238 O. C. GLASER. 



only way in which such a change can conceivably take place 

 seems to me to be by a very rapid increase in the number or the 

 size of the cells. An increase of the former sort can very readily 

 be noticed during the period when the spindle-shaped cells of the 

 early pre-cannibals (Fig. 1) change to the "cuboidal" cells of 

 the young cannibals (Fig. 2). If a similar increase takes place 

 when the fully gorged cannibals transform into veligers one 

 should be able to, find histological evidence of it. While the 

 original transformation may be accounted for on the basis of 17 

 mitoses per 339 nuclei, the second transformation, if mitosis is 

 the only method of division in normal cell differentiation, must 

 be accounted for on the basis of 1 mitosis in 1,751 nuclei. This 

 single case is absolutely the only indication of mitosis, that care- 

 ful and frequently repeated search through stages IV. and V. has 

 revealed. On the other hand I found in the same sections 1 1 1 

 cases of what I interpret as undoubted instances of amitosis. If 

 the amitoses do not account for the increase of cells needed to 

 explain the change from the spindle to the cuboidal shape, I 

 doubt very much if the all but total absence of mitosis accounts 

 for the facts. It is necessary to conclude therefore, either that 

 the only form of division seen in any quantity is responsible for 

 the assumed increase in cells, or that these enlarge and become 

 comparatively crowded. 



The crucial question then is, which of these two explanations 

 is correct ? Do the cells become crowded because they increase 

 in size, or in number or for both reasons ? The drawings show 

 plainly that the dorsal cells do increase in size during the meta- 

 morphosis ; they also show that this is not true of the ventral 

 cells. In addition to the enlargement of the liver cells it can be 

 shown that actual increase in the number of cells present, is a 

 factor bringing about the change from spindle-shaped to cuboidal 

 cells. 



Absolutely faultless series, or strictly comparable sections are 

 in several cases unavailable. Of stage I. for example, I have no 

 transverse sections, but a study of the differences between the 

 long axis and the short one of this larva makes 20 cells in a 

 transverse section midway between the extremities of the longer 

 axis, a safe estimate (see Fig. 2 and stage I., Fig. 9). In stage 



