armatures in the Hymenopterous genus Colhtcs. 81 



In certain cases (cf. my last figures on PI. VIII) a very 

 singular appearance is given to this segment by a tendency 

 which the lateral thickened portions or " costte " show to 

 dctacli themselves from the lobes and stand up as separate 

 tooth-like or spine-like processes. This leaves the lobes 

 comparatively unsupported (cf. fig. 83«), or they may even 

 disappear almost entirely, so that only a sort of skeleton of 

 the segment remains to represent it (fig. 35rt). One can 

 trace, I think, a regular gradation in this respect — the 

 likeness of the modified segment to the simple semi- 

 anmdus, from which it must have developed, decreasing 

 stage by stage, till it vanishes altogether. 



It is curious to note how in different genera of the 

 Anthophila nature seems to select different ventral 

 segments of the $ for pamdoxical specific modification — 

 the sixth in certain Osmias, the seventh in Colletes, the 

 eighth in Andrena, both the seventh and the eighth in 

 Prosopis — while in other cases none of these segments are 

 particularly characteristic, and it is tlie armature itself 

 wliicli suppUes the best specific characters (e.g. in Bombus 

 and Sphecodes). I have already noted that these modifica- 

 tions are not confined wholly to the concealed portions of 

 the abdomen ; and indeed I suspect that its entire structure 

 (the emhoitement of all its segments, the development of 

 ventral tubercles on some of them, the usual acumination 

 and occasional peculiar armature of its apex, etc.) depends 

 a good deal on the same sort of causes, which have diverted 

 the concealed ventral segments from their original function 

 as a part of the insect's integument, and made them a sort 

 of appendage to the armatura copulatrix. 



We come now to the " armature " proper, which com- 

 prises at least three obviously distinct portions, viz. its 

 cushion-like base — the cardo, and two objects, each 

 resembling a pair of forcipes, one within the other — the 

 outer of which (following Thomson and Saunders) I call 

 the stipitcs, and the inner the sagiltie. 



The cardo has little character, and need not detain our 

 attention ; but the stipitcs and sagittx are very differently 

 formed in different species, and undoubtedly give to some 

 armatures a peculiar and easily recognizable facies. Yet 

 it is often not easy to define exactly the characters on 

 which the peculiarity depends. This is partly because 

 these organs, being practically very irregular figures of 

 three dimensions, completely alter their appearance, when 



