198 Metrical Attributes of Wheat Plants 



with facility. These two reasons, then, led to its adoption. Overlap of 

 distributions is the form in which the difficulty of fluctuation makes 

 itself felt. To counter overlap, two wndely differing parent form.< were 

 selected, viz. the T. durum known as Kubanka (mean glume-length in 

 England£il0-5 mm.), and the distinctive T. polonicum or Polish Wheat 

 (mean glume-length in England £i 30-5 mm.). Greatly as these forms 

 diverged in glume-length, their distributions nevertheless showed a slight 

 overlap. In the F.,, the parental types reappeared in a "shifted" form. 

 With them, and intermediate between them, came a heterozygote. 

 Consequently the glume-length distribution of the whole F, took a 

 trimodal form which, defying analysis on the basis of rigid measurement, 

 enforced a reversion to eye-judgment. It was quite obvious that the 

 wide "'fluctuation'" in glume-length was due to the occurrence in the 

 parental and segregate populations of a certain number of very poorly 

 grown plants. To reject such plants would have been disastrous and to 

 include them was to confuse distributions and inhibit analysis. Plants 

 of this kind were poor in every _way and, roughly speaking, the less the 

 stature of a plant, the shorter were its leaves, its ears, its glumes, etc. 

 This very patent fact suggested that instead of making, for every plant, 

 an absolute measurement, there should be made some form of "com- 

 pensated" measurement. In short, every plant should be "handicapped." 

 "Length-of-rachis"' was selected as the basis of the "handicap." The 

 working hj'pothesis was that a big, thri^^ng. plant had long ears (i.e. great 

 rachis length) on which were borne proportionately long glumes. Per 

 contra, small plants would be small "all round" and it seemed not 

 unlikely that the ratio — length of glume : length of rachis —would 

 exhibit a smaller plant-to-plant fluctuation than would absolute glume- 

 length. To test the constancy of this ratio, then, became the first object 

 of investigation. 



A phenomenon designated by the term "shift" was observed in the 

 F2 of the Pohsh x Kubanka cross. It consisted in the appearance in F^ 

 of two groups of plants, each in number about a quarter of the whole F2 

 population, and having respectively a complete eye-resemblance to the 

 parental (F„) forms. In mean glume-length, however, the F^ Kubanka 

 type .slightly exceeded the F^, while the F^ Polish type was 25 per cent, 

 shorter than the Fq. Explanations of "shift" could be based upon 

 "modification by cros.sing" [cf. Ruggles-Gates(3)] or upon "minor 

 nmltiphnng factors," but yet another explanation seemed possible. The 

 embryos and endosperms from which grew the Fo Polish-type plants, 

 were nourished by F^ (i.e. heterozygous and intermediate) plants. 



