224 SPHAERODACTYLUS. 
time-honored usage have no diagnostic value whatsoever. The median groove 
in the rostral, the superciliary spine for instance are really generic and not specific 
characters yet their mention in specific descriptions has always been customary. 
The conditions in the supralabial series are so variable that no reliance can be 
placed on characters apparent there. So also, though in a less degree, the 
number and shape of the scales separating each supranasal from its fellow of the 
opposite side, is subject to much variation. The shape of head and body is 
useful to observe, although the tail varies widely in proportion to the body-length. 
Still it is generally longer and much more slender in females than in males. The 
best characters are to be found in the size, form, and arrangement of the scales of 
both the dorsal and ventral surfaces. Colour also as becomes daily more evident, 
is of far more value in diagnosis than has previously been supposed. Many 
features of pattern are strongly fixed. Lines may break up into series of spots; 
these may be many or few or the whole series may be absent but even a remnant 
is likely to be very definitely located with reference to the original complete 
pattern. Thus also markings, such as the spectacles so often seen on scapular 
or sacral regions, may or may not be present but if present their location is not 
haphazard, but is very distinctly fixed. The brilliantly cross-banded species 
occur in the Greater Antilles and Bahamas and curiously enough are found 
among the two extremes of the gerrus. They are either species with tiny granular 
scales or with enormous tectiform dorsals and among these two divergent groups 
sex-linked dichromatism has become well established. There are probably 
many more sex-linked colour-characters than are now recognized. Their dis- 
covery will further reduce the range of apparently fortuitous variability of each 
species. 
It is a somewhat surprising fact to observe how close is the correspondence 
between the distance from the tip of snout to the ear and the length of the fore 
limb. If the first distance is longer than usual so also the fore limb is likewise 
long. Whether this fact has any special significance does not appear. There 
may be some relation between the length of the arm and a corresponding head- 
length which makes more convenient the picking of small objects from the ground 
with the mouth or by using the thick fleshy and only slightly extensible tongue. 
While one occasionally sees a sphaerodactyl struggling to subdue a little moth, 
its wings flapping a most inconvenient if ineffectual protest, it is far more com- 
mon to find them eating ants. Little ants swarm in houses in the tropics as 
everyone who has lived there knows to his sorrow, and this may in some degree 
account for the lizards frequenting houses too. Ants creeping over smooth 
