The Effects of Aging on Hair–More Than Just Amount 454 The data of Table 1 summarizes that of Keogh and Walsh, comparing % any gray and % complete gray versus the “true color of the hair” categorized as Fair, Medium or Dark. The numbers of subjects are all relatively large, especially in the three lowest age groups which contain more than 300 subjects per group. The data of age groups 42.5 through 52.5 are all 205 subjects or larger however, in age group 55+ the number of subjects drops to 137 or greater, while in age group 62.5 and above the number of subjects in each hair color group are only 26 or fewer. Therefore, the data of this oldest age group are the least reliable. Table 1. Age and graying of hair from the work of Keogh and Walsh6 % Any Gray % Completely Gray Age Fair Medium Dark Age Fair Medium Dark 27.5 3.4 9.4 27.7 27.5 0.4 0 0.0 32.5 19.4 34.0 47.4 32.5 0.6 0 0.5 37.5 39.2 54.1 63.9 37.5 0.9 2.2 0.54 42.5 60.1 70.0 77.3 42.5 4.6 3.3 2.6 47.5 79.1 82.1 87.2 47.5 8.9 7.0 7.4 52.5 93.6 90.7 94.1 52.5 23.9 17.4 13.8 57.5 100 96.1 97.7 57.5 42.3 30.1 18.6 62.5 97.8 98.6 98.2 62.5 50.0 23.3 16.8 Model equations using JMP statistical software were used to calculate the above values from the actual data of Keogh and Walsh. In general, the calculated values are within ± 2% of the Keogh and Walsh values. All three % “any gray” models contain R2 = 0.99 or greater and p 0.0001. For % “completely gray,” higher order polynomial models were used, but were not effective for extrapolating to zero gray. An effective model could not be constructed for fair hair, therefore the Keogh and Walsh data are listed above for that group for percentage of “completely gray” for both medium and dark hair: R2 = 0.99 or better and p 0.02 or better. These authors analyzed the data by logistic regression analysis and concluded that at approximately age 49 (48.6) about 50% of this population has 50% gray hair. This statement has been widely misquoted to apply to all populations. It would appear that age 49 for 50% gray hair should apply to Caucasian populations, but, if we accept the conclusion of Tobin and Paus2 about graying in other