had shown that gall structures are significant measures of an inherited physiologic capacity of these insects. Again, the gall of teres occurs on the leaves of a mountain form of the Oregon white oak (Q. garryana semota) in the southern Sierras adjacent to the Central Valley of California (fig. 28), and all of our experience with the distribution of species had lent support to a corollary of the so-called Jordan's Law, to the effect that the most closely related species occur in adjacent areas. But teres (fig. 162) was a flea-like insect with short wings not a quarter of the length of those of clavuloides, and it had been placed in a genus which contained none but short-winged species. Only the hypopygial spines of the two insects were similar (figs. 188, 190), but we had found that these spines are of great phylogenetic significance; and considering the spines, the galls, and the distribution data, the conclusion seemed inevitable that clavuloides and teres were close relatives.
As we have extended our study to other Cynips, we have repeatedly disclosed similar relationships between many other short-winged and long-winged species, until we are forced to believe that the genus includes 42 subapterous forms which have originated more or less directly from long-winged stocks within the genus. Our bases for the recognition of these affinities are:
1. Close identity of galls, as already explained;
2. Occurrence in adjacent ranges, as follows from Jordan's Law and from our other cynipid data;
3. Close taxonomic affinities between hosts of the insects;
4. Possession of similar hypopygial spines, tarsal claws, and antennal counts, altho we shall show in a later paragraph that dissimilar spines are not evidence of lack of relationships; and
5. Utilization of the bisexual form (where known) as more primitive than the agamic form in these insects (Kinsey 1920:369).
A few typical instances will illustrate our use of these bases. Turn first to the case of the long-winged Cynips acraspiformis, which Weld (1926) recognized as a good species of our present genus, altho he was puzzled to observe that it had a gall (fig. 304) similar to that of a short-winged Acraspis! The map (fig. 59) shows the range of this species. From its distribution and near identity in all structural characters, the closest relative of the long-winged acraspiformis is certainly the long-winged expositor (fig. 340) of eastern New Mexico
