247 
those periods. We must use other evidence to support such a 
proposition. We cannot expect to find much evidence of Coc- 
cidae in the geological records, but we can reason from their 
high specialization that they appeared later than the more 
generalized Sternorhynchi. 
When considering the problems of phylogeny the possibilities 
and probability of parallel and convergent evolution must be 
constantly borne in mind. In every large group of animals there 
is evidence of such, and it is the first work of the phylogenist 
to decide where this has taken place. For this reason we must 
not base our conclusions upon one structure alone. But we 
must base our conclusions upon evidence, even if it be incon- 
venient. If evidence can be shown that the “filter” was once 
present in the Cicadellidae and has since been lost, then our 
task would be simplified; the Membracidae would then come off 
the same base and the Cercopidae would represent the more 
direct line. This is indicated by the thickened line in the dia- 
gram. But until such evidence can be produced we must take 
things as we find them and arrange our diagram accordingly. 
In drawing up diagrams Nos. 1 and 2 (Plates VII and VIII) 
I have tried to take the above remarks into consideration. ‘They 
do not indicate the time element, but simply try to express my 
conceptions of the morphological affinity of the families dealt 
with. As such they are liable to alterations and repairs, as they 
give way beneath the weight of accumulated knowledge. 
