402 
J. ARTHUR HARRIS 
^4 « S lO 12 14 IB IS 22 34 26 2S 3" 32 34 3« 3S 
Pods per plant. 
Diagram 3. Regression of seeds matured per pod on number of pods per 
plant. 
1,484 plants there are only 36 which produce more than 38 pods. To 
calculate means for arrays of such sizes is idle. Neither is it desirable, 
in view of the great differences in the range of variation from series to 
series, to combine classes in order to secure larger arrays. Any ob- 
jections based on the fewness of individuals in the arrays applicable to 
