552 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS ampies of application to human body odor measurement. The examples serve to illustrate only-not to prove efficacy of some specific formulations. GENERAL REMARKS The efflcacy of odor control can be measured by sensory or analytical tech- niques, or by a combination of both. Sensory techniques measure the overall change in odor sensation caused by the change in emission, e.g., a reduction in the intensity of odor, or a shift in the character of odor. Analytical techniques permit one to measure the reduction in the emission of malodorants. When fragrances are used to sup- plement the •reatment, the analytical approach cannot yet indicate the over- all sensory impact however, it is useful in following the process of the dissio pation of fragrances after they have been applied. In sensory techniques, substances reach a panelisgs nose by migration through the air in the form of vapors. Vapors may be channeled directly from the source, e.g., the axi]lar portion of a shirt, or else they may be temporarily stored in plastic bags-a method much in use in odorous air pollution stud- ies. Analytical techniques must be similarly applied to emissions in vapor form, to approach composition smelled by nose. Ratios of concentrations of various odorants above perspiration are different from the ratios in the condensed perspiration: less-polar odorants exhibit higher volatility from an aqueous phase than do more-polar odorants of the same vapor pressure in a pure state. Gas-ehromatographie (GC) analysis, with a vapor preeoneentration step to bridge the gap between the sensitivity of nose and the most suitable detector (hydrogen-flame ionization detector), is the most used analytical technique in investigating odors. In the sensory techniques, panelists serve as instruments. Their expecta- tions and biases may seriously distort their judgments. It is, therefore, pre- ferable to remove the odorous sample from its context to use panelists who do not know the nature of the sample, and wherever possible to devise sam- ple presentation techniques that minimize anticipation effects. The principles of panel selection and sensory evaluation have been de- scribed elsewhere (1). A few considerations deserve stressing. Familiarity with an odor that is tested is a double-edged sword. An expert becomes quite proficient in discriminating fine nuances of odor, which he has repeatedly studied. In evaluating the effleaey of an odor eontroI treatment, however, such familiarity becomes a liability: an expert tends to anticipate certain ef- fects. The same applies to panelists who are quite familiar with the particu- lar fragrance and its use in eosineries. An individual using a certain scented product would recognize its fragrance and would be biased in rating its

EVALUATION OF HUMAN BODY ODOR 553 effect in odor control. Well-known panel effects such as a desire to please the panel leader or to judge toward some expected result, etc., should be neutralized. SAMPLING FROM SOURCES When sampling from odorous emission sources, it is desirable to sample selectively, without interference h'om other sources. When the source is a specific part of human body, source-adapted confining devices are used. Figure 1 depicts devices which have been used for some sources. Under- arm in vivo may be confined by a polyethylene form (2,3). A mouthpiece is used to collect vapor samples h'om the mouth high-purity air is supplied through one tube, and the vapor is removed at a controlled rate through an- other, while a U-tube water level manometer is used to control breathing to maintain pressure in the mouth cavity at ambient pressure. When the subject manipulates his breathing in such way, the content of lung air in the sample stabilizes at about 50 per cent. Samples of skin vapors are taken using a Teflon • cup (4). Vaginal vapor samples are taken utilizing perforated Teflon inserts (5, 6). Sampling of vapors from the entire human subject is accomplished by placing him into a glass tube on a Teflon-lined stretcher suspended within the tube (7). With all such devices, air must be introduced at a rate commensurate with the sample removal rate and slow enough to prevent depletion of odorant content in the source. On the other hand, sampling still should be suffi- ciently fast so that an adequate sample size can be collected without gross inconvenience to the subject. Typical flow rates are on the order of 50 to 100 ml/min for localized sources, and 50 1./min for the subject in the tube the duration of sampling from the mouth can hardly exceed 10 to 20 min, while longer times are possible with other sources. The function of the confining devices is not only to temporarily isolate the source, but also to permit a more quantitative vapor sampling. Such a temporary isolation does not eliminate the presence of vapors of ambient origin in the source emission. Thus, in taking a sampling from the entire sub- ject, components characteristic of engine exhausts and mothproofing com- pounds are commonly present. Care therefore must be exercised in the evaluation of data, especially in the case of analytical gas-chromatograms. One particular problem in taking samples from human body sources is the presence of excess water vapor, which is close to the saturation point at 37øC. To avoid water condensation in the tube which leads from the source to the sample collector (such as an adsorbent or a bag), tubing should be •E. I. dupont de Nemours & Co., Wilmington, Del.