426 JOURNAL OF COSMETIC SCIENCE finish) cylinder parallel to the light source served as a mounting device for the tresses. Tresses of Piedmont hair colored for 45 minutes were combed and then tightly wound around the cylinder, resulting in a consistent and parallel alignment of hair fibers. A panel of 20 untrained evaluators was asked to evaluate the tresses on a scale of 1 to 5, where 5 represents the highest luster. RESULTS AND DISCUSSION REFLECTANCE MEASUREMENTS The spectral reflectance curve showing the fraction of light reflected at each wavelength is an important characteristic of the colored object and gives information about hue and saturation of color. The recorded reflectance curves for the hair tresses with different colors are shown in Figure 2. As expected, the reflection of the light of at least their own hue and absorption of complementary hues are observed for different colors. The refer- ence Piedmont hair has a high reflectance in the red, yellow, and green parts of the spectrum and a low reflectance in the blue region. For Piedmont hair dyed with Clairol Hot Red color, a steep rise in reflectance is observed at wavelengths longer than 600 nm. For the Piedmont hair dyed with Clairol Green Weed and Denim Blue colors, the reflectance spectra show pronounced maxima at around 530 and 480 nm, respectively. From the reflectance curves it is also important to note that for all colors the charac- teristic reflectance bands are broad (half width of more than 50 nm). This indicates the dullness of the hair hue. A weak second band in the reflectance spectrum can further complicate the observed hue. A reflectance tail at wavelengths higher than 650 nm is attributed to the reflectance of the substrate (i.e., Piedmont hair) from the fiber interior. The largest change in reflectance curves occurs between 0 and 5 minutes of dyeing time. Thereafter, the decrease in total reflectance is small, indicating that additional uptake of the dye is small. These experimental observations can be correlated with the properties of the individual components in commercial semipermanent dyes as follows: The com- mercial dyes employed in this study contain both neutral HC dyes and cationically charged basic dyes. The basic dyes have a high affinity for keratin fibers and are known to color the cuticle very rapidly, which is observed in the reflection spectra after five minutes of dyeing time. However, due to their high molecular weight, basic dyes are unlikely to penetrate into the hair cortex. In contrast, HC dyes, consisting predomi- nantly of aromatic amine, nitroaromatic amine, and anthraquinone derivates, have little affinity for hair. Even though the HC dyes penetrate into the fiber cortex during 45 minutes of dyeing time, the dye concentration in the cuticular sheath is much higher compared to the fiber interior. This was confirmed by optical microscope examination of the cross sections of red-colored hair, showing the typical ring-dyed cross section with the cuticular sheath as a dark red ring, predominantly produced by the basic dye, and a weak pink-colored cortex, resulting from the neutral HC red dyes. We note that dyeing with semipermanent dyes does not involve chemical reaction, but depends on diffusion of colorants from the solution into the hair. Therefore, the resulting coloration of Piedmont hair is relatively weak (see Figure 3A). Stronger coloration could be achieved by using bleached hair instead of Piedmont hair, as bleaching damage would make hair easily penetrable.

EFFECT OF HAIR COLOR ON LUSTER 427 8O 7O • 60 r- 50 m 4O o 2o lO 0 i 350 400 8O Piedmont i i i i i i l 450 500 550 600 650 700 750 Red (sp) Wavelength (nm) 80O 7O • 60 r- 50 m 4O 0 2O 10 Piedmont Blue (sp) 0 • i i i i [ i l 350 400 450 500 550 600 650 700 750 800 Wavelength (nm) B Figure 2. Reflectance spectra for Piedmont hair before and after coloring with (A) red, (B) blue, and (C) green semipermanent dyes. Arrows indicates the increase in dyeing time (5, 20, and 45 minutes). (Continued on following page) In reference (7), the use of CIELAB color parameters obtained from a diffuse reflectance spectrophotometer as a measure of luster is suggested: the luster scale is built by using the calculated total color difference (AE), where the zero-luster corresponds to the color values of the sample before applying the luster-improving treatment. In our study, when measuring the CIELAB tristimulus values (L*, a*, b*) for colored hair, the lightness indeed decreases and the total color difference increases for the same colored hair with increasing dyeing time. However, the luster scale built in this way does not correlate