206 PROBLEMS OF GENETICS 



" habitual " 19 with S. atra, three to nine larvae may be produced 

 at one spawning period, from 35 to 45 mm. long, with gills at 

 most 8 mm. long, and a tail-fin 2-3 mm. broad. Such larvae 

 are generally coffee-brown, or grey (instead of black), and show 

 other minor differences. 



The summary states that when grown to maturity they be- 

 come in their turn larva-bearing, and go into the water to bring 

 forth. Their young are more than two (3 to 5 being the numbers 

 observed) with a length of 33-40 mm. or of 21-23 mm. at birth. 

 They are light grey, spotted (mottled with lighter and darker 

 colour), have relatively short gills (8 to 9 mm. at most) and a 

 broad tail-fin (3 mm. wide). At metamorphosis they are rela- 

 tively long (44 mm.) and one of them had some yellow pigment. 



Here again this summary is, as a matter of fact, describing 

 the behaviour of two mothers, of which one produced three, and 

 the other five young. 



To my mind these experiments suggest that the reproductive 

 habits of both species, if closely observed, will be found to be 

 subject to considerable variation, and I think it not impossible 

 that each species is, especially in confinement, capable of being a 

 good deal deflected from its normal behaviour. Moreover, there 

 seems to me no great improbability in the idea that there is an 

 interdependence between the number of young and the stage of 

 maturity in which they are born. But, at the same time, the case 

 as told by Kammerer strikes me as proving too much. If each 

 species is so sensitive to conditions that the normal procedure 

 is gravely modified in one generation, and if that modification 

 can reappear in a pronounced form in the next generation without 

 a renewal of the disturbing conditions, it becomes extremely 

 difficult to understand how the regularity which each species is 

 believed to display in nature can be maintained. Surely both 

 species might be expected to be in confusion. From a passage 

 in Kammerer's earlier paper (1904, p. 55) on the subject, I infer 

 that he also would expect considerable irregularity in the natural 

 behaviour, but that he has not investigated the point. 20 



19 Throughout Kammerer's papers this is used almost as a technical term. 

 It means, I presume, that the feature was manifested more than once. 



20 It should be stated that the papers contain a quantity of detail, especially 



