THE CANADIAN ENTOMOLOGIST. 341 



Torre's subfamily 1 1 is Coelioxynas, including Coelioxys, which I regard 

 as a nudigaster, and Afiomobates, which I regard as a nudipede. 



From the analogy of Bombus and Fsithyrus, I claim that Coelioxys 

 is related to Megachile, Stelis to Aiithidium, Me/ecta to Anthophora, 

 Nomada to Andrena : not only, however, on this analogy, but also on 

 morphological grounds, from the venation and other characters. 



Assuming that the inquilines arose from some of their hosts, as 

 is certain in the case of Bovibus and Fsithyrus, we would expect the most 

 resemblance between host and inquiline in the recent cases of the highly 

 specialized bees, as Bombus and Fsithyrus, and the least in the 

 oldest cases of the least specialized bees, as Andrena and iVotnada. 



In Delia Torre's classification, as above stated, Fsithyr7is ioWov/s 

 Bombus, though in different subfamilies. I would give Fsithyrus the same 

 position, but put both in the same subfamily. Stelis follows Anthidium 

 in a separate subfamily, whence I would remove it to follow Megachile in 

 Megachilina?. Also Meleda and Epeolus follow Anthophora, but in a 

 separate family. I would separate Anthophora from Mellissodes, etc., and 

 put Meleda and Epeolus with it. 



Nornada must seem the most far-fetched of my cases. Although it 

 has a long first discoidal cell, I think other characters of the venation 

 separate it far from Meleda and Epeolus, especially the large stigma and 

 pointed marginal cell. I think Nomada is an ancient offshoot from 

 Andrena, and is not related to any other genus. Its differences from 

 Afidrejia and resemblances to other bees I hold are acquired, not 

 inherited. After Andrena I would place Farandrena, a more recent off- 

 shoot, and then write Nomada. 



In this connection I think the taxonomic proposition will hold that 

 an offshoot from a certain group is related to that group. It may acquire 

 resemblances to the other forms, but not relationship. 



That Mr. Ashmead is right in interpolating the inquiline bees among 

 the host bees is no doubt correct, but this has been done by Delia Torre 

 to such an extent as to destroy the contrast which exists between Mr. 

 Ashmead's arrangement and the old-fashioned and unnatural arrangement 

 of Schmiedeknecht. As in the Delia Torre arrangement, I hold that Mr. 

 Ashmead does not go far enough ; indeed, it seems to me that he refutes 

 his own scheme by the very arguments which he cites in defence of it In 

 his section III. Schmiedeknecht arranges certain bees whose differences 



