It is certainly desirable to find tests of the accelerationist hypothesis. The usual test is the value of the coefficient of "expected inflation," for which the proxy is lagged inflation, in a regression explaining money-wage inflation. Professor Fellner's interest in a more direct test is understandable. But, as he is quite aware, simple correlations like his (2) and (4) are not decisive. The trouble is correlation (5), the positive association of AU and U. This leaves us uncertain whether observed acceleration is to be explained by AU in conventional Phillips nonaccelerationist terms. Further calculations are required.