RECORDING EXPERIMENTS 39 



to make numerical count in accordance with the 

 statistical method adopted by workers who are 

 experimenting on a limited scale. 



The second reason is that even if such a count, 

 showing the exact number or percentage of seed- 

 lings with different combinations of traits, were 

 attempted, it would be unavailing unless vast 

 companies of seedlings were preserved for the 

 term of years necessary to bring them to fruitage. 



When one is concerned solely with numbers, 

 or with such tangible qualities as color of hair in 

 the case of Professor Castle's guinea pigs, or 

 color of feather with Professor Davenport's 

 fowls, it is an easy matter to check results, be- 

 cause the creatures under investigation manifest 

 the qualities that are being tested from the 

 moment of birth, or develop them at a very 

 early age. 



But with plants the case is obviously different. 

 Whereas we may judge something as to the 

 character of fruit that a seedling will ultimately 

 bear from observation of the seedling itself, yet 

 for purposes of scientific record such predictions 

 would be considered as worse than worthless. To 

 know what percentage of seedlings of a given 

 generation have really progressed toward the 

 ideal of a sugar prune that will ripen in August 

 instead of September let us say, it would be 



