950 
Journal of Agricultural Research voi. xxm, No. xa 
the segregation and recombining of characters makes the method of 
hybridization with the subsequent selection of seedlings a more promising 
method for potato improvement. It has frequently been assumed that 
sufficient variation existed in all varieties to justify the grower in attempt¬ 
ing improvement by asexual selection. 
METHODS AND MATERIALS 
In the fall of 1917 six lots of Early Ohio potatoes were secured from 
growers in different parts of Minnesota. Five of the lots were from 
growers who had practiced little or no selection. The sixth was from a 
grower who has practiced continuous mass selection of tubers for vigor 
and type in the same seed stock for approximately 20 years. In 1919 
a seventh lot of Early Ohios was obtained from a grower who had prac¬ 
ticed continuous mass selection of tubers for vigor and type in the same 
seed stock for 21 years. The growers differed in their ideas of what 
constituted the ideal type of the Early Ohio variety and had selected 
toward divergent types. Attention is called here to the fact that 
these two seed stocks were known to be separated for more than 20 
years. This would seem to allow ample time for strains to develop 
within the seed stocks if there was a tendency for this condition to occur. 
The various lots represented distinct regional types found within the 
state. While it was recognized as a possibility that these seed stocks 
mi ght not be asexual progeny of the same variety, it was still considered 
that a study of them would be of interest in determining the value of 
selection as practiced by the growers. 
EFFECT OF SELECTION AND ENVIRONMENT ON THE POTATO PLANT 
EFFECT ON PRODUCTIVITY 
Results are presented in Table I showing the original source of the 
seed stock, the years in which the lots were tested, and their yield for 
each year at the places tested. The exact number of years the seed stock 
was grown at the place from which it was obtained is not definitely known, 
further than that lots 2 and 9 were grown at the places indicated for 20 
or more years. 
The 1917 test was carried on at University Farm, Duluth, Grand 
Rapids, and Crookston. The yields as presented in Table I, column 
III, are computed on the acre basis. The actual plot grown for each lot 
and locality consisted of a 4-rod row plot. The experimental error in 
such a test is large and the yield differences of the six lots grown are 
probably not significant as no lot was a consistent high or a consistent low 
yielder. In column IV are shown the results of the yields for 1918. 
These yields represented plots of three 4-rod rows computed to the acre 
basis. The three 4-rod rows represent seed stocks of each lot from three 
places—Duluth, Grand Rapids, and Crookston. The results are similar 
to those obtained in 1917 in that no lot gave a consistently high or low 
yield. In column V are shown the results obtained in 1919. The size 
of the plots at University Farm and Grand Rapids were the same as in 
1918. At Duluth the yield given is the average yield computed to bushels 
per acre of two plots, each plot consisting of two 4-rod rows. As in 1917 
and 1918, no lot gave a consistently high yield. Lot 7, however, gave a 
consistently low yield. The cause of this low yield is not clear. It could 
