18 



nary length, while the other antenna consisted of 9 joints, the 

 terminal one seeming to consist of two fused together. Another 

 female had 9-jointed antennae, but again the terminal joint was 

 fully twice the normal length, with the same appearance of two 

 joints being fused as is exhibited where there are 10, and a con- 

 striction as if indicating an 11th joint, while still another female 

 had 9-jointed antennae, with an indication of a tenth. The males 

 all have 9-jointed antennae. It will thus be observed that the fe- 

 male does not agree with Mr. Coquillett's description of this 

 species. Hardy described Dichelomyia (Cecidomyia) rosarum &s 

 having 14-jointed antennae, so our species does not fit his descrip- 

 tion, though, as stated by Mr. Theobald, aside from the number of 

 antennal joints there is no perceivable difference between them. 

 We have shown that the number of joints in the female antennae 

 of the American species is too variable to be considered a specific 

 character. Are not those of Hardy's species equally variable, 

 and did he not describe a variation instead of the normal? How 

 far can the number of antennal joints be relied upon in sepa- 

 rating the species of Cecidomyiidcel These are questions that I 

 am unable to answer, but they have a decidedly important rela- 

 tion to the solution of the problem of the specific identity of 

 the insect under consideration. 



The larvae from which all my material was reared, closely 

 correspond with the description given by Hardy, but unfortu- 

 nately he could not say whether it was his C. rosarum or C. rho- 

 dophila that developed from them. 



HABITS OF THE BRITISH SPECIES. 



The habits of Dichelomyia rosarum are given in "Die Rosen- 

 schadlinge," p. 272, by Friedrich Richter v. Binnenthal; and 

 by Rubsaamen, in "Biologisches Centralblatt," Vol. XIX., Nos. 

 16, 17, and 18. Mr. Theobald informs me that he has found 

 it attacking roses in England, both in rose-houses and in the 

 open. In the open, it attacks the dogrose, Rosa canina Linn., 

 the commonest rose in Britain, which grows rapidly and luxuri- 

 antly in hedges, thickets, and various dry places in every part 

 of the country. "The gall is formed by the edges of the leaflet 



