MUSEUM OF COMPARATIVE ZOOLOGY. 23 



tidium c, these three groups are represented by retinulae 1 and 2, 3 and 

 4, and 5 and 6. The fourth group is composed of the single retinula 

 number 7. 



The four groups thus defined are identical with the ones which pass 

 through the four openings in the basement membrane. In Figure 21 

 the four openings which surround the cross (x) can be designated from 

 their positions as the upper, lower, right-hand, and left-hand openings. 

 The upper and lower openings each present the transverse section of 

 four large fibres. The right- and left-hand openings are each occupied 

 by three large fibres. The source of these fibres can be ascertained by 

 comparing Figures 20 and 21. In ommatidium c (Fig. 20) retinula? 

 5 and 6 unite with retinulse 1 and 2 of ommatidium d, and thus consti- 

 tute the four fibres which pass through the upper aperture (Fig. 21). 

 In a similar way, retinulse 1 and 2 of ommatidium c unite with 5 and 6 

 of an ommatidium which lies below c, and pass as four fibres through 

 the lower opening (Fig. 21). Eetinulee 3 and 4 of ommatidium c 

 (Fig. 20) unite with retinula 7 of an ommatidium to the left of c, and 

 emerge as three fibres through the left-hand opening (Fig. 21). Retin- 

 ulse 7 of c, and 3 and 4 of b (Fig. 20), unite and form the three large 

 fibres of the right-hand opening (Fig. 21). This plan of distribution is 

 repeated in each ommatidium, and thus brings about the groups of 

 three or four fibres which occur in each opening through the basement 

 membrane. The groups of fibres are distinguishable for only a very 

 short distance below the basement membrane. The individual fibres of 

 each group soon separate, and in the deeper part of the optic nerve they 

 never again present this grouping. The description of the termination 

 of the fibres will be deferred to a later part of this paper. 



The relation of the rhabdome to the cone-cells in Homarus has al- 

 ready been described. That they are separate structures, as Schultze 

 and Grenacher have asserted, I believe there can be no doubt. I have 

 seen nothing which favors the view held by Patten, namely, that the 

 rhabdome is an enlargement in the proximal part of the cone-cells. In 

 Homarus the rhabdome has the general form of a spindle. In trans- 

 verse section, however, it is not circular, but square. Its four sides are 

 thrown into ridges, the crests of which extend across the sides at right 

 angles to its longest axis. The inner face of each proximal retinula is 

 thrown into corresponding undulations. The rhabdome and retinula 

 are so adjusted to each other that a crest on one fits into a furrow 

 on the other. (For a similar condition in Penasns, compare Patten, 

 '86, Fig. 72.) The retinulse and rhabdome are thus intimately bound 



