Eleven years of field experiments on manuring sugar beet were used by the late E. M. Crowther to compare methods of analysing soils for soluble P and K; this paper reports his work.

The experiments were divided by soil analyses into groups of equal numbers of sites; average crop responses were used to value the analytical methods. Such tables of average data overvalue soil analyses because each method was misleading in a small proportion of the fields used. A quantitative way was developed of assessing the gains from using soil analysis in planning fertilizing and of comparing analytical methods. The total profit from uniformly manuring all the soils examined was compared with the profit made by using analyses to select a proportion only of the soils to be manured; the total amount of fertilizer used was the same with each way of planning manuring. The most efficient analytical method gave the most profit.

For phosphate, differential manuring of sites selected by soil analysis was more profitable than uniform manuring, with all the methods of measuring soluble P that were tested. The best method used a rapid extraction with dilute hydrochloric acid, but extracting with water (calcium bicarbonate solution was used for calcareous soils), or with citric acid solution, was nearly as effective. These three methods, using little solvent relative to soil, were more useful than methods using larger volumes of dilute sulphuric acid, dilute acetic acid, and a lactate solution.

The sugar beet responded more often to K than to P manuring, so there was less chance of making extra profit by using analyses to detect the richer soils where either no K fertilizer, or small dressings only, should have been given. Using the citric acid method of measuring soluble-K to separate the soils into groups for differental manuring was more profitable than giving uniform dessings to all fields; acetic acid was less effective than citric acid, and the hydrochloric acid method gave no advantage over uniform manuring with the heavier rate of potassium. Water-soluble K measurements were worse than acid-soluble values; using them to predict responses and manuring would have given less profit than uniform manuring at the heavier rate used.