SYSTEM OF INSECTS. 411 
low the second method lately mentioned, and consider 
them as arranged in parallel series. 
In studying the analogies that take place between in- 
sects themselves, we should always bear in mind that our 
inquiry is not concerning an affinity which demands a 
correspondence in various particulars that are not neces= 
sary to constitute an analogy; as, for instance, that there 
should be a mutual imitation in all the states of any two 
insects. Wherever we discover a marked resemblance 
between two perfect insects, there is a true analogy, 
though their metamorphosis may differ ; and where there 
is not that resemblance, though the metamorphosis may 
agree, there is no analogy. In fact, insects are some- 
times analogous in their fst state and not in their last ; 
and at other times analogous in their dast and not in their 
Jirst; but the analogy is most perfect when it holds in 
all their states: it then, indeed, almost approaches to an 
affinity. ‘They may also be analogous to each other in 
their habits and economy, when there is little or no re- 
semblance in their form; and, vice versa, be analogous in 
their form and not in their habits. So that different sets 
of analogies may be assumed as foundations for different 
systems. Thus Mr. MacLeay assumes the metamor- 
phosis as the basis of analogy between the corresponding 
Orders of Mandibulata and Haustellata*, while M. Sa- 
vigny compares the perfect insects”: the result therefore 
differs in some instances. I shall now lay before you 
in a tabular view their plans and my own. 
* Hor. Entomolog. 456. Comp. Linn. Trans. xiv. 67—. 
> Mem. sur les Anim. sans Vertebr. 1. i, 20—. 
