A common seed stock for each of four varieties of barley was increased at five locations in 1959 and again in 1960. In 1960 and in 1961, the five seed lots of each variety were tested at several locations. There were significant differences in yield between crops of the same variety produced from different seed lots. In 1960, differences ranged from 9 to 16 per cent, and in 1961 from 3 to 13 per cent. They were such that differences between varieties could be changed materially and yield rankings could even be altered by using seed lots propagated at different locations. There was some suggestion that chemical content of seed from different sources was associated with crop performance at other locations although this was not established.This study serves to emphasize the extreme importance of seed source and the necessity for controlling this or any other non-genetic source of variability in comparative tests.