210 BIOLOGICAL LECTURES. 



The birds which perished are at a decided disadvantage because 

 of their smaller representation, yet there are many more " freaks" 

 among them than among the surviving birds. 



If it is thought that the association of the larger number of 

 extreme variants with the eliminated birds is merely a matter 

 of accident, we will not stop to argue the matter, but will apply 

 the same test to the birds that remain after these extreme 

 examples have been removed. We find even after the removal 

 of these twenty-three examples, that extreme examples of the 

 second order, indicated by the exponent 2, show the same tend- 

 ency to occur more frequently among the eliminated birds. 



The longest birds now are 166 mm.; the shortest, 153 mm. 

 Of the former, Nos. 22, 24, 28, 32, 35, perished, and No. 18 sur- 

 vived ; of the latter, Nos. 45 and 62 perished, and Nos. 35, 54, 

 and 55 survived. 



If we count the times that the exponent 2 occurs in the 

 tables, we shall find that there are ten birds of extreme abnor- 

 mality of this second grade which survived, while there are 

 twenty of the same grade which perished. 



These figures indicate that the amplitude of variation of the 

 surviving birds is less than that of the birds which perished. 

 Were we to attempt the arrangement of the data into curves 

 of distribution, the curve representing the distribution before 

 the storm would be found to have a broad base, whereas the 

 curve representing the distribution after the storm would be 

 found to have a narrow base, for the eliminative process con- 

 centrated its energy on the individuals which occupied extreme 

 positions. 



Lest there remain some doubt as to the importance of this 

 eliminative process, and of its efficiency in exterminating ex- 

 treme variants, let us examine our figures again and see whether 

 the group of birds which has already contributed thirty-four of 

 the extremes of variation has still an excess of variability. 



If we count up the exponents (3) of this third order of 

 variable individuals, we find that the birds which survive give 

 eleven examples, whereas those which perished give twenty-one. 

 It appears unnecessary to carry our investigations further 

 along this line, for our results point always in one direction. 



