HAIR VOLUME AND TEXTURE 347 Table II Experiment III: Image Analysis Body vs Panelists' Visual Assessments and Panelists' Visual-Plus-Textural Evaluation Image analysis Panelists Panelists body (ratio) (visual body) ("visual + feel") Most body A* (1.51) A I A* 4/ B* (1.36) B B I C* (1.04) C* C Least body G* (0.53) G* G* * Significantly different at p = 0.05 level. If the data are separated by sex of panelist, there is no significant difference in the evaluations of the twelve female and nine male assessors. When panelist rating data are correlated with image analyzer values, non-parametric analysis shows Rho = 0.85, p 0.001 (twenty-one panelists). This is a highly significant relationship given that the image analyzer is measuring tress "volume" (really a 2-D image) but panelists are incorporating true volume and visually apparent textural components into their assess- ments. The surface treatments in Experiment III, while providing for a wide range in hair texture, also resulted in relatively large differences in hair volume between treatments, particularly for the pomade (G) versus the quaternized protein (A) (a threefold differ- ence). In a subsequent experiment (IV) the hair surface treatments, B, C, D, E, and F (see Experimental section) were chosen to provide a range of hair body and texture, but with a smaller volume difference than for Experiment III (approximately a 1.6-fold difference between largest and smallest volume changes see Table III). Three replicates per treatment were used, and the sequence of evaluations was as described previously for Experiment III. Table III tabulates the image analysis data and panelists' visual ratings for the five sets of treated tresses for Experiment IV. The panelists were more discriminating in their assessment of visual hair body than image analysis in this experiment. Since the hair volume differences in this experiment are relatively small, the panelists may be building other parameters besides volume into their visual evaluations, e.g., stiffness, dryness. Nevertheless, a significant correlation between panelist rating data and image analyzer volume values shows Rho = 0.94, p = 0.0001 (non-parametric Spearman). Table III Experiment IV: Image Analysis Body vs Panelists' (Visual Body) Treatment Rankings Image analysis Image analysis Panelists (body) (hair body ratio) (visual body) Most body Least body D* 1.08 -+ 0.04 D* E I 0-95+0.13 E I B 0.95 --- 0.07 B F I 0.71 --- 0.13 F* C 0.65 - 0.16 C* * Significantly different at p = 0.05.

348 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS IMAGE ANALYSIS VS BODY ("VISUAL + FEEL") In Experiments III and IV the final hair body assessment by panelists included both visual and tactile evaluations. This was done to examine the contribution of texture to hair body. The question arises of how different is a "visual" vs "visual + feel" panelist evaluation of body and how does "visual + feel" evaluation agree with the image analysis body measurement. Inspection of the data for Experiment III (Table II) shows the same rankings from image analysis body measurements and visual assessments (by panelists) and "visual + feel" assessments (by panelists). However, statistical analysis of the data shows subtle differences in the ability to distinguish among these three eval- uations, with image analysis body assessment being the most sensitive of the three evaluation methods for these four treatments. The data from Experiment IV are slightly different (see Table IV). Although the rankings of the image analysis evaluation and the panelists' visual assessments are in the same order, the visual assessment appears to be more sensitive. However, the textural assessment causes a reversal in the body ranking of treatments B and F in this experiment (although it should be noted that this reversal is not statistically significant by the "visual + feel" evaluation of body). Table V compares Spearman rank correlation coefficients from the two experiments and shows highly significant correlations of visual body and image analysis or "visual + feel" body and image analysis in both experiments. However, the correlation coefficient does decrease for Experiment IV when the textural component of body is included. This result may be explained by the relative volume and textural changes of the treatments being considered. In Experiment III, where the textural component did not alter the rank order of the body measurements, the relative volume difference between treatments was almost twice that of Experiment IV (see Table V) when the texture component began to play a role. This observation suggests that image analysis is a good measure of hair body even when there is a relatively large textural difference between hair samples. However, when large textural differences are being examined with relatively small volume changes of the order of 1.6 or less, then the relationship between image analysis of hair body and the perception of hair body may weaken. BODY ASSESSMENT BY MEN AND WOMEN An analysis of body "visual + feel" ratings by men and women from Experiments III and IV suggests that hair treated with the particulate treatment B are assessed differently Table IV Experiment IV: Image Analysis Body vs Panelists' Visual Assessments and Panelists Visual-Plus-Textural Evaluation Image analysis Panelists Panelists body (visual body) ("visual + feel") Most body D* D* D* v B F ] F* B C* Least body C C * Significantly different at p = 0.05 level.