138 Annals Entomological Society of America [Vol. VIII, 
have served to unite groups otherwise discordant. A careful 
study of the proboscis, or the scales might serve to give clues to 
phylogeny. 
In the Coleoptera, tarsal and antennal characters have been 
used yet mostly in vain, the elytra, a character peculiar to the 
group, should be investigated. If we consider groups as large 
or larger than families we notice that specialization has not 
proceeded along one definite line, but the line of specialization is 
continually changing, and often accompanied by other mod- 
ifications. Each change in the line of specialization marks the 
limits of a group of greater or less extent; one structure having 
reached a certain stage marks time while other structures are 
modified. 
The Mantispide have peculiar front legs, having reached a 
certain development this structure remains fairly stable, while 
other structures develop. The Limnephilide in the Trichoptera 
are an example of stability in venation; generic characters are 
largely to be found elsewhere, while in the Sericostomatide 
venation continues to vary and aid in defining genera. 
In the Diptera the Muscide, Tachinide, Dexide and Sar- 
cophagide were defined by bare or pilose antennz, yet genera 
with pilose antennz occur in various related families. An 
accrescent character of these groups appears to be the chaeto- 
taxy, and this has been used to indicate a new classification of 
these families. 
Structural and other characters may be roughly grouped 
into two sections, adaptive, that is those which have been 
influenced by environment and habits, and atavic, or those 
which are of no use to the insect, and persist because they are 
not in the way, and have a long history back of them. The 
adaptive characters are of use in small groups to indicate 
affinity, but soon break down when applied on a larger scale. 
Thus two eyeless species occurring in the same caves may be 
closely related when belonging to one genus, but most such eye- 
less cave insects are not related. Asa whole adaptive characters 
are of little use in tracing relationships. It is the atavic, or 
accompanying characters, not related to a life-habit, that are 
the best for indicating affinity. All insects have many points 
of structure or color that are of no use to them. Many of these 
