Sects. 12, 13. 89 sequence and that subgiant and fainter stars in globular clusters have ultraviolet excesses. When dealing with stars whose physical properties are imperfectly under­ stood, such as in globular cluster stars, we cannot rely too heavily on the empiri­ cal calibration by the kinds of stars used to define Fig. 5, to determine their true, unreddened U-B, B-V curve. But if by a combination of arguments, principally the reddening in the region of the stars we do known about, we can assign a fairly probable unreddened U-B, B-V curve to a group of stars about which we know little, the argument may be turned around. In this case some information may be gained about the energy envelope of the stars by examining the differences between the normal two-color index curves for the unknown group of stars compared to the known. In general there seem to be two possible causes for different stars defining different normal sequences in the U-B, B-V plane. One, the relative energy distribution in the continuum in the U, B and V photometry bands are different. An example of this is the effect of the Balmer depression in supergiants. This, of course, requires deviation from black body radiation curves for one or both groups of stars. This cause seems to be the dominant effect for very blue, hot stars where the depression of the continuum by absorption lines is at a minimum.