The procedure that will here be followed consists essentially in a calibration and an extrapolation. The temperature scale from to is regarded as known from spectrophotometric data. Within this range, the theoretical and observed maxima are compared. The possibility of finding a value of appropriate to a given atomic state is next examined. Finally, a method of estimating will be justified for the cooler stars, within the limits of accuracy permitted by the data, and will be extended by simple extrapolation to the formation of a temperature scale for the hotter stars, where the temperatures cannot be safely estimated from the color indices. The salient point is that complete absorption will occur for any line at a depth that is inversely proportional to the abundance of the corresponding state of the atom. No light in this wave-length reaches the exterior from any lower level, and the deepest level from which the line originates therefore forms a lower boundary to the effective portion of the atom in question. The “effective level” from which a line comes is probably best regarded as the level at which the effective atoms above the “lower boundary” have their median frequency. Clearly the partial pressure will differ at different effective levels, and thus abundance has a direct influence on the appropriate value of [Pg 135] the partial pressure. The theory with which we have so far been concerned deals with the excited fraction of the total amount of the element which is present. A knowledge of this quantity suffices for specifying the variation of intensity for the lines of any one element. But the absolute abundance of a given atomic state varies jointly with the fractional concentration of the appropriate state and the total amount of the element present. Now, for the first time, the absolute abundance of different atomic species becomes of possible importance, as a factor affecting the depth from which radiation corresponding to the given atom will penetrate. Fowler and Milne[410] rightly claimed that their method of maxima eliminated questions of relative abundance, “if can be regarded as known ... [and constant]. The proper value of must be a function of the abundance of the atom in question relative to free electrons.”