318 S. WATASfJ. 



rods witli !i cone in tlie centre are the two essential factors in the 

 neurot'pitlioliiil layer of the human retina. In the absence of 

 enouixli comparative data in Arthropods at present we have to 

 dwell largely on the analogy suggested in tlie other gron])s of 

 animals. Whatever be the views as to the fundamental homology 

 of the ommatidium of Limulus to a structural unit of the sensory 

 part of tiio human retina, a superficial rescinblanco of the one to 

 the other is certainly very strong. The structural reseml)lance 

 is paralleled by a physiological one. The place where tlie light 

 acts in the visual end-organ of Arthropods and of man may 

 alike be considered as consisting of a number of definite groups 

 of cells, each group being a morpiiological and a physiological unit ; 

 or, in other words, the sensory part of the retinae in both cases 

 consists of a mosaic of several sensitive spots. The image formed 

 on such a surface is therefore a mosaic one, whether in an 

 Arthropod or in a Vertebrate. 



Fundamental as this arrangement appears to be in the human 

 retina, these two factors are liable to variation in their I'elative 

 distribution in diflerent Vertebrates. In fact, the variation takes 

 place between the two extremes where the rods alone exist on 

 one hand and where the cones alone constitute the essential part 

 of the retina on the other. Thus, according to Schultze, "either 

 form of percipient element (rod and c-one) may be represented by 

 the other " in the Vertebrate. This range of variability in the 

 distribution of the cones and rods occurs even in a single group 

 of Vertebrates, as in mammalia, sliowing that the variation in 

 the distribution of the essential factors, even within a tolerably 

 well circumscribed group of animals, is sometimes quite exten- 

 sive. The group of Arthropods is a heterogeneous one, and I 

 see no a priori objection to believing in the existence of a phe- 

 nomenon analogous to what we find in Vertebrates, viz. that the 

 two percipient elements represented by the central and the 

 peripheral cells in the ommatidium of Limulus may be difler- 

 ently represented in diflerent Arthropods. 



There is no doubt whatever that the retinula cells are homo- 

 logous throughout the Arthropods. In fact, in most Artiiropods 

 which I have examined no other elements but the retinulae 

 have any connection with the optic nerve fibres, and they often 

 undergo an enormous development and acquire most complicated 



