Modern photoelectric techniques yield magnitudes and colors of stars with accuracies of the order of a few thousandths and a few hundredths of a magnitude respectively. Hence for star clusters it is possible to derive highly accurate color-magnitude arrays since all of the members of a cluster may be considered to be at the same distance from the observer. It is much more difficult to do this for the nearby stars where all of the objects concerned are at different, and often poorly determined, distances. If one depends upon trigonometric parallaxes, the bulk of the reliable individual values will refer to main sequence stars, and while the mean luminosities of brighter stars are given reasonably well by this method, the scatter introduced into a color-magnitude array by using individual trigonometrically determined luminosities could obscure important features. Somewhat similar objections could be raised against the use of the usual spectroscopic parallaxes which also should be quite good for the main sequence but undoubtedly exhibit appreciable scatter for some, at least, of the brighter stars.