Supplementary irrigation experiments were conducted on a Clansthal sand and a Windermere clay. Irrigation treatments were based on the application of 1 inch of water when the soil moisture deficit reached 2·5, 3·5, 4·5 and 5·5 in. in 7 ft of soil on the sand, and 1·00, 1·75, 2·50 and 3·25 in. in 2 ft of soil on the clay. Each experiment included a dryland control and soil moisture was estimated weekly in each plot with a neutron probe. The results showed that there were significant linear increases in yield with increasing quantities of water applied. The crops exploited soil moisture to a depth of 6 ft in the sand and 3 ft in the clay, even with the maximum water treatments. Under dryland conditions, on the sand the crop exploited soil moisture to the wilting point only in the surface foot of soil, but on the clay the crop extracted moisture held at tensions greater than 15 bars in the top 3 ft of soil. The conclusions are that, where available water for supplementary irrigation is limited, heavy soils should be irrigated in preference to sandy soils, and that the most efficient use of both rainfall and irrigation water is achieved when 1 cusec is used to irrigate about 200 acres.