202 EXPERIMENTAL FARMS 



9-10 EDWARD VII., A. 1910 



the best strain not boinc: made nntil after all the most promising strains have been 

 propagated and studied for several years. 



Both systems of selection have been used in the work of the Cereal Division, but 

 the method of repeated selections for the improvement of varieties has been abandoned. 

 A fevp years ago, when plots produced by the two methods were sometimes grown side 

 by side, the difference in uniformity of appearance was strikingly in favour of the 

 plots which had each been bred from a single plant. 



The greatest danger in any form of repeated selection is that desirable qualities 

 may be lost in one direction while a gain is being made in another. If, for instance, 

 we always select the largest seeds, these may be the product of the plants with the 

 smallest heads, and we may in time materially reduce the productiveness of the 

 selected grain. If the largest heads are chosen, these may come from plants with 

 unduly long straw, which may be undesirable. Earliest heads, if we are selecting for 

 earliness, may be from plants of otherwise poor quality. Again, in wet seasons one 

 necessarily selects those individuals which can best withstand an excess of moisture 

 and in dry seasons one must choose the opposite class, the work of one year thus 

 conflicting with that of another. The danger of selecting false heads or kernels of an 

 undesired variety in mistake for unusually large heads or kernels of the desired sort 

 is very great, when the work is being done by any one but a well-trained specialist. 



The advocates of repeated s-election may object, however, that undue emphasis 

 is being laid on the dangers of this method, and that in actual practice it has been 

 found to give excellent results in some cases. Of course it is true that striking 

 improvement can easily be produced by the selection of heads, for instance, even for 

 a single season, provided that the grain with which the experiment is started is badly 

 mixed. Strictly speaking, however, this is not improvement but purification, and such 

 instances can scarcely be said to furnish a fair argument. Purification of seed is of 

 great importance, but a fair test in regard to the improvement of a variety cannot be 

 begun until after pure seed has been obtairLed. 



One other kind of instance should be mentioned. If for a long series of years we 

 carefully select any particular type of head or seed we may filially reach a point where 

 the greater part of our selected crop consists of the descendants of that plant which 

 in the first year was the most productive of all those of the desired type. In other 

 words, the selection of best typical heads year after year may ultimately bring us very 

 close to the point which we could have reached by the other method of selection in a 

 much shorter time and with very much less labour. 



While fully admitting the value of the ordinary method of the selection of heads 

 for the purpose of purifying mixed grain and for the maintenance of the seed in a 

 high state of purity, it does not seem to me to be the best way to bring about real 

 improvement in any variety. The selection of good typical heads, rather than of 

 heads which are in any way unusual, seems to be the safest and best plan for farmers 

 who wish to keep their seed quite true to name. If a strain of different type is being 

 sought for it is best to breed a number of separate strains, each started from a single 

 plant which appears to possess the desired qualities. One can thus usually obtain, at 

 a single step, some distinct and permanent advantage and can make* sure before the 

 final choice of one particular strain is made that in effecting an improvement in ono 

 direction nothing essential has been lost in other ways. 



The work which has been done in the Cereal Division with Red Fife wheat may 

 serve as an illustration of the practical value of this method. It was desired to obtain 

 a selected type of this wheat which should ripen earlier than the original variety but 

 still retain its striking ability to produce strong flour. Early-maturing heads were, 

 therefore, picked out, and from each of these a new strain was produced. For several 

 years the new strains were studied in the fields, and having been at last reduced to 

 four they were subjected to milling and baking tests. As a result of these tests it was 

 made clear that the selections B, H and M, which are still being grown, are genuine 

 Red Fife so far as baking strength is concerned. Strain M does not show any par- 



