306 MONTGOMERY. [ VoL. X11. 
drapagus canadensis, D. franklini, and in Bonasa, Lagopus, 
Tympanuchus, and Pediecetes. Now it is interesting that the 
subgenus Dendrapagus of the genus Dendrapagus is limited 
to North America, and its species are variable in regard to the 
number of the rectrices ; while such variation apparently does 
not occur in subgenus Cazace of the same genus, which inhabits 
both Eurasia and North America. Of the other genera, Bonasa 
and Lagopus have the same geographical distribution as Canace, 
while Z7ympanuchus, Centrocercus, and Pediecetes are limited to 
North America. Those individuals evincing this numerical 
variation of the rectrices must have been influenced by a con- 
siderable change of environment, since the variability has 
continued through a large number of generations. Now we 
can hardly suppose that this change of environment has occurred 
within the areas occupied by these variable species in North 
America, since in that case we must suppose that the other 
species occupying the same areas must have become similarly 
modified. Accordingly the species of Dexdrapagus which pre- 
sents this variation must be of comparatively recent occurrence 
within its present habitat ; and its variability would be caused by 
having changed its former habitat within a comparatively recent 
period, by migrating from its former area (Eurasia ?) to its pres- 
ent geographical position in America. On the other hand, the 
non-variable North American species of this genus (namely 
canadensis and frankliniz) must be regarded either as not 
having experienced such a change of environment, or as 
having experienced it in a much remoter period, having in 
the meanwhile restored the correlation of their organs, 2.e. 
adapted themselves to the new environment. Regarded from 
this standpoint, the variation in the number of the rectrices 
may serve to elucidate the origin and present distribution of 
the Zetraonide. 
These considerations are offered as mere suggestions for 
explaining how variation can serve as a criterion of development. 
We have found that the data of variation offer a certain mean 
for determining whether the development is progressive or 
regressive, depending upon the fact whether the variations are 
above or below the normal. When the phenomena of varia- 
