350 Willis .— The Sources and Distribution of the 
the country, at least not very far away. And often the local distribution 
enables one to decide. 
The same criticism applies to the remark lower down on the same page, 
that ‘ we are familiar with many species, the range of which is widely dis¬ 
continuous’. Here, again, the total number, though in itself considerable, 
is really very small when compared to those with continuous ranges. Again 
he says, dealing with the species of Ceylon (wides) which have discontinuous 
ranges and have co-types in Assam, and New Zealand wides which have 
co-types in South America, that a little more dying out would result in the 
production of forms definitely endemic in one of their present areas. Quite 
true, but he has to show that such a thing as dying out without change 
of conditions can occur. On my view a species may be killed out by 
submergence, great climatic change, or other catastrophe, but will continue 
to spread (following age and area) in the regions where it survives, so long 
as conditions there remain unaltered. In the case of the New Zealand 
species, the antarctic land that once connected them to South America has 
either disappeared or become incapable of supporting plant-life, and in the 
case of Ceylon and Assam there is good reason to suppose that the two were 
for some time separated by an arm of the sea, while in recent times, at any 
rate, the intermediate country has been dry, Ceylon and Assam being both 
wet. Though cut off from their co-types, the species of Ceylon and New 
Zealand behave in those countries exactly like the other species found there, 
and are arranged in graduated series in ‘ wheels within wheels It is 
impossible to reconcile the idea that many endemics are dying out with the 
regular graduation of species shown by my figures. One cannot conceive of 
species dying out in such a regular way, whether they have or have not 
wides beside them, whether they are endemic species in wide genera, or 
endemic genera, whether endemic with large or with small area, and the rest 
(cf. the distribution map of Doona ( 18 , p. 14) and those given above). 
Can Dr. Sinnott produce, on the hypothesis of Natural Selection, any 
shadow of a reason why Ranunculus Lyallii , perhaps the finest of the 
Ranunculi, should be confined to South and Stewart Islands in New 
Zealand, while R. acaulis and R. rivularis (wides) range New Zealand from 
end to end and reach the Chathams, and several endemics range into the 
North Island also ? Other reason, that is, than the mere fact that such is 
the case. The natural selectionists assume, without facts to go upon, that 
this and other endemic species are dying out by reason of unsuitability. 
But in no single case can they say with certainty that a species is really 
dying out under unchanged conditions. In this particular case, for example, 
some would probably maintain that it was dying out, others would 
vehemently deny it. They cannot in the least define a size of area above 
which species are to be regarded as growing, or below which as dying out, 
and would apparently prefer to see the subject of the distributional areas 
