1915} Tracing Relationships of Insects 137 
tendencies. The positive characters are most useful in delimiting 
genera (and other groups) but because of their constancy of little 
value in tracing relationship. It is to these accrescent characters 
that we should look for phylogeny. _ 
If several species of a genus A have spines on the vertex, and 
an allied genus also has spinose vertex, it is not likely that these 
spines in genus A will indicate relationship; but if in a series of 
genera with bare vertex, there is a genus in which spines are 
present, then the arrangement and size of these spines may 
indicate phylogeny. Take for example the spider genus Tetrag- 
natha; it has a peculiar character in the enlarged, much-toothed 
mandibles; a study of the increased modification and armature 
of these mandibles will afford clues to relationship of the species. 
Formerly I and others have used variations in eye-position as 
group characters, but these same variations in eyes occur in 
allied genera and so may occur in various parts of Tetragnatha 
irrespective of phylogeny. 
Therefore, to my mind the best way to get at the relation- 
ships of the species of a genus, or the genera of a family, is by 
tracing the development of some character peculiar to the series; 
an accrescent character, found in varying stages of development 
in the group, but not found in allied groups, particularly groups 
that may be considered ancestral to the group in question. 
There are many prominent cases where, I believe, primitive 
characters have deceived systematists. For example, in spiders 
the cribellum and calamistrum are primitive characters, and , 
occur in groups otherwise widely separated. Several arachnol- 
ogists have insisted on grouping these forms together, thus 
producing a most heterogeneous assemblage, whereas if they 
would ignore these primitive characters, and study the accres- 
cent development of some peculiar character of spiders they 
would reach a better knowledge of their phylogeny; the male 
palpi are just such a character. 
Another case is the pronotum in Hymenoptera extending to 
the tegulza; Ashmead put the social and fossorial wasps together 
on this account; the character occurs elsewhere in the Hymenop- 
tera, and therefore cannot be depended upon to indicate affinity. 
In the Lepidoptera various systems have been based on the 
possession of some primitive character, thus the case-forming 
habit of larve, the jugum, mandibles, number of anal veins, etc., 
