Abstract
The value of oil analysis tests (five each for nitrogen, phosphorus and potassium) was assessed by correlating test value with yield data from 87 field PK fertilizer experiments using maize (Zea mays L.). Methods were selected on their ability to predict correctly crop response to applied fertilizer using the appropriate critical test values. Total nitrogen was as good as any other test for available soil nitrogen, and using a critical test value of 0·17 per cent was able to give 74 per cent correct predictions of response. Equally reliable estimates of response to nitrogen fertilizer could be obtained from the percentage base saturation of the oil; a critical test value of 65 per cent base saturation allowed 77 per cent correct predictions. A modified Bondorff method (l g soil shaken for 16·5h with 50ml 0·2N H2SO4) is recommend d for soil phosphorus; this method, and also Olsen's 0·5 M Sodium bicarbonate extract gave 80 per cent correct predictions. The critical test value is 35 p.p.m. P for Bondorff' method and 2·5 p.p.m. for Olsen's. Exchangeable potassium removed in a neutral 1 NH4OAc leachate is suggested for routine measurements. The critical value of 0·16 mequiv. K per 100g soil gave 90 percent correct predictions. These method are recommended for use in routine analysis of farmers' samples in the region as a basis for a fertilizer advisory service.
