380 



the processes which are attached to the fenestrate membrane. 

 I was unable to observe how the nerve fibre penetrates to 

 this portion, i.e., whether it forces its way through the cyto- 

 plasm of the cells of the perioptic layer, or whether, 

 entangled as it is in the fibrillae of the cells, it works its way 

 just below the cell membrane, and passing round to the 

 opposite side, enters the process to the fenestrate membrane. 

 It is certain, however, that a fibre from a nerve cell does 

 eventually work its way into one of these processes. This is 

 clearly shown in fig. 74, where the nerve cell gives off a long 

 fibre which communicates with the ommatidium, and is 

 entirely enveloped by the cell process which meets the fene- 

 strate membrane. TTie cells of the perioptic membrane must, 

 therefore, be regarded as functioning, also as neurolemmae. 

 Moreover, it follows that, as a single cell gives off processes 

 towards a number of ommatidia, a single perioptic cell must 

 act as neurolemma for a number of distinct nerve fibres. I 

 have not been able to see distinct instances of this in my 

 preparations, partly because, in pupae a little older, when 

 this process has been completed, the cells of the perioptic 

 membrane have cohered closely together, making further 

 observations on this point impossible; but the fact that all 

 the ommatidia, several of which were supplied with processes 

 from a single perioptic cell, later have nerves entering them, 

 leaves us no alternative but to accept this view (compare, 

 however, fig. 75). The nerve cells which have entered the 

 perioptic membrane can often be seen to give off a distinct 

 process backwards towards the brain; but I was quite unable 

 to trace any of these fibres to their termination. 



The coherence of the cells of the perioptic membrane, 

 which takes place in pupae a few hours later, gives the struc- 

 ture a much firmer appearance, the loose branching network 

 of the pupating larva being transformed into a fairly thick 

 pavement membrane. At this stage leucocytes are occasion- 

 ally seen between the perioptic membrane and the optic disc. 

 What their function is I am not able to say. 



When pupae about twenty-four hours old are examined, a 

 further development of the perioptic membrane is seen to have 

 taken place. The ' 'neurolemmal" processes are no longer visible ; 

 probably the best interpretation which can be placed on this 

 is that their disappearance is only apparent, and that they 

 have now assumed their true function as neurolemmae and 

 have closely enveloped their respective nerve fibres (fig. 75). 

 These fibres can be seen communicating with every omma- 

 tidium, but they are so exceedingly minute that the non- 

 appearance of a neurolemma as distinct from the nerve is 

 only to be expected. I am also unable to say where the nerve 



