EVALUATION OF ANTIPERSPIRANT ACTIVITV 757 The device used to refine these displacements or errors from the desired information is based on the realization that the slopes of the ratio lines obtained from the tests are the factors to be compared. The slope of the line is independent of its placement on the graph. Figure 3 illustrates how this comparison may be accomplished graph- ically. One draws new lines respectively parallel to the test lines and Figure 7. Absorption cup in place passing through the origin. The change in ratio from point a on the parallel to the control ratio line to that of point b on the parallel to the test ratio line may immediately be read. Perspiration rate x" has been reduced to x' by the application of a product to site A. Values x" and x' taken from Fig. 3 may now be used to calculate the absolute reduction of perspiration, as in equation 2. The cm•ses of these additive shifts in position of perspiration ratio lines are not easily defined. Tentative data suggest that variations in

758 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS the initial wetness of the skin during the test may be the cause. It is surprising how long it takes to dry wet skin even with a very dry gas stream. Capillary creepage of liquid moisture from the wet side to the dry side under the edge of the separation in the cup may be another cause. If this is so, then any method utilizing absorption cups will be subject to this error. Applying grease to the cup can possibly eliminate Figure 8 Heating chamber this creepage, but great care must be used not to smear the test area with it or the results will be compromised. The additive error becomes considerably more serious as the test period is reduced or as the rate of perspiration utilized for the test is de- creased. However, if the test period is lengthened, one loses the ability to follow the relative change in product effectiveness with time. If heavy perspiration rates are used, a washing away of the active material renders the test inaccurate. Since the procedure to be described can separate the error from the desired data, a short test period does not im- pair the results. Consequently, this procedure can both follow a product effect with time and measure, if any, a wash-out effect as well.