F. L. Englbdow and J. P. Shelton 203 



Both for Polish, aud Kubauka, ghiiiie- and rachis-leugths are greater 

 in the case of two-ear than in the case of one-ear plants; and since the 

 dif^'erence is more marked for rachis than for glume, the "ratio" is 

 lower for the two-ear plants. 



A difference is, perhaps, to be anticipated for, within limits, the 

 more vigorous the plant the more tillers it produces: and the facts 

 displayed in Table IV make it appear that greater vigour is evinced in 

 "all-round"' form — not only more tillers but bigger ones (larger glumes 

 and rachis). This "co-fluctuation" of the attributes rachis-length. glume- 

 length, and number of tillers, encourages one to think that in such 

 attributes is to be found a means of estimating degree of growth and of 

 comparing and contrasting different pure lines. There may still be room 

 for hope but no progress is possible until "fluctuation" has been dealt 

 with and, as far as this case goes, the mathematical handling of attribute 

 measurements (use of ratios, etc.) has been valueless. 



It has been inferred that observation should be confined to the main 

 ear of the plant and this inference, if valid, is not without interest from 

 the point of view of yield-investigations. When, in seeking higher- 

 yielding forms, an F2, is raised from two parental strains, it is almost 

 essential to cast out what are believed to be the "inferior" segregates. 

 Even if there be no casting of Fj plants, the process must be applied 

 to the resulting F^ families for otherwise available time and ground- 

 space become inadequate. In actual practice " casting" by eye-judgment 

 has attained some very conspicuous successes but there is now a tendency 

 to try to substitute an accurate "method" in its place. If those attri- 

 butes which are considered to govern yield behave in a way analogous to 

 that described for glume-length, etc., then the best tiller and not the 

 whole plant should be the basis of estimation. In the field, however, 

 small tillers as well as large go to make a crop, and whatever form of 

 judgment of F2, plants is evolved, it must of course, in some way, pay 

 due regard to tillering power. 



§ V. The Relation of Weight of Seed Sown 

 TO THE Resulting Plant. 



It has been explained in the introduction that there existed a special 

 reason for attempting to determine the influence upon the plant of the 

 weight and composition of the seed from which it grew. Apart from this, 

 in a great mass of literature [for a very full summary see Kidd and 

 West (8)], there is evidence which justifies the suppo.sition that much of 

 the troublesome "fluctuation" is induced by lack of uniformity of seed. 



