Table III Recommended Sample Size for Polarized Light Shine Tests Comparison Reich-Robbins TRI Stamm Guiolet Bosse-Nova Tech. Virgin MB Bleached MB Virgin MB Bleached MB Virgin MB Bleached MB Virgin MB Bleached MB Virgin MB Bleached MB Untreated Leave-In conditioner 6 9 6 8 6 6 6 7 6 8 Untreated XM Conditioner 663 3109 3282 7594 40 272 67 388 115 103 Untreated 2 in 1 19 63 27 109 22 13 22 17 12 25 Untreated Phenyl trimethicone 6 6 8 6 6 6 6 6 6 6 Phenyl trimethicone Leave-in conditioner 6 6 6 6 6 6 6 6 6 6 Phenyl trimethicone XM Conditioner 6 6 8 6 6 6 6 6 6 6 Phenyl trimethicone 2 in 1 7 6 9 6 6 6 6 6 6 6 Leave-In conditioner XM Conditioner 6 9 6 8 6 6 6 7 6 7 Leave-In conditioner 2 in 1 6 8 6 7 6 6 6 7 6 7 2 in 1 XM Conditioner 24 63 27 110 94 14 84 18 23 277 2008 TRI/PRINCETON CONFERENCE 121

JOURNAL OF COSMETIC SCIENCE 122 the technical measures of luster perform well for larger changes and for more global comparisons between colors. What we fi nd remarkable about the results reported in Figure 6 is that the human perception is quite reproducible, both in the repeatability of individual comparisons and in the consistency of changes in shine with damage across multiple hair colors. COMPARISON OF PERCEPTION TO TECHNICAL MEASURES The change in perceived shine with damage is consistent across the hair types studied. The change in measured luster, however, is not so consistent, neither across hair types nor between types of damage (Table II). Given the lack of consistency, it is diffi cult to recom- mend one formula over another for the prediction of perceived change in shine. In this situation, there are only a few ways forward. We could accumulate more experimental results on more hair types with more repeated measures and hope that consistency begins to emerge, or we could begin a search for a new and more relevant measurement. CONCLUSIONS We have examined the change in several technical measures of shine with several standard hair treatments and sources of hair damage. We demonstrate the use of several, well- characterized control treatments (Figure 1) and recommend experimental sizes to be used for specifi c comparisons (Table III). These controls and methods reported in this paper should provide consistent results in the technical measure of shine by polarization imag- ing. We further have attempted to connect these measurements (Figure 2) to newer work on the origins of perceived shine based on the histogram skew (Figure 3), and while the utility of these methods is not immediately obvious, the fact that this metric gives more systematic variance with hair lightness is intriguing. Finally, we compare perceptual results on shine comparisons (Figure 6) to technical mea- sures (Figures 3,4 Table II). Much previous work has shown that these measurements are consistent with perception for large changes in shine (7). Here we focus on the small changes associated with between treatment comparisons or damage. We conclude that the problem of quantitatively measuring the physical drivers of hair shine at the level of between treatment comparisons is not well captured by technical measures used here. Based on our preliminary unpublished work and other references (12) we expect that in addition to lightness, color based signals are critical cues for quality of color as well as for shine. Future work will focus on these. REFERENCES (1) R. S. Hunter and R. W. Harold, The Measurement of Appearance (J. Wiley & Sons, New York, 1987). (2) K. Ward et al., Ieee Transactions on Visualization and Computer Graphics, 13, 213 (2007). (3) R. F. Stamm, M. L. Garcia, and E. J. Fuchs, J. Soc. Cosmet. Chem., 28, 571 (1977). (4) C. Reich and C. R. Robbins, J. Soc. Cosmet. Chem., 44, 221 (1993). (5) R. Dror, A. S. Wilsky, and E. H. Adelson, J. Vision, 4, 821 (2004). (6) E. H. Adelson, in Human Vision and Electronic Imaging VI, B. E. a. P. Rogowitz, Ed. (SPIEE, San Jose, CA, 2001), pp. 221.