56 BULLETIN OF THE 



cells. I believe they represent the nuclei of the corneal hypodermis. 

 The retina proper is probably separated from this hypodermis by delami- 

 uation ; at least, the corneo-conal membrane is formed at a stage slightly 

 older than that last mentioned, and, judging from the appearances at 

 this stage, its formation is not accompanied by any folding of the hypo- 

 dermis or retina, but is the result of a differentiation in place. Unfor- 

 tunately, none of the specimens which I studied showed any steps in the 

 formation of the corneo-conal membrane, and I am therefore uncertain as 

 to the exact method of its growth. 



Of the two membranes in the basal portion of the retina of Gammarus, 

 presumably only one corresponds to the basement membrane of other 

 Crustaceans. The position occupied by the two membranes, as well as 

 their structure, serves to indicate which is the true basement membrane. 

 At first sight one might suppose that the capsular membrane, at least 

 in its proximal portion, corresponds to the basement membrane, but this 

 interpretation is not probable, for the reason that the capsular mem- 

 brane is not pierced by the fibres of the optic nerve, a characteristic of 

 the true basement membrane of the eye. I therefore believe that the 

 intercepting membrane, since it is perforated by these fibres, is the homo- 

 logue of the basement membrane, and that that portion of the capsular 

 membrane which might be regarded as a basement membrane is in 

 reality merely the cuticular sheath of the optic nerve. 



So far as I can foresee, the only objection to be urged against this 

 interpretation of the intercepting membrane is found in its relation to 

 the retinular nuclei. These nuclei in the eyes of almost all other Crusta- 

 ceans lie on the distal side of the basement membrane. Granting that 

 the intercepting membrane is the basement membrane, one must admit 

 that in Amphipods they lie on the proximal side of this membrane. 

 This admission might at first sight appear to offer an obstacle to the 

 homology which I have suggested ; but it can be made with consistency, 

 I believe, provided one can show that the position of the retinular nuclei 

 is not necessarily fixed. That such is the case is evident from the fol- 

 lowing facts. In Decapods the retinular nuclei usually occupy a position 

 in their cells distal to the rhabdome. In Porcellio, as Grenadier ('79, 

 Taf. IX. Fig. 96) has shown, they have a more proximal position, lying 

 in the same transverse plane as the rhabdome itself. In Serolis they 

 are midway between the rhabdome and the basement membrane. These 

 conditions show, I believe, that the retinular nuclei ma}^ occupy very 

 different positions in their cells, and that the step from the condition 

 shown in Decapods to that shown in Serolis is not greater than that 



