379 



optic disc, so that a very distinct basement membrane can 

 now be seen underlying it. This membrane connects the 

 internal ends of adjacent pigment cells, from which it has 

 doubtless been secreted. These pigment cells surround the 

 ommatidia fairly closely, but are not in direct contact with 

 them, and they secrete the- basement membrane in such a way 

 that a hole is left at the base of each ommatidium, thus per- 

 mitting the easy entrance of a nerve towards the rhabdome 

 cell. This basement membrane is the fenestrate membrane 

 of the eye, and will be referred to as such hereafter. It 

 undergoes scarcely any visible change during the rest of larval 

 and pupal life. 



The cells of the inflected ectodermal layer (which may, 

 for convenience, be spoken of as the p'riopfic memhrane, to 

 show its relation to Hickson's Periopticon) now undergo a 

 remarkable process of branching (fig. 74), the branches being 

 of three types : (1) those which unite the cells with similar 

 processes from other cells of this layer, (2) those which connect 

 the cells with the fenestrate membrane, and (3) those which 

 grow in towards the brain. 



The cells themselves are large, with a distinct but irregular 

 nucleus; the branches which connect neighbouring cells 

 together are not very numerous. The second type of process 

 is very remarkable, and is seen to join up, each, with the 

 base of a group of pigment cells, and several such processes 

 may be seen coming from a single cell of the perioptic layer. 

 Since these processes thus fuse in reality with the circum- 

 ferences of the holes in the fenestrate membrane at the bases 

 of the rhabdomes it would seem probable that they are 

 hollow; later events show that this must be so. Tlie third 

 type of branching is also very remarkable ; this consists 

 essentially of a great massive outgrowth of fibres towards the 

 adjacent cortex of the brain. This great fibrillar mass from 

 the inner side of the cells of the perioptic. layer now enters 

 the more ventral portion of the brain (having apparently 

 broken away its own somatopleure and the splanchnopleure 

 of the brain) and gradually terminates amongst the cortical 

 cells comprising it. At the same time these cortical cells 

 become active, and, dividing mitotically, begin to proliferate 

 and to migrate outwards in the meshwork of fibrillae, towards 

 the optic disc. 



^Tlie function of the perioptic membrane is thus to form 

 a kin^'^f neuroglia to support the nerve cells of the optic 

 ganglion ; but it seems to have a second function, namely, to 

 act as what must essentially be regarded as a neurolemma^. 

 From the above description it follows that no nerve fibre can 

 penetrate to the rhabdomes unless it can enter the cavity of 

 l2 



