inc


f.  Respiration volume both maximum and minimum was significantly

different only as an effect of the duration of the stimulus.

In both instances, the greatest change was observed with the longer

duration.  Until the possible interference of an artifact is corrected,

this data, although it does show some changes, does not appear to be

very informative.

g.  Galvanic skin resistance, both hypothetically and from the litera-

ture reports, should be the most interesting and responsive measure.

Indeed, there does not appear to be any treatment or subject who does

not show some response according to the measures applied in the test.

One may easily postulate that an early indication of unrest could

well be detected in palmar sweating.  Since this data is recorded

as a ratio of the duration of reduced resistance as compared with

the duration of the experiment up to the time of the return of the

concentration to 80% of the maximum concentration, the artifact of

time is not present.  The only significant effect is the interaction

of concentration, duration, and odorant.  Examining the data presented

in the graph (Figure 5), it would appear that there is a general

rise between the intermediate and the high concentrations.   The

                                                           Arthur DLittldnc


apparent difference between the means for the two durations of

amyl acetate is probably not significant.  The primary interaction

is indicated in the data for tetrahydrothiophene where the maximum

change occurs with the intermediate concentration and the long

duration, but occurs with the maximum concentration for the short

duration exposure.  One can rationalize that  the  intermediate

concentration and long exposure to tetrahydrothiophene allows memory

to associate the odor with gas leaks and fire, producing some alarm

reaction.  Whatever the rationale, influence of interactions on the

response is of considerable interest and worthy of further exploration.

The difference in the response to amylacetate as compared with

tetrahydrothiophene is the effect which was sought in the program's

                                                           Arthur D Littldnc

                                Figure 1


                               Figure 3

                        MEAN MAXIMUM HEART RATE   '

                   (After 80% Maximum Concentration)
          120 sec
                     15 sec.
                                         15 sec.
                                                          Concentration, Duration

                                                              and Odorant

                                                                 Interact Jointly
                                                       120 sec.
              012           012

               Log Concentration             Log Concentration

                 Amyl Acetate                    Thiophan

                            Log Concentration

                          0        1        2
                   	I	,	,	
                              Figure 5

                      GALVANIC SKIN RESISTANCE
                 (During 80% Maximum Concentration)

                 Concentration, Duration and Odorant
                         Interacting Jointly




15 sec
                      •   120 sec.
           Log Concentration

             Amyl Acetate
                         All means  +0.44
                                         15 sec.
                       0        1
                     Log Concentration

                                                               Arthur D Little, Inc

Erland Jonsson:

On annoyance reactions observed within human populations

In several countries survey investigations have been made to study how
people perceive and react  to bad smelling air pollutants.  The aim
of these investigations has in most cases been to get information
about the seriousness of the smell nuisance problem; if the exposure
can be regarded as so strong that corrective action is called for.

With one exception - the Eureka-study (1) - no studies have compared
quantitative odor exposure data to community reaction data.  However,
there are a few studies where community reactions in areas with as-
sumed different exposures  to the sources of the odors (2, 3, 4, 5).

In surveys made to get data on the effects of offensive odors exposing
human populations different reactions have been studied.  Most of those
reactions have been annoyance reactions, that usually means that the
exposure has not been a toxic one.   An annoyance reaction to odorants
was at the Third Karolinska Institute Symposium on Environmental
Health (6) defined as the  extent to which people report being bothered,
disturbed or irritated.  It is always an unfavorable reaction of which
the subject is aware.

Reactions studied in different odor surveys

In most surveys one or more of the following five kinds of reaction to
offensive smells have been studied:

-  Reports of interference with the everyday activities of the exposed

-  Reports of feelings of annoyance caused by offensive smell,

-  Reports of physical symptoms of physiological changes,
                                                                           •fl O<

      -   Reports  of  actual complaints  to  an  authority,

      -   Reports  of  various forms  of direct  individual  action  to modify  the
         environment other than through complaints.

      Concerning  the first factor  the  surveys  have shown  that  the  following
      interferences  are reported:

      -   with relaxation,  rest  and sleep,

      -   with entertainment of  friends and relatives.'

      These interferences  cause feelings  of  annoyance,  which was the second
      of the dependent variables.   This variable may be defined as  the extent
      to which people report a  dislike to being bothered, disturbed or
      irritated by the exposure.

      Reports of  physical  symptoms - the  third variable - appear only to a
      relatively  small extent.   Questions about the existence  of nausea,
      vertigo,  headache, breathing difficulties etc.are, however,  as a rule
      included in the questionnaires used.

      Experience  in  Sweden suggests that  only  a very small percentage of the
      population  ever take any  action  with respect to the annoyance with
      smells in terms of making this known to  the authorities.  For this
      reason readiness to  complain - the  third of the mentioned kinds of
      reactions - has been defined as  the willingness to:

      -   visit an official,

      -   assist in the organisation of community action groups for  abatement
         of the smell,

      -  write or  telephone to an official,

-  attend  a meeting,

-  sign  a  petition.

The willingness  is defined by  reports  of  the  following reactions:

-  ever  having felt like  complaining,

-  actually having complained,

-  being very likely  to complain, pe.rhaps being  likely to  complain.
   but not sure, or probably not being likely to

Concerning the last of the dependent variables - reports of various
forms of direct  individual action other than  complaining - the  following
forms of direct  individual action maty,  be noted:

-  installation  of air conditioning so that windows can be closed,

-  moving  out of the  area,

-  shutting of windows,

-  increased drug consumption.

Variations in reactions in a population that  are not caused by
variations in the exposure

In several investigations differences  in annoyance reactions have
been established between different categories of  interviewees,  separated
by demographic variables such as sex,  age, civil  status, occupation
etc.  and by individual attitudes towards and  dependence on the  pollu-
tion source (6).  These "irrelevant" factors make it difficult  to
establish an unequivocal correlation between  dose and  response.  It
will seem reasonable  to ass.ume that the risk  of  influence on the


reactionsjfrom "irrelevant" factors is least when the exposition is
very strong or very weak, respectively.  Under all circumstances, the
population in which correlations between exposition and reaction -
if any - have been found, must be minutely defined.
Results of some investigations of annoyance reactions caused by offen-
sive smells

Information to what extent the odor problem exists in industrialised
communities is very incomplete.  However, in 1964 an investigation
was made in Sweden using a sample of about 2000 persons, representa-
tive for the country.  The purpose was to throw light on the problem
to what extent people were bothered by noise and air pollution.  In
the investigation 13% reported bother from evil smell in the ambient
air (2).

Studies concerning bad smell from sulphate cellulose industries have
been made on a number of occasions.  Among these, an interview study
covering about 400 persons selected at random from the population
living in the vicinity of (within 20 km of) a sulphate cellulose factory
was carried out at Monsteras in Sweden in 1959 (3).  The main purpose
of the investigation was to study the inconvenience caused by air
pollution from the factory.

"The sulphate smell" bothered 36% of those interviewed, and one-third
of those bothered described their discomfort as severe.  Despite the
complaints only 5% considered it disadvantageous that the factory was
located in the area.  Comparisons between different parts of the area
investigated indicated the existence of a correlation between the
position of the interviewee's residence in relation to the factory
and the comparative incidence of persons bothered.

The investigation showed that certain groups of people were more dis-
comforted than others.  Women were more bothered than men and younger
more than older.  It was also evident that persons bothered by the

sulphate odour complained to a greater extent of the odour from a
sulphate cellulose factory in the area investigated than those not
bothered by the sulphate odour.

A study of the same kind as that at Monsteras was made at Morrum,
Sweden, in 1963.  It was an unmasked interview survey of annoyance by
offensive smells, made among a population in the vicinity of a recently
opened sulphate cellulose factory.  A relatively large proportion of
the population had signed a petition to the public health authority
complaining of the offensive smell from this factory.  The material
comprises about 400 persons selected at random with those having
signed the petition represented to a disproportionate extent (4).

Of the population in the area investigated, 10-15% suffered major
annoyance from the factory while the corresponding figure for those
living in the immediate vicinity of the factory was just under 20%.
It should be noted, however, that the purpose of the investigation was
not masked, so the degree of annoyance found is probably rather too

Among those having signed the petition complaining of the existing
conditions, 47% reported that they had not experienced any annoyance
from the flue gases at the time of the investigation.  Only 31% re-
ported major annoyance.  On the other hand, almost 50% considered the
smell to be much the same as when they had signed the petition.  The
petition must accordingly be rejected as a reliable measure of the
extent of the annoyance.

The results also show that the annoyance derives from factors other
than the exposure to the factory.  Thus, annoyance is reported more
frequently among those with previous respiratory and cardio-vascular
diseases, with a propensity to neurosis, sensitivity to aircraft noise,
propensity to displeasure with other factors of community life and no
connection with the factory.
                                   95                                   127

       Two  investigations made  on  similar  lines  as  those of Monsteras and
       Morrum were  undertaken at Clarkston, Washington  (5) in 1965, and at
       Eureka,  California in 1969  (1).   The purpose of  the latter  investigation
       was  to relate  quantitative  data  describing exposure to quantitative
       data describing  community reactions.   The American investigations
       establish  about  the  same bother  frequencies  as the Swedish.

       As regards oil refineries data are  available only from Swedish investi-

       In connexion with the planning of oil  refineries, the population around
       already  existing plants  in  Denmark, Norway and Sweden were  interviewed
       by persons using idiomatically identical  questionnaires, in 1965 and
       1968 (7).

       The  results  show that only  moderate annoyance reactions were present
       around modern  refineries.   The oil  refinery  disturbance factors were
       mainly considered less bothering than  normal background disturbing
       factors  such as  traffic  noise.

       Summarising  comments

       The  effects  of offensive odours  on  the exposed population dealt with in
       this paper and which have been studied at the investigations under reference,
       are  as a rule  based  only on the  exposed individual's experience of
       conditions.  This fact reduces the  possibilities to establish objective
       data about reactions of  interest.   It  is  therefore desirable that very
       soon light is  thrown on  the problem which physiological reactions
       are  connected  to the annoyance reactions.

       The  empirical  investigations accounted for show  that the odor problem
       concerns relatively  many human beings.  Particularly near sulphate
       cellulose  factories  a large proportion of the population is bothered.
       The  investigation results have,  however,  been of limited value for
       establishing air quality criteria as the  reactions studied  have not


been related to the exposure conditions.  Later studies, e.g.  the
Eureka investigation,indicate, however, that it should be possible to
relate quantitative response data to quantitative exposure data.


      1.  Jonsson, E., Deane, M. and Sanders, G.: Community reactions to
          odors  from pulp mills - A pilot study in Eureka, California,
          Paper  presented at the Third Karolinksa Institute Symposium on
          Environmental Health, Stockholm, June 1970, (Unpublished).
      2.  Jonsson, E. and Sorensen, S.: Forekomsten av bullerstorningar i
          samhallet - Tva enkatundersokningar -. Nord. Hyg. Tidskrift,
          Vol. XLVII, 1967, 21-34.
      3.  Friberg, L., Jonsson, E. and Cederlof, R.: Studier over
          sanitara olagenheter av rokgaser fran en sulfatcellulosafabrik
          (I). Nord. Hyg. Tidskrift, Vol. XLI, 1960, 41-49.
      4.  Cederlof, R., Friberg, L., Jonsson, E., Kaif, L. and Lindvall, T.:
          Studies of annoyance connected with offensive smell from a sulphate-
          cellulose factory. Nord. Hyg. Tidskrift, Vol. XLV, 1964, 39-48.
      5.  Nahum  Z.Medalia: Community Perception of Air Quality: An Opinion
          Survey in Clarkston, Washington (Cincinnati, Ohio: U.S. Public
          Health Service, 1965).
      6.  Jonsson, E.: Annoyance Reactions to External Environmental Factors
          in Different Sociological Groups.  Acta Sociologica, vol 7 -
          fasc.  4, Copenhagen 1964.
      7.  Rylander, R. and Sorensen, S.: Omgivningshygieniska konsekvenser
          av ett planerat oljeraffinaderi pa Lysehalvon.  National Institute
          of Public Health, March 1969.

                   Draft of Working Paper
          Studies of Public Opinion of a Traffic Odor
                        Presented at
April 26-30, 1971 Odor Symposium - Cambridge, Massachusetts
                        Presented by
                      Karl J.  Springer
       Manager, Vehicle Emissions Research Laboratory
                 Southwest Research Institute
                     San Antonio, Texas
                            99                                   131


    The control of an air pollutant such as malodorous traffic odor depends

on three "musts. "  First there must  be a way to measure and express the

odor in simple,  understandable and repeatable terms.  Second, there must

be a way to effect a reduction in the odor that is  feasible economically.

Finally,  the reduction in odor must do something positive as far as the

public is concerned.  Work has been in progress by government,  industry,

and independent laboratories on all facets  of these three"musts. "  The

Vehicle Emissions  Research Laboratory at Southwest Research Institute

has been privileged to work on all three at the same time under the

guidance of the National Air Pollution Control Administration (now the

Air Pollution Control Office of the  Environmental Protection Agency).

    At last year's  meeting in Stockholm, the working paper based on work

done at Southwest Research was concerned with  correlation of instrumental

measurements with sensory panel ratings.  In an attempt to extend the use-

fulness of the  human panel, several components  in diesel exhaust, some

odorous and some not,  were related to sensory ratings.  The odor ratings

were  from a trained panel of ten persons using the PHS quality/intensity

kit, better known as the  Turk kit, shown  in Figure 1, as the primary

reference.  This subjective quantitative-qualitative method has been well

described in the literature and has formed the basis for most government -

       sponsored diesel odor research for the past five years.  The standards

       are a "diesel composite" or "D" in intensities 1 through 12, a  "burnt-

       smoky" or "B" in intensities 1 through 4, an "oily" or "O" in intensities

       1 through 4,  an  "aromatic-aldehydic" or "A" in intensities  1  through

       4, and a "pungent" or "P" in intensities  1 through 4.  The "D"  series

       describes diesel odors most comprehensively in that it is composed of the

       four sub-odors or odor qualities.  The quality/intensity kit is designed

       so that the twelve odor intensity steps of the "D"  series ("D"-l is weakest,

       "D"-12 strongest) are perceived linearly by the observer.  Concentration

       of odorant in each "D" bottle is twice that in the next weaker one.  The

       "D" is the primary intensity standard,  and the others are used to describe

       the quality of the odor.

          This working paper describes a pioneering effort to acquire  public

       opinion of diesel exhaust in terms of the  Turk kit panel ratings of the

       same odor.  The intent of this effort is to relate and  close the loop between

       technology, measurement and response.  In doing so, additional  insight

       is gained that can be used in consideration of control effectiveness, pre-

       paration of standards and measurement methodology.  For purposes of

       this description, the paper  is divided into two sections, dealing first

       with the dose and then with  the response.

       The Dose

          To  adequately describe  the dose, the method of  presentation and

the means by which the participants were asked to express their opinion

is just as important as what the odor was.  Several approaches were con-

sidered in the effort to determine public opinion of diesel exhaust odor.

The idea of "canning"  diesel odors was attractive because it allowed

use of a national probability sample, but it could not be used because

diesel odors cannot be preserved.  Another idea involved use of the Turk

kit for sampling by the public, but this idea was rejected because the odor

standards do not smell enough like diesel exhaust.  The most plausible

idea,  and the one finally adopted,  was to construct a mobile laboratory

for odor  evaluation which would incorporate a "live" odor source,  that

is, an operating diesel engine.  This method presents odors which are

realistic and which have not had time to change, but it also introduces

mechanical complexity and less precise participant sampling measures.

    The  design of the  mobile  odor evaluation laboratory resembles very

closely that of the stationary odor measurement room  in the SwRI

Vehicle Emissions Research Laboratory.  Figure 2 is a schematic of

the mobile laboratory  showing compartment boundaries  and location of

major items of equipment.  The trailer was divided into the front

mechanical compartment,  the central odor presentation room, and the

rear engine/generator compartment.  The basic idea was to sample  con-

tinuously raw exhaust  from the engine stack and mix this sample with

  diluent air in ratios from about 50:1 up to several thousand to one.
       The diluent air was passed through a prefilter, a particle filter,
  a charcoal filter,  and more recently a scrubber-humidifier before being
  mixed with small  amounts of diesel exhaust to help assure consistency
  in odor levels.  All portions of the odor presentation system which
  come into direct contact with odor samples before they reach the parti-
  cipants are made  of relatively inert materials, usually either Teflon*
  or stainless  steel.
       Early in this  research effort,  it was agreed that odors presented to
  participants  should simulate those found in the exhaust of typical muni-
  cipal buses,  since these urban odors seemed to be of most concern to
  the public.  Since  over 80% of municipal buses are powered by six-
  cylinder two-stroke diesel engines, it was decided to use a multicylinder
  two-stroke diesel  operating at moderate speed and load as the odor source.
       The exhaust odor from two engines,  a Detroit Diesel 3-53 and a 3-71,
  were evaluated by the odor panel.  Followinga number  of evaluations at
  various speeds and loads,  the powerplant/odor source chosen for the
  mobile laboratory was a 3-71 "E"  series, two-valve, two-stroke engine
  driving a 35-kilowatt generator at  1200  rpm.
       The environmental control systems are located primarily in the front
    Reg.  Trademark, E.I. duPont de Nemours & Co.

mechanical compartment of the mobile laboratory.  The criteria for the

odor room air were a temperature of 75°F, relative humidity of 50%,

odor background so low as  to be undetectable, and a static pressure in

the room slightly above atmospheric to prevent odors from leaking  into

the room.  All makeup and recycle air passed through a prefilter,  a

particulate filter, and a charcoal filter before entering the odor pre-

sentation room.

    The interior of the mobile laboratory is finished with colorful,

attractive materials, as a part of the total effort to  make the participants

comfortable and at ease.   Figure 3 is a view of the odor evaluation  room

looking toward the front of the facility. Figure  4 is another view of the

odor evaluation room, this time looking toward the rear of the facility.

Figure 5 shows  the typical  position of a participant preparing to evaluate

an odor.  Figure 6 is a view of the Sniffmobile from the right rear,  and

Figure 7 shows  the left side.  Figure 8 shows the mobile laboratory on-

site in San Antonio, with the patio umbrella and  chairs  which were used

as a "staging" or "preinterview" area  by the  survey crew.  These items

were brightly colored and formed a nice place to talk to passers-by and

keep up with data tabulation.

The Response

    The use of a questionnaire  to aid in eliciting and recording public

      opinion on diesel exhaust odor was judged to be preferable to the alter-

      native of verbal personal interviews.  This decision was reached  because

      it is more economical to have group interviews using a self-administered

      questionnaire than to have one-to-one verbal interviews, because a

      questionnaire can "ask" questions of each participant in an identical

      manner, and because structured responses were  considered desirable.

      Structured responses lend themselves to analysis more readily than res-

      ponses which are made up of the participants'  comments about the odors

      given in their own words.  Proper questionnaire design can do much toward

      reducing participant bias by using neutral phrasing in instructions and, in

      the case of a multiple-choice questionnaire, by allowing each person to

      choose answers from the same lists.  National Opinion Research Center,

      subcontractor to SwRI on this research program,  has a well-known

      capability in the areas of public opinion  survey work and questionnaire

      development.  The expertise of the NORC staff was utilized throughout

      the questionnaire development and the field survey.

           In general terms, the major objectives of the questionnaire were to

      obtain informed consent, to  obtain personal information, to obtain an un-

      biased opinion of a set of diesel exhaust odors, and to obtain certain

      supplementary information from each respondent  which might aid in

      interpreting his or her  response to the odors.  In addition, it was decided


that the questionnaire should be as simple and brief as possible because

persons of all educational backgrounds were expected to participate in

the survey.  Words which might bias the participants,  such as  "diesel, "

"bus, " "truck, " and "air pollution" were intentionally avoided, both on the

questionnaire and in verbal contact with the public at the test sites.  A

copy of the questionnaire used in the  1969  survey is included for reference

as the next page of this paper. The front of the sheet contains  personal

information questions,  and the back is used for odor testing.  The consent

of the people to participate in the  survey was  obtained after informing them

of what they would be required to  do.  If anyone expressed reservations,

that person was  asked to exclude himself or herself from  the group.  The

multiple choice questions and the  theory behind their use came directly

from NOR C, whose  experience showed that people are much less likely

to balk at answering personal questions  if  they do not have to be too specific.

    By all standards, the most difficult portion of the questionnaire to

design was that section which deals with response of the general public

to a series of diesel exhaust odors.  The idea of using  something other

than descriptive words on a scale came  from an informal  discussion with

Dr. Amos Turk  some 2 years before this project was begun. He related

that a food company  had once determined public opinion of a food item

served in a cafeteria by using a questionnaire having five  human faces


This survey is to find out your opinion of a common odor typical of U.S. cities. You will be asked to sniff
the odor three times. The odor has no known health hazard and your participation is completely voluntary.

For each item, circle the one code number which fits you best. All answers are confidential You will
never be identified.
     YOUR SEX:
         Male  .
     YOUR AGE:
          15-24	1
         25-44	2
         45-64	3
         65 or more	4
         Excellent	1
         Good	2
         Fair	  3
         Poor  	4
         8 yrs. or less  	1
         Some High School	2
         Completed H.S	3
         Attended College	4
     Employed	1
     Housewife	2
     Student	3
     Retired	4
     Other	5

     Professional, business
       owner or manager	1
     Clerical, office or sales	2
     Skilled or semi-skilled
       wage earner	3
     Other	4

     Under $4,000	1
     4,000-6,999	2
     7,000-11,999	3
     12,000 or more	4
     Regularly .  .
     Never .  . .  .

Next, a series of different odor levels will be presented. Please check the appropriate box under the figure which best expresses your feeling.
Test 3
            Are any  of these odors so bad that someone
            should take steps to reduce them?
                                 How often do you experience odors like these?
                  No—None of them   	

                  Yes-Test 1   	
                  Yes-Test 2

                  Yes-Test 3
                                      Very often

                                      Fairly often


                                      Never   . .  .

whose expressions ranged from pleasure to displeasure.  The proposal

of this research project suggested the use of a similar method to eli-

minate bias due to inconcistency in word usage among the general populace.

Matching one's feelings about something with one of several representative

facial expressions is perhaps even more accurate than free-form verbal

description of those feelings, and it is certain that the  structured responses

generated by the former method lend themselves  more readily to quanitia-

tive interpretation.

    As the idea of cartoon faces  evolved,  a conference with NOJRC was

held out  of which arose a suggestion to add bodily action to the faces.

This idea allowed a deeper interpretation of responses to  odors,  e. g. ,

the 5th cartoon character is  not only displeased with the odor he perceives,

but he is also taking steps to physically  remove himself from the odor

source.  Many versions of these  cartoons were  tried in an effort to space

them equally along a  scale of objectionability, that is, to make the step

changes  in reaction depicted by the cartoons equal.  A descriptive word

was added under each cartoon figure only after the characters were in

essentially final form.

    The  last two questions on the back of the questionnaire were added

with very limited applications in  mind.  The question dealing with action

on odor abatement was  intended for use with a person's responses to

odors to examine that person's  sense of values regarding action on odors.

Pre-survey validation of the survey questionnaire was accomplished in

two separate steps.   Step 1 determined how a sample of people related

the five cartoon characters to the concept of objectionability.  Step 2

involved the matching of descriptive words with the established cartoon


    NORC personnel compiled population distributions for each of the

five cities to be sampled according to sex, age, education, and income.

These  distributions were based on I960 census figures and were restricted

to areas within city limits.  Upon examination of the distributions,  a

meeting was held in the city to be sampled with SwRI personnel, the Pro-

ject Officer,  and local air pollution control personnel in attendance.  Out

of this meeting came the suggestions of local air  pollution control people

as to areas which should be searched for test sites,  and the candidate sites

were subsequently evaluated for presence of required foot traffic, adequate

space for mobile laboratory access and parking,  and presence of desired

characteristics of the population.  During the conduct of the survey itself,

people were approached by members of the survey crew and asked to

participate in the survey.  Quotas were based on  a sample size of 500 per-

sons per city, but 600 to 700 persons were routinely surveyed before all

the quotas had been filled.

     Calibration of the Sniffmobile required that sample flow to each parti-

cipant be balanced and equalized and that the concentrations in each cone

be identical.  Calibration also meant that the trained odor panel had to

define the relationship between sampling rate and odor level before and

after operation at each survey site.  Figures 9-13 show the "D", "B",

"O",  "A", and "P" panel ratings  during the May-to-October 1969 testing

period.  The essentially flat curves indicate the ability of the trained panel

to respond almost identically in terms of the Turk kit to the same odor at

infrequent intervals.  This is the first known instance of  such an experi-

ment with the bottle rating method.

Results  - 1969 Survey

    Over 3000 quota-sampled individuals in five  cities, three or four  sites

per city,  were included in the survey of 1969;  The cities were San Antonio,

Chicago,  St. Louis, Philadelphia, and Los Angeles.  San Antonio and  Los

Angeles  represent two extremes of air pollution and public awareness,

while the other three seemed to be somewhere  between in this regard.  To

discuss the  1969  survey,  let's consider first the  most important  result, the

cartoon responses for the three odor tests  of "D"-2 (Test 1), "D"-4 (Test 2)

and "D"-6 (Test  3), in turn.  Figure 14 is  a frequency polygon for the three

tests,  with a mean response of 2. 44 (Test  1),  3. 37 (Test  2), and 3. 96 (Test

3).  The numerical result is based on assigning the cartoons a number
                                 113  '

      from 1 to 5,  1 being pleasant and 5 being most objectionable.  Figure 15

      shows the range of cartoon responses in terms of the "D", "B",  "O",

      "A",  and "P" ratings.  The participants considered the stronger odors more

      objectionable than weak ones  and, this is shown by the smooth, easily de-

      fined curves.  One question on the back of the questionnaire asked, "Are

      any of the odors so bad that someone should take steps to reduce them?"

      In answer to this question,  the survey  results looked like the top curve in

      Figure 16.  This result,  not unexpected, agrees somewhat with the inde-

      pendent calibration of the cartoon on a 0 "pleasant" to 100 "most objection-

      able" scale.

          At this point, we have a good indication of people's reaction to a live,

      realistic traffic odor and some insight on its relative objectionability.

      Of course, we have also discovered the variation by city and site on odor

      response,  effect of sex, age, health, occupation, employment, income,

      and smoking.  Some variation in odor response from city to city can  be

      found which remains unexplained by consideration of variation in socio-

      economic characteristics of the several populations or other quantitative

      factors.  In particular, San Antonio participants seemed to find the ex-

      haust odors less  objectionable and Los Angeles participants seemed to find

      them more objectionable  than did the Chicago, St. Louis,  and Philadelphia

      participants,  whose responses were close to the average.  Participants

      and passers-by in Los Angeles were very quick to associate the survey



with air pollution, but those in San Antonio had little to say about pollution,

and again the participants in the other three cities fell between the ex-

tremes.   San Antonio has the cleanest air of the cities surveyed, and Los

Angeles is at least the best known of the five cities for polluted air if it

is not in fact most heavily polluted.  It is  quite possible that a relationship

exists between response to diesel exhaust odors (or, for that matter,

almost any pollution-related subject) and public awareness  of and attitudes

toward air pollution, both of which depend on the severity of the local

pollution problem and the type and amount of publicity given it.

    The effects of education, employment, and income,  commonly known

as "socioeconomic" categories  on odor  responses were quite similar.

The more educated, the more highly skilled, and the more  affluent the

participant, the more objectionable he found the odor.  Older people

(over 65)  found the odor less  objectionable,  as  did those who judged them-

selves in  poor health.   These two factors  seemed to be related and tend

to support the idea of decreased sensory perception with age.

1970 Survey
    Public opinion of the "D-2" odor level (Test  1) was adequately de-

fined by the 1969 survey.  Public opinion of the "D-4" and "D-6" levels

(Tests 2 and 3,  respectively) could not be taken as absolute but had to

be used with the full knowledge  that each participant's response might

have been influenced somewhat by his prior exposure to one or two other


    With this in mind,  an additional 2100 quota-sampled individuals were

included in the 1970 survey.  The main objective was to elicit responses

from singly presented odors and to  inquire further about the objection-

ability of these odors.  Some slight changes were made to the front side

of the questionnaire, shown on the next page, but the cartoon method

of rating was  retained, and more and different questions were employed

in an attempt  to gauge the public objectionability of the odors.  Improve-

ments in the Sniffmobile included installation of additional diluent air

purification equipment in order to control himidity and assist in approaching

a "D-O" odor level.  A nominal "D-2", 3,  4, 5, and 6 were presented,

each at  a separate site,  and the results are plotted in Figure 17.  The

almost linear response is similar to the  1969 survey except the results

above about "D-4" are lower,  indicating  a possible "history" effect during

the 1969 survey.  Below about "D-4", the 1969  results were lower than

1970.  This may be  attributed to the completely different public attitude

about air pollution promulgated by public officials and the media since

January 1970.  This plot also illustrates  that the public does not center its

responses given an adequate means  of expression such as the cartoon scale.

    Figure  18 is a  plot of the fraction of  participants who perceived an

                                                          Form Approved
                                                          •Budget Bureau No. 85-S69017
                         IMPORTANT! PLEASE READ.

 This survey is to find out your opinion of a common odor typical of U.S. cities. The odor has no known
 health hazard and your participation is completely voluntary.

 1.   YOUR SEX:
         Male .  .
         Female .
25-34 ....
65 or over . . . .
. . . n

. . . .D
    8 yrs. or less	
    Some High School	I	I
    Completed H.S	LJ
    Some College .......  I	I
    Completed 4 yrs. College . .  .  I	I
         Professional, business  .  .
          owner or manager
         Clerical, office or sales .  .
         Skilled or semi-skilled  .  .
                                                     wage earner
                                                   Other  . .  .
    Under $4,000	CD
    10,000-14.999  ....
    15,000 or more	
NAPCA Form No.
HQ 27 (4/69)

        Next, an odor will be presented.

        1.    Did you smell anything?
             THEN ANSWER QUESTIONS 2, 3, AND 4.
        2.    How often have you experienced this odor?

                 Very Often	
                 Fairly Often .

             If you have experienced the odor, where?

                 Both Indoors and Outdoors  .
                                    If you were to experience an odor just
                                    like  this outdoors, would you find  it
                               4.    If an odor just like this occurred  out-
                                    doors,  should  someone take  steps to
                                    reduce it?

odor and found the odor objectionable according to question no. 3 asked

after the odor test, "If you were to experience an odor just like this

outdoors, would you find it objectionable?"  In addition, the fraction of

respondents who perceived an odor is also plotted.  These results appear

as expected at the noticeable odor levels of "D-2" to "D-6" and improve

the confidence in the  1969 survey data.

    As a separate item, an attempt was made to ascertain public opinion

of diluent air devoid of diesel exhaust or diesel odor. After substantial

system clean-up and  purge, a dusty  residual odor remained,  not diesel

nor fuel, that was slightly noticeable.   Though the participants,  a group

of over 500, were handled the same way as in higher level tests,  a large

percentage (60%)  perceived an odor and a higher percentage than expected

(about  50%) of those who said they perceived an odor, termed it objection-


    Our analysis of the 1970 survey is  incomplete at this writing, but

several points may help to explain the final test results.  First, the pre-

sentation of such an extremely low level odor actually confused a number

of the participants.  They were  originally asked to sniff an odor and then

a sample containly nearly no odor was presented.  It should be noted that

at odor levels below "D-2", the trained odor panel agreement  begins to

scatter substantially.  "D-2" is a faint  odor and the difference between

      "D-l" and "D-2" is discernible mainly to the trained panelist.  It

      appears that the experiments at the near zero odor level are really incon-

      clusive and that the sniffmobile test methodology is not suited for ac-

      quiring opinion to "no-odors. "  In fact,  this "no-odor" test seems to ask

      a completely different question of the research, the answer to which is

      certainly beyond the scope of the current effort.

          In summary,  ratings of dies el exhaust odors by the public can be

      related to the intensities and qualities which those odors are judged to

      possess by a trained odor evaluation panel,  according to the standards

      of the Public Health Service Quality/Intensity Kit.  This result makes

      possible the evaluation of the effectiveness of odor control methods in

      terms of the approximate extent to which their  use would reduce ad-

      verse response to dies el exhaust odor.


                                        Dilution Air
                                                                  Fume Removal

                                                                    Dilution Air

              Fume Removal


                 Diluted Exhaust


                                        QOOO  OOO  OOO
                                              ^^— Stools                                     
                                 May 9
                                                             Test 3-Signal Pressure • 18 in. Hg
               Test 2-Signal Pressure - 10 in. Hg
              -e-	e	
                                                             Test 1-Signal Pressure - 2 in. Hg
June 27       July 24       August 29    October?. 1969

         Date of Calibration

  a: 1
                                                     Test 2
                                                    - -- ©
       5/9/69   6/27/69  7/24/69  8/29/69  10/7/69
                  Date of Calibration
Figure 10.  Initial, Interim, and Final
             Sniffmobile  calibrations by SwRI
             Odor  Panel  -  "B" Ratings
                                                                 5/9/69  6/27/69   7/24/69  8/29/69   10/7/69
                                                                            Date of Calibration

                                                              Figure  11.   Initial, Interim,  and Final
                                                                            Sniffmobile calibrations by SwRI
                                                                            Odor Panel - "O" Ratings

                                                            (0 ,
                                                            cc 1
                                                                                                                   Test 2
                                                                  5/9/69   6/27/69  7/24/69  8/29/69  10/7/69
                                                                              Date of Calibration

                                                                Figure 13.  Initial,  Interium, and Final
                                                                             Sniffmobile calibrations  by SwRI
                                                                             Odor  Panel -  "P" Ratings

                                                                              Test 2
                                                           Questionnaire Cartoons


B 1

                                                                                                                      1  i
                                                         Average Odor Test Response
              AND QUALITIES


            is "8
            I c
            ? s
            I &
            .* s
            £ •
            c    30


                                                      50 I
              40  o



   Test 3—Three
Times Test 1 Intensity
  158                                   126


I   3


               2              34               5

                           Nominal "D" Rating of Odor Presented

~ 0.8

                             . Perceived Odor
•Found Odor Objectionable
               2.3              4              5

                           Nominal "D" Rating of Odor Presented


                              OR OBJECTED TO ODORS


              By R. David Flesh
        Copley International Corporation
      For presentation to the Symposium on
     Dose-Response Relationships Affecting
     Human Reactions to Odorous Compounds
           Cambridge, Massachusetts
                April 26, 1971

                      OF ECONOMIC COSTS DUE TO ODORS

               By R. David Flesh, Copley International Corporation

       Whenever the actions of private enterprise cause a divergence between private

and social costs, the government should intervene. This principle of welfare economics

was developed and first expressed by A. C. Pigou about 40 years ago.   It is based on the

premise that an excess of social costs over private costs means that market forces alone

are insufficient to make private enterprise bear the total responsibility for its actions.

In the language of economics, such a condition is known as an external diseconomy.

       It is the role of the economist, working under Pigou's principle,  to measure

external diseconomies and thereby determine the extent of governmental intervention

necessary to supplement market forces.  Yet, under existing knowledge, social costs

are usually difficult and often impossible to measure.  Those that can be measured are

termed economic costs.

       The social costs of odors remain in the identification stage.  Except for one

possibility, measurements cannot be made with any degree of confidence.  The single

possibility is that of property value differentials.  Numerous opinion surveys of home-

owners have been conducted with the common result that odors are perceived to be a

principal effect of air pollution and that, other things equal,  the effect ultimately mani-

fests itself in the form  of reduced property values. This supports the underlying

hypothesis of the property value approach to the  measurement of social costs, which


maintains that such effects are capitalized negatively into the value of land and any

lasting improvements fixed thereon.

       Under a present study of the social and economic impact of odors, methods

are being developed to measure property value differentials due to odors. Initially,

these methods will be applied to neighborhoods of single-family dwelling units.  They

will benefit particularly from the recent work of two teams of investigators: Ronald

G. Ridker-John A. Henning and Robert J. Anderson, Jr.-Thomas D. Crocker.

The Ridker-Henning Study

       Economic theory suggests that property values might be statistically explained

by regressing neighborhood, occupant, and property characteristics against the market

price of the property.  The first published study to include  an index of air pollution as

an independent variable was undertaken by Ridker-Henning  in the St.  Louis Metropolitan

Area.   They selected census  tracts to serve as neighborhoods on the assumption that

the census tracts were homogeneous with respect to the variables relevant to the study.

Both cross-sectional and time  series analyses were performed.

       In the cross-sectional analysis, almost 94 percent of the variation in average

single-family dwelling unit values between neighborhoods was explained by a regression

equation  using 12 independent variables:  a measure of air pollution,  median number of

rooms (a proxy for house size), percent of all units recently built (January 1950 to

March 1960), units per square mile (a proxy for lot size), distance in time from the

central business district,  accessibility to highways and major thoroughfares, school

quality, percent of workers in blue collar occupations, persons per unit, percent of

units occupied by non-whites,  whether census tract was in Illinois or Missouri (a dummy

variable  to account  for differences in property taxes),  and  median family income in 1959.


The variable representing air pollution was based on milligrams of sulfur trioxide per

100 square centimeters per day averaged for one year.  The regression coefficient of

that variable indicated that, other things equal, average single-family dwelling unit

values would decline $245 for every increase of 0.49mg SCL/lOOcm^/day.

       The time series study was confined to only one neighborhood. During 1962,

residents of a quite, middle-class neighborhood in St. Louis began to complain about

the choking gases from a plant that had just been taken over by a metal fabricating

company. The odors were said to cause nausea, headaches, and other discomforts.

In order to assess the effect of air pollution  on property values apart from changes

in general market conditions, a control neighborhood similar to the affected area was

selected. A regression index of recorded property sales was computed for each neigh-

borhood and compared over a 9 year period as shown in Table I.

Table I.  Regression index of property values in St. Louis, 1957-1965.
Regression Index
Area Control Area
— •
Affected Area
of Sales
Control Area
* P is the probability that the observed difference between the index values for the
  affected and control areas  for a given year is due to sampling error (two-tailed
  test). The null hypothesis under test is that the true difference is zero.
Source: Ronald G. Ridker, Economic Costs of Air Pollution (New York:  Frederick
        A. Praeger,  1967).

       Ideally, the indices for the affected and control areas should not have been

significantly different prior to 1962.  Beginning in 1962, however, significant differences

should have appeared to reflect the effect of air pollution.  As can be seen in column P,

the ideal pattern was interrupted in 1959 and 1960 by divergences which are significant

to the 1 and 3 percent levels, respectively. In addition, the effect of air pollution does

not appear until 1963.  Ridker-Henning could not find evidence to explain these unexpected

divergences and did not comment on the apparent lag of the effect until 1963.

       To indicate the dollar losses that may have been caused by air pollution, Ridker-

Henning multiplied the average percentage differences in the indices for 1963,  1964, and

1965 by the average property value in the affected area for the same period. The figure

used to represent property value in the neighborhood under investigation was $12,100,

the median value of single-family dwelling units in the neighborhood found in the U. S.

Censuses of Population and Housing:  1960, St. Louis, Missouri-Illinois.  Applying this

figure to the indices produced a mean property value for the three year period  of $11,302.

The mean percentage differences between the affected and control areas for the same

period was 9.6 percent.  Together, these figures  suggest that homeowners in the affected

area suffered an average loss of $1,085 to the value of their homes.  Since there were

1,068 single-family dwelling units in the affected area in 1960, the total property-value

loss was estimated at $1,158,780.

       The time series analysis was based on so many assumptions, that Ridker-

Henning claimed little confidence in the results.  Most disturbing was the failure of the

cardinal assumption "other things equal" by the unexplained divergences between the

affected area and control area indices in 1959 and 1960.  The same cause(s), along


with air pollution, may have been present to some extent in 1963,  1964, and 1965.

The Anderson-Crocker Study

       Cross-sectional studies were more recently performed by Anderson-Crocker

in three standard metropolitan statistical areas—St. Louis, Kansas City, and Wash-

ington.   Except for air pollution,  observations on all neighborhood, occupant, and

property characteristics were taken from 1960 U.S. Censuses of Population and

Housing.  The air pollution data consisted of average annual sulfation and suspended

particulates readings taken during the mid-1960's at various locations in each city by

the National Air Pollution Control Administration.  Anderson-Crocker proceeded with

their analysis on the assumption that the several years difference  between the socio-

economic data and the air pollution data would cause the impact of air pollution to be

somewhat understated.

       Seven independent variables were used in the regression equation:  the two

measures of air pollution,  median family income, proportion of housing units classed

as dilapidated, proportion of housing units more than twenty years old in 1959, pro-

portion of occupied units inhabited by non-whites, and distance from the central business

district.  At least 75 percent of residential value differentials were explained by these

variables in St. Louis, 80 percent in Kansas City, and 69 percent  in Washington.

       The equation provided figures corresponding to what Anderson-Crocker called

"marginal capitalized" property values.  The highest marginal capitalized damage

figure obtained for suspended particulates  was for the predominantly single-family

dwelling units census tracts in Washington, where the addition of 10 mg/m  /day to a

relatively low concentration of suspended particulates caused an average reduction of

    $700 in the market value of owner-occupied homes.  For St. Louis and Kansas City,

    the comparable figures were $119 and $126, respectively.  However, the average

    concentrations of suspended particulates in these two cities were substantially higher

    than in Washington.

            The highest marginal capitalized damage obtained for sulfation was $912 for

    each additional 0.1 mg SO^/100cm2/day. This figure was obtained for owner-occupied

    homes in the portion of Kansas City lying in Kansas, where average concentrations of

    sulfation were extremely low.  hi St.  Louis, where such concentrations were quite high,

    the comparable figure was $158.

            Over the ranges of sulfation and suspended particulates observed by Anderson-

    Crocker, marginal capitalized damages and the responsiveness of damages to air

    pollution appeared to decline with increases in average pollutant concentrations. That

    is, total damages seemed to increase at a decreasing rate and the proportionate change

    in damages relative to the proportionate change in pollution concentrations appeared to

    decline with increases in average pollutant concentrations.

    Preliminary Findings of the Present Study

            Based on the subjective nature of odor evaluation and the types of neighborhood,

    occupant, and property data available, two methods of measuring property value dif-

    ferentials due to odors seem promising. Both methods are applicable when odors have

    been identified—by neighborhood opinion surveys or public complaint records-- as the

    principal cause of air pollution annoyance to homeowners.  The first method is currently

    being tested.  The second is untried.  It evolved from discussions with real estate ap-

    praisers relative to gathering data for the first method and from experience with the



use of rating scales in other fields of research.

       The "Other Things Equal"  Approach.   This approach is virtually identical

to the Ridker-Henning time  series analysis. The only difference is that the control

neighborhood must be carefully located either adjacent to the source of odors in the

affected neighborhood or adjacent  to a very similar source of odors.  In either case,

the control area must be upwind of any odors.  The intent of this strategy is to elimi-

nate as many factors other than odors as may contribute to property value differentials.

       The preliminary findings obtained from affected and control neighborhoods

adjacent to two major oil refineries in the Los Angeles Metropolitan Area indicate

that the method is a workable one. Selling prices for a sample of single-family dwell-

ing units sold in each of the neighborhoods for the years 1966,  1968,  1969, and 1970

were obtained from local boards of realtors.  As shown in Table II, a divergence of

the annual averages of these selling prices began in 1969.  The divergence occurred

within one year after a new, but inadequately controlled sulfur plant began operations

within the refinery complex adjacent to the affected area.
Table II.  Average selling prices of single-family dwelling units in neighborhoods
adjacent to two major refineries, 1966 through 1970.
Affected Neighborhood (AN)
Selling Change from
Price Previous Year
$26,200 -- %
29,300 +5.9**
29,100 -0.7
30,000 +3.0
Control Neighborhood (CN)
Selling Change from
Price Previous Year
$23,300 — %
25,700 +5.2**
27,100 +5.4
29,000 +7.0
Change from
Previous Year
" %
 *  First three quarters only.
**  Average change over the previous two years.
n.a., data not available.

       There is no doubt that the substances released into the atmosphere by the new

sulfur plant were perceived as obnoxious odors by the residents of the affected neigh-

borhood.  The annual number of odor complaints from these residents to local authorities

increased by a factor of 10 between 1968 and 1970.  An opinion survey was conducted in

the area in December, 1970. Sixty-three percent of the adult residents interviewed said

they were bothered by odors in their neighborhood,  considered the presence of the odors

to be  a serious problem, and believed the odors originated at the refinery. Thus, if

other things remained equal during the period shown in Table II, then about $1,800 of

average property value differential in 1969  and about $3,000 of average property value

differential in 1970 would be due to the presence of odors in the affected neighborhood.

An examination of other variables must be completed, however, before these findings

can be considered conclusive.

       A major drawback to the method was found to be the time required to locate

suitable neighborhoods for the study. Although several cities were visited, Los

Angeles was finally chosen because of the constancy of meteorological parameters,

especially wind direction.

       The Problem Rating Scale Method.  Under this approach, an average property

value differential between a neighborhood affected by odors and a similar, but relatively

odor-free neighborhood would be computed  using data obtained from local boards of

realtors,  the Society of Real Estate Appraisers, or the U. S. Bureau of the Census.

Of course, depending upon whether the researcher is interested in an immediate situa-

tion or long-term trends, such differentials may be computed for the current year or

for the past several years.


       Following these mechanics, a series of discussions would be held with highly

qualified real estate appraisers, who normally serve both areas. The appraisers

would be asked to identify all of the factors they believe may have contributed to the

differential.  The purpose of these discussions would be to confirm the existence of

factors that typically contribute to such differentials and, more importantly, to include

other factors that are unique to the neighborhoods under investigation. In general, such

factors would arise from differences in neighborhood, occupant, and property charac-

teristics .

       Adjustments  of the computed differential for differences in occupant charac-

teristics may be avoided through careful selection of the control neighborhood.  The

principal characteristics to match with the affected neighborhood would be median

family income.  Adjustments for property characteristics may be necessary depending

upon the average age,  size, and construction quality of the single-family dwelling

units in each area.   Fortunately, adjustments for property characteristics usually are

not difficult to perform. Real estate appraisers  are specially skilled  in such work and,

therefore,  would be  of valuable assistance to the researcher.

       Adjustments  for differences in neighborhood characteristics  are much more

difficult to accomplish, since they are grounded on the  subjective evaluations of both

buyers and sellers.  Buyers and sellers may evaluate all neighborhood characteristics

known to them as a set,  rather  than individually.  If so, the  "other things equal"

method would be the only method to measure property value differentials due to  odors.

Taking an optimistic position, the author offers the following  approach to measure

differences in neighborhood characteristics in general and differences due to odors

   in particular.

          A randan sample of homeowners in the affected and control neighborhoods

   would be asked to indicate on a scale of, for instance, one to ten, how much each

   of the neighborhood characteristics (identified by the appraisers as possible contrib-

   uting factors)  represents a problem to them. A weighted average of the distribution

   of the ratings would then be computed for each factor in each neighborhood using a

   standard formula such as
                               £    f.
                              1=1    l
   where X  = the weighted average of the distribution, x^ = the ith rating and f. = the

   frequency of responses to the ith rating.  For example,  assume that only three neigh-

   borhood characteristics had been identified as possible contributing factors and that

   interviews of 100 homeowners in each neighborhood had produced the distributions of

   problem ratings given in Figure 1.  The weighted averages of these assumed distri-

   butions are given in Table HI.

Figure 1. Distribution of problem ratings of three neighborhood characteristics resulting
from hypothetical interviews with homeowners in a neighborhood affected by odors and a
relatively odor-free neighborhood.	

Neighborhood              Distribution in                       Distribution in
Characteristic        Affected Neighborhood*                Control Neighborhood*
Presence of Odors
in the Neighborhood

  # Responses         5  10  15 25   25  15   5      5  15  25 25  15  10   5
              I	1	1	1	I—I	1	1	1	1   I	1	1	1	H
       Rating 12       4      6       8      10  12       4       6      8      10
Distance to Central
Business District from
the Neighborhood

 # Responses 5   20  50  20   5                     20  50  20  10
              I	1	1	1	1	1	1	1-
       Rating 12       4      6       8      10  1.   2       4       6       8      10
Quality of Schools
Serving the

 # Responses  60 30  10                             515  30  30  15   5
                  1	1	1	t—I	1	hH	1   I
       Rating 12      4       6       8      10   12       4      6       8      10

Explanation of Rating Scale: "1" represents no problem; "10" represents a serious problem.

*Assumes a total of 100 interviews completed in each neighborhood.



  Table III» Weighted averages of the distrubutions depicted in Figure 1.	

  Neighborhood               Affected               Control
  Characteristic           Neighborhood (AN)    Neighborhood (CN)   Difference (AN-CN)

  Presence of Odors              6.2                  3.8                  +2.4
  in the Neighborhood

  Distance to Central             3.0                  2.2                  +0.8
  Business District from
  the Neighborhood

  Quality of Schools              1.5                  3.5                  -2.0
  Serving the

  Totals                       10.7                  9.5                  +1.2
         The net figure, +1.2, would represent the total contribution of differences in

  neighborhood characteristics to the average property value differential, while the net

  figure, +2.4, would represent the contribution due to the greater impact of odors on

  the affected neighborhood.  If the average property value differential had been computed

  to be $1,800 and if differences in property characteristics had been determined to

  account  for $1,000 of the differential, then differences in neighborhood characteristics

  would account for the remaining $800 of the differential.  By simple proportion, the

  presence of odors would reduce the average value of single-family dwelling units in

  the affected neighborhood by $1,600 relative to the average value of comparable units

  in the control neighborhood.


         It was Pigou's principle that whenever the actions of private enterprise cause

  a divergence between private and social costs, the government should intervene.  Yet,



the measurement of social costs is usually difficult and often impossible for the economist

to perform.  Except for one possibility, so it is with the social costs of odors.  The single

possibility is that of property value differentials.

       Based on the subjective nature of odor evaluation and the types of neighborhood,

occupant, and property data available, two methods of measuring property value differ-

entials due to odors seem promising.  One of these methods is currently being tested in

the Los Angeles Metropolitan Area. It requires careful location of an odor-free neigh-

borhood near an odor-affected neighborhood so that any property value differential found

between the areas may be attributable to the presence of odors.  This method is work-

able, but difficult to apply generally because of the many constraints necessary to main-

tain "other things equal."

       The second  method is untried.  As in the first method,  an average property value

differential between two neighborhoods would be computed using secondary data.  Follow-

ing this, a series of discussions would be held with highly qualified real estate appraisers

in order to identify all the factors that they believe may have contributed to the differential.

The factors would be categorized as arising from differences in neighborhood, occupant,

and property characteristics.

       Adjustments of the computed differential for differences in occupant and property

characteristics would be readily accomplished using secondary data.  Adjustments for

differences in individual neighborhood characteristics such  as the presence of odor would

be approached through a survey of homeowners in the neighborhoods under investigation.

To substantiate the  validity of this method, it would be necessary to determine whether

the opinions of homeowners are representative of the opinions of home buyers and sellers



as they negotiate together.


       This paper is based, in part, on a study being performed under Contract

No. CPA 70-116 with the Environmental Protection Agency, Air Pollution Control

Office.  The author is grateful to Mr. R. Paul Weddell for his contribution of

information and to Dr. Donald G. Gillette and Mrs. Marian O. Doscher for their

comments leading to the final  draft.


1.     A. C. Pigou, The Economics of Welfare (4thed.; London:  Macmillan,


2.     Ronald G. Ridker and John A. Henning, "The Determinants of Residential

       Property Values with Special Reference to Air Pollution," Review of Economics

       and Statistics, XLK (May,  1967), pp. 246-257, and Ronald G. Ridker,

       Economic Costs of Air Pollution (New York; Frederick A. Praeger, 1967),

       chapters 6 and 7.

3.     Robert]. Anderson, Jr. and Thomas D. Crocker, Air Pollution and Housing:

       Some Findings, Paper  No. 264 (Institute for Research in the Behavioral,

       Economic, and Management Sciences,  Herman C. Krannert Graduate School

       of Industrial Administration, Purdue University, 1970).

                                                PRELIMINARY  DRAFT
               Health Effects  of Pulp  Mill  Odor'in
                      Anderson,  California
              Margaret Deane and John R.  Goldsmith,  M.D

          California State Department of  Public  Health
              Bureau of Occupational  Health  and
                   Environmental Epidemiology
For presentation at the Conference on the Dose-Response  Relationships
Affecting Human Reactions to Odorous  Compounds,  April  26  -30,
Boston, Massachusetts.


      During late summer of 1970 the California Air and Industrial

 Hygiene Laboratory carried out an environmental survey designed to

 measure the presence and intensity of pulp mill odor in the Anderson-

 Cottonwood area of northern California.   Both chemical and organo-

 leptic measurements were used.  In spite of several limitations,

 this  location seemed appropriate for carrying out a health survey

 in relation to exposure to pulp mill odor.  A previous study in a

 northern coastal area of California (Eureka) had been conducted

 during the summer of 1969 to estimate annoyance reactions  to pulp

 mill  odor and to test their relationship to several measurements

 of exposure to odor.  Hovever, in the Eureka study questions concern-

 ing physical health were asked only of respondents who indicated

 that  they were "very much" or "moderately" "bothered "by the odor.

 Furthermore, the health question was phrased "Do you get  any of the

.following symptoms when you are bothered by the odors?"  The results

 have  limited value for comparing the health of individuals experienc-

 ing different levels of exposure to odor since responses  were not

 obtained from individuals who:

      1.   were not "very much" or "moderately" bothered by  the odor;

      2.   lived in relatively odor-free areas;

      3.   did not attribute their symptoms to the odor.

 In addition, because the questionnaire obviously dealt .with reactions

 to the odor, the responses might have been biased because  of the

 respondents' attitudes toward the odor or their desire to  force

 community action against the source.  A  better measure of  health

 ought to be obtained by a general health survey .of all members of


    a population sample chosen so as to represent varying exposures to

    odor.   The questionnaire or interview should make no reference to

    the odor although the respondent might be given an opportunity to

    attribute his symptoms to odor.


         The specific aim of the Anderson study was to determine whether

    community exposure to odor from pulp mills has any effect on health

    measurable by the type of personal interview used.  Implicit in the

    design of such a study is the measurement and "control" (during

    analysis) of other relevant factors which may affect the health

    responses to the questionnaire.   A similar study is being carried

    out in the same area to determine whether satisfactory results can

    be obtained from a postal questionnaire.


    Description of Study Area

         According to an estimate made in 1967, the incorporated City

    of Anderson had a population of 6,137 persons.  The sampling frame

    for the study was comprised of 1,2^6 households, including most of

    the incorporated area as well as some unincorporated districts.

    The town is located at the extreme north  of the Sacramento Valley.

    Unlike Eureka, which experiences the moderating effect of the Pacific

    Ocean, Anderson is exposed to seasonal extremes of temperature and

    rainfall.  Also in contrast to Eureka, which is exposed to a sea-

    sonal  shift in wind direction, Anderson is characterised year around

    by winds from the northeast in the morning, shifting to the southeast

    in the afternoon.
ISO                                18

     As mentioned above, several limitations of the area were recog-

nized with respect to population surveys.  These included the relative-

ly small size of the community, the small proportion of the population

which lives near the pulp mill, and the predominantly rural character

and poor housing of the area closest to the mill compared with the

areas with less exposure to the pulp mill odors.

Selection of Exposure Areas

     Before the environmental survey, three areas of Anderson had

"been defined as representing three levels of presumptive exposure

to odor on the "basis of topography, prevailing winds, and distance

from the mill.  (Figure l)  These were subsequently confirmed by the

environmental measurements.  (Figure II)  Although these measurements

were made in August and the health survey was carried out in November,

the seasonal factor was not felt to be of importance because the

prevailing wind pattern is essentially the same all year round.

     To take advantage of the diurnal wind pattern, Cottonwood measure-

ments were made in the morning and Anderson measurements were made

in the afternoon.

     The two areas farthest from the mill in the Anderson sector

were selected to represent moderate and slight exposure to the odor.

The areas closest  to the mill in both the Anderson and Cottonwood

sectors were combined and expanded slightly to represent the greatest

exposure to the odor.


        Preliminary scouting of the three "exposure" areas suggested

   that significant differences occurred in housing, type of neighbor-

   hood, including population density, and possibly in level of income.

   It was not feasible to characterize these subareas with any degree

   of precision, but some stratification seemed advisable.  Consequently,

   each "exposure" area was subdivided into three subareas, character-

   ized predominantly by scattered rural housing, central town-type

   housing, and suburban tract-type housing.  The rural housing was

   defined primarily on the basis of sparsity of settlement, but ap-

   peared to represent a relatively large proportion of housing of

   poor quality, frequently located on property which included farm

   out-buildings and livestock.  The suburban tract-type housing was

   on the outskirts of town and was largely "California ranch" style.

   The housing in the central areas was usually more modest and was

   located in the older, central part of town where a grid pattern of

   streets prevailed.

        The "exposure" areas will subsequently be referred to as

   Areas I, II and III, representing high, moderate, and low exposure

   to pulp mill odor, and the subareas will be designated as rural,

   central, and tract.

        Unfortunately, from the viewpoint of study design, Area I

   (high exposure), is sparsely populated and fell entirely within

   the rural subgroup.  The highest rate of housing vacancy appeared

   to occur here.  Areas II and III  are comprised of all three resi-

   dential subgroups.

        The sample sizes shown below ( Table 1 )  were chosen so as

   to permit some comparison within residential subgroups as well as

182                                20

between totals for each of the three areas.  The high sampling ratios

may have been a disadvantage because they increase the possibility

of bias resulting from pre-interview discussion of the survey among


     A sampling frame vas constructed for each subarea by listing

all houses which appeared to be occupied.  If houses included in

the sample were found to be unoccupied at the time of interview,

they were replaced by sampling randomly between the "interviewed"

houses on either side.

     Sampling was systematic with an independent random start in

each subarea.  Alternate households were designated to have a male

or female respondent interviewed.  If a household did not have a

respondent of the designated sex, one of the other adult members

was interviewed.

Questionnaire Design

     The questionnaire included information regarding date of birth,

sex, marital  status, occupation and place of employment, length of

residence in  the area, pre-existing medical conditions, smoking

habits, and exposure to specific industrial hazards which might

affect health.  The basic questions of the British Medical Research

Council's Questionnaire on Respiratory Symptoms (1966) were included

as a measure  of chronic effects.  This was followed by three sets

of questions  concerning symptoms which it was believed might be

associated with exposure to pulp mill odor.  The respondent was

first asked whether he has each of these symptoms frequently, occasion-

ally, or hardly ever.  For each symptom experienced frequently or

occasionally, he was then asked whether there was anything in parti-

cular which seems to bring it on, and whether he had it during the

                                21                                 183

   past tvo weeks.   He was  also asked whether he had been sick at any

   time during the  last two weeks,  and whether he would say that  his

   health was excellent,  good,  fair,  or poor.  It proved to be cumber-

   some and time-consuming  to ask this much detail about the symptoms

   on the list, but no estimate existed of the frequency of these symp-

   toms in the general population,  and we hoped to obtain adequate

   frequencies of at least  some of  the responses to permit statistical

   testing of the results.   It  was  also felt that the respondent  should

   be given an opportunity  to attribute symptoms to the pulp mill odor

   although the odor was  not mentioned at any time during the interview,

   The interview was introduced to  the respondent as a general health

   survey and no reference  was  made to pulp mill odor anywhere in the



        Interviewing was  carried out  by two part-time and four full-

   time interviewers, including two staff members who participated in

   other aspects of the survey.  Training began on the Wednesday  even-

   ing preceding the field  work and continued through part of the fol-

   lowing Saturday.  A preliminary  description of the study and run-

   through of the questionnaire and instructions was followed by  demon-

   stration interviews by staff members and practice interviews by

   staff members and trainees.

        Standard training tape  recordings of demonstration interviews

   of the MRC part  of the questionnaire were also used.  The most

   intensive part of the  training consisted of one-to-one interviewing

   practice with the training staff,  followed by playback of tape re-

   cordings and group discussion.   The final phase of training included

   "real" interviews on members of  a  community not being included in

184                              22

the survey proper.  Completed' questionnaires were edited by staff

members and discussed with the trainees.

     The initial plan required each interviewer to interview a

proportionate number of respondents in each area, and to spread the

interviews systematically by date in each area.  The former should

have equalized interviewer differences which were not eliminated

by the training sessions so that they would not appear as area dif-

ferences; the latter was to insure that increasing community aware-

ness of and discussion of the content of the study would occur at

the same rate in all areas.  In practice, it became increasingly

difficult to maintain this schedule beyond the first few days  of


     Each interviewer was usually seen by a staff member at least

once a day. for collection of completed interview schedules, assign-

ment of new interviews,  and discussion of problems.   Some preliminary'

editing of interview schedules  was done on a current basis  so  that

omissions or errors could be corrected while field work was still

in progress.


     Table 2 shows the prevalence of cough, phlegm,  and shortness

of breath defined from responses to the MRC questions,

demonstrating differences by area and sex.In all three exposure

areas, cough and phlegm were reported by a larger proportion of men

than women; in all three areas  shortness of breath was reported

more frequently by women.  A consistent trend can be seen in preva-

lence of all MRC symptoms among men, increasing from the area of

greatest exposure (Area l) to the area of least exposure (Area III),


   but some inconsistencies in prevalence appear among women, mainly

   "between Areas II and III.  Chi square tests for trend  showed the

   exposure area differences to "be significant only for phlegm grades

   1 and 2 combined and persistent cough and phlegm among men.

        The proportion of men and women in each area reporting each of

   a set of sixteen symptoms frequently or occassionally  and within

   the last two weeks are shown in Tables 3 and it.  Tests of statistical

   significance have not been completed.

        In order to explore the possibility that apparent area differ-

   ences merely reflect the influence of background variables, chi

   square tests were done to test the relationship between the responses

   to the health variables and the background variables.  Cough and

   phlegm showed significant relationships to several of  the "background

   health" variables and to smoking and type of housing area (whether

   rural, tract, or central town).  Since no dates of occurrence were

   obtained for the background health variables (which were phrased

   "have you ever had?"), it cannot be determined whether the relation-

   ships to these imply a possible effect before exposure to pulp mill

   odor or whether they imply a current relationship.  Among present

   cigarette smokers, the amount smoked did not show a significant

   relationship to the MRC questions  (Table 5).

        Smoking habits (never, ex-, and present smokers)  were found  •

   not to differ significantly by exposure area, so they  were not

   taken into account in the analysis by area.

        The effect of subarea was tested by combining data for Areas

   II and III (Area I had only rural housing) and stratifying by rural,

   tract, and central town housing.  The significant relationship to



cough and phlegm suggests that apparent exposure area differences

could be due to differences in housing.  In order to test this,

responses to the MRC questions were tabulated to show exposure area

comparisons for rural subareas only.  These are shown in Table


     Similar significant relationships were found between various

background variables and frequency of 16  listed symptoms as well

as incidence within the last two weeks.  Occupation and industry

showed significant relationships to nervousness, insomnia and pal-

pitations.  Marital status was significantly related to nervousness

and fatigue.  Comparisons for rural subareas are shown in Tables 7, 8

     Eight individuals attributed at least one symptom specifically

to odor in the air.  These included headache,  insomnia, sinus con-

gestion, eye irritation, burning or irritation of the nose, runny

nose, and cough.  An additional seventeen attributed at least one

symptom to air pollution without specifically  mentioning odor.

Discus sion

     Although the area comparisons are inconclusive, they suggest

that the interview as structured is a useful method for further

studies of the health effects of exposure to pulp mill odor.

Furthermore, relationships among the health variables themselves give

some evidence for the validity of the questions.  For example,

responses to the question, "Would you say that your health in general

is excellent, good, fair or poor?", showed a highly significant

relationship to many of the other health questions.

     The apparent effect of type of residential area indicates the

importance of taking into account socio-economic variables in making

  area  comparisons.   Larger  sample  sizes would increase the possibility

  of  determining  the  effects  of  the background variables.

       Further  analysis  is being planned in conjunction with two later

  phases  of  the study:   the  additional health survey by postal ques-

  tionnaire  and a study  of annoyance reactions to odor similar to that

  conducted  in  Eureka.   Multivariate analysis is planned to examine

  further the  effect  of  the  background variables.


     Acknowledgements are made to George  Sanders  for  data  from  the

environmental survey, and Madeline Thresh  and  Linda  Scott for  assis-

tance in carrying out the field vork and  data processing.


                               5^'     y V
                            ft vY. --•'">}
                            K7 i. '  '^T^>>

                                      FIGURE II
                                 DRJRNAL FREQUENCY •(%)
1 300-3 329
1 '100-1'! 29
3 430- M59
150 0-1 5 29
1600-1 629
] 930- 1959
2130-21 59
'i 1 .7
59. -1
1 2.5
. 16.V
                        Total mun'ocr ofmaloclor detections
       (Overall frequency--	x 100)
                        Total number of MJC;isuiT.mcnts

                          MAXIMUM MALODOR CONCENTRATION
                                   pp'o  ODOR AS  CH?.SH
                     . .             I *               *J

                                       ANDERSON '
1500 -1529
2000-20 29
21 30 ?J 59
33.7 .
67 /j
-13. S
8.2 .
34 .4
3,681. 4

                                 Table 1
                     Sample Sizes and Sampling Ratios

                   by Exposure Area and Housing Subarea
Popul. =   115

 Ratio =   1:1

Sample =   100
Popul. =    67

 Ratio =   1:1.3

Sample =   52
Popul. =   154

  Ratio =   1.3

Sample =   51
Popul. =  232

 Ratio =  1:4.6

Sample =  45
Popul. »   298

 Ratio =   1:5.4

Sample =   50
Popul. =  0
Popul.  =  97

 Ratio =  1:1.9

Sample =  50
Popul. =  383

 Ratio =  1:7.5

Sample =  50

                                          Table 2

                           Prevalence of Respiratory Symptoms

                                     Anderson,  1970
dumber of Observations
Cough Grade 1 or 2
Cough Grade 2
Phlegm Grade 1 or 2
Phlegm Grade 2
Persistent Cough & Phlegm
Shortness of Breath Grade
2 or Greater
Shortness of Breath Grade
3 or Greater
     Cough -
         Grade 1
         Grade 2

     Phlegm -
         Grade 1
         Grade 2
           Cough first thing in the morning or during the day or at night
           on most days for as much as three months each year.

           Cough first thing in the morning and during the day or at night
           on most days for as much as three months each year.
           Phlegm first thing in the morning or during the day or at night
           on most days for as much as three months each year.

           Phlegm first thing in the morning and during the day or at night
           on most days for as much as three months each year.
     Persistent Cough and  Phlegm;   At  least  Grade  1  of  both  cough  and  phlegm.

     Shortness  of Breath -

         Grade  2;   Shortness  of breath when hurrying on  level  ground  or walking
           up a slight hill.

Grade 3;   Shortness of breath walking with other people  of own age on

                    level  ground.

          Table 3
Prevalence of Symptoms Reported
  Frequently or Occasionally
        Anderson, 1970
Number of Observations
Sinus Congestion
Eye Irritation
Shortness of Breath
Nose Irritation
Runny Nose
Chest Pain
48 67 62
41.7 40.3 22.6
41.7 29.9 22.6
10.4 19.4 12.9
37.5 22.4 24.2
6.2 10.4 6.5
8.3 10.4 11.3
12.5 1.5 3.2
4.2 1.5 0.0
16.7 11.9 16.1
39.6 26.9 33.9
16.7 16.4 12.9
18.7 13.4 12.9
8.3 7.5 1.6
33.3 20.9 21.0
14.6 7.5 8.1
39.6 20.9 21.0
52 80 90
67.3 63.8 58.9
51.9 50.0 35.6
25.0 28.8 33.3
50.0 46.2 48.9
15.4 10.0 22.2
21.2 18.8 25.6
11.5 13.8 10.0
0.0 3.8 6.7
28.8 18.8 17.8
38.5 35.0 34.4
17.3 20.0 17.8
26.9 22.5 22.2
7.7 11.2 4.4
32.7 32.5 22.2
9.6 12.5 10.0
21.2 21.2 20.0

                                   Table  4
                  Prevalence  of Symptoms Reported Frequently  or
                    Occasionally and During the Last Two Weeks
                                Anderson, 1970
Number of Observations
Eye Irritation
Shortness of Breath
Nose Irritation
Runny Nose
Chest Pain




                            Table 5

            Chi Square Test for Relationship  Between
             MRC Questions and Background Variables

                          Anderson, 1970

Chest Illness
Past Three Years
Bronchiti s
Pleur i sy
Hay Fever
Nev ,Ex ,Pres
Smoking Cigarettes
Quest ions
Cough Grade
x2 d.^.
9U** 1
13** 2
11*** 2
56* 2
96 2
87** 1*
61+ 1*
oit* i*

d.f .
Cough &

v d. • f •
7. 51***
 *  Significant at  the  5$  level.
**  Significant at  the  1%  level.

                                 Table 6
                     Prevalence of Respiratory' Symptoms
                           Rural Subareas Only
                              Anderson, 1970

Number of Observations
Cough Grade 1 or 2
Cough Grade 2
Phlegm Grade 1 or 2
Phlegm Grade 2
Persistent Cough & Phlegm
Shortness of Breath
Grade 2 or Greater
Shortness of Breath
Grade 3 or Greater
15. 1*















11*. 3

ho. 2






17.8 •





  Grade 1
  Grade 2
Cough first thing in the morning or during the day or at night
on most days for as much as three months each year.

Cough first thing in the morning and during the day  or at night
on most days for as much as three months each year.
  Grade 1:   Phlegm first thing in the morning or during the day or at night
            on most days for as much as three months each year.

  Grade 2:   Phlegm first thing in the morning and during the day or at night
            on most days for as much as three months each year.

Persistent  Cough and Phlegm:  At least Grade 1 of both cough and phlegm.

Shortness of Breath
  Grade 2:   Shortness of breath when hurrying on level ground or valking up
            a slight hill.

  Grade 3:   Shortness of breath walking with other people of own age on level

            Table 7
Prevalence  of Symptoms Reported
   Frequently or Occasionally
      Rural Subareas Only
         Anderson, 1970
Number of Observations
Sinus Congestion
Eye Irritation
Shortness of Breath
Nose Irritation
Runny Nose
Chest Pain




            Table 8
Prevalence of Symptoms Reported
Frequently and Occasionally and
   During the Last Two Weeks
      Rural Subareas Only
        Anderson, 1970

Number of Observations
Sinus Congestion
Eye Irritation
Shortness of Breath
Nose Irritation
Runny Nose
Chest Pain



              Summary of technical problems
                       Lars Friberg
As a basis for this presentation,  I have read several of
the working papers in advance. This has given me the oppor-
tunity to evaluate the evidence that has been presented on
dose-response relationships and to summarize some of the
problems existing in regard to the dose as well as in regard
to the response.

Although there is undoubtedly a considerable amount of data
available on odors and different types of response to odors,
I will start out with a pessimistic statement. If, when re-
ferring to dose-response relationships, we mean quantitative
associations between either odor intensities or qualities,
or concentrations of odorous substances on the one hand, and
well established adverse effects in human beings on the other,
very few data are available. Response to odors, however, must
not necessarily mean adverse long-term effects in an exposed
population, but may mean, for example, the possibilities for
human beings to detect odors or to evaluate odor intensities
or qualities. If,  under such circumstances, a dose would be
defined as a concentration of a single substance or a combi-
nation of substances in relation to exposure time, more data
will be available.

In this paper, I will start with a discussion of the response,
then discuss the dose and finally talk about the dose-response
relationships. Focus will be on issues, which may have bearings
on practical problems, and on possible ways for responsible
authorities to approach the various odor problems.


         The  response

         Wa will  have  to  distinguish  between  a  response  to  odors
         as such  and  to odorous  substances. The title  of this  con-
         ference  is dose-response  relationships affecting human  re-
         actions  to odorous  compounds.  Of  course,  it  is  not my inten-
         tion to  change the  terms  of  reference  for this  meeting,  but
         my feeling is, that  we  should  primarily discuss odors and  not
         odorous  substances.  If  we include effects of  odorous  sub-
         stances,  when we embark on dose-response  relationships,  we
         would have to discuss things like effects of  high  concentra-
         tions of hydrogen sulfide on the  central  nervous system  as
         well as  on the respiratory system. We  might wish to include
         effects,  such as chronic  bronchitis  or eye irritation of sul-
         fur  dioxide  ozone,  since  also  these  substances  are recognized
         as smelling  at certain  concentrations. Hundreds of other sub-
         stances  should be included;  each  substance having  a more or
         less specific effect on human  beings.

         The  most common  adverse effect of odorous substances  are
         annoyance reactions  caused by  the odor itself.  If  certain
         substances,  H_S  or  mercaptans  for example,  in concentrations
         found in the  ambient air,  give rise  to adverse  effects  caused
         not  by the odors per se but  by a  direct toxic action,  such
         effects  and  dose-response relationships should  be  considered
         with regard  to the  particular  substances  in the same  way as
         has  been done in air quality criteria  documents for e.g.  sulfur
         oxides and oxidants.

         Disregarding  the more unspecific  annoyane reactions,  very
         few  studies  have been carried  out with the  aim  of  evaluating
         whether  or not long-term  exposure to odors  can  give rise to
         chronic  disease. No  doubt, acute  effects  in the form  of  nausea
         or even  vomiting are often reported  at interview surveys.


Furthermore, in David Kendall's report we have evidence of
reactions from the autonomic nervous system. The meaning of
such reactions for possible long-term effects are not well
understood. John Goldsmith's report, which I have, unfor-
tunately, only seen in an abstpact form, indicates that per-
sistent cough and phlegm were more frequent in areas close
to a pulp and paper mill than in control areas. If this is
an effect of odors, and not a toxic effect of an odorous sub-
stance or possible substances occurring together with the
odorous substances, this would be the first evidence of a
well-defined medical effect of exposure to odor. One would
like to find a plausible explanation, though, how such effects
could arise. Reactions from the autonomous nervous system
might be important in this connection.

The more unspecific annoyance reactions have been documented
in various countries in different exposure situations, e.g.,
around pulp mills, around areas spread with manure and from
motor exhausts in cities. Such reactions must be evaluated
with great caution. City dwellers complain heavily of the
air pollution situation. There are good reasons to believe
that exposure to odors is a major cause of such complaints,
but exposure to other agents is certainly also important.
I may in this connection refer to dose-response relationships
established in the US air quality criteria document on parti-
culate matters.

Annoyance reactions are not easily evaluated, besides we
have no clear indication of possible medical implications
in the long run of such reactions. From the methodological
point of view, there is a great need for evaluating existing
methods, e.g. survey technics,  as well as establishing new
methods. Petitions and spontaneous complaints are notorously

     unreliable  as  has  been  examplified  in  a  presentation by
     Erland  Jonsson earlier  today.  Such  questions  will be dealt
     with  at a meeting  to  be held  at  the Karolinska  Institute
     in  Stockholm this  September and  sponsored  jointly by the  US
     and Swedish Environmental  Protection Agencies.

     Concerning  the long-term medical implications we  are in the
     same  position  as when evaluating effects of traffic  and air-
     craft noise.  We are well aware of the  fact that people  are
     annoyed, often to  a considerable degree, but  we do not  know
     enough  of possibly more serious  endpoints.

     Whatever the  medical  endpoints may  be, the first  question
     to  be answered by  agencies responsible for the  control  of
     air pollution  is the  following:  are health effects in the
     form of only annoyance  reactions, with possibilities of other
     and more severe reactions  in  the long  run, sufficient evidence
     for the taking of  administrative steps in  order to reduce the
     exposure to odors? In my opinion this  question  should be  an-
     swered  with yes.

     The dose

     If  the  answer to this question is yea, the next step to con-
     sider is whether or not it is  possible to  describe the  dose.
     It  would certainly be an overstatement if  we  said that  good
     methods, suitable  for field work, are  available to describe
     the dose under all circumstances occurring in everyday  life.
     On  the  other hand, I  think it  is fair  to state  that  very  often
     we  have the technical know-how and  also  the technical facili-
     ties  to describe the  dose  either based on  sensory evaluations
     or  chemical analyses. The  practical implications,  however,  are
     often formidable due  to the many dimensions involved. In  several

working papers, it has been well documented that the dose
often should not be expressed  only as whether or not and
how often it smells. The hedonic aspects of odors as well
as intensity above threshold are of importance.

Whether or not it smells can be decided by means of odor
threshold studies. For single substances we often have good
information about the critical concentration for detection.
The information is much more scarce for mixtures of different
substances. However, I can see no major difficulties, which
could not be overcome when working out quantitative relations
between odor thresholds and mixtures, although such studies
may be time-consuming because of, among other things, the
nature of the Stevens exponents, as discussed by Amos Turk.

It is possible to carry out odor threshold studies at the
source as well as in the ambient air using sensory methods.
Generally, chemical analyses at the source should present
no serious problems. The situation is another one at chemical
analyses of the ambient air, as meaningful analyses must have
a high enough sensitivity to measure threshold concentrations
over very short periods of time, probably seconds, because of
the short response time of the human nose.

The frequency of odor is another important criterion. It
goes without saying that any adverse response within an ex-
posed population must be correlated to some degree with the
number of times (and the duration of each time) odor thres-
holds are exceeded. It raises considerable problems to eva-
luate this. One can use observers, located at several places
in the area to be observed, or one can carry out sensory ana-
lyses of the ambient air at different times and at different
places. Practical problems will often make such type of ana-


lysis impossible, If we get instruments which can register
odorous substances in low enough concentrations over short
periods of time and if we know how the odor thresholds change
with different composition of the air to be analyzed, we
could, of course, use chemical analyses. Such instruments are
still lacking even if progress seems to have been made con-
cerning analyses of certain sulfurous compounds.

One way that has been 'used in Sweden merits some consideration.
I am referring to the method of estimating how often odor thres-
holds can be expected to be exceeded based on intenstity or
concentrations of certain substances in emitted stack gases.
Ulf Hogstrom and Thomas Lindvall have reported on the theore-
tical background and the practical results achieved up until
now from studies around Swedish pulp mills. By and large, this
method seems promising and the agreement, found between predicted
and observed odor frequencies, has been reasonably good. Ulf
Hogstrom's statement, that the method can be modified to give
more sophisticated answers concerning e.g. intensities above
thresholds, should make the method still more valuable. There
are several methodological problems remaining though, for ex-
ample, in connection with chemical reactions of odorous sub-
stances after the emission with the possible change in odor
intensities when the polluted air reaches ground level.

Despite limitations that still exist in the method referred
to, it has been used by agencies responsible for air pollution
control in Sweden. Health authorities, in two communities where
pulp mills are located, have ruled as a condition for the ope-
ration of the mills that gases emitted from the main stack
must not have an odor strength corresponding to a dilution
factor above 10,000. This would mean a predicted odor frequency


of only a few hours per month at ground level at any place
around the industries. These regulations have been set without
having the exact dose-response relationships between odor fre-
quency and annoyance reactions.

The Swedish Environmental Protection Board, which is mainly
in favor of this type of approach, has not taken any definite
stand on how to act in the long run. For the time being, they
require that the concentration of hydrogen sulfide in the
                       *                       3
emitted stack gases beJiret--btghQr than 10 mg/m- . This would
correspond to a dilution factor of about 10,000, provided
hydrogen sulfide is the only cause of the odors, which fact,
however is very dubious. The reason why the Environmental
Protection Board has not yet officially adopted a standard,
based on odor threshold measurements of the stack gases, is
a practical one, namely the difficulties involved in checking
the operations on a continuous basis.

In collaboration with the Environmental Protection Board,
The Karolinska Institute will carry out a study this fall,
covering the possibilities of predicting odor intensities
based on sensory as well as chemical analyses of the emitted
gases around a pulp mill. The outcome of that study will
probably be of considerable importance for future emission
standards for odors from pulp mills.

Dose-response relationships

Here I am only talking about the situation where the dose
is defined as a concentration of substance, or substances,
or sensory evaluated, and where the response is an adverse
medical effect within the exposed population.

   -/—i                 "\                                     ,
 be lower* than in ma/m  ±Y\ 99 % of the time for new units ano
 lower th^n 10 mq/m  in 90 °4 of the time fore old mnits.

      As was mentioned  in the  introduction, such relationships
      have not been shown for  any odor. We have a few  points  on
      the dose-response curve  for certain odors. I am  thinking
      particularly of the situation around the pulp mills. At
      interview surveys in Sweden, about 25-35 percent of  the
      population, exposed within a radius of about 10-15 miles
      from the factory, considered themselves bothered by  the
      smell. Generally  speaking, there was a gradient  with regard
      to distance from  the factory.  Similar results have  been re-
      ported from the Eureka study in California. Odor threshold
      studies were not  carried  out at the same time but from  studies
      at another time,  there is reason to believe that the odor
      strength of the stack gases would correspond to  a dilution
      factor of about 100,000.

      Up until now the  data do  not show how any adverse reaction
      within the population would change if the dose was doubled
      or halved. Furthermore,  available data refer only to one
      particular exposure situation. We know that emitted  gases
      can differ considerably  in composition from one  mill to an-
      other and from time to time. However, we know nothing about
      the importance of the hedonic aspects. This also means  that
      an acceptable dose of odors from pulp mills may  differ  from
      what can be acceptable for other types of odors. If  we  limit
      ourselves for the moment  to pulp mill odors - or to  any well-
      defined point source - it would be possible to carry out epi-
      demiological studies, which would give more or less  complete
      dose-response curves. It  would then be possible  to predict
      which changes in  response could be expected, if  the  dose is
      decreased with a  certain  factor. What is needed, is  money
      and manpower in order to  carry out studies around different
      point sources. The dose  could be evaluated either by means
      of emission analyses in  connection with meteorological  dis-

persion calculations or by means of direct evaluations of
the dose in the ambient air. For the time being, the first
mentioned approach seems to be the most promising one.

Apart from being time-consuming, we must appreciate that
such studies will not necessarily give data which are valid
for long periods. It has been shown that annoyance reactions
are subject to the influence of several factors not related
to the dose. The attitude towards the source of emission seems
to be of particular importance. In Sweden, Erland Jonsson
and Stefan Sorensen have shown, in a field experiment, that
annoyance reactions from aircraft noise diminished with about
50 percent if the exposed population were given a positive
attitude towards the source (in this case air traffic) as com-
pared to those given a neutral attitude.

No doubt the attitude of the people towards the need for
protection against adverse effects from the environment will
change considerably within the coming 10-20 years. In his
working paper, Trygg Engen states that there should be no
difficulty in teaching a person to experience the smell of
skunk as something pleasant. He may be right, but generally
speaking, I am convinced that it will go the other way, meaning
that what is acceptable today, might well be considered un-
acceptable within not too distant a future. Available dose-
response relations will then not be valid. This is one reason
why every effort possible should be focused particularly on
the dose, as a concentration today will mean the same con-
centration several years from now. Besides, the possibilities
of the human nose to detect - and hopefully also to make hedo-
nic evaluations - in controlled laboratory experiments will
not change considerably for long. If information on the dose
is available, there will thus always be the possibility of
establishing certain standards for odors based on known in-

      tensities  and  qualities.

      For odors  from motor vehicles  the  situation  is  somewhat
      different.  Dose-response  relationships  where  the  dose  is
      the odor from  one  single  source,  of  course,  cannot  be
      established. Theoretically,  sensory  analyses  can  be carried
      out on  odors in the  ambient  air  in the  cities and related  to
      medical effects. As  has been mentioned  already, the urban
      community  contains so many  different substances that the
      difficulties to relate any  effects with exposure  to odors
      might be overwhelming. The  American  approach  to study  the
      source  seems to be the best  way.  The question of  whether or
      not it  will be possible to  find  any  index  substance1 for odor
      is  still left  open.   Under  all circumstances, it  would be
      possible to use odor measurements  when  trying to  reduce odor
      and change  the odor  quality  by the construction of  new types
      of  motor.  There are  all reasons  to pay  much  attention  to odors
      from motor vehicles.  People  are  complaining  about odors and
      not only diesel engines are  powerful producers  of odor, but
      regular gasoline powered  engines  have exhausts  with odor
      thresholds  corresponding  to  dilution factors  of 5,000-10,000
      as  seen in  Swedish studies.

      One could  continue discussing  dose-response  relationships
      for several other  types of  odors.  I  will just briefly  mention
      the odor from  manure. This  has become a great problem  in
      Sweden  because of  the need  for having e.g. hog  raising farms
      fairly  close to populated areas.  Since  this  is  not  a single
      source  emitter of  odors,  the only  approach here seems  to be
      to  evaluate the effect of different  types  of  treatment by
      means of odor  intensity and  odor  quality studies.  Such studies
      are under  progress in Sweden.


As should have been obvious from what has been said, there
is not much hope for reliable dose-response associations
for odors in general within the near future. For certain
types of odors such associations could be achieved provided
money and manpower are made available. This refers to single
source odor emitters,  such as pulp mills. Other approaches
seem more appropriate  for other types of odors, e.g. from
motor vehicles and manure spreading. The first thing would
be to get information  on the order of the problem, for instance,
by means of interview  surveys. Next would be to study the possi-
bilities of reducing odors at the source. As soon as a technic
is worked out that fulfills the odor reducing requirements and
which is within reasonable technical and economical limits,
such a technic should  be required.

         Summary of the Problems in Odorous Air Pollution in Japan

                                 by Takeo Suzuki

     In Japan the investigation of odor is still in its initial stage, major

effort is being paid to the development of method of measurement of odorous

gases and only a very fev studies have been made of the dose-response

relationship in the exposure to odor.  On the other hand since many petitions

of odor have been submitted to the government, the particular effort has to be

given for the control of odor by the government.  The folloving is a review of

problems of odor in Japan.

1.  Features of problem of odor in Japan

     Japan occupies small land and furthermore as 80$ of the land is mountain

remaining only 20$ is available for human activity.  Further Japan has population

of more than one hundred million, about 40$ of the population live in large cities

and industry has been rapidly progressed in the large cities and their outskirts.

Despite these situations Japanese cities have no veil arranged zoning plan, so

that industrial plants and residences are located close together.  In addition

insufficient investment has been made for public utilities such as sevage system.

These situations lead to community - as veil as local - air pollution.  In

Japan odor is one of the major air pollution.  The autoexhaust introduces

conspicuous urban air pollution but people in the city has no special concern

vith mal-odor arising from the autoexhaust.  Rather the discharge of odorous

gases from industrial production at the plants and other facilities causes the

nuisance for their neighbors.

     The recent statistical results of the recent objections and petitions as

regards environmental pollution all over Japan are shown in Table 1.  Total

number of objections and petitions arising from air pollution and odor in 1969

                                       41                                       £13

are 7,558 and 7,983 respectively.  In recent times the petition most frequently

occurred from noise and vibration and it is followed by those due to order and

general air pollution.  Needless to say the frequency of petition does not

alvays represent the level of injury and effect, and absolute number of the

episodes, but it is useful for obtaining the relative pattern of the above

levels and episodes.  For example, the number of petitions of air pollution

has not appreciably been changed since 1966 but their number due to odor

increased by tvo or three times during the period of 1961 to 1969.  There has

been a tendency for older Japanese to endure malodor but nov people actively

express their objection against  nuisance.  In addition, the generation of nev

type of odorous compounds as described belov is enhanced.  These facts tell us

why the frequency of petition has been rapidly increased.

2.  Major malodorous compounds and their sources in Japan

     The odors to which people make objections are generally low concentration

and multi-components of gases.  Although naturally the odor problem can not be

solved by the control of each one of the component gases, the gases are listed

in Table 2. which are noted as the major components of malodor as the result

of the recent development of the odor measurement techniques and the surve in

the area where the odor creates serious social problems.  The major sources of

odor in Japan in recent times are petroleum chemical plants, oil refineries,

kraft pulp mills, fish and animal meal plants, live-stock yard, night soil plants

and incinerators.

     Most of the petroleum chemical plants, oil refineries and kraft pulp mills

are big industrial plants.  Particularly petroleum chemical  and oil  refining

industries have rapidly developed in recent ten years and new industrial cities

are being constituted around the core petroleum chemical and oil refining industries,

Odor is one of the most important social problems arising from air pollution

in these new industrial cities.  The odor in combination with the community

air pollution enhances the number of citizens who complain various symptoms.

Some people complain their chronic bronchitis, emphysema, asthma and cardio-

vascular diseases in conjunction vith the odor.  Therefore if no measure is

taken for the problem of odor the citizens have the feeling that the community

                               	&***%  ft/ &y1ft~Ja v~io£~-
       One of the most notable features  of problem  of  odor  in  Japan  is  the

  malodorous gases released from the following treatments and  processes in

  small- and middle-sized plants and other utilities.   Processes  of  the intestines

  and bone of fishes, feather, animal's  fat and bone,  swine, poultry, the treatment

  of faeces of hens, treatment of lees of soy-been  sauce, and  vegetable protein,

  drying of organic compounds, incineration of animal  body, manufacturing of fish

  oil, varnish and fats, industry related with the  production  of  plastics, drying

  of various organic solvents, rubber manufacturing plants  and the production of

  inorganic and organic chemicals.  These processes are usually conducted in

  small- and middle-sized industries and since these plants are located near
  residential district they create:  serious nuisances  in many  places.   These

  small- and middle-sized plants are concentrated in the outskirts of large cities

  and people living in these areas have  rather conservative point of view and their

  relationship vith their neighbors are  one of the  characteristics of Japanese.

  Such sense is different from the thinking of people  in nev industrial  cities

  having petroleum industry.  Thus the objections and  petitions from people of old

  industrial towns are less frequent despite the absolute concentration  of the

  odorous gases is in higher level.  However, as their thinking is being changed

  the incidence of odor in these areas gradually results in a  social problem.

  3.  Reaction of people against odor

       In Japan very few studies have been made on  the dose-response relationship

  in exposure to odor.  In recent times  the dose-response relationship  is conceived

  in the development of application of olfactory methods which are employed as one

  of the method of measurement of odorous substances.  The  olfactory methods

  include odor-free room, odor free room fitted with windows for  sniffing,

  obfactometer (Nader's method) and so on.
216                                     44

     In Japan the experimental studies of threshold values for olfactory

perception of primary odorous substances have been performed and one of them

will be shown in this paper as one example.

     Dr. Matsushita of National Institute of Industrial Health made experimental

determination of threshold valurs of nine odorous substances for 15 students of

21 to 24 years old by means of 500 ml of vial.  Fig. 1. shows one of his

experimental results.  N-heptyl alcohol from 6.3 x 10~  to 8 x 10   ml/ml

in concentration is tested in his experiment.  All fourteen subjects detect

odorous above 10   ml/ml solutions of n-heptyl alcohol and their numbers of

subjects who can detect odorous decrease rapidly with the reduction of the

alcohol concentration.  Finally three or four subjects only detects odorous

less than 10   ml/ml.  The perception threshold value is determined to be

2 x 10~  ml/ml from the curve in Fig. 1.

     In this way Matsushita established the population threshold values for

olfactoly perception of Japanese for primary odorous substances as shown in

Table 3.

     These studies and the like are being progressed by Matsushita and several

other investigators and based on these investigations the effort has been

started for the determination of dose-response relationship among people living

in the district where there is an ordor problem.  However no substantial results

have be gained so far.

4.  Control of odor in Japan

     Although no progress has been made of the investigations of odor a.nd its

related fields, as described above the odor easily results social problem, so

that local agencies made effort to evaluate the intensity of odor by means of

olfactory method and made application of the intensity scale to the technical

guidance of odor control.  In the application of olfactory method usually only

a fev professional panel members can be employed and thus layman has to be

incorporated as a panel member.  These procedures are satisfactory for

administrative measure of local agencies to some extent.  But as these

measures are not substantial, Japanese Government now presents the "lav for

abatement of odor" to the Diet.

     The lav for abatement of odor is directed tovard the control of emission

of odorous substances resulting from the operations of plants and utilities.

The lav does not purpose to the control of autoexhaust emission.  The lav

designates 13 gases such as ammonia and methyl mereaptan as those under control

and designates the area vhere it is enforced.   It is further planned to establish

environmental and emission standards of the odorous gases.  It seems to me that

the lav is directed tovard the educational effect rather than its strict


5.  Conclusion

     The general problems of odor in Japan are reviewed in the preceding sections.

In Japan the studies of odor vhich is currently at vork are exclusively those

of its measurement techniques and the health effect of odorous substances is a

subject vhich has to be studied in the future.

Table 1. Trends of objections and petitions caused by  eavircnnsnta
         po^.^ ution anci nuisance in iiapan.

Air pollution
Vater pollution
Lr.nd subsidence
Ccl or
Industrial waste

Total ' • .




of petition

s rec



eived V,
, 314 \
,796 J

' Qcv




r :.- O

, s*j^

                                  According to the  statistics  of
                                  Ministry of Eor.e  Affairs
 Table 2. Gases vhich are designated as important  odorous  gases
          in Jar,an.
    !  Sulfur-bearing compounds
      Nitrogen-bearing compounds
      Gas os
Methyl aereaptan
Ethyl rsercaptan
Dimethyl sulfide
Liethyl sulfide
Hydrogen sulfide
Dimethyl amine
Ethyl amine
Triaethyl anine
Butyric acid

                      C   if
         1.  Relation between concentration of  n-hcptyl alcohol  in  solution and  number
             of correct response by serial  test (fourteen subjects):
Tnblc  3«  Population threshold values for olfactory perception for  pri.-nary odours.
                                    jQdour quality
                                                   Threshold value* jConc
  n odour
1 Cnt-.-.plior
! Cyclohcxanol
15-Mydroxy pcntadecanoic acid lactone
n-Hcptyl alcohol
Lina'.yl acetate
Ethyl mcrcaptan
Methyl n-nonyl ketonc
1.2 xlO-«
2.4 xlO-'**
1.2 xlO"
5.4 xlO-«
S.3 xlO-«**
4.-1 xlO-'**
 * Unit used is as g/inl for 15-r.ydroxy pcr.tudccanoic acid lactonc and p
** Mean of two observed values.


                        D. S. Earth, Ph.D.
                  Environmental Protection Agency
                   Air Pollution Control Office
                 Bureau of Air Pollution Sciences
To be presented at the Conference on The Dose-Response Relationships
Affected Human Reactions to Odorous Compounds.   Sponsored by the
Environmental Protection Agency, Air Pollution Control Office,
April 26-27, 1971, Cambridge, Massachusetts.


   Administrative Mechanisms Available for Control of Odorous
        Compounds under the U.S. Clean Air Act as Amended

                           D. S. Barth


     The Clean Air Act as amended, P.L. 91-604, states as one  of its

purposes, "to protect and enhance the quality of the Nation's  air

resources so as to promote the public health and welfare and the productive

capacity of its population."  It goes on to define specific authorities

granted to the Administrator, Environmental Protection Agency, which  he

may utilize to achieve the purposes of the Act.  In brief some of these

authorities are:

     1.  Designation of Air Quality Control Regions, issuance  of Criteria

     and Control Techniques Documents and promulgation of National Ambient

     Air Quality Standards.

     2.  Promulgation of National Standards of Performance for New

     Stationary Sources

     3.  Promulgation of National Emission Standards for Stationary Sources

     of Hazardous Air Pollutants

     4.  Promulgation of National Emission Standards for Motor Vehicles

     5.  Regulation of fuels and fuel additives

     6.  Issuance of National Aircraft Emission Standards

     7.  Setting of Aviation Fuel Standards*

     In the following a brief discussion of each authority will be given

with emphasis placed on the possible applicability to the control of

odorous compounds.
*This authority is granted to the Administrator,  Federal  Aviation
Administration, but it must be based on recommendations of the
Administrator, EPA.

General Discussion of Information Needs to Develop Adequate Controls
for Any Air Pollutant
     In genral adequate information must be available in the following
subject areas in order to establish a basis for control of any air
     1.  Effects
     2.  Sources
     3.  Measurement techniques
     4.  Control technology
Effects information should include adverse effects on human health or
welfare as a function of pollutant concentrations and averaging times
and as modified by meteorological factors, presence of other pollutants
and sensitivity of the exposed receptors.   All significant sources,
both stationary and mobile must be identified.  Adequate measurement
procedures of known precision and accuracy must be available for both
source emissions and ambient air quality.   And finally control technology
must be available and possible control efficiences must be known for
control of most, if not all, significant sources.
     When the information described briefly above is available then it
becomes possible to develop a strategy to bring a given air pollutant
under adequate control.  The selection of a specific strategy for control
should take into consideration the following factors:
     1.  Available authorities for control which are applicable.
     2.  Controlling the objectionable pollutant to acceptable levels
     at the earliest possible time.

     3.  Total administrative and operational costs associated with

     different approaches.

Odorous Compounds as a Special Case

     The special case of odorous compounds as air pollutants must now

be considered.  Of the authorities for control of air pollutants cited

previously some are not applicable for odorous substances.   In particular

the effects of odorous substances are not deemed to meet the definition

required to qualify odorous substances as "hazardous air pollutants."

Furthermore there is no evidence now to indicate that regulation of

fuels and fuel additives or setting of aviation fuel standards would be

necessary to abate objectionable odors.  No further discussion will

be given of the cited control options which are not judged  to be

applicable to the control of odorous substances.  The remaining control

options, all of which may be applicable, will now be discussed in more


Criteria and Control Techniques Documents

     The Administrator, EPA, is directed by P.L. 91-604  to  publish Criteria

and Control Techniques Documents for those air pollutants

     "(A) which in his judgment have an adervse effect on public health

     and welfare;" and

     "(B) the presence of which in the ambient air results  from numerous

     or diverse mobile or stationary sources."

Air quality criteria for an air pollutant "shall accurately reflect the

latest scientific knowledge useful in indicating the kind and extent

of all identifiable effects on public health or welfare  which may be

expected from the presence  of such pollutant in the ambient air,  in



varying quantities."  Information on control techniques "shall include

data relating to the technology and costs of emission control.  Such

information shall include such data as are available on available

technology and alternative methods of prevention and control of air

pollution."  The Act states that "effects on welfare include, but are

not limited to, effects on soils, water, crops, vegetation, man-made

materials, animals, wildlife, weather, visibility, and climate, damage

to and deterioration of property, and hazards to transportation, as well

as effects on economic values and on personal comfort and well-being."

     For those materials for which Criteria and Control Techniques

Documents are issued the Administrator is directed to prescribe National

Primary and Secondary Ambient Air Quality Standards which are defined

as follows:

     (1)  "National primary ambient air quality standards shall be ambient

     air quality standards the attainment and maintenance of which in

     the judgment of the Administrator, based on such criteria and allowing

     an adequate margin of safety, are requisite to protect the public


     (2)  "National secondary ambient air quality standards shall specify

     a level of air quality the attainment and maintenance of which in

     the judgment of the Administrator, based on such criteria, is

     requisite to protect the public welfare from any known or anticipated

     adverse effects associated with the presence of such air pollutant

     in the ambient air."

     To date Criteria and Control. Techniques Documents have been issued

and National Primary and Secondary Ambient Air Quality Standards have

been proposed for sulfur oxides, particulate matter, carbon monoxide,

photochemical oxidants, hydrocarbons and nitrogen oxides.  Once these

standards have been promulgated, which must be no later than April 30,

1971, the States have nine months in which to develop and submit an

implementation plan designed to achieve the air quality standards within

a rather tight time schedule after approval of their plan--3 years after

approval for primary standards and a "reasonable time" for secondary standards.

     From the discussion above it is clear that this control option

would be feasible and applicable for odorous substances.  In other words

we could prepare and issue Criteria and Control Techniques Documents for

Odorous Substances and then subsequently promulgate National Ambient

Air Quality Standards which would lead to implementation plans for

control.  The advantages and disadvantages of this approach will be

addressed later.

Standards of Performance for New Stationary Sources

     "The term  'standard of performance' means a standard for emissions

of air pollutants which reflects the degree of emission limitation

achievable through the application of the best system of emission

reduction which (taking into account the cost of achieving such reduction)

the Administrator determines has been adequately demonstrated."  On a

tight time schedule set forth in the Act the Administrator is required to

list and then publish standards for categories of new sources which he

determines "may contribute significantly to air pollution which causes

or contributes to the endangerment of public health or welfare."


        In addition the Act stipulates that the Administrator shall prescribe

        regulations which shall establish a procedure under which each State

        shall submit to the Administrator a plan which establishes emission

        standards for any existing  source in the same category of sources  for

        any air pollutant which is a non-criteria pollutant and which has not

        been listed as a hazardous air pollutant, and to which a standard of

        performance would apply if the existing source were a new source.

             Placing categories of industrial sources of odorous substances

        on the list and then publishing standards of performance for such new

        sources would lead to control of both new and existing sources through
        the application of the best system of emission reduction which has  been

        adequately demonstrated.   Some advantages and disadvantages of this

        option will be presented later.

        National Emission Standards for Motor Vehicles and  Aircraft

             For these authorities it is envisioned  that control  needs will  be

        based on desired air quality for significant pollutants  to  assure

        protection of health and welfare.   To the extent that  odorous  substances

        are emitted from motor vehicles  and aircraft this is  the  only  applicable

        authority for control.   This authority can and would be used in conjunction

        with whatever controls  are established for stationary  sources  of odorous


        Discussion of Alternate Control  Options

             For the control of stationary sources of odorous  air pollutants

        there are two possible approaches  as indicated previously.   Criteria  and

        Control Techniques Documents could be issued or Standards of Performance


for new sources and in turn emission standards for existing sources

could be promulgated.  There are advantages and disadvantages inherent

to either approach.  A major problem to either approach is the lack

of availability of an objective measurement technique for the wide

variety of odors which are objectionable to many humans.   Proceeding

via the Criteria Document route would lead to obtaining some control

over all sources simultaneously.  However, since the National Ambient

Air Quality Standards would probably be secondary rather than primary

ones, achievement of the standards could be delayed for a "reasonable

time," whatever that time might be.  This approach would also require

the development of a nationwide air quality surveillance system to

measure odorous air pollutants.  In addition the built-in delays in

proceeding from development of criteria documents to achievement of

ambient air quality standards might delay the starting time by which

odorous pollutants could be placed under control.

     The use of the"Standards of Performance" control option has the

advantage of requiring the use of best adequately demonstrated control

technology regardless of any air quality considerations.   Thus there

is no requirement for nationwide air quality surveillance networks.

This approach would lead to the earliest control for the most objectionable

sources of odorous air pollutants.   One of the disadvantages is the fact

that it would take many years to bring all sources under control because

not all categories of sources can be listed at one time.   Also, particularly

for some existing sources, the best adequately demonstrated control

technology may not be adequate to abate the citizen complaints.


     At this point in time we are leaning toward the use of the

Standards of Performance route.  Hopefully information presented
and developed at this symposium will assist us in coming to a firm

decision as to the best control option for control of odorous air

pollutants from stationary sources.   For control of these materials

from mobile sources we can proceed in a clearcut fashion as soon as the

technology has been developed for measuring the emissions precisely

under an adequate testing procedure.  Also the controls for mobile

sources may be set stringently enough to insure protection of the

public welfare from objectionable odors.