226 
SYDNEY J. HICKSON. 
upon this group into a separate paper ; but I have said 
sufficient to show the relation that exists between the peri- 
opticon and terminal anastomosis of Crustacea and the homo- 
logous structures of the Hexapoda. 
Before leaving this part of my subject I must refer to some 
observations I have made upon the development of these 
structures in the bee. 
In a young bee, some time before it emerges from the cell, 
the fibrils passing from the epi-opticon to the peri-opticon do 
not cross one another (fig. 6, Nf.), but the decussation takes 
place at a later stage in a manner I have been unable to follow. 
I imagine, however, that is due to a shifting of the position of 
the fibrils at their origin from the epi-opticon, in a manner 
somewhat similar to that which occurs in Vertebrata when the 
roots of the posterior spinal nerve shift from the neural crest 
to a lateral position. 
At first the fibrils pass straight from the epi-opticon to the 
eye, but an anastomosis soon takes place, and the limits of the 
peri-opticon are indicated by two rows of nerve-cells, by the 
presence of well-marked transverse and looped anastomosing 
fibrillae, and by a thickening of many of the original fibrils 
(fig. 6). As development proceeds the thickened fibrils split 
up into masses of neurospongium, joined together by the trans- 
verse anastomosing fibrillae, and each of these forms one of the 
elements of which the peri-opticon of the adult is composed. 
Before the bee is fully developed these elements stand some 
little distance apart from one another, as they do in Musca, 
but the spaces between them are gradually filled up by the 
development of new elements and the growth of those already 
formed. 
To recapitulate, then, I have shown that in the young Peri- 
planeta the optic nerve-fibrils which leave the peri-opticon pass 
without decussating to the ommateum; in the adult Periplaneta 
there is a partial decussation ; and that in Nepa there is no 
decussation, but the anastomosis is complicated by the presence 
of looped and transverse anastomoses. In Musca the fibrils are 
split up into little cylindrical blocks of neurospongium, which 
