8 /. H. Pozvcrs 



preformed in the larva, and to be due to conditions affecting larval 

 development. I have measured non-mutilated larvae of so s,ur- 

 prising proportions that the width of the head entered the total 

 length little more than 4.5 times, while in very slender larvae the 

 width of the head may enter the total length nearly eight times. 

 This range of larv^al variation probably approaches nearer the 

 limits of the species than does that given above for adults. I have 

 handled more larvae and have collected more extreme types. Yet 

 still more slender larv^ae than the lirhit mentioned above probably 

 exist or might be produced experimentally. 



This range of larval variation as given above is the more re- 

 markable in that it does not to any great degree include sex dif- 

 ferentiation, as do the measurements given for adults. Most of 

 the larvae I have mentioned have been too young to admit easily 

 of sex determination. . When kept until sexual maturity is 

 reached, some time after metamorphosis, it turns out that the 

 more robust (not larger) larvae become more frequently females; 

 the more slender (longer and on the whole larger) become males. 

 Yet this does not hold for larvae grown under certain exceptional 

 conditions, nor for all larvae grown under normal conditions. In 

 my last season's catch I secured beautiful larvae grown under 

 typical conditions among which undeveloped males showed a 

 breadth of head entering the total length but 5.86 times; while 

 a slender female larva now in my aquarium shows a width of head 

 entering total length 7.85 times. Variation in the general body 

 form, then, is not primarily or chiefly a matter of sex. Reference 

 may here be made to plates I and II and also to plates VII, VIII, 

 and IX. These figures were not chosen with especial reference 

 to illustrating this part of the text, but are useful in this connec- 

 tion: figure I, plate I, shows a very robust form; also figures i, 2, 

 and 3 on plate II ; figures 2 and 3 on plate I are typical specimens, 

 especially 3 ; figure 4, plate II, is a slender type. The later plates 

 show mainly more extreme animals developed under less typical 

 conditions. 



Turning now to the causes of these general variations. I find, 

 by field study as well as by observations upon developing animals 

 in my experiments upon metamorphosis and coloration, that they 



204 



