The relation between the revised types of the B2-G0 dwarfs and color is shown in Figure 1. An approximate calibration of the luminosity classes is given in Figure 2. While any definitive calibration requires the use of many more stars than are considered here, we do not think that any of the curves should be in error anywhere by much more than a half-magnitude.
Since about a year was needed for the making of the photographic prints for the Atlas, there is a difference in epoch of that time between the classification as illustrated there and as expounded here. It was unavoidable that certain improvements and alterations should have suggested themselves in the interim. These have been incorporated in the text; and there are therefore several discrepancies between the Atlas plates and the text. In all such cases the text is to be taken as final, and the data on the Atlas prints should be altered to agree with the outline. The most important of the changes has been the shifting in spectral type of two standard stars. These are µ Peg (Plates 36, 41, 44), whose type should be changed from G5 to G8, and 2 UMa (Pl. 37), whose type has been altered from F8 to F6.
The characteristics of the system described here can be summarized as follows: The two-dimensional classification can be used to describe accurately the spectra of the normal stars brighter than the eighth apparent magnitude. Since this includes all but a very small percentage of the total number of stars brighter than that limiting magnitude, it is possible to derive from the extension of the classification to fainter objects certain general information concerning the distribution in space of the stars absolutely brighter than the sun.
In the course of the investigation several interesting details have been noted. Among the Be stars very broad absorption lines have been observed, which suggest maximum stellar rotational velocities somewhat higher than those found earlier. The most striking example of this is the star Per. Others stars having lines suggesting higher rotational velocities than the Bnn star, UMa, are Oph, 25 Ori, and Mon A.
Figure 1. Color equivalents of B2-G0 main-sequence stars. The photoelectric color indices of Bottlinger (above) and Greenwich gradients (below) are plotted against the spectral types of the present Atlas. The stars included are those of luminosity classes IV and V which appear to be definitely less than 100 parsecs distant from the sun. The same stars are plotted in the two diagrams for types earlier than F5. Stars of class V only are shown for classes F6-G0. The multiple system UMa has not been plotted.
The two relationships between color equivalent and spectral type are not similar; a simple change of zero point and scale will not suffice to change one color system to the other. There is a marked depression in the curve for the early A stars in (a) which is not present in (b). The curve in (a) is definitely concave upward from B8 to F5, while it is sensibly linear in (b). This difference is interpreted as an effect of the hydrogen lines on the violet wave lengths for the photoelectric color indices. The same effect is present to a varying degree in other catalogues of color equivalents. The two straight lines connect the centers of gravity at B8-B9 and F0-F5.
In the G and K stars other spectral features appear to affect observed color equivalents. In particular, the strong absorption due to CN in giants tends to increase the color differences between giants and dwarfs observed with short base-line photoelectric color indices. In the K stars of high-luminosity a heavy absorption extending toward the violet from the vicinity of 4300 cannot fail to have an appreciable effect on colors determined in this region.
Figure 2. Preliminary calibration of luminosity classes in terms of visual absolute magnitude.
Also of interest is the discovery of similar spectral peculiarities in several G- and K-type high-velocity giants. The high-velocity stars Lep, Boss 2527, Leo, and probably a Boo have similar peculiar features. The most striking of these on low dispersion is the abnormal weakness of the CN absorption extending toward the violet from 4215.
When carefully calibrated, the luminosity classification should allow the determination of accurate spectroscopic parallaxes on low-dispersion plates of stars of all classes from O9 to M2 (with the possible exception of classes B8-A2).
The spectral classification defines with accuracy a system of color standards which can be used in investigations of interstellar absorption and determinations of systematic errors in spectral classification of faint stars. It should be emphasized that the actual features used for classification are dependent on the dispersion used and that some or most of the criteria listed here might be unsuitable for use on spectra having greatly different dispersion.
We wish to acknowledge our indebtedness to the following persons: to Dr. Struve for making the publication of the Atlas possible; to Dr. Joaquin Gallo, director of the Astronomical Observatory of Mexico at Tacubaya, for the loan of a number of objective-prism plates; to Dr. A.H. Joy, of Mount Wilson, for determining the spectral types of several M giants which we have used as standards; to Dr. A.N. Vyssotsky, of the Leander McCormick Observatory, for several discussions of the problem of spectral classification; and to Dr. G.P. Kuiper for a discussion of the dynamical parallax of Cygni. We are also indebted to the following persons for taking a considerable number of the spectrograms used in the investigation: Mrs. Frances Sherman Bailey, Dr. J.A. O'Keefe, Dr. L.R. Henrich, Mr. W.P. Bidelman, and Mr. Frank R. Sullivan. All the photographic prints for the Atlas were made by Miss Kellman and Miss Phyllis Anderson.