110 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS But that example merely serves to illustrate, by reductio ad absurdum, a principle that is worthy of more serious attention. An ordinary bar of soap, without any doubt and beyond any challenge, must act as a deodor- ant. This would not justify the labeling of Lux as a deodorant soap, unless it can be shown that it has greater properties in the reduction of perceptible unpleasant body odor than have soaps made up of other ingredients. In the same manner, the deodorant claims for Dial soap are completely un- justified unless it is demonstrated that such a soap diminishes the body odor more than does Ivory, Lux, Palmolive, or some other product. As a beginning, I should like to suggest, merely as a point of departure for further discussion, and not as an end point in any investigations, that the deodorant be defined as a product that effectively diminishes the per- ceptible unpleasant odors more than does a comparable material used or sold for other than deodorant purposes. Such a definition does not mean that Arrid must be more effective than Stopette, or vice versa, because they are both used for their deodorant value. It does mean that Dial, to justify its claims, must be more effective than Lux, because the latter is sold and used as soap. Now, how can one determine whether there is such an effective reduction of odor? Up to the present time, two general methods have been pro- posed. I should like to classify them as the indirect and the direct. Tug Im)•RgcT MgT•or) The indirect method consists of experimentally demonstrating that the alleged deodorant causes certain biological and/or chemical changes to take place within the body or on its surface, and then to draw from such experimental evidence the conclusion that ipsofacto such changes cause re- duction of body odor. The two changes of body condition that have been interpreted in such manner are the reduction in the quantity of perspiration, and the reduction in the bacterial count of the perspiration. The indirect method of establishing that a deodorant is effective assumes that there is and must be an inviolable and intransigeant ratio between such a biological state and the quality of odor. Tug D•Rgcx Mgx•or) The direct method of determining the efficiency of a deodorant relies on actual olfactory examination either of a subject or of a substance that he has worn. This method can be divided into two general systems: 1. NumericalMeasurement System: Under such a method, the sample of material collected from the breath, perspiration, or elsewhere, is tested on an instrument, at the end of which an olfactory judge is smelling. The instrument can be constructed along several different lines, one of the most popular being based on an air-dilution technique. It is not my wish in this

TESTING THE EFFICIENCY OF DEODORANTS 11! article to examine the relative merits of the different instruments, particu- larly inasmuch as I believe all of them have been overrated, but I should like to call attention to the statement of KillJan, that during extensive investigations of cutaneous odors, the metallic osmoscope was found to be "less reliable than the unaided nose" (1). Inasmuch as most of the testing on chlorophyll as well as on other al- leged deodorants has in recent years been conducted on such instruments, I should like to emphasize a few facts about them: (a) The instruments, contrary to some popular myths, all depend on a human subject who must finally make a choice. It has not yet been dem- onstrated that there are any ways of diminishing the variability between one human nose and another, which has been such an obstacle in odor clas- sification and other olfactory work. To state, as has been done, that a thermometer or a spectrophotometer likewise depends upon a human being to read the results is to ignore the overwhelming evidence that the eye is more reliable than the nose in giving reproducible results. (b) The reliability of such instruments in giving results that are repro- ducible from one judge to another or even by the same judge from one oc- casion to another has definitely not been demonstrated. (c) There is no information whatsoever on the standard deviations ob- tained in the use of such instruments, the margins of error, and the possi- bilities of obtaining the given results by chance alone. Nevertheless, the numerical measurement system has one advantage which must not be gainsaid and cannot be denied. It permits the examina- tion of a person under one situation in comparison with the same person under another situation. It enables one to judge the mouth odor of A, who may be using a chlorophyll-containing toothpaste, not against B, who may be using the penicillin-containing paste, but against himself, A, the following week or month, when he has been switched to the penicillin. It enables the right arm of A, under which Dial soap may be used, to be compared with the same man's right arm, on another occasion, when Life- buoy may be used, instead of making it necessary to compare the right arm of an individual with the left arm, thus bringing up the possibility that there may be extreme differences of activity of the sweat glands under each arm. This is not an asset without a drawback. Let us not overlook the fact that A on one occasion may be more unlike A on another than A might be unlike B or, even more possible, that the right armpit of A may be more dissimilar from the same armpit a week or two later than it would be unlike his left armpit on the same day. All in all, the numerical evaluation of olfaction would be ideal, in that it would reduce the results to a mathematical equation, but it is completely unacceptable until such time as its reproducibility has been established and the margin of error computed.