8 JOURNAL OF COSMETIC SCIENCE A B C D •· · ··· ·••1••·•• 11 ••1• •·····••1 11 ••···••1 1••1111•1 1 10 20 30 40 50 Time/min Figure 4. A HPLC chromatogram of fatty acid-ADAM derivatives in integral lipids. A: palmitic acid. B: stearic acid. C: MEA. D: tricosanoic acid as an internal standard. goids (data not shown). The absence of peaks at ca. 11 min in Figure 6 indicates that human hair contains little CERs with phytosphingosine. Both CH and CERs were calculated as the sum of the extractable lipids and the integral ones, since it is unclear whether they should be chemically distinguished from each other or not in the case of CH and CERs different from MEA, which is clear in its chemical state (6). STATISTICAL ANALYSIS OF THE LEVELS OF THE HAIR LIPIDS The variation of the hair lipids among individuals is shown in Table I. An average of the level of each lipid was in the order of (FAs) (WEs) (HCs)(CH) (SQ) (TGs) (MEA) � (CERs). The major hair lipids can be assigned as FAs and WEs since their sum accounted for more than 85% of the total lipids (average 23.5 mg/g hair). The percent­ age of coefficient of variation was in the order of (TGs) (SQ) (WEs) � (FAs) ===:= (HC) (total lipid) (CH) (CERs) (MEA), thus indicating that their levels of TGs, SQ, WEs, FAs, and HCs among individuals fluctuate more than those of the total lipids. Relationships between the levels of each lipid were analyzed (Table II). There was a significant positive correlation (p 0.01) among the levels of SQ, WEs, and FAs, one of which is shown in Figure 7 A. A significant positive correlation (p 0.01) between the levels of CH and CERs was also observed (Figure 7B). Further, there was a significant negative correlation between the levels of SQ and CH (p 0.01), SQ and CERs (p 0.05), WEs and CH (p 0.01), WEs and CERs (p 0.01), TGs and CH (p 0.01), TGs and CERs (p 0.05), FAs and CH (p 0.01), and FAs and CERs (p 0.01) (Table II). A representative negative correlation between WEs and CH is shown in Figure 7C. While HCs were weakly correlated only with CERs (p 0.05), MEA exhibited no significant correlation with any other lipids.

HAIR LIPID COMPOSITION 9 A I � I • . . . - .,.J . - I 5 10 1 I 0 15 Time/min Figure 5. A GC chromatogram of cholesterol in extractable lipids. A: cholesterol. A principal component analysis (PCA) for the levels of eight lipids was performed. Factor loading, eigenvalue, and contribution obtained by the PCA are shown in Table III. The first two components accounted for 60.4% of the total variation, with 42.3% in the first component (21) and 18.1 % in the second component (22). The values in 21 increased in accordance with the increase in the levels of CH and CERs and the decrease in the levels of SQ, WEs, FAs, and TGs. The values in 22 mainly increased in concert with the increase in the level of HCs. Therefore, 21 was interpreted to imply a negative corre­ lation between one lipid consisting of SQ, WEs, FAs, and TGs and another lipid consisting of CH and CERs (Table II). 22 was also interpreted to imply a definite contribution to the level of HCs. A two-dimensional projection of the PCA, which implies a total variation for the hair lipid composition of 44 Japanese females, is shown in Figure 8. The individuals who were characterized by the lower levels of SQ, WEs, FAs, and TGs, and by the higher levels of CH and CERs, were situated at a positive value of 21 (Figure 8). On the other hand, individuals who were characterized by the higher levels of SQ, WEs, FAs, and TGs, and by the lower levels of CH and CERs, were situated at a negative value of 21 (Figure 8). Based upon Figure 8, the individuals were symbolized according to their age (Figure 9). Young individuals (ages 1-10) tended to locate at the right-hand side, and older