58 /. H. Powers 



or abraded, some having been, it would seem, nearly swallowed 

 before meeting with the resistance, no doubt, of some friend who 

 had just gone before. The ordinary specimens were mutilated 

 much less frequently. Such facts as these give added reason why 

 large larvae of this special type are never numerous. They also 

 established a fair presumption as to the cause of the form : all 

 of the broad-heads were cannibals, and of most heroic type, 

 while even the transition specimens showed the same habit. 

 Others did not. 



Nevertheless, cannibalism might still possibly be a result, 

 rather than the cause, of the form. I attempted to raise a num- 

 ber of the cannibals and semicannibals in order to see whether 

 they would prove subject to still further modification. Although 

 I had reared many hundreds of ordinary specimens, the broad- 

 heads proved a wholly new problem. They killed or injured 

 each other ; they suffered at once in a way very probably due to 

 change in depth of water, the animal losing its power of equi- 

 librium and control of its hydrostatic apparatus, and worse still, 

 many could not endure either running water or the sudden daily 

 transfer from aquarium to fresh tap water. Rapidly grown lar- 

 vae with delicate skins are frequently killed by this, the transfer 

 to fresh water causing a rapid blistering of the epidermis about 

 the gills, head, and even tail fin, until portions of the surface m^y 

 look like coarsely beaten white of egg. Lastly, specimen after 

 specimen succumbed to a disease, evidently bacterial, which has 

 often given me great trouble, but to which the cannibals were 

 quite unusually susceptible. So delicate, indeed, did these speci- 

 mens prove that, despite my utmost efforts, all died save one. 

 This specimen proved interesting. It was in a starvation experi- 

 ment to test the susceptibility of the type to metamorphosis. 

 Confined with another of its own type and two ordinary larvae 

 of its own length, it killed its companion, but could not catch 

 either of the others, both of which became extremely wary. 

 Within ten days starvation induced metamorphosis in both of the 

 latter, &nd six days later they had practically completed the proc- 

 ess, the cannibal meanwhile showing no sign of transformation. 

 It then consented to eat liver and become a moderate feeder. 



254 



