294 PLANT-BREEDING 



On this principle of correlation a definite method of 

 searching for monstrosities has been founded. Years ago 

 I was impressed by the fact that deviations in the seedling 

 plants are often followed by abnormalities in the aclult 

 state. By this means, the cotyledons of the seedlings may 

 give indications of what is to be expected. Now it is as 

 easy to inspect thousands of germinating plantlcts as it is 

 difficult to cultivate the same number to their full develop- 

 ment. But by choosing the stray seedlings with three 

 seed-leaves or with connate cotyledons, the cultures may be 

 effectively reduced, the chance of producing monstrosities 

 on a given space being proportionately increased. 



Leaving the abnormal characters, we have now to con- 

 sider a group of observed cases of correlations in ordinary 

 deviations. I choose my instances from among the agri- 

 cultural plants. Fruwirth has studied, the relations in 

 the development of leguminous crops, at the agricultural 

 station of Hohenheim in Wurtemberg. He observed a 

 parallelism between the increase of the total harvest and the 

 weight of the pods, the number and the weight of the seeds. 

 Less intimately, but still clearly correlated with these char- 

 acters was the average size of the seeds and the average 

 height of the whole plants. Gwallig observed that large 

 kernels are as a rule richer in albuminous substances 

 than small ones. This relation has been found to hold 

 good especially for peas and lupines, both belonging to the 

 same family and often cultivated as forage crops. 



Flax has been studied in this respect by Schindler, who 

 stated that the length of the fibers and their total amount 

 are correlated with the total height of the plant, measured 

 from the root to the highest capsule. The thickness of the 

 stem is inversely correlated with its height, higher plants 

 having thinner stems. Therefore, for the' harvest of the 

 fibers the thinner specimens and strains should be preferred. 



