SKIN CONDITION MEASURED BY SONIC VELOCITY 13 panelists were then randomly divided into four groups and issued products which they were instructed to use as they normally use a Hand and Body Lotion. Grading, using both techniques, occurred again on the third and twenty-eighth day of the study. On each grading day the panelists were instructed not to use the product before examination. At the conclusion of the study each participant was asked a series of questions concerning the current condition of her skin and her perception of the effectiveness of the treatment she received. The visual and sonic grades were analyzed for the 3- and 28-day examinations using day 0 results as a covariate with the procedures of the Statistical Analysis System (36). These data are set out in Table III. After three days, results similar to the observations of the Table III Visual and Sonic Velocity Results of Ad Lib Hand Study Treatment Day 3 Results Day 28 Results Mean Mean Mean Mean Panelists' Sonic Velocity Visual Grade Sonic Velocity Visual Grade Self-evaluation (Meters/Sec.) (Overall Dorsal) (Meters/Sec.) (Overall Dorsal) after Study•' A 91.96-1 B 98.23 -1 C 102.80 ] D* 105.91 '] 1.23] 94.00- I 0.39'] 2.39- 1.39[ 95.99_l 0.611 2.73.] 1.17.] 100.64] 0.73.] 3.04 1.69 '] 102.60 1.17 ] 4.77 *Indicates placebo group. Day 0 sonic velocity used as covariate. Brackets enclose means not significantly different at 95% confidence level. ñGrade Scale: 1 = good to 7 = poor. previous forearm study were noted in that products A, B, and C were all found to be significantly different from each other and the control treatment based on sonic velocity measurements, while visual grading indicated A, B, and C were not significantly different from each other but were superior to the control. Again the elastic modulus of hand skin treated with active products decreased (softened) relative to the control hands. After 28 days of ad. lib. product usage, significant visual improvement was noted between product A, products B and C, and the control placebo. The sonic technique indicated that at this point in the study, products A and B significantly reduced the calculated skin elastic modulus relative to product C and the placebo. As before, the absolute ordering of the products by both methods is consistent, although product C was found to be not different from the control with the sonic measures. Examination of the data gathered from the panelists at the conclusion of the study (day 28) in response to the question "OVERALL, how would you rate the skin condition of your HANDS?" is also set out in this table. The panelists used a 7-point scale (! = very good to 7 = very poor). The treatment ordering produced by the panelists' responses is consistent with both the dermatologist's grades and the sonic velocity measures. Significantly, the product ordering produced by the sonic velocity measures on day three of the study (well before any active product differences were noted visually) predicts the final ordering produced by both the 28-day visual grades and the panelists' self-assessment.

14 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Overall, this study reaffirmed the observation of the forearm test. Significant sonic velocity differences are achieved earlier than significant visual dij•rences. Further, the differences thus seen with sonic measures are consistent with the longer term visual and consumer-perceived effects. 3. STATISTICAL ANALYSIS OF ELASTICITY/VISUAL/CONSUMER RELATIONSHIPS Because of the large base size of the Hand Clinical Study it was possible to perform a detailed statistical analysis examining the relationships between visual, sonic, and consumer-treatment evaluations. With the procedures of the Statistical Analysis System (36), correlations between these parameters were established and the amount of the variances predicted, assuming linear relationships, were calculated. Higher order interactive and nonlinear relationships between the variables are not reported, as the resulting components lacked clear physical meaning or interpretation. Other workers (19) have observed an age-related increase in the elastic modulus of skin. Averaging sonic and visual data for the right and left hands of each panelist in the Hand Clinical Study, a correlation over all panelists between age and the elastic modulus calculated using the measured sonic velocities on test day 0 resulted in a correlation coefficient of q-.43, indicating that age alone explains 19% of the variation. Initial skin condition would also be expected to affect the day 0 sonic response, and the correlation coefficient between the visual grades and the calculated moduli on day 0 was found to be +.35, explaining 12% of the variation. Because the correlation between age and the day 0 visual grades was very low (equal to q-.23) it is evident that these factors (age and visual grade) are almost orthogonal in their ability to explain the elastic moduli derived from the sonic measurement. That these factors are not more predictive may rest in the fact that, as discussed above, changes in the elasticity calculated from measured sonic velocities are associated with a lower limit in changes in the elastic modulus of the outer cornified skin layer. The technique used by these other investigators to detect the age-related elasticity differences measures the absolute elastic modulus of a larger volume of skin. There is no reason to believe that the absolute sonic velocity measures the absolute elasticity of skin only that, all else being equal, for a given individual, a change in the sonic velocity squared is proportional to a change in the outer layer elasticity. Calculation of the first order partial correlation coefficient (37) between the sonic velocity-derived elasticity change after 28 days and age, with the visual skin condition change after 28 days removed, indicates less than 5% probability that a correlation exists (the partial correlation coefficient is -.08). Thus, changes in elasticity measured in this experiment are age independent. Finally, the relationships between the visual and sonic measures and consumer information have been examined. Again, assuming a linear model, the correlation coefficients between the panelists' responses to four questions concerning the performance of the products tested in the clinical study and the current condition of their hands, and the visual and elasticity parameters determined at various times during the study, were calculated. Table IV sets out the variances explained in the consumer responses for each of these questions by each of the measurements. The table also lists the significance between the r 2 values at c• .05 calculated using the method of Walker and Lev (38) for the various measures for each question.