MUSEUM OF COMPARATIVE ZOOLOGY. 51 



Suppose the individual ommatidia of the three rows to be arranged in 

 reference to one another as the four in Figure 55 ; i. e. the ommatidium 

 of one row covering the open spaces between two ommatidia in the 

 adjoining row. Under these conditions hexagonal facets would be pro- 

 duced. But now imagine the middle row to move in the direction of 

 the arrow through the distance of half the thickness of an ommatidium. 

 After the completion of this movement, the ommatidia of the middle 

 row are directly opposite the ommatidia of the adjacent rows. With 

 this arrangement the ommatidia can be grouped in square blocks of 

 four, nine, etc. This is the grouping which obtains in the adult retina, 

 and the facets resulting from it are square in outline. In the more 

 primitive arrangement, that with hexagonal facets, the ommatidia can 

 also be gi'ouped in blocks of four, nine, etc. These blocks, however, 

 are never square in outline, but lozenge-shaped (Fig. 55). 



The process of rearranging the ommatidia is accomplished at a period 

 in the growth of the young lobster much later than that at which the 

 nerve-fibres have arranged themselves in relation to the openings in 

 the basement membrane. Such being the case, one would expect to 

 find in the deeper part of the adult retina traces of the more primitive 

 arrangement. This is seen, I believe, in the lozenge-shaped outline 

 which grorps of four ommatidia present in the deeper part of the retina. 

 A good instance of this is to be seen in Figure 14, although it is ap- 

 parent in almost all of the transverse sections which include the proxi- 

 mal retinulffi. One might say that the ommatidia were rooted to the 

 basement membrane when the hexagonal system prevailed, and that the 

 rearrangement fully affected only their free distal ends. 



The movement suggested as a means of rearranging the ommatidia 

 not only explains the new position of the ommatidia, but also accounts 

 for the situation in which the nuclei of the distal retinulae occur. In 

 Figure 55 each ommatidium is surrounded by a circle of six nuclei. 

 These belong to the distal retinulse. Instead of describing them as 

 being in circles of six, it might be said that there is one nucleus at each 

 corner of every cone. Thus, cone x (Fig. 55) has nuclei 1, 2, 3, and 4 

 at its four corners, and cone y has in a corresponding way nuclei 5, 6, 7, 

 and 8. If the cones were to move as I have already described, and 

 were to caiTy their surrounding nuclei with them, the result would be 

 that nucleus 5 would come to lie next to nucleus 2, and 7 next to 4, and 

 so on. In other words, a pair of nuclei would occupy each space be- 

 tween the adjoining angles of four adjacent cones. This is the position 

 which the nuclei of the distal retinulsR occupy in the retina of an adult 

 lobster (Fig. 5). 



