294 



one plant species, or a few closely allied plants, and the resultant gall 

 is characteristic of the insect. Certain species that live in resin of 

 trees have the lateral abdominal spiracles vestigial and the apical pair 

 very much enlarged. 



Pupa. — It is not a simple matter to indicate characters by means 

 of which the pupae of this family may be distinguished from those of 

 allied families because there are within its limits species that differ 

 greatly in their larval habits, some living in hard, woody galls, and 

 others in decaying fungi, manure, mines in leaves, or under bark, the 

 nature of the material in which the pupa is found having a consider- 

 able influence upon the structure of the head. I have found it true in 

 other groups that when the species have in the pupal stage to force 

 their way through either hard earth, wood, or the enclosing cocoons 

 of some host, the head-capsule of the pupae is armed with a number 

 of sharp thorns well suited to the purpose of boring a way out so that 

 the imago may be able to emerge. On the other hand, species that 

 either do not penetrate deep into soil or wood, that are aquatic, or nor- 

 mally frequent moist ground, do not have cephalic thorns. From my 

 limited data this rule appears to hold good in the Cecidomyiidae, the 

 species that must bore their way out of galls having the head armed 

 with sharp thorns, while those that are found in fungi, manure, or 

 decaying wood have no such armature. According to my present data, 

 the presence of cephalic thorns may be depended upon to separate the 

 species possessing them from other Nematocera, no other family of 

 which has such thorns, but the species without such thorns are diffi- 

 cult to separate from those of Sciaridae. The comparative sizes of 

 the first and second tarsal joints of Cecidomyiidae serve to separate 

 the pupae of most of this family from those of Sciaridae, the basal 

 one being much shorter than the second in the former while in the 

 latter it is distinctly longer. 



Imago. — The antennae differ considerably in the number of their 

 segments, ranging from 1 6 to 30 in different genera, and not infre- 

 quently differing in the sexes of the same species. Usually the an- 

 tennal segmentation is very distinct, the segments being often nodose, 

 or even binodose, and armed with whorls of long hairs. In 

 Cecidomyiinae looped hairs which have no free end are present on the 

 antennae, being undoubtedly sense organs. The legs are usuallv long 

 and slender, and in some groups the basal joint of the tarsi is verv 

 much shorter than the second. The wings are characterized bv the 

 small number of the longitudinal veins, and the absence of cross-veins 

 except near base; the surface is usually pubescent or covered with de- 

 pressed, scale-like hairs. 



