1915] Tracing Relationships of Insects 139 
characters are variable, and one must endeavor to find by an 
examination of a long series of at least a few species what 
characters are constant. 
Atavic characters usually exist unchanged through a long 
series, so they are of no use (or little use) in tracing affinity 
within a genus. They are of most use in indicating the relation- 
ships of genera and families, and especially where insects have 
acquired a number of striking adaptive characters, some of 
which may be those of convergence and tend to conceal the 
true affinities. 
Other points might be brought out, but at present I desire 
to impress upon systematists that atavic characters should be 
sought in the broader fields of classification, while in many 
studies, particularly in genera, accrescent characters should be 
considered, while the use of primitive and of adaptive characters 
should be avoided, or used only in connection with the others. 
