6o /. H. Pozvcrs 



and show the beginnings of cannibahstic development. But re- 

 lapse was almost certain to follow, a relapse not into slow growth 

 again, but into metamorphosis. The mere beginnings of canni- 

 balism do not, as does its fuller development, predispose the ani- 

 mal to the retention of the branchiate type. One of these results 

 of perhaps a week of cannibalism is shown as figure 4, plate IX, 

 figure 5 showing the form and the maximum size of the ordinary 

 larvae from which figure 4 was produced. It will be noticed 

 that the anterior part of the head has undergone a considerable 

 lateral expansion. Just such larvae as figure 4 frequently occur 

 among ordinary larvae in ponds, and, although not surprising, 

 are instantly recognizable. They are undoubtedly due to the 

 same causes which produced them in my experiment. 



Another trial in the same pond enclosure showed much greater 

 results; but in this I began with young larvae (hardly more than 

 6 cm. in length) taken from clear water, which already showed 

 marked beginnings in the cannibalistic habit, as well as partial 

 assumption of the form. With the young amblystomas were also 

 introduced in this case several hundred equally large tadpoles of 

 Rana. Thus launched, the results were varied. The majority of 

 the young cannibals were slowly, and evidently with the utmost 

 reluctance, starved out of their practice, returning to normal 

 feeding and normal form. A very few developed into transi- 

 tional types, becoming several times larger than the ordinary 

 run but evidently carrying on the cannibalistic mode of life in a 

 desultory fashion. The head of one of these is shown in figure 

 3, plate IX. Only a half dozen of the incipient broad-heads 

 stanchly maintained their mode of life despite the muddy water. 

 Their development was remarkable. Their growth was aston- 

 ishingly rapid ; in a few weeks they were double the length and 

 many-fold the bulk of their mates, which were of equal size when 

 the experiment began. . Their heads assumed nearly, though not 

 quite the extreme of type. Most interesting of all was their 

 steadily developing appearance of leanness, well-nigh of emacia- 

 tion, with gills diminutive and back fins very low, conditions 

 which, as I have said, indicate semistarvation in ordinary larvae. 

 By the time, however, that the largest of these cannibals had 



256 



