52 J. H. Powers 



terest has been in tracing them to their causes, which prove to b<' . 

 so simple and uniform in character. 



VARIATIONS CAUSED BY CANNIBALISM 



I now wish to speak of an instance of variation in this species 

 which affects not one organ only, but many, if not all the organs 

 of the body, and which would be interesting for its amount alone, 

 even if its cause were unknown. I am able, however, after much 

 search and experiment, to demonstrate its cause, and thus to 

 confirm and cast much light upon the variations discussed hith- 

 erto. I have already referred in some sense to this variation, in 

 that I included it in the general estimates of total variation at the 

 Beginning of the paper, and in that the adult profile in figure 3, 

 plate III, is to be classed partially with animals now to be dis- 

 cussed. But the type now to be spoken of has not previously 

 been considered. I will again introduce it, not by measurements 

 and statistics, but by a direct reference to plates VII, VIII, and 

 IX. Figures i and 2, plate VII, show profile and dorsal views 

 of at remarkable larva (unfortunately not very well preserved, 

 having been rolled in a cloth), which I secured several years ago. 

 Twelve other specimens secured from the same source in the 

 same season resembled this specimen in varying degree, two or 

 three being nearly identical. I had no idea at that time of the 

 rarity of such specimens ; but since then I have tried in vain to 

 duplicate them in all points of peculiarity. A glance at the fig- 

 ures will show the remarkable character of the specimen, espe- 

 cially when compared with figure 3, which represents a typical 

 larva of the same length and somewdiat similar preservation. I 

 need hardly call attention to the prodigious width of the head, or 

 to the distance between the orbits of the eyes (still greater, pro- 

 portionately, in some other specimens than in the one figured). 

 or to the length of the head, one-half greater than that of many 

 normal larvae of the same total length. It is more necessary to 

 point out the fact that the deviation here from the normal type of 

 amblystoma is in reality greater than a casual glance would indi- 

 cate, the skeletal proportions being modified much more than the 



248 



