Experiments at three sites in 1993, six sites in 1994 and eight sites in 1995, mostly after oilseed rape, tested effects of previous fertilizer N (differing by 200 kg/ha for 1993 and 1994 and 300 kg/ha for 1995) and date of sowing (differing by about 2 months) on soil mineral N and N uptake by winter wheat cv. Mercia which received no fertilizer N. Soil mineral N to 90 cm plus crop N (‘soil N supply’; SNS) in February was 103 and 76 kg/ha after large and small amounts of previous fertilizer N respectively but was not affected by date of sowing. Previous fertilizer N seldom affected crop N in spring because sowing was too late for N capture during autumn, but it did affect soil mineral N, particularly in the 60–90 cm soil horizon, presumably due to over-winter leaching. Tillering generally occurred in spring, and was delayed but not diminished by later sowing. Previous fertilizer N increased shoot survival more than it increased shoot production. Final shoot number was affected by previous fertilizer N, but not by date of sowing. Overall, there were 29 surviving tillers/g SNS.

N uptakes at fortnightly intervals from spring to harvest at two core sites were described well by linear rates. The difference between sowings in the fitted date with 10 kg/ha crop N was 1 month; these dates were not significantly affected by previous fertilizer. N uptake rates were increased by both previous fertilizer N and late sowing. Rates of N uptake related closely to soil mineral N in February such that ‘equivalent recovery’ was achieved in late May or early June. At one site there was evidence that most of the residue from previous fertilizer N had moved below 90 cm by February, but N uptake was nevertheless increased. Two further ‘satellite’ sites behaved similarly. Thus at 14 out of 17 sites, N uptake until harvest related directly and with approximate parity to soil mineral N in February (R2 = 0·79), a significant intercept being in keeping with an atmospheric contribution of 20–40 kg/ha N at all sites.

It is concluded that, on retentive soils in the UK, SNS in early spring was a good indicator of N availability throughout growth of unfertilized wheat, because the N residues arising from previous fertilizer mineralized before analysis, yet remained largely within root range. The steady rates of soil mineral N recovery were taken as being dependent on progressively deeper root development. Thus, even if soil mineral N equated with a crop's N requirement, fresh fertilizer applications might be needed before ‘equivalent recovery’ of soil N, to encourage the earlier processes of tiller production and canopy expansion. The later process of grain filling was sustained by continued N uptake (mean 41 kg/ha) coming apparently from N leached to the subsoil (relating to previous fertilizer use) as well as from sources not related to previous fertilizer use; significant net mineralization was apparent in some subsoils.