558 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS T VESSEL ! I •3, 'l ,•õ I •" UNKNOWN 5 6 7 8 5•o I 2•oot,,,PP,• •-BUTaNOL I 16 62 250 IOOO Figure 3. Odor intensity re{erencing scale based on 1-butane] (bottom port on represents air flow distribution scheme in the dynamic dilutiqn binary scale olfactometer) Ioe one would simply smell this area ooe the shirt, he would deal with the "as-is" undiluted intensity of the odor. This aspect is best represented by measurements ooe an undiluted sample. On the other hand, odor dilution threshold comparison ooe two samples would give an indication on how far from the shirt the odor would be noticeable. Although both are somewhat related, the two values describe functionally different odorous properties of perspiration. The odor threshold can be measured by a great variety of methods (19). The problem with the threshold measurements is that the odor threshold of a substance is not an exact property, such as its density or boiling point, but depends on the method of odor presentation, statistical design, sensitivities, and motivations of panelists, etc. Method used for threshold measurements in room air will not yield functionally correct values for the same odorant in thc vapor space above axilla. Methods that use small volumes suffer from wall adsorption losses. Methods that use fast volumetric flows of air usually will be unsuitable for the evaluation ooe odors considered cosmetic defects, since the sample consumption rate would be prohibitively fast. Similar problems exist in odorous pollution measurements on samples taken to an evaluation laboratory. Recently, for these, (oe0, oe1) a dynamic

EVALUATION OF HUMAN BODY ODOR 559 ß .'. :: v !'•'-'Y' ... . . , ,• •-• ß -•3:.--' .7.:::..' .... : ....•...•:-ff• •'• . -. :.....:.:..•..•:,: .•...:.• •?"•' --.• •.• . :•.. .... :. ß ...:7::•:[•:.•:::::•.•. .:• ß ..:• :•,- ::.:& :.• .- .-½ .•: •! "'•.• ß ,.-.7:" Figure 4. Evaluation of odor threshold using Dynamic Forced-Choice Triangle olfac- tometer (teflon bag with sample of odorous air is in metal cylinder at right) forced-choice triangle olfactometer has been developed. Figure 4 depicts the device and the odor threshold evaluation process. Each sample dilution level is presented at flow rate of 500 ml/min from a glass sniffing port, and is accompanied by two more ports delivering only air at the same flow rates. The three ports are mounted in a circular arrange- ment in a plastic cup, connected to the sample dilution module by a flexible tubing, which carries inside Teflon tubing and electrical wiring. There are 5 such cup assemblies, providing all together 5 different sample dilution lev- els. The panelists proceed from the most diluted sample level toward the most concentrated sample set. It each level, they are required to select the "odd" port-that is different in sinell from the two others such design is known as forced-choice triangle design (I). The panelists signal their choice by de- pressing the electrical signal button next to the selected port. Lights on a panel seen only by the panel leader announce the panelist's choice. Un- intentional steering of the panelist toward some specific choice is impossible in this scheme, in contrast with presentation designs where one stimulus at a time is offered.