97 



are only partially connected, being free at the apex 

 though firmly attached towards the scutellum. In 

 every instance, however, even where they are united 

 throughout their entire length, a little force will succeed 

 in separating them, showing their structure, as I have 

 indicated in the diagnosis, to be sub-connate rather than 

 connate. But that it does require force to effect the dis- 

 junction, when they are really in the condition described, 

 is proved to a demonstration to any one who has seen 

 the remains of the insect beneath the slabs of stone on 

 many of the small adjacent islands where it most 

 abounds, or drifting about over the surface of the rocks, 

 under which circumstances I have observed them in 

 immense numbers, apparently the accumulation of two 

 or three generations, which the violence of the elements 

 had not been able to sever. It is rare in the sylvan 

 districts to find them joined ; nevertheless such is some- 

 times the case, thus proving that the peculiarity is not 

 actually essential, but merely one which it is the ten- 

 dency of the species to assume, and which is more 

 developed in some specimens, and under certain condi- 

 tions, than in others.* ' 



But by far the greatest amount of variability to which 

 insect structure is liable, is presented by the wings, 

 especially the metathoracic ones. The wings, indeed, 

 unless I am much mistaken, are essentially (as compared 

 with other primary details) organs of variation, capable 

 of being more or less developed, according as the several 



* Insecta Maderensia, pp. 56, 67. 



F 



