F. L. Engledow and J. P. Shelton 201 



Table 11. Coefficients of Variation for 01 Nine-Length, Baehis-Lengtli and 

 the Ratio of these two quantities in Polish and Kiibanhi ivheats. 



CiiefRcicnt. of vaiiation of 



Variety iiiid No. of ears , — 



no. of plants ])er plant (ihinie-lengtli Rachis-length Ratio 



Poliali (:)1<;) 1 ]7-13±0-47 25-77±0-74 l.'-)-47 -l:0-4:? 



Polish (110) 2 S-5()±0-39 13-8.3±0-l!4 10-S:?:i 04!1 



Kuhanka (20")) 1 10-I8±0-34 20-43±0-71 18-L'niO(>3 



Kiihanka (107) 2 8-40 |:0-.3n Id-SSi-O-SO 12-.58-J:0-.^lt 



It is therefore to he concluded that this particular "compensated" 

 ratio possesses no value either to the geneticist for a critical study of 

 variation or to the plant breeder for the discrimination of closely re- 

 sembling agricultural strains. 



Despite this record of failure, it is felt that "compensated" observa- 

 tions {i.e. some form of ratio) ofi'er still the only alternative to "abso- 

 lutely (unattainably?) uniform conditions of growth" as a means of 

 removing the masking effects of "fluctuation."" It is not to be expected 

 that the random choice of two characters — e.g. leaf-width and grain- 

 length — will serve as a basis for compensation or handica]) and ])rovide 

 a con.stant ratio. Manifestly, in seeking constancy, one should endeavour 

 to find two "lengths" (or other attributes) which are determined by 

 the same causes and during the same time-period. In glume- and 

 rachis-lengths, it would seem, these requirements are as likely to meet 

 fulfilment as in any pair of attributes of the wheat plant to which one 

 could point. This granted, there follows the regrettable but not surprising 

 conclusion that "lengths'" are such vague expressions of the real 

 "nature'" of a plant variety, of its physiological activities, as to be of 

 little value in attempts to determine accurately its modes of inheritance. 



The employment of "ratios"' by genetici.sts has usually been dictated 

 not so much by a desire to achieve "compensation"" as by the necessity 

 of devising representations for elusive attributes like leaf-shape, etc. 

 The work of Martin-Leake (S) upon Cotton, of Balls (6) upon the same 

 plant, and of Groth(7) upon the Tomato afford examples; but in all 

 these cases fluctuation has played its customary havoc. 



In outline, the "handicap"" principle for plants is strictly analogous 

 to that followed in flat-racing. Horses are handicapped so that they 

 may afford an exciting "bunch" at the winning-post. It is required to 

 "bunch"' metrical observations upon a plant population {i.e. to minimise 

 fluctuation) but it seems probable that a really successful "handicap"' 

 will have to be a complex one. It will, in fact, have to be applied not 



