The effects of the applications of P on the sorption of newly added P differed between sites. On 5 of the 8 sites, the slope term (a) of the fitted equation decreased with increasing levels of previously applied P. Consequently, the P buffering, measured as P sorbed between 0·25 and 0·35 mg P/L, decreased. When the decrease in P buffering was large, the indirect measures of buffering capacity (oxalate-extractable aluminium, oxalate-extractable iron, and the pH of the soil measured in sodium fluoride) also decreased.

On the other 3 sites, the slope term did not decrease significantly. Indeed on one site there was a small increase. On these 3 sites, the effect of the P application on the sorption curves was largely shown by increases in the intercept term (q). The indirect measures of P buffering sometimes increased with increasing application of P.

On all sites, both the P sorbed at 0· 3 mg P/L and the values of PRI decreased with increasing P applications. This is because their values are determined by the values of both a and q. We argue that the decreases in P buffering observed on 5 sites would increase the effectiveness of subsequently added fertilisers and would limit the ability of such soils to act as sinks for phosphate disposal.

We suggest that the differing behaviour on the other 3 sites may have occurred because there was little continuing reaction between the soil and P.