A technique is, described for securing rapid approximate tests of significance in comparison of 2 samples equal in size. Observations are ranked after pooling and, for unpaired experiments, the probability of a total sum of ranks equal to or lower than the sum of the lowest sample is determined. For paired expts. the differences are assigned rank numbers neglecting signs and the probability of a sum equal to or less than that of the ranks corresponding to the negative differences is determined. Tables for the probabilities are given.