MUSEUM OF COMPARATIVE ZOOLOGY. 57 



the division and differentiation of cells. The two cone-cells in the 

 simpler type would remain unaffected in Mysis ; in the Decapod they 

 would divide once, thus producing a cone of four cells. By division, 

 the five retinulse of the Amphipod would become the ten retinulse of 

 the higher type. These ten retinulse are differentiated into two sets, 

 eight proximal retinulse which surround the rhabdome, seven of them 

 possessing nerve-fibres, and two distal retinulse which envelop the cone, 

 but have no nervous connections. If the ten retinulse of the higher 

 type were developed from the five retinulse of the lower type, it follows, 

 since all of the five retinulse possessed nerve-fibres, that those of the 

 ten retinulse which do not possess nerve-fibres must be considered as 

 degenerate. The degenerate cells of the higher type of ommatidium 

 are consequently the eighth proximal retinula and the two distal retin- 

 ulse. The structural condition of these three cells favors this view. 

 The eighth proximal i*etinula is identifiable ouly through its nucleus. 

 Each distal retinula, as I have previously described, possesses a proxi- 

 mal fibre. This fibre passes through the basement membrane with the 

 nerve-fibres of the proximal retinulse, but it is very much smaller than 

 these fibres, and terminates without reaching the optic ganglion. This 

 condition is easily interpreted as one of degeneracy. 



The process by which distal and proximal retinulse were probably differ- 

 entiated from unmodified retinulse is easily suggested. A characteristic 

 distinction between the ommatidia of the higher and lower Crustacea, 

 as, for instance, between Gammarus and Homarus, is that in the former 

 the cone and rhabdome are very close to each other, whereas in the 

 latter the cone proper and rhabdome are separated by considerable 

 space. Without attempting to assign physiological reasons, the state- 

 ment may be made that it seems necessary that the sides of the cone 

 should be sheathed with pigment. In the ommatidia of Gammarus this 

 is accomplished by the distal ends of the five retinulse ; these reach be- 

 yond the rhabdome and envelop the cone. They thus form a funnel, in 

 the large central cavity of which the cone rests, while the neck of the 

 funnel is occupied by the rhabdome. Imagine now the division of 

 these five retinulse into ten, and the separation of the cone from the 

 rhabdome. It is easy to understand how two of the ten retinulse may 

 retain their connection with the cone, while the remaining eight adhere 

 to the rhabdome. The two retinulse connected with the cone would lose 

 their nervous function and become simple pigment-cells. The retinulse 

 which remain attached to the rhabdome would continue as perceptive 

 cells. The separation of the rhabdome and cone not only offers an 



