Winter wheat grown following potatoes on a sandy loam at Woburn in 1978–9, 1980–1 and 1981–2 was compared with that on a clay loam at Rothamsted in 1978–9 and 1980–1, and on a silty clay (alluvium) at Woburn in 1981–2. The cultivar was Hustler in the harvest years 1979 and 1981 and Avalon in 1982. On each soil in each year multifactorial experiments tested effects of combinations of six factors, each at two levels.

The best 4-plot mean grain yield ranged from 89 to 11·1 t/ha during the 3 years; it was smaller on the sandy soil than on the clay soil in 1979, but larger on sand than on the clay in 1981 and 1982. Until anthesis the number of shoots, dry weight and N content of the wheat giving these best yields were less on sand than on clay. Unlike grain weight, straw weight was always less on sand.

Sowing in mid-September instead of mid-October increased grain yield on clay in each year (by 0·4·0·7 t/ha) and increased yield on sand only in 1981 (by 1·6 t/ha). Early sowing always increased dry weight, leaf area, number of shoots and N uptake until May. The benefits were always greater on clay than on sand immediately before N fertilizer was applied in the spring and usually lessened later on both soils.

Aldicarb as an autumn pesticide increased grain yield of early-sown wheat on both soils in 1981 by lessening infection with barley yellow dwarf virus. Aldicarb increased yield on clay in 1982; it also decreased the number of plant parasitic nematodes.

Wheat on sand was more responsive to nitrogen in division, timing and amount than was wheat on clay. In 1979 yield of wheat on sand was increased by dividing spring N between March, April and May, instead of giving it all in April, and in 1982 by giving winter N early in February. In 1981 division and timing on sand interacted with sowing date. Yield of early-sown wheat given N late, i.e. in March, April and May, exceeded that given N early, i.e. in February, March and May, by 1·4 t/ha; single dressings given all in March or all in April also yielded less than the late divided dressing. Yield of later-sown wheat given all the N in April was at least 1·2 t/ha less than with all N given in March or with divided N. In all years treatments that increased yield usually also increased N uptake. Grain yield on clay was never affected by division or timing of spring N or by application of winter N. This was despite the fact that all treatments that involved a delay in the application of N depressed growth and N uptake in spring on both sand and clay. The mean advantage in N uptake following early application of spring N eventually reversed on both soils, so that uptake at maturity was greater from late than from early application. Increasing the amount of N given in spring from the estimated requirement for 9 t/ha grain yield to that for 12 t/ha increased yield in 1982, especially on sand. The larger amount of N always increased the number of ears but often decreased the number of grains per ear and the size of individual grains.

Irrigation increased grain yield only on the sandy soil, by 1·1 t/ha in 1979 and by 07 t/ha in 1981 and 1982. The component responsible was dry weight per grain in 1979 and 1982, when soil moisture deficits reaching maximum values of 136 and 110 mm respectively in the 2 years developed after anthesis; the component responsible was number of ears/m2 in 1982 when the maximum deficit of 76 mm occurred earlier, in late May.