204 Metrical Affn'hntes of Wliiat Plants 



As a contributory test of this point, correlations were evaluated between 

 weight of mother-seed and glume-length, rachis-length, and their ratio 

 in the resulting plants. Preliminary investigation having shown an 

 absence of conolatioii for a population of jiinnts having different numbers 

 of ears, attention was confined to single-car plants. Talilc \' contains 

 the results. 



Table \ . Correlation between Weight of Mother Seed and the Characters of 

 the Resulting Plant. (Only one-ear plants are included.) 



Varii'tj' and 



Correlation between weight of mother seed and 



no. of plants Glume-length Rachis-length Ratio 



Polish (.316) 0-O(55±OO38 0043-0038 OOOOi — 



Kuhanka (20.5) 0-262i0-a44 0-1.57±0-04li 0003±0-047 



The supposed relation between weight of mother-seed and glume- 

 length, etc., is thus emphatically negatived. Mere weight of seed is not — 

 save at the lower extreme — likely to be of great physiological importance 

 and the belief in its importance (vide literature above mentioned) is, 

 perhaps, largely due to the form, of ex])eriment often adopted — the 

 removal of a portion of the endosperm, work with non-pure lines, 

 fewness of observations and so on. No embryo during germination uses 

 the whole of the reserve food with which it has been provided and 

 reserves beyond a certain amount must be simply "surplus." That 

 "quality" of endosperm may be important still remains a possibility 

 and evidence exists to this effect. This possibility is the only remaining 

 basis for the explanation of "shift" of which an account is given in the 

 introduction. With the general principle that good, sound, seed must be 

 sown to reap a good crop, there is common agreement, but beyond this, 

 even in genetic work and in face of the danger of fluctuation, there seems 

 no need to go. The weighing of a vast number of seeds in order that 

 seeds all of the same weight may be .sown, seems, from the facts above 

 recorded, not to be worth while. 



Conclusions. 



The conclusions which follow hold in strictness — failing further test — 

 only for the year, the locality, and the wheat varieties concerned in the 

 investigation. Since, however, the residts iwe fairly emphatic, they 

 seem likely to prove applicable in principle to other circumstances. 



(i) Glume-length and rachis-length in both Polish and Kubanka 

 "Wheats are very highly correlated. 



