Varying amounts of potassium were applied five times a year to a permanent pasture on sandy soil, a different part of the field being chosen on each occasion. The potassium availability of the soil at the time of potassium application was kept at approximately the same level. The effect and after-effect on yield and K20 uptake of the potash fertilization is studied. The uptake-yield curves, the fertilizer rate-uptake curves and the fertilizer rate yield curves of this first to fifth cut are shown in the first, fomth and second quadrant of figure 1. It is found that : a the relation between uptake and yield differed in successive cuts, but was the same as regards the effect and after-effect of the potassium dressing ; b the effect of potash fertilization in autumn depended to a large extent on the withdrawal of potassium with previous cuts ; c the existence of a causal relation between temperature and the effect of potassium fertilization is not proved ; d the availability of potassium at a K-value of about 16 was almost sufficient to enable any growth to take place ; e the uptake from soil was not determined by growth ; the recovery of the potassium fertilizer was to a large extent dependent on growth ; g except for the first cut, the uptake was greater than that needed for reasonable growth ; h the most favourable treatment would have been a dressing of about 200 kg K20 per ha in spring only.