﻿64 PROCEEDINGS OF THE NATIONAL MUSEUM vol. 102 



from 38 to 44 in the males. We cannot divide the series into species 

 at some points arbitrarily chosen. And we know that the variation 

 in this respect may be quite considerable. 



It seems clear that Packard's species A. aequalis, lucasanus, and 

 newherryi cannot be upheld. All these forms ought to be included 

 in A. longicaudaius, which was described first. 



In this species I also include Rosenberg's species A. oryzaphagus 

 and biggsi. These two forms differ from each other in two points, 

 according to the description. The former species should have 35 

 body-rings (34 +i or 35 in paratypes examined by me) and be small; 

 the latter should have 36 rings (35 +i or 36 in paratypes examined by 

 me) and be larger. Considering the continuous series of variation of 

 the number of rings in American forms it seems difficult to accept a 

 certain number as a specific character. Similar numbers are observed, 

 even in other samples, as table 8 demonstrates. One of the types of 

 A. aequalis has exactly the same number as A. biggsi (table 8). The 

 taxonomical value of the differences in size is doubtful. The forms 

 in question are probably nothing more than populations with a pro- 

 nounced parthenogenetic tendency, counterparts of which are to be 

 found in the European species, A. cancriformis. In all respects, they 

 join the series of variation in other American forms. Parthenogenetic 

 populations probably also occur in the Galapagos Islands, from which 

 locality I have seen lots (U. S. N. M. Nos. 83031, 82033-5, and 84240) 

 totafing several hundred specimens, all female. Table 8 lists other 

 localities from which only females have been found, but the material 

 is too scanty to permit any conclusions about the possible existence 

 of parthogenesis. 



I have not had access to specimens of A. domingensis Baird from 

 Haiti or A. guildingi Thompson from St. Vincent Island (Packard, 

 1883, p. 326), and so I cannot say anything of the taxonomic status 

 of these forms except that no facts in the short descriptions contradict 

 an assumption that they belong to A. longicaudatus. 



Apus longicaudatus, in this sense, is very variable. So are other 

 species of the genus, too, according to my experience. In one respect 

 the variation seems to be remarkably great here, and that is the total 

 number of rings in the females (34 +i to 43). It is true that I have 

 not observed such a great range of variation in other species (see table 

 4), but we must not forget that only relatively limited material of the 

 other species is considered here. How great a variation they really 

 have remains an open question. 



The future may show into what units A. longicaudatus can be 

 divided. Not even their taxonomic value can be predicted, 

 though, from the standpoint of our present knowledge, it seems un- 

 likely that they will be species. 



It seems clear that A. longicaudatus is closely related to several 



