onwards) stars.

The table that follows contains, in concise form, the chief features by which the type stars of each class are to be recognized, although it is again emphasized that these were not actually measured as criteria for the Draper classes. The lines characteristic of each class serve, however, to specify its degree of thermal ionization.

The homogeneity of the spectra in a given class is striking, and the fact that large numbers of stars display exactly similar spectra has a significance—considered in another chapter[504]—to which the classification problem cannot do more than call attention. The similarity of the spectra becomes the more striking when it is remembered that the range of conditions embraced within any one class is very wide; the ratio in mean density may be as great as 10[505] between stars of the same class but of differing absolute magnitude.[506]

The close spectral similarity between, giants and dwarfs, in spite of the great differences in physical conditions, should not, however, be misinterpreted. The observed facts are in exact accordance with what might have been anticipated. In the first place, thermal ionization is governed by the surface gravity, and only indirectly by the mean density.[507]

[TABLE XXX]

Class Characteristic Lines
OHHeSi+++C++N++He+
HHeSi+++He+O+
HHeSi+++ O+*
H He*
HHe
HHeSi+Ca+
HHeSi+Ca+Mg+
HHeSi+Ca+Mg+
H* Si+*Ca+Mg+
HCa+ Mg+*CaFe
HCa+Mg+CaFe
HCa+CaFeTi
HCa+CaFeTi
HCa+CaFeTiG bd.
HCa+CaFeTiG bd.
HCa+CaFeTiG bd.
H Ca+*Ca Fe* Ti* G bd.*
HCa+ Ca*FeTiTiO₂
HCa+CaFeTiTiO₂
HCa+CaFeTiTiO₂

The division into “arc,” “spark,” and “superspark” is clearly shown by the table. Maxima of the lines which are used as criteria of class are marked with an asterisk.

It is shown in [Chapter III] that the range in surface gravity is far smaller than the range in mean density. Secondly, the basis of the classification has been shown to be the degree of thermal ionization.[508] Granted that the value of the partial electron pressure is low enough, in dwarfs as well as in giants, for thermal ionization to predominate over ionization by collision, a mass of gas will pass through the same succession of ionization-stages with changing temperature, whatever the surface gravity. Any given stage of ionization will, however, be reached at a lower temperature, the lower the pressure, since, as pointed out in [Chapter X],[509] lowered pressure tends to increase the degree of ionization, and will help to produce a given degree of ionization at a lower temperature.

The Draper system takes no direct account of temperature. It classifies purely by degree of ionization, and therefore, as it relates to atmospheres in which the surface gravities differ widely, it will produce classes that are not homogeneous in temperature; dwarfs will be hotter than giants of the same spectral class. Fowler and Milne[510] anticipated a difference of from 10 to 20 per cent, and differences in this sense and of this order actually occur.[511] Physically it seems to be more important to class together stars having the same atmospheric properties than stars at exactly the same effective temperature, although the latter might conceivably be better suited to some purposes.

Although giant and dwarf stars may be found with very similar spectra, it is well known that they display important differences for individual lines, and these differences have formed the basis for the estimation of spectroscopic parallaxes.[512] If the spectrum of a giant star is compared with the spectrum of a dwarf of the same temperature, the two will be found to differ. The line-intensities in the spectrum of the dwarf will place it in a spectral class nearer to the red end of the sequence—if the giant is of Class