112 Glume- Length and Grain-Length in Wheat 



cent, of the single measurements deviate from the means of the sets of 

 eight to which they belong, by no more than + 0*4 mm. The accuracy 

 of measurement is only + 0'5 mm., and consequently one glume measure- 

 ment per ear was considered to be justifiable. 



III. The Inheritance of Glume-Length. 



The symbols " P" " /," and " K " are used to facilitate reference. 



Thus " P " implies " Polish " or " of Polish type." 



Fq . P designates a population of pure " parental " Polish type, the 

 type which was employed as one parent in the cross. 



^0 • Mp (glume) denotes the mean glume-length of a population of 

 parental Polish plants. 



Po ■ o'p (glume) denotes the standard deviation of the glume-length 

 frequency distribution for such a population. 



" K " is similarly used in the case of " Kubanka." 



" / " implies the heterozygote form which, as later appears, is roughly- 

 speaking intermediate between "P" and " K'\ 



The glume-length distributions in successive generations were as 

 follows : 



Both parent varieties were grown and measured in 1913 and 1914. 

 For comparative purposes the results of 1914 are by far the more useful, 

 for they are based on much greater numbers of observations, and the F^ 

 was. also grown in that year, the conditions being precisely similar. 

 Summarised, the results were : 



Glume-lengths of Parent Varieties (F^ . P and F^, K) of the Gross. 



That the variability of Polish is great and that of Kubanka small 

 becomes clear from an inspection of Tables I and II. 



During 1914-19, the F^ measurements were lost. From memory it 

 can be stated simply that the F^ mean lay somewhere between the 

 parental means — probably it was close to the arithmetic mean of the 



