MUSEUM OF COMPARATIVE ZOOLOGY. 159 



pigment-cells, all the cells in an ommatidium are ectodermic in origin ; 

 the accessory pigment-cells are probably derived from the mesoderm. 

 Of these five kinds of cells, the granular bodies probably do not repre- 

 sent the accessory pigment-cells, for in fully developed eyes the latter 

 lie on both the distal and proximal sides of the basement membrane, 

 whereas the granular bodies are found only on the distal side of that 

 structure. The granular bodies, then, more likely represent one of the 

 four remaining elements, all of which naturally occur only on the distal 

 side of the membrane. It is not probable that the granular bodies 

 represent the cells of the corneal hypodermis, for these produce the cu- 

 ticula of the retinal region, and if they have any representatives, those 

 representatives must be the distal layer of unmodified hypodermal cells 

 already indicated in the retinal thickening. The position of the granular 

 bodies, therefore, precludes their representing corneal hypodermis. If 

 then the granular bodies are not accessory pigment-cells nor corneal 

 hypodermis, they must be either distal or proximal retinuleo or cone- 

 cells. In a previous paper I have given reasons for considering the 

 proximal and distal retinuke as both originating from a commou group 

 of cells, the retinuke. These are essentially sensory in function, as con- 

 trasted with the cone-cells, which are merely dioptric. The question 

 then narrows itself to this : Are the granular masses clusters of dioptric 

 cone-cells or sensory retinulse 1 



In determining to which of these two groups of cells the granular 

 masses belong, the relation which the latter sustain to the fibres of 

 the optic nerve would doubtless be of great importance, for the nerve 

 fibres in fully developed eyes are known to terminate in the retinuke, 

 not in the cone-cells. Unfortunately, the histological condition of my 

 material was such as to preclude the possibility of determining this 

 question. 



The fact that each granular mass contains several nuclei clearly indi- 

 cates that it consists of several cells. The number of cells in each 

 mass, judging from the number of nuclei, varies from one to about five, 

 the more usual number being three or four. When one compares the 

 condition of intimate fusion which the cells of each mass present with 

 the normal condition of the retinulse and cone-cells, the masses must 

 certainly be admitted to resemble more closely the cone-cells. More- 

 over, the number of cells in each mass, although variable, is nearer to 

 that of the closely united cone-cells than to that of the retinulae. Not 

 only do the number of cells involved and the intimacy of their fusion 

 favor the idea that each mass represents a degenerate cone, but the 



