JOURNAL OF COSMETIC SCIENCE 90 Table I Typical Data from SPF Testing on 10 Subjects MED (s) SPF Test Test Subject no. Untreated Standard Material Standard Material 1 15 245 990 16.3 66.0 2 19 310 827 16.3 43.5 3 15 281 863 18.7 57.5 4 15 225 792 15.0 52.8 5 15 245 750 16.3 50.0 6 15 245 653 16.3 43.5 7 15 245 863 16.3 57.5 8 15 281 750 18.7 50.0 9 15 245 750 16.3 50.0 10 19 356 1093 18.7 57.5 Average SPF (n=10) 16.9 52.8 SD 1.32 6.97 Standard error 0.42 2.20 t-Value (one-tailed) 1.833 1.833 A 0.76 4.04 SPF value 16 48 MED (s) - Minimal Erythemal Dose (seconds). decreases, the SD in the numerator decreases, and the number of subjects in the denomi- nator increases. All three values contribute to a decrease in the value of A as subjects are added. An example of this is shown in Table II, which contains the same data as in Table I, only doubled, so that the data from each subject is listed twice. The data for subject 1 is the same data as for subject 11. In this example, the average SPF is 52.8, the same as in Table I. Subtracting the A value of 2.62, results in a label SPF value of 50, which meets the desired target label SPF value. Interestingly, according to the Final Rule, the label SPF of a sunscreen can increase by increasing the number of valid subjects. Increasing data to 30 subjects by adding the same group of 10 again fails to result in signifi cant additional benefi t because the rate of change in the t-value, the SD, and the number of subjects decreases. With data from 30 subjects, the A value reduced from 2.62 to 2.09 and the label SPF remains at 50. Example 2. A second example of the effect of subtracting A from the average SPF occurs when all 10 subjects in the clinical trial to determine SPF return an identical SPF, in this example an SPF of 30. In this instance the SD is zero, so the A value is zero. Subtracting the A value of zero from the average SPF value results in a label SPF value of 30. If one of the subjects returned an SPF value higher than 30, e.g., 34.5, then the SD would change from zero to 1.42 and the value of A would become 0.82. Subtracting the A value of 0.82 from the average SPF value of 30.5 results in a label SPF value of 29. The label SPF value for the product is 29 despite the fact that not a single subject returned an SPF value below 30. This example shows that, according to the Final Rule, the label SPF value of a formulation can decrease by increasing the individual SPF values of valid subjects.

CHOOSING AN EXPECTED SPF VALUE 91 Table II Typical Data from SPF Testing on 10 Subjects Twice MED (s) SPF Test Test Subject no. Untreated Standard Material Standard Material 1 15 245 990 16.3 66.0 2 19 310 827 16.3 43.5 3 15 281 863 18.7 57.5 4 15 225 792 15.0 52.8 5 15 245 750 16.3 50.0 6 15 245 653 16.3 43.5 7 15 245 863 16.3 57.5 8 15 281 750 18.7 50.0 9 15 245 750 16.3 50.0 10 19 356 1093 18.7 57.5 11 15 245 990 16.3 66.0 12 19 310 827 16.3 43.5 13 15 281 863 18.7 57.5 14 15 225 792 15.0 52.8 15 15 245 750 16.3 50.0 16 15 245 653 16.3 43.5 17 15 245 863 16.3 57.5 18 15 281 750 18.7 50.0 19 15 245 750 16.3 50.0 20 19 356 1093 18.7 57.5 Average SPF (n=20) 16.9 52.8 SD 1.28 6.78 Standard error 0.29 1.52 t-Value (one-tailed) 1.729 1.729 A 0.50 2.62 SPF value 16 50 MED (s) - Minimal Erythemal Dose (seconds). The results of these two examples led us to consider the consequences of the expected SPF value in SPF testing according to the Final Rule. EXPECTED SPF VALUE AFFECTS THE LABEL SPF Consumer Product Testing Company analyzed data from 10 clinical trials evaluating SPF according to the Final Rule (6). The purpose of the analysis was to investigate the rela- tionship between the expected SPF value and the label SPF value using real data from these 10 clinical trials. The expected SPF on these trials were 15, 20, 30, and 50. From these data, we analyzed the SD values that were used to calculate A values, which were subtracted from the average SPF value. The calculated A values were used to forecast ex- pected SPF values. The question we asked was “What Expected SPF value should be