104 
Journal of Agricultural Research 
Vol. XXV, No. 2 
the row next to a mosaic hill. The other was obtained at Aroostook 
Farm in a rogued plot about 15 meters from the nearest unrogued 
mosaic hills. In each place 10 hills were dug on each of five dates—• 
August 2, 11, 20, 30, and September 9. The progeny of the five groups 
of hills were mosaic respectively in 67, 68, 70, 89, and 93 per cent in 
the laboratory stock and in o, 7, o, 2, and 26 per cent in the farm stock. 
Therefore, in the laboratory hills, next to mosaic ones, there was a high 
percentage of mosaic infection by August 2 and probably only chance 
escaping of inoculation prevented all hills from becoming diseased by 
August 30. In the farm plot, 15 meters from mosaic hills, there was 
little infection until September 9, and then much less than at the labora¬ 
tory. The two sets of hills were not only of the same variety but to a 
large extent of the same strain. Evidently, proximity to mosaic hills 
greatly increased infection, apparently through easier dispersal by 
virulent insects. The small amount of infection that occurred at the 
farm and that was manifested chiefly in the last harvest probably was 
due to virulent insects, soil and other factors appearing to be negli¬ 
gible as possible causes. 
An experiment parallel to the preceding consisted in removing one 
tuber from every one of 30 hills in each of the two places on each date 
except the last, when every hill was dug usually with more than one 
tuber. The results were the same except that the percentage of infec¬ 
tion generally was higher. Another parallel experiment consisted in 
selecting apparently healthy hills next to mosaic ones in a field, at the 
farm, that contained Green Mountain plants with 80 per cent mosaic. 
One hundred pounds of tubers were dug on August 7, and a like quantity 
again on September 8. The progeny of the two lots were, respectively, 
46 and 84 per cent mosaic. These results are similar to those of the 
series of harvests made at the laboratory. 
In 1920 conditions of uncontrolled field transmission were presumably 
varied by the selection of hills at different times and at different places 
on the same farm. The data are presented in Table XXVII. 
In field A, proximity to mosaic hills increased the amount of infection 
and proximity to leaf-roll hills was necessary for infection. The sudden 
increase of infection from the August 12 harvest to that of August 23 is 
to be explained as follows: In this field certain recommended methods of 
spraying for aphid control were being tested and meanwhile the approxi¬ 
mate number of aphids was determined and recorded frequently. About 
95 per cent of these insects were potato aphids (Macrosiphum solani- 
folii Ashm.). These were present by July 17, increased most rapidly 
during the second week in August until there were over 2,000 to a plant on 
the average, and decreased rapidly and disappeared during the third 
week in August mostly because of a fungus disease that followed the 
inception of a period of cloudy, humid weather. The aphids of the 
other species (determined as Aphis sp. by Dr. Edith M. Patch), were 
extremely localized on occasional plants, being apparently slow in 
dispersing. Presumably, the greatest dispersal of aphids occurred 
about August 13, when they were most numerous and when the second 
harvest of tubers was completed. Apparently the high level of tuber 
infection was reached by August 23, in the healthy hills adjacent to dis¬ 
eased ones, indicating a 10-day period to be necessary for diffusion of 
the virus to the tubers if the chief cause of infection were the potato 
aphids. This is the time for such diffusion as shown on page 97. The 
