548 Journal of Agricultural Research voi. xix, No. n 



stable. Green Mountain, as grown at Bozeman, has rather strong degen- 

 erate tendencies, but Early Six Weeks would be classed as almost as 

 stable as Rural New Yorker, since no rapid decline in yield followed selec- 

 tion based upon tuber characteristics. Since in this work every effort has 

 been made to overcome the possibility of error, a discussion of methods 

 will be in order. 



Tuber units were selected from the bin in the spring of 1916. In each 

 variety these tubers were cut to a uniform weight by slicing off the stem 

 end. The Green Mountain tubers were cut to 7 ounces, the Rural New 

 Yorker to 8>^ ounces, and the Early Six Weeks to 6}^ ounces. The 

 remaining portion of the tuber was then quartered lengthwise. Since 

 1 91 6 seed pieces have been cut with a special cutter which cuts a seed 

 piece approximately hemisperical in shape, iX inches across the face 

 and X inch deep. Such seed pieces weigh approximately }i ounce, or 

 about 1 1 gm. Because of variation in depth of eyes, these seed pieces will 

 differ somewhat in weight, the maximum variation within a variety usu- 

 ally not exceeding 2 gm. and in many cases not exceeding i gm. Groups 

 of 20 seed pieces seldom show variation of over 15 gm. While these seed 

 pieces may appear small, they have apparently given just as strong plants 

 as seed pieces cut in the usual manner and weighing iX ounces. Seed 

 pieces have been planted by hand and covered to a uniform depth of 3 

 inches. Single-plot tests have been used entirely. Such tests carried on 

 over a series of years should give just as reliable results as duplicate or 

 triplicate tests conducted for a shorter period. Duplicating plots com- 

 plicates the work of seed selection. It is not safe to select seed from any 

 one plot to continue the line the following season, and it is difficult to 

 select a small com.posite sample from two or three plots. It is for this 

 reason that the single-plot test was decided upon. 



The 1 91 6 plots consisted of 4 hills, and the 191 7 and 191 8 plots of 20 

 hills. The latter is the standard-sized plot adopted here for this 

 work. The selections from each variety are arranged in a group with 

 from 9 to 12 plots in a row and from 9 to 12 rows^ There is no space 

 between plots in the rows. The control plot consists of a single row near 

 the center of the group running the length of the group of plots. Hills are 

 planted 15 inches apart in rows 3 feet 9 inches apart. Seed is chosen 

 from the plots before the tubers are assembled, and in this wa)^ it is 

 secured from different portions of the plot. To illustrate, from a standard- 

 sized plot 5 seed tubers are selected. Four seed pieces from each 

 tuber are planted the following season in adjacent hills, and when this 

 crop is harvested, effort is made to select one seed tuber from each group 

 of four hills, or from each unit, as they are called later. The plots 

 receive rather deep cultivation, are well ridged, and are commonly irri- 

 gated once. At any rate they all receive the same cultivation and irriga- 

 tion. All hills are thinned to single stems as soon as the vines 

 are strong enough to pull, or from five to six weeks after planting. 



