POTATO DEVELOPMENT WORK IN WISCONSIN. 41 



This is perhaps most easily accomplished by resorting to the tub- 

 er-unit method, that is dividing each tuber lengthwise into four 

 equal parts. The four parts of each tuber are planted consecu- 

 tively in the row and a slightly wider space left between each set 

 of four plants. If the small tubers are planted split them in two 

 but it will be necessary to note the number of tubers planted by 

 both methods or else it may not be possible to identify the pro- 

 geny of each tuber. A close study of the growing plants will very 

 probably reveal quite striking differences between the progeny 

 of the various selections. It is quite likely that some selections 

 will be noted in which the plants are extremely variable in size 

 and vigor such selections should be marked for rejection. Note all 

 those which show uniformity in size and which most nearly con- 

 form to the varietal type. Harvest each selection by hand and re- 

 tain, only those for further trial which most nearly meet the ideal 

 of the variety both in type, color, size and productiveness. 



The further culture of these selections consists in planting 

 them according to the most approved methods in the increase 

 plot. If at the end of the second season you have a single strain 

 left which has met your most exacting demands you have reason 

 to rejoice over your success for I am convinced that there are 

 relatively few strains within a variety which are markedly more 

 productive than ordinarily well selected stock. The great value 

 of selection according to my mind lies in the elimination of the 

 weak, unproductive plants rather than in the isolation of high 

 producing ones. 



One of the best known experiments on hill selection is that re- 

 ported by Waid in bulletin 174 of the Ohio Agricultural Experi- 

 ment Station. The data given by Waid shows that the average 

 weight of ten high yielding hills from the 1903 crop was 2.38 

 Ibs. while that from ten low yielding hills was 0.72 Ibs. The av- 

 erage of the progeny of these ten high and ten low yielding hills 

 for the seasons of 1904, 1905, and 1906 are shown to be 1.38 Ibs. 

 and 0.73 Ibs. respectively while that from unselected or the gen- 

 eral run of seed stock was 1.10 Ibs. Two deductions might be 

 made from these results one that the high yielding hills had not- 

 transmitted this character, if compared with their initial pro- 

 duction, for we have a falling off in yield of over 72%. If on the 

 other hand we compare the behavior of the high yielding hills 

 with un-selected stock we note a gain of over 25% in favor of the 

 selections evidencing a distinct gain through selection. In some 



