ON VARIATIONS. 341 



ture of the beach where his crabs were collected, in 

 that it looks due south and is uncovered for hours daily, 

 was any lower in 1898 than in 1893, and also he found 

 that crabs gathered in January were no narrower 

 fronted than those gathered in August, as they ought 

 to have been on Cunningham's hypothesis. 



The proof of the existence of Natural Selection 

 really centres upon the proof of a selective destruc- 

 tion or death rate. If among any group of organisms 

 the eliminated individuals can be measured and exam- 

 ined, as well as the survivors, and if it be found that 

 these two divisions differ in their mean characters, then 

 Natural Selection must have been at work. Very 

 likely the parts or organs measured do not represent 

 the characters upon which the selective process had 

 been acting, but are merely correlated with them. 

 But that is no matter. The offspring of the survivors 

 will have different average qualities from those of the 

 previous unselected generation, or the race will be- 

 come modified by Natural Selection. 



Unfortunately in the majority of cases, as in Pro- 

 fessor Weldon's crabs, it is impossible to get hold of 

 the eliminated individuals, and hence the proof of 

 Natural Selection is rendered much more laborious, 

 and at the same time more open to possible source of 

 error. In the case of the (introduced) English spar- 

 row (Passer domesticus), however, Bumpus * has suc- 

 ceeded in obtaining the desired material. One hun- 

 dred and thirty-six of these sparrows were collected 

 after a very severe storm of snow, rain, and sleet in 

 North America, and of these 72 revived, whilst 64 

 * Biol. Lectures, Wood's Holl, 1898, p. 211. 



