560 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS 3 2 DIRECTLY INTO BAG ' l PADS INTO BAGS i i I I o T MIDNIGHT MIDNIGHT TIME, HOURS AFTER WASHING NON-MEDICATED SOAP, AXILLAR ODORS Figure 5. Change in odor threshold of consecutively exposed underarm pads as function of time after washing with nonmedicated soap The dilution factor increases by a factor of 3 per level. Panelists' judg- ments are statistically combined and yield a group threshold termed ED50 (effective dosage at the 50 per cent level). Since all needed dilutions are ob- tained by a continuous steady-state dynamic dilution simultaneously, pane- lists can proceed with evaluation without waiting for any adjustments. One sample can be evaluated by a panel of 9 within 12 to 15 min, and about 10 min are needed to change to another sample. An example of ED50 determinations applied to study of underarm per- spiration odors is shown in Fig. 5. The objective of the above was to explore how the underarm odor changes during the time between when one first washes with a nonmedicated soap (Ivory©*) and abstains from washing for several days. Cotton cheesecloth folded patches, 5 x 5 cm in size, were taped to both underarms of the same subject and were removed after durations of exposure indicated bv the length of the segments in the figure. Exposed patches were stored in s'mall plastic bags in a refrigerator until all samples were collected. Twenty-five hours after washing had taken place, a sample of underann vapors was also collected directly into a 15 1. Teflon bag, using procedure shown in Fig. 2C. *Procter and Gamble, Cincinnati, Ohio.

EVALUATION OF HUMAN BODY ODOR 561 For the evaluation of the dilution quotient (22), which is numerically equivalent to ED50 and indicates how many volumes in total one volume of odorous emission must be diluted to reach group odor threshold, both left and right patches were placed in 15 1. Teflon bag and 15 1. of nonodorous air were added (using a peristaltic pump and a wet-test gasometer at the intake into the pump to measure the volume of air added). The bags were kept at room temperature for i hour, to permit an equilibration of vapors be- tween the patches and the air in the bag. Bags were then placed into the device shown in Fig. 2D and the odorous air expelled into the olfactometer shown in Fig. 4. Nine panelists then were judges, and all the samples were evaluated in the same session. The plot of Fig. 5 indicates that during the first few hours the perspira- tion odor was at level below group threshold (only 4 out of 9 panelists-a statistically inadequate number-identified the undiluted-sample-containing port correctly). The odor peaked after 8 hours, then more or less stabilized at a somewhat lower level. The sample taken from the underarm and put directly into the bag showed a higher odor level this may be expected, since some loss of odorosity results when odorants from pads distribute between the pad and the air in the bag while with the sample taken directly into the bag with a preflush of the bag with the sample, such loss is greatly reduced. Since the pad method is logistically simpler, and would still indicate odor control efficacy on a relative basis, it is possibly preferable. The significance of differences between the samples can be tested statisti- cally. If the same panel has been used for all samples, the t-test-by-Difference, which compares judgments of each panelist separately, is a suitable proce- dure (1). Odor Character A cosmetic treatment may not influence odor intensity, but can modify the character of odor, either by changing the relative concentrations of mal- odorants, or by adding fragrances. "Before" and':"after" samples can be compared by a variety of methods (23). Most of the methods require that samples that are to be compared be available simultaneously. Some methods, however, use reference samples, or a series of semantic descriptors, which in essence serve as references. These permit the evaluation of samples which are not available simultaneously. The procedure is less discriminating than a direct comparison, but is much simpler logistically. In cosmetics dealing with human body odors, a wide spectrum of odors may occur-from most unpleasant to quite pleasant. Only a few odor descrip- tor scales encompass a wide selection of descriptors.