By inspecting Palomar Sky Survey prints, van den Bergh (1960a, b) made the important discovery that the appearance of the spiral arms was a steep function of the absolute luminosity of galaxies. He showed that galaxies of the highest luminosity have the longest and most highly developed arms, whereas fainter systems such as NGC 5204 (Plate 3) have poorly developed arms.
The result is equivalent to stating that the appearance of galaxies varies systematically as one proceeds from left to right within the scatter of redshift-apparent-magnitude diagrams for field galaxies (see, for example, Figures 6, 7, and 9 of HMS).
By inspecting the character of the spiral arms alone, van den Bergh was able to divide Sb types into five luminosity classes (I, I-II, II, II-III, and III), which subsequent calibration showed were ~ 0.5 mag apart in <Mpg>. The Sc-Irr systems could be divided into eight half-classes (I to V with intermediates).
van den Bergh's system is two-dimensional. It retains the Hubble types a, b, and c as a division along a linear sequence, and adds the luminosity class as the second parameter.
Additional symbols are used such as + and - (following Holmberg) to divide the Hubble classes more finely, n for nebulous arms such as NGC 3623 (plate 4) and NGC 4569 (Plate 4), an asterisk (*) for patchy arms such as NGC 157 (plate 4) and 4088 (plate 5), and t for indications of ``tidal'' distortion. Extreme characteristics are noted by double symbols; incipient features, by (n), (*), and (t).
van den Bergh notes that most of the Hubble-type examples usually illustrated in textbooks, in the Hubble Atlas, and in plates 1-8 are supergiant galaxies, and that the early Hubble classification system defined by such bright galaxies cannot generally be applied to dwarf galaxies without modification. He emphasizes that all dwarf and sub-giant galaxies classified in HMS (1956) are of type Sc, which shows that few if any dwarf Sa and Sb systems exist. This fact is recognized in van den Bergh's system for Sb systems where only class I, II, and III galaxies are present, whereas class V (dwarf) galaxies exist in the Sc to Irr classes.
The validity of the luminosity classes is shown by the clear separation of the classes in the redshift-magnitude relation (van den Bergh 1960a, fig. 4; 1960b, fig. 4). Van den Bergh's preliminary calibration of <Mpg> for each class was based on this separation, adopting H = 100 km s-1 Mpc-1. The calibration is reproduced in table 1. The dispersions of <Mpg> were found to be 0.3-0.4 mag for the well-defined cases (van den Bergh 1960a, table 4), as obtained by comparing differences between the Mpg predicted from the luminosity classification and known values for particular galaxies in groups and clusters.
|SbII||-19.4||Sc and Irr II||-19.4|
|SbII-III||-18.6||Sc and Irr II-III||-18.9|
|SbIII||-18.0||Sc and Irr III||-18.3|
|Sc and Irr III-IV||-18.0:|
|Sc and Irr IV||-17.3:|
|Sc and Irr IV-V||-16.1:|
|From Pub. David Dunlap Obs., Vol. 2, No. 6, 1960.|
The importance of the luminosity classification, regardless of its calibration, is that relative distances can be obtained to large numbers of field spirals within an accuracy of r / r = 0.461 M. This error is small when (M) is, say, 0.4 mag that arises from classification errors and from true cosmic spread. The distances can be changed to absolute values once H0 is accurately known. (Table 1, based on an assumed value of H0, is subject to modification. A calibration in 1972 by Sandage and Tammann gave H0 55, which would require the values in table 1 to be about 1 mag brighter). On a distance scale with H0 = 100 km s-1 Mpc-1, van den Bergh has given absolute moduli for many Shapley-Ames galaxies in his reclassification catalog (van den Bergh 1960).
The van den Bergh classification is defined by the many type-examples given in his catalog, as classified from the Palomar prints. Because of the burned-out nature of many galaxies on these prints, he could usually make no distinction between S0 and E systems, and many flattened systems were classed E with flattenings ranging to 8. Such large flattenings do not, of course, exist in the standard system because E galaxies are never flatter than E7, and very few of these exist (cf. Sandage, Freeman, and Stokes 1970). Many of the flattened E8 so classified in the catalog are actually edge-on spirals and S0 galaxies, which, for this reason, are not represented in the DDO Catalogue. Comparison of the DDO types with other classification lists is made by de Vaucouleurs (1963b, table 10 and fig. 5).