F. L. Engledow 119 



On expectation, these constituents are K, I, and P, and in order to 

 determine their measurable characteristics, all plants of all F^ . I families 

 (between family No, 51 and family No. 265 both inclusive) were 

 re-measured and re-classified as K, I, or P. From a sample of 1190 

 plants thus obtained the following values were calculated : 



Measurable Characters of the K, I, aaid P Glume Types bred 

 from F^ . / plants. 



Year Generatiou 



1914 Fs.KexF2.I 



Fs.IexF^.I 



F3.PexF.2.I 



Comparison of these values with those given under K (p. 117) and 

 P (p. 117) show that for the mean glume-lengths : 



[(^3 . ^ ex Jf; . iT) - (^3 . i<^ ex ^2 ./)] = 11-98 - 11-81 = + 0-17, 

 [(^3 .PexF^.P)-iF,.P ex F., . /)] = 24-66 - 24-68 = - 0-02. 



That is to say, although "shift" was first exhibited by the P 

 descendants of heterozygous plants (viz. hy F^.P ex ^i) it is not again 

 exhibited by the P descendants of F^ heterozygotes (viz. F^.P ex F^.I). 

 It appears that the reduced or "shifted" value of Polish glume-length 

 cannot be further shifted by either 



(a) Selfing— for F^.P exF^.P (mean) :D.F^.P. 



(j8) Hybridisation, as in the zygosis performed hy F^.I plants — for 

 Fs.PexF^.I (mean) -a. F,.P ex F^. P. 



V. The Inheritance of Grain-Length. 



It was originally intended to deal with grain-length entirely by 

 measurement but the principle of metrical type-standards proved in- 

 applicable just as it did in the case of glume-length. In fact the break- 

 down was more complete, for the parental {K and P) mean grain-lengths 

 lie, comparatively, very close, and the overlap of their distributions is 

 considerable. From the early stages of the work upon F^ it was quite 

 certain that "P" grains were never associated with "K" glumes or 

 vice versa and after extensive trials of sorting grains into types, it was 

 concluded that in F2 three types only were found (K, I, and P), and 

 that on any plant K glumes were invariably accompanied by K grains and 

 similarly for / and for P. Consequently in F^ and ^3 the classification 

 by glumes was accepted as a classification by grains. Biffen [(1), p. 38] 

 remarked upon the relationship between grain and glume. 



