In a recent article in this Journal, Professors Soldofsky and Miller (hereafter referred to as S & M) present ex post annual holding period return data for a spectrum of long-term securities ranging from government bonds to speculative common stocks for the years 1950 through 1966. They utilize this data to appraise the "risk-return" tradeoff which prevailed over this seventeen-year period and several subperiods. Finally, S & M present "annual yield profiles," or discussions of the impact of changing patterns of real and financial relationships upon realized security yields during these years. In preparing their ex post yield series, S & M have performed a scholarly task of substantial value to students of finance. However, S & M's contribution towards eliminating the paucity of empirical evidence regarding the tradeoff between return and risk on long-term securities, as well as their related conclusions, may be questioned on both analytical and theoretical grounds. After having discussed the nature and some limitations of their data (but not the necessarily tentative relationship between ex post returns and ex ante expectations), S & M present the geometric mean annual return and associated standard deviation of annual returns for each of their fifteen risk classes of long-term securities for the 1950-1966 period and several subperiods. Their risk classifications are based upon agency ratings of security quality. S & M then regress the standard deviations of return from the various classes of securities on the mean rates of return; plots of the resulting equations constitute their risk-premium curves. From this analysis they conclude: (1) approximately 91% of the variation in the standard deviations of yield among these instruments was explainable by reference to their mean yields; (2) over this period of time a zero rate of return would have been associated with a standard deviation of returns of 3.5%, or conversely, a risk-free (zero standard deviation)... [ABSTRACT FROM AUTHOR]