The results of 216 field experiments with sugar beet (1957–69), testing K and Na fertilizers, were re-examined. There were only small increases in yield with K or Na on peat or peaty mineral soils but on mineral soils, 100 kg K/ha without Na increased sugar yield by about 0·4 t/ha and 100 kg Na/ha without K increased yield by 0·6 t/ha. Although the average response per kg of nutrient was approximately in inverse proportion to their atomic weights, K and Na were only partially interchangeable; their relative effects differed from field to field and year to year.

K seemed most effective in years with more than 80 mm rainfall in March–April but responses were halved when rainfall was 50 mm or less. Sowing date and the interval between giving fertilizer and sowing also influenced response to K; with late-sown crops, K fertilizer given within two weeks of sowing was usually ineffective. Response to Na differed much less with season so Na was on average more effective than K, especially in dry springs. It seems, therefore, that instances of near-perfect substitution between K and Na fertilizers result from chance combinations of weather and crop husbandry.

Since in some experiments yields from a combined dressing exceeded those from either nutrient alone, and as the weather is largely unpredictable, both K and Na fertilizers should be given on all mineral soils, except those very rich in K, to be certain of avoiding yield loss. Texture and exchangeable soil K and Na also affected both the amount and rate of response, but did not much influence optimal fertilizer dressings and so are not very useful for making differential fertilizer recommendations.