1895.] NATURAL SCIENCES OF PHILADELPHIA. 163 



densely stained, and while still showing a paired condition are no 

 longer in contact, and no longer have their adjacent faces flattened 

 as when united, as in the preceding section. One may now count 

 eight cone cells arranged in four groups, of two each, around a 

 single very distinct rod, Fig. 2. This arrangement is at first some- 



Fig. 2. 





what difficult to make out, but it requires only a short time for the 

 observer to convince himself that a regular arrangement of the rods 

 and cones also obtains at this lower level. This evidence conclu- 

 sively establishes the fact that every one to five rod- cells in the eye 

 of a larval salmon are surrounded by eight cone-cells that are 

 closely opposed or united at their apices into four pairs of two each. 

 At a slightly deeper level the bodies of the pairs of cones separate 

 still more, but traces of a regular pattern can still be made out at 

 a third level. 



I have stated in the title of this paper that there is an arrange- 

 ment of retinal cells of fishes that partially simulates that of the 

 retinal elements in the compound eyes of Arthropods. This definite 

 proviso is necessary because the comparison between the groups 

 consisting of four or five rod-cells and four double cones in the 

 retina of the salmon cannot be exactly homologized with the cell- 

 groups known as ommatidia in the Arthropod eye. In the latter each 

 ommatidium is distinct from its adjacent fellows and none of its cellu- 

 lar elements enter into the formation of the contiguous ommatidia by 

 which it is encircled. In the retina of the larval salmon the case is 

 very different and quite peculiar. Of any one group consisting of 

 five central rod-cells and four double cone- cells, the latter also enter 

 into the formation and form part of eight adjacent similar groups of 

 cells. Into four of the eight adjacent groups two double cone-cells 

 enter, and, alternating with these are four other groups into which 

 only a single pair of the double cone-cells of the central group enters. 

 The comparison is therefore only a partial one. It is obviously absurd 



