44 
as the larva needs protection. Larvae that feed exposed have very 
often no trace of them generally when the larva is protected by colour, 
but on the other hand we find exposed larvae with bunches of setae on 
the warts. In concealed feeding larvae, we usually find the tubercles 
simple, carrying perhaps only a single hair or perhaps none at all. 
But not only do the tubercles offer variation in the number and 
variety of setae but they vary in position on which we bane our character, 
with different changes of skin. This is to my mind the weak spot in 
the classificatory value of this character. Larvae are not bound down 
to any particular number of instars and if in the penultimate change, 
the position of the tubercles were different to the ultimate, a change 
brought about to undergo one change less in this direction would 
quickly alter the former conditions. Over and above this the 
prespiracular tubercle is never constant. From the moderate constancy 
of the position of tubercles i, ii, iii, and iv, Dyar made six large 
groups, but the number of these groups is so small that it is obvious that 
many widely divergent smaller groups must be lumped together. The 
structure, therefore, of the larva does not lend itself for the formation of 
families. Neither does the structure of the ovum or pupa offer 
itself for family formation. There is not primarily a sufficient variety. 
But if the embryological stages are not found to be of value for family 
demarcation, they are of the utmost importance for determining the 
phylogeny of the families. I have been pointing out the claimed 
value of the spiracles in the larva. These I personally believe to be of 
not much consequence and for this reason. The larva, unlike the 
ovum and pupa, has an infinitely greater struggle for existence and 
therefore has to undergo a corresponding number of modifications. 
This we find to be the case and although the covering or cocoon of 
the pupa has very many modifications, the actual pupa itself has no 
power per se of modifying its conditions, the larva forming the cocoon 
before histolysis has commenced. Although the pupal state follows 
the larval, embryological conditions hold to a much greater degree in 
the former. From this reason it appears that we should place very 
much greater value on ovum and pupa than upon the larva embryo- 
logically, that is, in studying the phylogeny. Dr. Chapman, who I 
suppose must have taken this view, has in his very valuable paper on 
the phylogeny of the lepidoptera treated upon the ovum and pupa only 
and left out the larva. Tutt, in his British Lepidoptera appears to 
take the same view. Speaking of the larva and pupa he says : “ It is 
necessary therefore in dealing with these stages (larval and pupal) 
to bear in mind two points :—(1) Whether the similarities which one 
sees are phylogenetic .... or (2) Whether they are oecological 
in their origin and due to a similar relationship of the animals to their 
organic and inorganic environment. The characters manifest in the 
egg state must almost of necessity belong to the first division ; those 
in the active larval (considered as an embryonic) condition may belong 
to the first or second.” Mr. Tutt does not suggest to which the pupa 
belongs. It is undoubtedly to the first along with the egg. Thus, 
although he does not say so in so many words, I think it is fair to 
assume that he gives much greater weight to the ovum and pupa than 
to the larva in determining the phylogeny. It would be presumption on 
my part to speak fully on the phylogeny of the lepidoptera with so 
little knowledge and after such work as Dr. Chapman’s on this 
