2l8 BIOLOGICAL LECTURES. 



where albinism twice occurs, it affects in one case a bird 

 marked by four other abnormalities (No. 21), and in the other 

 a bird (No. 52) already cursed by six abnormalities, the most 

 miserable individual in the entire collection. 



While we have shown that the birds which perished have 

 certain average structural peculiarities which distinguish them 

 from the survivors, and that the intensity of selective elimina- 

 tion has been felt most by birds of extreme structure, it remains 

 to be shown that a general instability of structure is as char- 

 acteristic of the birds which perished as a general stability of 

 structure is characteristic of those which survived. 



If we had sufficient data, this fundamental difference in the 

 two groups of birds might be indicated by curves of distribu- 

 tion, one curve narrow and elevated, showing that its compo- 

 nents are closely crowded around an ideal mean, the other 

 broad and low, showing that its components are relatively 

 indifferent to any ideal. But in the absence of sufficient data 

 to illustrate the differences in this manner, we can arrive at a 

 numerical result equally instructive by another method. 



Having determined the ideal means for the several characters 

 in each group of birds, we can then find the distance that each 

 individual departs from this ideal. By adding these degrees 

 of departure in respect to the several characters, and dividing 

 by the number of individuals, we shall have numbers which 

 represent the average departures from the ideal means. These 

 numbers will be large if the members of a group of birds show 

 a general tendency towards disregard of the ideals, and they will 

 be small if the birds tend to crowd around the ideals. If all 

 the birds actually attain the ideals, the number will be zero. 

 This is simply following out the principle that one man at 

 the end of a ten-foot lever can do as much work as ten men 

 at the end of a one-foot lever. A bird removed ten units from 

 the mean exerts the same divergent influence upon its group 

 that ten birds would exercise if removed one unit. 



The results of this test, numerically expressed in Table V, 

 are most instructive. In every case but one the numbers indi- 

 cating the average departure from the ideal mean are smaller 

 for the birds which survived, and thus indicate a general tend- 



