628 [December 



yellow. The number of antennal joints, too, in grossularifx is said to 

 be only 12, instead of 13 or 14; but that may very probably have arisen 

 from the scapus being counted as only one joint instead of two. (See 

 above, p. 557.) Loew, for what reason he does not state, perhaps be- 

 cause the verticils are not mentioned in the description, thinks that 

 Fitch's species '' ought, as it seems, to be referred to the subgenus ^.s- 

 ^Aonrfy^i'a," which has no verticils at all S 9 • (-£^?p^.-A^.^4.pp.7andl76.) 

 But Fitch refers his species to Cecidomi/ia, which he would scarcely have 

 done if it had been totally without verticils, unless he had at the same 

 time stated the fact of there being no verticils. I suspect it is a Dlplo- 

 six, and that the $ only was known to the describer, who says not a 

 word about the sexes in his description. 



The subgenus Diplosls is circumscribed as having " 26-jointed % an- 

 tennae with sometimes one additional rudimental joint;" (^Dipt. JV. A. 

 p. 176;) but from cai'efully examining the dried specimens, I am pretty 

 well satisfied that in atrocularis, as well as in scptcm-maculata n. sp.. 

 the antennae S are only 2o — 25-jointed. Since in the subgenus Cerldo- 

 myia the number of antennal joints % is confessedly very inconstant, not 

 only diflFering in different species, but varying even in the same species, 

 and actually in the right and left antenna of the same individual, it 

 seems but agreeable to what I have called the Law of P^quable Varia- 

 bility, that it should be somewhat similarly inconstant in the S of the 

 allied subgenus Dlplosis. The same observations apply in a less de- 

 gree to the 9 antenna, which, as stated in the description, is in atrocu- 

 laris properly speaking 13-jointed, though it is limited subgenerically 

 as being •• 14-jointed with sometimes one additional rudimental joint." 

 The number of joints being so very much smaller in 9 than in % Dl- 

 plosis, we cannot expect to find the range of variation so extensive in 

 the 9 as in the % . (See above pp. 556-7.) " The number of the joints 

 of the antennae," says Loew, " is of higher value among the Gall- 

 gnats, for the distinction of species, than for that of genera, since almost 

 every genus comprises species with different numbers of joints of the 

 nntennse." {Dipt. N. A. p. 179.) We see the same thing in Ct/ni- 

 pidse. {P. E. S. P. II. pp. 460-1.) 



E. D. ATRICORNIS n. sp. (Dried.) % Differs from % of atrocularU 

 only as follows : — l.s^ The antennae are twice as long, instead of half 

 as long again as the dried body, conspicuously stouter, about 24-jointed. 



