It must be admitted that in many cases we have an apparently cumulative effect of selection, and it seems almost impossible to draw the line between improvements created by continuous selection of slight individual variations within the race or the selection of those plants which are mutations. In the case of the gooseberry, tomato and many other plants, the fruits have been increased in size grad ually, until they are now four to eight times that of the original wild fruits. Much of this increase in size has of course been accompanied by hybridi zation between different wild species and different races of the same species which have been mixed together, yet it is a cumulative gain in size, as none of the wild types ever produce fruits nearly so large as those of the cultivated races that have been developed. Practically the entire development of the tomato has taken place within the memory of men now living, and in this case the develop ment has not been accompanied by hybridization of different species but by the selection of different races within the species and the hybridization of these races. One of the experiments conducted by DeVries with corn is of interest in this connection. This experiment was undertaken for the purpose of increasing the number of rows of kernels on the ear. The corn used in the selection averaged twelve rows at the time the selection began. After seven generations of selections from ears which bore the largest number of rows, the mean was raised to twenty rows. In the first year of the selec tion the variation in number of rows ranged from 8 to 20. In the seventh generation of selection the variation in number of rows ranged from 12 to 28. This shows clearly the increase in the number of rows and the development of an apparently new race by simple selection. However, when the selec tion was discontinued the improvement or new character was soon lost.
The majority of new races produced as a result of selection are due, without much doubt, to the choice of mother plants showing marked variations which we would term mutations, and which are referred to by gardeners ordinarily as sports. In reviewing the history of cultivated varieties, one is surprised at the large number of varieties, which have had their origin in this way. Many of our apple, pear and peach varieties are simply accidental seedlings which have sprung up in fence corners or door-yards, and a number of our wheat, tobacco and cotton varieties have been developed by selection from certain individual plants that have attracted attention because of the exhibition of superior qualities. It is probable that a large number of these accidental and selected varieties, particularly in the case of apples and pears, are really the results of accidental hybridization, and the same may be true of many wheat, corn and cotton varieties. Yet there are many cases in which the mutations or extreme variations cannot be traced back to hybridization. In the production of the Cupid sweet-peas, for example, the first small dwarf plant of this type was found growing in a row of the Emily Henderson, which is one of the normal climbing forms of the sweet-pea. At
that time no other dwarf type of the sweet-pea was known, and this variation. therefore, cannot be accounted for as due to hybridization with some other dwarf form. It is impossible to account for these striking variations which sometimes occur, but it is important that all plant-breeders be on the lookout for the occurence of new types and variations of this sort.
The writer has been asked frequently whether it is possible to select a plant so highly that it will not revert to the original mother type. Experience would indicate that when the mother plant from which the selection is made is a true mutation, like the sweet -pea mentioned above, the type will maintain itself even after the selection has been discontinued, and indeed this practically the only real criterion as to whether a new race has been produced. For example, in the case of the corn mentioned above as selected by De Vries. that in seven years had been increased from 12 to 20 in the number of rows to the ear. DeVries found that it required only about three years of cul tivation without selection to fall again to the original aver age of 12 to 16 rows. In a case like this it would seem, therefore, that no distinctly new character had been added as a result of selection, but that the average of the race had been increased by the continuous selection under isolation, and that when the different individuals were al lowed to breed together freely, without selection, the mean of the race, as a whole, was again quickly reestablished.
Systematic methods of selection, or pedigree breeding.
Two distinct methods of selection are in use, which are termed (1) the nursery method, and (2) the field method. The nursery method, which was used first by Hallet about 1868, so far as the writer is informed, consists in cultivating each plant under the most favorable conditions possible for its best development. By this method, with wheat, for example, Hallet pursued the policy of planting the individuals in squares a foot apart, which would give the plant abundant opportunity for stooling, and also enable the investigator to distinguish clearly each individual plant. In more recent years this method has been strikingly em phasized by the work of Professor Hays, at the Minnesota Experiment Sta tion, who, at the same time, has modified the principle somewhat into his centgem r method (Fig. 82). In Profes sor Hays' method, the progeny of each plant, presumably about one hundred individu als, are grown together in a small plat or centgener, the individuals being planted four to six inches apart in the case of wheat and small grains.
The field method, which was emphasized by Rimpau about 1867, and has been used by many investigators, consists in selecting from plants grown under normal conditions. The argument for this method is that the plant will show what it will do and its true worth only when it is grown under the method of ordinary field culture. Roth of these methods depend on progressive or cumulative selec tion, the building up and adding together of small improvements.