W HILE collecting and preserving several hundred specimens of seven species and subspecies of West Indian celesti, and handling and observing as many more during several years of field work, certain characters which are frequently listed as specific appeared to me to be of generic value. The specific characters which may be relied upon are: (1) arm length compared to distance from snout to ear; (2) number of undivided lamellae under longest toe; (3) scale count, mental to vent; (4) scale count around midbody; (5) number of pairs of chin shields; (6) character of keels and striae on dorsals; (7) position of angular subocular between upper labials; (8) average adult size; and (9) color pattern. Frequently one only of the above characters will suffice to classify a specimen. There are two groups of celesti, one having arms longer than the snout-toear distance, and the other, equal or shorter. This proportion is evident at any age (size) and immediately separates the long-armed C. duquesneyi, C. occiduus and C. o. hewardi from the short-armed C. maculatus, C. plei, C. crusculus crusculus, C. c. molesworthi and C. c. cundalli. The number of undivided subdigital lamellae under the longest toe is correlated with arm length, and is mentioned here for the sake of those who, in the past, have confused the short-arm-and-fewer-lamellae group with the young of the long-arm-and-more-lamellae group. Members of the long-arm group live in stone walls, cliffs, heaps of coconut husks or similar dry habitats and make their escape by running. Members of the short-arm group live beneath objects or among dead leaves and escape by wriggling. Their habitats do not necessarily overlap. All stages of growth of each group were found, but no intermediates between the two. The number of lamellae does not change with growth. The number of scales between the mental and the vent is a more accurate
Read full abstract