418 



THE AMERICAN NATURALIST [Vol. XLV 



with grain. This last determination was made by the alcohol 

 method. The upshot of this elaborate study is to show that the 

 shape of the grains as measured by the length/width index is a 

 very significant factor in determining how wheat will grade 

 according to commercial standards. It is shown that "a differ- 

 ence of at least as high as three pounds per bushel may exist 

 between different pure-bred wheats having identical average 

 kernel-volume and kernel-weight. ' ' The final conclusion is that 

 the percentage volume of grain in a packed measure would be a 

 much more just and scientific basis for market grain grading 

 than the present system of test bushel weight. This paper illus- 

 trates in a very striking way how the scientific method can solve 

 in a precise and final manner a practical i-ommercial problem. 



Lill has made a quantitative study of the relation of size, 

 weight and desirability of kernel to germination in wheat. His 

 data indicate that germination capacity is not correlated with 

 size of kernel, but is correlated with density of kernel. No bio- 

 metrical analysis of the data is attempted. 



Waldron has made an interesting and significant biometrical 

 study of the correlation between weight of grain and other plant 

 characters in oats and wheat, using his own measurements for 

 the former cereal, and published data for the latter. He shows 

 that in oats the mean grain weight per head is nc<jat\vehf corre- 

 lated to a rather high degree with (a) number of grains per 

 head, (6) length of head and (c) length of culm. This obviously 

 leads to a somewhat paradoxical result, namely, that when 

 plump, heavy seed is sown, it is seed which is taken from mother 

 plants which are hdoir the average in size and yield. Yet care- 

 ful experiments, covering a period of years, have shown that 

 planting heavy seed gives increased yields. In other words, a 

 practise which amounts to continued selection of the poorer 

 yielding of plants as parents results in increased yield in the 

 progeny. This paradoxical result needs analysis by careful 

 pedigree breeding. 



Clark has published a general biometrical study on variation 

 and correlation in timothy, the material being gained in connec- 

 tion with the extensive breeding experiments with this grass 

 which have been in progress for some years at Cornell Fniversity. 

 The point of chief interest and novelty in the work is that each 

 of the 3,505 plants which furnished the data was under observa- 

 tion during three consecutive years. The material thus gives 

 seme basis for an estimation of the relative influence, on the one 



