Tables 2 and 3 contain experimental estimates of the power of the para- metric test for n = 30 and n = 60 respectively. Each case represents 100 repli- cations of the experiment for a particular choice of a sample of x’s. For each case we rank the powers for the five values of k, the number of omittedcentral observations. If Rendall’s coefficient of concordance W is calculated for each table, duly taking into account the number of ties, we find W= .379 for Table 2