26 B. GRASSI AND A. SANDIAS. 



development^ but may present characters peculiar to themselves, 

 and differing from those of the perfect insects (e. g. the long 

 outstanding abdominal setse and black maculation in Termes). 

 Special and still more pronounced characteristics are exhi- 

 bited by the other larvse of arrested growth, that is the castes 

 of soldiers and workers. The proof just given that one, the 

 neoteinic caste, is dependent on nutrition, is enough to suggest 

 on a priori grounds that this may be equally true of the 

 others. 



Strictly a priori the soldiers may be regarded as further 

 differentiated workers, and indeed at the beginning of their 

 development they exhibit but a single known characteristic, 

 the enlargement of the head as in the worker. The soldier 

 therefore begins by possessing the characters of the 

 worker. In other words, we may say that an individual 

 destined originally to become a perfect insect is capable of 

 undergoing a paranomalous development, of acquiring 

 certain characteristics which, by absolutely destroying its 

 fertility, no longer allow of its becoming a perfect insect. If 

 the differentiation of the head is arrested when certain 

 structural modifications have been acquired (simple increase 

 in its size, a special form of pronotum, &c.), and the individual 

 is limited further to uniform growth in stature, except for the 

 augmentation in the number of antennal joints, we have a 

 worker. But if, instead of this limitation, the mandibles and 

 labrum become elongated while the maxillse and labium do not 

 alter, then we have a soldier. In short, worker and soldier 

 follow a common path for some distance ; at a given point 

 one, the worker, continues along it, while the other, the soldier, 

 diverges from it. And with this are connected the facts that 

 young soldiers and workers are indistinguishable, and that 

 certain Termitidse (Calotermes) possess soldiers, and others 

 (Anoplotermes) workers only. 



But these inductions all require the control and support of 

 direct observations ; and mine indeed are not quite complete, 

 as I have been unable to keep Termes alive for a sufficient 

 time. 



