In the present paper it is tried to trace the fate of fertilizer phosphorus in soil by comparing the analyses of soils from treated and untreated plots of field trials. This indirect approach cannot be expected to provide exact values, but it is likely to give an approximate answer. The results reported above do not in any marked degree change our present conception of the forms in which fertilizer phosphorus accumulates in soils. In the acid soils studied (pH 4—6.4 in 0.02 N CaCl2) superphosphate tended to increase the fractions which were extracted by NH4F or NaOH. Hyperphosphate phosphorus was mostly found in the acid-soluble fraction. During a longer period of dressing with phosphate an increase in the organic phosphorus content of a peat soil could be detected. In the incubation experiments the mineralization of organic phosphorus occurred at a higher rate in the samples from the plots treated with superphosphate than in those from the untreated one. It might be supposed that the organic phosphorus mineralized mainly originated from the plant residues. It seems that the fractionation method developed by CHANG and JACKSON (4) for the estimation of discrete forms of soil phosphorus is not quite satisfactory for tracing the fertilizer phosphorus in soils recently dressed with phosphates. In particular, it may be fallacious to conclude that the fraction extracted by NH4F would only represent phosphorus bound to aluminium and its compounds. At least in the absence of soil, a large part of phosphorus in dicalcium phosphate dihydrate falls into this fraction, and also a small amount of hyperphosphate phosphorus may be found in it. The test values for »available» phosphorus showed the effect of fertilizers in accordance with previous observations (9, 13). Acetic acid soluble P revealed the treatment with hyperphosphate, but only slightly the application of superphosphate. The test value for the sorbed P of BRAY and KURTZ (2), or phosphorus extracted by 0.03 N NH4F-0.025 N HCI, distinctly indicated the addition of superphosphate, and also to some extent the presence of hyperphosphate phosphorus. Thus, even these results furnish a supplement to the data (11, 19) which prove that acids are not recommendable for the estimation of »available» phosphorus in Finnish soils. Probably, the use of ammonium fluoride would give a more reliable picture, provided, it is on the whole possible to characterize the phosphorus condition of soil with this kind of test values.