84 



THE AMERICAN NATURALIST [Vol. XL VIII 



TABLE I 







' i • » 











18 







I \- 



E — 



fl 



1 



2 - 



5 



- 5 



- 1 



- 2 



- 17 



65 mo 



40 107 134 

 :.3 117 17;; 



49 105 105 



17 44 11(1 

 6 8 16 

 UI7 126 1 12 



57 42 

 17ii in.-, 



126 56 



os 32 

 96 86 

 71 11 1 

 is;; 



25 51 



19 

 ;;: 



is 



3 

 69 

 15'. 

 63 



« 



6 11 1 

 3 2 — 











168 



549 

 606 

 367 



631 

 603 

 633 

 565 



4 lj — 



132 81 23 

 20 6 2 

 127 126 101 



5; 2, 



55 



17 





1 



TABLE II 

 Weight of Seeds which Failed to Germii> 





Series 4 5 6 7 









A 4 



c \ — 

 D 1 — 

 E — 

 F — 



// — 



/ 1 — 



16 49 63 

 2 1 12 



— 2 52 104 117 84 36 3 — 2 



37 88 127 93 37 11 — • — — 



115 210 180 91 23 4 1 



13 46 84 125 153 114 49 16 10 











558 

 400 



399 

 637 

 613 



of the seeds which germinated. 4 Table II gives the same 

 distributions for the seeds which failed to germinate. 

 The physical constants 5 with their probable errors are 

 given in Tables III-IV. 



Taking the differences, germinated less failed, in order 

 to have the positive sign if elimination tends to increase 

 mean weight or variability of weight and the negative 

 sign if it tends to decrease these constants in the popula- 

 tion of seeds which grow as compared with those which 

 fail, I find the differences shown in Table V. 



4 When the plantlets were about three inches high the labels for seeds 

 which had failed to germinate were collected. The distributions for the feeds 

 which had germinated were then obtained by subtraction from the weight 

 seriations prepared before planting. Some of the plants subsequently died. 



s Sheppard 's correction was applied to the second moments. 



