With the addition of the galaxies listed in Table I, an extensive body of material becomes available on the system described in Paper I. At this stage, therefore, the question can be asked: How closely do galaxies classified in the various subdivisions conform to the definitions outlined in Paper I?
In any attempt at classification, three stages can be recognized: (1) The setting up of a system of categories that are defined in an approximate manner - according to certain preconceived ideas; (2) the classification of a body of specimens into the categories outlined; (3) the description of the categories in terms of the specimens classified in each.
In the course of the progression outlined above, developments and modifications in the original scheme result of necessity. An interesting example of this is furnished by the evolution of the Harvard system of spectral classification into the system of the Henry Draper Catalogue. Some categories tend to disappear; others are modified in nature; and, most important of all, the definitions of the surviving groups undergo a certain amount of systematic change.
A dogmatic insistence on the exactness of the preliminary definitions of the classification system is therefore inadvisable; a valid and precise definition of an empirical system of classification can only be in terms of the observed properties of the specimens classified in each category. As the amount of observational evidence increases, the ideas on which the new classification is based are subjected to increasingly critical tests.
The original basis for the Yerkes form system was the spectral classification of galaxies of Morgan and Mayall. 2 In the latter, groups of galaxies were classified into spectroscopic categories of A to K. The galaxies of earliest spectral type were described as "A-systems"; there was no category of "B" galaxies included in the classification. The ultraviolet region of the spectrum of these "A" galaxies generally contains strong absorption lines of hydrogen, and it was considered that the principal contributors to the violet spectral region are main sequence stars of type A.
Spectrograms of higher resolving power obtained more recently at the McDonald Observatory by Mayall and Morgan show that galaxies classified in the "A" category vary greatly among themselves: in some cases (NGC 4490) it seems that B8-A5 main sequence stars are probably responsible for the principal contribution to the light in the ultraviolet region of the spectrum; at the other extreme are galaxies similar to NGC 4214, whose spectroscopic characteristics in the ultraviolet resemble rather closely those of the inner parts of the Orion Nebula. 3 At an intermediate position are galaxies similar to NGC 4449, where the ultraviolet line of He I at 3820 is present. 3 In such systems as NGC 4449 it seems likely that most of the ultraviolet light originates in B1-B3 stars; in this case, there seems to be a resemblance to the population of the Large Magellanic Cloud. 4
Now, if this wide range in spectroscopic appearance is to be taken account of in a classification system of forms, further subdivisions in the system are called for; however, at the present time, it does not appear to be practicable to distinguish the three above-mentioned categories of galaxies of early spectral type from each other by their form characteristics alone. We therefore, in the form classification procedure, continue to describe all three categories as "a-systems." At this current stage in the classification development, the category "a-systems" refers to galaxies whose stellar populations, as observed in the ultraviolet region, range from a situation approximately similar to that of the Orion Nebula region to a population in which the principal contribution to luminosity originates in A- or even early F-stars.
In view of the modifications outlined above - and of the fact that further modifications are almost certain to be made in the future - the justification for the carrying out of the reclassification of forms of galaxies in Papers I and II might be called into question. The present justification lies in the fact that the new classification does effect an approximate separation of galaxies according to their stellar populations; the "a" galaxies do have populations rich, by luminosity, in early type stars and gas; the "k" galaxies have a population rich, again by luminosity, in yellow-giant stars; and the galaxies of intermediate form type tend to have mixtures of the two extreme categories above mentioned. The final decision will depend on how useful the new classification proves to be in the future.
I wish to express my thanks to Dr. I. S. Bowen, Director, and to the Committee of the Mount Wilson and Palomar Observatories for the opportunity to carry out the classification of galaxies given here.
The investigation was supported by a grant from the Office of Naval Research.