56 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS PREPARATION FOR SCANNING TRANSMISSION ELECTRON MICROSCOPY Sample preparation for electron microscopy consisted of staining the tresses for two hours in 1.3% OsO4 in 0.067 M cacodylate buffer. Hair segments, 1 cm in length, were cut from the root and tip regions and embedded in Spurr's resin. Transverse sections approximately 120 nm thick were cut with a diamond knife in an LKB Ul- tratome NOVA ultramicrotome and collected on 200-mesh copper grids containing a formvar-carbon support film. The grids were then post-stained with uranyl acetate and lead citrate for image enhancement. A thin layer of carbon was evaporated onto the specimens to reduce electrostatic charging in the electron microscope. The specimens were examined in an Areray 1200B scanning electron microscope op- erated in a transmission mode at 30 Kv with a final objective aperture of 200 •m. Photomicrographs were taken with Polaroid Type 52 black and white film. A total of 160 photomicrographs were taken at random of the cuticle and outer portion of the cortex at a final print magnification of 5000. IMAGE ANALYSIS OF ELECTRON MICROGRAPHS The image analysis process consisted of, first, defining the types of ultrastructural changes observed. The cuticle and cortex regions were analyzed separately. In the cuticle, round holes or voids originating in the endocuticle were differentiated from linear voids which occurred between cuticle layers. In the cortex, irregularly shaped voids were analyzed. A Quantimet 900 Image Analysis System was then used to quantirate the voids ap- pearing in the electron micrographs. Once detected, several parameters were measured: the mean area of individual voids (•m2), the total projected area of voids (•m2), the % voids (the total projected area of voids divided by the total area scanned), the total number of voids and the size distributions of the mean areas. STATISTICAL ANALYSIS The experiment has two factors, each at two levels: (1) location of analyzed region, root versus tip and (2) treatment, no shampooing versus shampooing. Four independent test conditions were examined with a sample size of 40 each thus, a two-factor analysis of variance with no interaction model was considered. However, the equal variance assumption was violated and, in order to employ the analysis of variance technique, the four test conditions were combined into one sample and assigned ranks. Since all comparisons involved testing one group mean to another, the Wilcoxon rank-sum test was a plausible test statistic. Because the analysis of variance technique does not apply to the Wilcoxon test, an alternative is provided by Fisher and Yates' normal scores test which transforms the ranks to standard normal values. These values were then analyzed by the analysis of variance technique. It should be noted that there were no gross violations of assumptions using the transformed variates. All t-values obtained from the analysis of variance techniques were adjusted downward to reflect the Pittman efficiency of the Wilcoxon test to the Student t-test. This adjustment employed the lower bound of X/0.864 as a multiplying factor for all t- values in an effort to be conservative in light of the transformations that were employed.

QUANTIFICATION OF ULTRASTRUCTURAL CHANGES IN HAIR 57 RESULTS AND DISCUSSION SUBJECTIVE ANALYSIS Electron microscope examination of the transverse hair sections showed striking differ- ences between the root and tip ends of the hair fiber. Hair from the tip end of the fiber exhibited discrete holes or voids in the endocuticle region of the cuticle measuring 0.1 to 0.2 p,m in diameter. In the cortex, there were numerous voids which appeared to be associated with the intermacrofibrillar matrix, nuclear remnants, and cell mem- brane complexes. Hair from the root exhibited fewer holes in both cuticle and cortex regions of the fiber. Whether the observed voids are originally present in the hair fiber as such or represent latent ultrastructural modification which becomes overt during the techniques of sample preparation is not known and is one of the unresolved problems of microscopy. How- ever, these voids are clearly a manifestation of changes occurring in fiber ultrastructure as a direct result of the weathering process. A subjective analysis of these observations was conducted which consisted of a blind ordinal rating of the photomicrographs for the degree of ultrastructural modification. A scale of 0 to 3 was used, with 0 representing the least modification. After identifying the photomicrographs, the average rating for each treatment group was determined. The root or non-weathered region received an average rating of 0.5 for both non- shampooed and shampooed treatment groups. In other words, no differences were seen which could be related to the multiple shampooing treatment in the root region. In the tip or weathered region, there was a statistically significant increase in the disrup- tion observed compared with the root region. The ratings were 2.4 for the non-sham- pooed and 2.6 for the shampooed hair. The increase from shampooing, however, was not statistically significant. IMAGE ANALYSIS Weathering Efj•cts. Statistical analysis of the objective comparisons between root and tip regions of the hair were in agreement with the subjective analysis that was performed. At a greater than 95% confidence level, all measures of ultrastructural disruption, except mean area of cortex voids, increased from the root to the tip (Table I). In the cortex, the mean area of voids showed no difference between tip or weathered hair and root or non-weathered hair. This indicated that the size of the voids which appeared Table I Ultrastructural Changes as a Result of Weathering* Total Number Mean Area Location % Voids of Voids of Voids (•m 2) IRoot 0. 365 33 0.189 Cortex LTip 1.050 76 0. 176 /Root 0.0335 5 0.00710 Cuticle [Tip 0.412 35 0.0358 * Each value represents the median of 40 measurements.