1910.] Improvement of Crops by Selection. 381 



and grown under precisely similar conditions, all inferior 

 plants were moved and the remaining individuals carefully 

 examined during the winter, as to the plant as a whole, as 

 to the head, and as to the grain. From the best were selected 

 only those which produced three stems ; from those only the 

 primary ears were taken, and all the ears were weighed on 

 a delicate balance. All under the average were discarded. 

 Of the residue, those with the densest ears were retained, and 

 from this final residuum only the grains from the middle of 

 the ear were preserved for sowing. In the following year the 

 same strict selection was repeated. At the end of five years 

 the result was nothing but disappointment. Chevalier barley, 

 for example, was only 1 J per cent, denser than at the begin- 

 ning. What was the reason ? In nearly all cases the plots 

 had been sown with grain which was derived from a group 

 of ears similar in appearance,- and, therefore, supposed to be 

 of the same type. It was noticed that the progeny of these 

 was never homogeneous, but the plants varied in leafage, 

 in length of straw, in maturity, &e. 



The Single Ear the Unit of Selection. — In 1890 or 1891, 

 however, a few small plots among more than a thousand 

 were found to be of the same type ; each plant was precisely 

 similar to its neighbour on the plot. Reference to the elab- 

 orate records showed that the homogeneous plots were the 

 progeny of a single head or panicle, and it was therefore 

 concluded that cultures can only be pure when started from 

 a single ear. In the following year over 1,000 plots were 

 sown with seed from single ears, and the result was con- 

 clusive. With very rare exceptions all the offspring of a 

 single plant were exactly alike. Of 422 plots of oats, 397 

 were uniform, and only 25 were heterogeneous, and those 

 exceptions might have been hybrids or imperfectly pollinated. 

 Thus was the principle firmly established that the starting 

 point or unit of selection is the single ear or head. Further 

 investigation brought out the fact that in an ordinary 

 field of wheat, oats, or barley, were dozens of different types, 

 most of which would breed true. The next step was to dis- 

 cover the superior types or those specially adapted for special 

 conditions. 



The Association of Characters.— The discovery of the 



