JOURNAL OF COSMETIC SCIENCE 100 A class value of I was the lowest degree in curl and VIII was the highest. Statistical detail about the classifi cation procedure can be found in earlier publications (5,6). GEOMETRIC ANALYSIS Geometrical measurements of hair from each subject were taken using a laser scan mi- crometer (Mitutoyo, Kanagawa, JPN) at 22 ± 2°C and 45 ± 5% relative humidity (RH). Dimensional values for a single fi ber were obtained by taking the average of 10 cross- sectional areas along the length of the fi ber that was 3 cm in length. Ellipticity was de- termined by dividing the major axis by the minor axis. Results from 50 fi bers per subject were averaged. MECHANICAL ANALYSIS The tensile properties of dry hair from all of the subjects were determined using a Minia- ture Tensile Tester 675 (Dia-Stron Ltd., Hampshire, UK) using a strain rate of 12 mm/ min and a gauge force of 1 g. The environmental temperature and humidity was 22 ± 2°C and 45 ± 5 RH%, respectively. All tensile properties were measured on the fi rst 50 hair fi bers that exhibited normal failure profi les. Young’s modulus and break stress were nor- malized using the cross section of each fi ber. THERMAL ANALYSIS For each country represented, hair from fi ve random subjects that originated from the same country was cut into a fi ne powder and blended. High-volume DSC pans were charged Figure 2. Segmentation tree that partitions the hair into 8 curl classes.

2008 TRI/PRINCETON CONFERENCE 101 with 6.5–7.5 mg of this blended hair and 50 μl of deionized water was added. The sam- ples were allowed to equilibrate overnight at 22 ± 2°C. All samples were then thermally characterized using a Q100 DSC (TA Instruments, New Castle, DE). A total of 5 blended samples that represented each country were run, and the denaturation temperature (Td) and change in enthalpy (ΔHd) were determined by averaging the 5 values. CHEMICAL ANALYSIS Twenty-three to twenty-fi ve individuals from the 5 different countries were used for two- dimensional polyacrylamide gel electrophoresis (2D PAGE) experiments. Three to fi ve fi bers were selected from each individual and were delipidated in ethanol (25 ml) and in cyclohexane (25 ml) for 20 minutes, successively. The hair was allowed to air dry at ambi- ent temperature and cut into a powder. The hair powder was placed in an extraction solution which contained 7 molar (M) urea, 2 M thiourea, 50 millimolar (mM) 2-Amino- 2-(hydroxymethyl)-1,3-propanediol (trizma), 50 mM dithiothreitol (DTT), and 0.1% Triton ×100 with doubly distilled Millipore water at 40°C for 18 hours. The keratin was post-translationally modifi ed with iodoacetamide in a tris solution prior to dialysis. The protein solution was dialyzed for 3 days in 3500 molecular weight (MW) cassettes sus- pended in water. The extraction solutions were lyophilized to yield white to off-white keratin. Protein was rehydrated in a solution containing 7 M urea, 2% 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), 2% sodium dodecylsulfate (SDS), and 0.2% 3-10 Biolyte ampholytes. The protein solution (125 μl) was loaded onto 7 cm im- mobilized pH gradient (IPG) 3-11NL support strips and then active rehydration was performed at 50 V overnight. The fi rst dimension was run on a PROTEAN IEF cell (Bio- Rad, Hercules, CA) with a fast gradient voltage of 8000 V at 10000 V-hr. Both current and temperature were kept constant at 50 μA/gel and 20°C, respectively. The second dimension was performed with Novex Sharp Protein Standard MW markers (Invitrogen, Carlsbad, CA) on 10% bis-tris Invitrogen gels in NuPage MES buffer at 200 V using a XCell SureLock system (Invitrogen, Carlsbad, CA). After each run, the gels were fi xed in an aqueous solution containing 10% acetic acid and 40% ethanol prior to staining with Sypro Ruby stain (BioRad, Hercules, CA) using stan- dard techniques. The gels were destained in 7% acetic acid and 10% ethanol with deion- ized water prior to imaging with the BioRad Molecular Imager Gel Doc XR System (BioRad, Hercules, CA) interfaced to Dell using Quantity One software. Three gels were performed on each sample to assure reproducibility. STATISTICAL ANALYSIS The statistical analyses to determine differences in hair from different countires for the geometrical and mechanical data were based on LSD (22), Duncan (23) and Scheffe (24). Group differences were determined by considering the methods that were consistent in their statistical output. Contrast tests (independent t-tests) (25,26) were used to show ethnic differences in the thermal properties of hair. A Levene test of variance homogeneity (27) showed that ΔHd