MUSEUM OF COMPARATIVE ZOOLOGY. 123 



of five elements is quite as close as that between the cells in cones con- 

 taining only four elements, I believe that the additional cell, which has 

 increased the number of segments from four to five, has been derived by 

 the division of one of the original four cone cells, and not from an extra- 

 ommatidial source. 



Another instance of this kind occurs among the Isopods. The cones 

 in this group, it will be remembered, are usually each composed of two 

 segments. According to Beddard's figures ('90, Plate XXXI. Figs. 1 

 and 4) in Arcturus, hovrever, they occasionally consist of three segments, 

 and in Asellus aquations, according to Sars ('G7, p. 110), although three 

 of the four cones in each eye are composed of only two segments each, 

 the fourth regularly contains three. The size of the segments in the 

 fourth cone differs ; two are small, and together their bulk about equals 

 that of the third, and the last is approximately of the size of a segment 

 in one of the other cones. If we attempt to explain the condition of the 

 cone composed of three segments by supposing it to have been produced 

 by adding to the normal pair of cone cells a single cell from some source 

 external to the ommatidium, we are met with the difficulty, that what 

 is apparently the added cell — the larger one — resembles more closely 

 a segment in the other cones than do either of the two remaining cells, 

 although the latter must on this assumption represent the original seg- 

 ments. If, however, we imagine the small segments to have arisen by 

 the division of a single larger one similar to the large one which I'emains 

 in the cone, the relation of the resulting segments both in size and num- 

 ber is a perfectly natural one. This explanation} therefore, seems to me 

 to be more probable than the former. For these reasons, I believe that 

 an increase in the number of cells in an ommatidium takes place by the 

 division of the cells already forming a part of that ommatidium, rather 

 than by the importation of new elements hitherto foreign to the om- 

 matidium. 



The conclusion which I would draw from the preceding discussion is, 

 that there are at least three means of modifying the numerical formulae 

 of ommatidia, all of which involve only the cells primitively belonging 

 to the ommatidium, and therefore do not necessitate the introduction 

 of new cells from extra-ommatidial sources. They are cell diff"erentia- 

 tion, cell suppression, and cell multiplication. 



Having now determined the means by which the cellular structure of 

 the ommatidia in living Crustaceans is modified, we are j5repared to ap- 

 proach the question of the structure of the primitive ommatidium. If 

 it could be shown that ommatidia were modified only by increasing the 



