NOVEMBEB 29, 1912] 



SCIENCE 



751 



The numbers are smaller in the widely 

 quoted observations of H. C. Bumpus on the 

 sparrows which survived in a lot that were 

 picked up after a severe storm at Providence, 

 R. I. By using only four classes, however, the 

 survival curve is so inclined that we may 



other measurements to percentage of length. 

 The result shows a much greater influence of 

 selection than that shown by the absolute 

 measurements. The selection is also seen to 

 be secular (Fig. 5) rather than periodic, as 

 Bumpus concluded from the absolute measure- 



safely draw conclusions. The conditions in ments. The survival curves are shovni for the 



tOSd 52 48')4'W34 312Si*20 A (2 S 4 4 S II It 20 2t Zi 31 3k40 



SO SO iS SO SS 50 27 S14144-Z1 Zf 1841 IS 2S 33 

 1 I 1. 2 3 a II II mt ziz7H2t IS n II s 3 X I a 



Fig. 1. The polygon of frequency of the fron- 

 tal breadth (dotted line) of Carcinus mcenas in 

 Weldon's experiment, with the survival rate (solid 

 line). First line of numbers, deviation; second 

 line, survival rate; third line, numbers. 



'2131 



Fig. 2. The survival polygon in the Weldon 

 crab experiment plotted in eight classes. First 

 line of numbers, deviation; second line, survival 

 rate. 



this study probably do not show the full se- 

 lective effect of the catastrophe, for we here 

 contrast those which perished with the dis- 

 abled which survived when protected. The 

 sparrows which had succeeded in finding suit- 

 able shelter would probably have shown a still 

 greater difference from those that died, could 

 they have been obtained. Bumpus confined 

 himself to a study of the absolute measure- 

 ments, but since there was a selection as to the 

 size of the birds (Fig. 4) I have reduced the 



./a 70 -40O ^tS 'I7JS 



MJ tS.S 3I.S m 



Fig. 3. The survival polygon in the Weldon 

 crab experiment plotted in four classes. First 

 line of numbers, deviation; second line, survival 

 rate. 



;51 /53 /S4 /5S (St /J7 /S8 151 /bO M Ihl 11,3 M IkS life 'W 



o 4t loo SI M II IS .to ts 31 n _j!^_i° n ,n o 



O Z h 3 10 II T 13 10 rt 13 13 II S b b ; 



Fig. 4. The polygon of survival rate of the 

 length in the Bumpus sparrow case. First line of 

 numbers, length in millimeters; second line, sur- 

 vival rate; third line, number of individuals. 



sake of direct comparison in one diagram, 

 using as ordinates the average of the respect- 

 ive classes. 



Curiously enough, while the males show a 

 well-marked secular selection in the same di- 

 rection in each measurement, the females are 

 either only slightly secular or indifferent. In 

 some cases, like the measurements of the 

 femur and humerus of the males, the differ- 

 ences are very great, all of the longest fourth 



