Tran^icfionfi of the 'Royal Soc'ipff/ of South A frirn. 
to a distance of 12 miles there is nothing but sand hills or sand hummocks 
ad uifinitum, and during the dry season they become carpeted with flowers 
after the first winter rains. Ograbies, 15 miles north from Port Nolloth, is 
partly roclcy, partly sandy. From Ograbies the sand ceases, and. is replaced 
fo Anenous railway station (50 miles from Port Nolloth), where the heavy 
ascent of the mountain begins, by rocky, grassy or bushy ground," 
Fi\)iii tliis it is clear that what is wanted is intensive study in a neigh- 
bourhood wliere two or more forms are known to occur. Tlie records are 
not always as trustwortliy as could be wished owing to the insufficient locali- 
sation of their captures l>y collectors. 
In the opinion of the junior author it is not at all clear that environment 
has been an important factor in the distribution of these several forms of 
the genus Agama, or that it is this adaptation which has led to evolution on 
so many minor lines. 
If natural selection were the process whereby the original homogeneous 
(this is an assumption) stock branched out into the various forms which 
segregate in areas of more or less different environmental conditions, it 
follows, of course, that the differences ])etween varieties acnJeata, cllsfanti, 
etc., are adaptative. So far as can be seen, however, there is no evidence 
that the difference l)etween var. disfanfi and var. acvleata (for example) 
has anything to do with, or at any rate is the direct result of, a difference 
in environmental conditions, taking these to include food, enemies, etc., as well 
as the natui-e of the surroundings. The fact that specimens from the Water- 
berg division of the Transvaal, well within the area of var. distanti, show 
such a wide range of variation in the direction of other allied forms, would 
seem to prove that natural selection has not operated strongly in regard to 
these characters. 
In order to conceive the drift of evolution in these lizards, it is 
necessary to picture to ourselves a hypothetical ancestral form of Agama, 
and basing our views on considerations derived from a general survey of 
the family Agamidae, the following characters may be postulated for this 
ideal prototype : 
(1) Body moderately elongate, neither compressed nor depressed, covered 
with scales of equal size. 
(2) Head not broader than long, not or but moderately depressed, 
covered with subequal, non-tuberculous scales. 
(3) Ear-opening large. 
(4) Nostril lateral, not tubular. 
(5) Limbs moderately elongate, covered with equal scales. 
(6) Toes rather elongate, subcylindrical, very unequal in length, fourth 
the longest, fifth much longer than the first. 
(7) Tail cylindrical, much longer than head and body, covered with 
equal scales not forming whorls, 
