-6 (126) 22 -------- 21 (130). This scale reduction follows the method proposed by Dowling (1951b: 133) in which the numbers on the mid-line represent the number of scale rows, upper figures refer to the right side of the snake, and figures in parentheses indicate the number of the ventral scale (counted from the anterior end of the series), thus marking the position of the addition or reduction. Addition of a row is shown by a plus sign and the number of the row, whereas reductions are shown by a minus sign and the number of the row that is lost or by a plus sign between the number of two rows that join. According to Dowling, variation in number of dorsal scales characterizes the few genera and species of snakes in which it has been studied. The time and difficulty involved in ascertaining the number of scales explain why it has not been widely used in classification. [Illustration: FIG. 2. Number of ventral scales in 48 female and 34 male _A. p. leucostoma_.] Ventral scales on 34 males averaged 134.4 (128 to 139), and on 48 females 133.5 (128 to 137) (Fig. 2.). Barbour (1956:34) found an average of 135.3 ventral scales on 64 males and 44 females, and Gloyd and Conant (_loc. cit._) found an average of 134 for both males and females. The average for the eastern cottonmouth obtained by Gloyd and Conant, however, was 137 ventrals in both sexes. Some of my counts were made before I knew of the standard system of counting ventrals proposed by Dowling (1951a:97-99), in which the first ventral plate is defined as the most anterior one bordered on both sides by the first row of dorsals. Therefore, some inconsistencies may exist in my counts. Where differences occur, Dowling's method probably will indicate the presence of an additional scale, since it appears to begin farther anteriorly on the average, than I began counting. [Illustration: FIG. 3. Number of caudal scales in 44 female and 34 male _A. p. leucostoma_.] TABLE 5.--Caudal Scale Combinations in 95 Cottonmouths. U = Undivided; D = Divided.