ON VARIATIONS. 345 



these characters were partly of blastogenic, and not 

 wholly of somatogenic origin; and this could scarcely 

 fail to be the case. Several of the changes in char- 

 acters, especially of the total length, weight, and femur 

 length, might possibly be present, on an average, to just 

 as marked an extent as in their parents (the survivors 

 of the previous generation); for though the characters 

 would tend to undergo reduction by^ virtue of their 

 " regression towards mediocrity," yet they would 

 tend to be enhanced by reason of the fact that more 

 of the extreme individuals (which would be of propor- 

 tionally greater weight in determining the characters 

 of the next generation) had been weeded out than 

 of the mediocre ones. Bumpus does not give any 

 details as to the way in which the sparrows were 

 collected, and whether the sample obtained was repre- 

 sentative of all the sparrows in the area in question. 

 Supposing it were not, and it included only spar- 

 rows which were exposed to the force of the storm 

 through failing to get adequate shelter, then, of course, 

 the average change produced in the characters of the 

 next generation would be much less than that suggested 

 by the above figures. 



Professor Weldon * has adopted a very ingenious 

 method for determining the presence or absence of Nat- 

 ural Selection in the case of a certain terrestrial mol- 

 lusc, Clausilia laminata. The outer layer of the shell in 

 this and other pulmonates is secreted by the growing 

 edge of the mantle once and for all, and it undergoes 

 practically no subsequent change. The upper whorls 

 of an adult shell therefore afford an unaltered record 



* Biometrika, i. p. 109, 1901 



