248 JOURNAL OF COSMETIC SCIENCE Table I Precision of the Device Output by Repeated Measurement on Three Subjects Device output (average) Standard deviation [mV] [-] Subject A (n = 20) 29.48 1.510 Subject B (n = 20) 17.53 1.045 Subject C (n = 20) 21.80 0.796 Although a difference is seen in the high scores of a few samples, measurements by the device and the sensory score for each group are also congruous. The correlation of the measurements obtained by the device and the sensory score supplied by the expert assessor is shown in Figure 20 for all subjects. Since the evaluation scores are saturated near both ends of the scoring range (1 and 7), it becomes such a curve. Both results were seen to be highly correlative, demonstrating that measurement by this device is useful in human hair luster evaluation. CONCLUSION A new and unique device, which can measure human hair luster easily and accurately, was developed. The sensory score obtained by the expert assessor and the measurements by this device were highly correlated. A more precise evaluation of hair luster is made possible by this device, which also offers the following advantages: ß It enables direct measuremen,t without cutting the subject's hair. ß It is not influenced by the subject's hairstyle and/or hair color. sensory score : 2 sensory score : 3 device output : 17.5 device output : 21.0 (a) permed (b) straight (short) v. sensory score : 4 sensory score : 6 device output : 29.5 device output : 42.3 (c) straight (medium) (d) straight (long) sensory score : 2.5 device output : 22.5 (e)colored sensory score : 3 sensory score : 3 device output : 28.0 device output : 24.3 (f) bleached (g) male Figure 17. Examples of the measurements. sensory score : 5 device output : 41.5 (h) male

DEVICE TO MEASURE HAIR LUSTER 249 5O 4O 30 20 10 0 . $ dyed or bleached hair ¸ not dyed hair - ¸ I I I I I 1 2 3 4 5 6 7 sensory score I' 95% confidence intervals Figure 18. Validation of the measurements by this device unaffected by hair colors. It requires only about 0.2 seconds to register measurement, is quick, and is less burdensome to the subject. It is a small and simple device. Therefore, measuring by this device will be useful not only for research, such as vali- dation studies on hair care products, but also for sales promotion at cosmetics shops. •' 50 40 • 30 • 20 o.,,• • lO • 0 $ penned hair . O short hair ß medium length hair . Along hair - O I i I I I 1 2 3 4 5 6 7 sensory score I' 95% confidence intervals Figure 19. Validation of the measurements by this device unaffected by hairstyles.