Gérard de Vaucouleurs's revised Hubble system is described in a lengthy Handbuch der Physik article published in 1959. It was not the only revision of the Hubble system available at the time: Hubble himself had been working on a revision since the 1930's, but died before completing it. In 1961, Allan Sandage published Hubble's revision in the form of an atlas. Both systems are still in active use today, and there are good reasons for this: the original Hubble system provided the first ordering of galaxies that represented more than just smooth changes in the properties of optical morphology. As data began to be accumulated for large numbers of galaxies, many physical parameters (e.g. colors, HI content, concentration indices, surface brightnesses) were found to correlate with these changes, implying that the Hubble sequence contains information on the basic physics of galaxies.
Although de Vaucouleurs's revision followed Hubble's framework in placing galaxies along a sequence of "stages" and in recognizing bar and ring morphologies, his view of the spread in morphologies at a given stage was clearly different from Hubble's: instead of a multi-pronged fork, de Vaucouleurs proposed to describe galaxy structure with a three-dimensional classification volume whose principal axis was the stage and whose secondary axes were the "family" and the "variety." These latter characteristics refer, respectively, to the presence or absence of a bar and of an inner ring. Continuity was implied along each axis. To highlight this, de Vaucouleurs introduced special notation to allow for transition cases. For family, SB would still denote barred spirals while SA would denote non-barred ("ordinary") spirals. Intermediate barred cases are denoted SAB, SAB, or SAB, depending on the degree of development of a bar. The same kind of notation was used for variety: pure ringed systems were denoted (r) while pure spiral (s-shaped) systems were denoted (s), with objects of intermediate characteristics highlighted by a compound notation such as (rs), (rs), and (rs).
For stage, de Vaucouleurs adopted two refinements. First, transition stages, Sab and Sbc, between the original Hubble spiral categories Sa, Sb, and Sc were recognized to eliminate the need for notation such as Sa+, Sb-, etc. Second, more stages were added along the Hubble sequence beyond Sc to allow for intermediate spiral stages between Sc and Irr I. De Vaucouleurs adopted Shapley and Paraskevopoulos's (1940) suggestion of the need for an Sd classification (very late, small-bulge systems like NGC 7793), and identified the Magellanic Clouds as an even later stage along the spiral sequence, Sm. Transition stages were denoted as Scd and Sdm. The true Irr I galaxies were denoted Im. S0 galaxies were divided into three categories, S0-, S0°, and S0+, but these do not correlate exactly with the three S0 categories in Hubble's revision (Sandage 1961). As in Hubble's revision, S0 galaxies are placed between E and Sa galaxies on the stage sequence, the transition stage to spirals being denoted S0/a. The various combinations and permutations of symbols possible gave de Vaucouleurs's classification system a complexity unprecedented in galaxy morphology studies, but as he noted, symbols can be dropped according to the amount of detail visible in an image or the desired level of complexity.
The availability of two revised Hubble systems means that care must be exercised in interpretation of published Hubble types, particularly the stage parameter where the stage "Sc" in the Hubble revision encompasses galaxies classified over the wide range of stages, Sbc-Sm, in de Vaucouleurs's revision. The arrangement of the late spiral stages and the validity of de Vaucouleurs's extensions have been repeatedly demonstrated from the correlations found between type and a wide variety of measured parameters (de Vaucouleurs 1977; see also Figure 1). The extensions are therefore fully justified and well-founded, and form one of the major advantages of de Vaucouleurs's revision over Hubble's revision.
Figure 1. Dependence of several photometric parameters on the stage in de Vaucouleurs's system. Here HI refers to the hydrogen index and µ'eo is the mean surface brightness within the effective isophote. Based on new data from the Third Reference Catalogue of Bright Galaxies. Triangles are median values, while circles are means. Stages Sd and Sdm are combined here.
Care must also be exercised in the interpretation of the part of the classifications involving bars and rings. Recently, much research has focussed on these secondary structures; on their classifications, independent observers are not very consistent. For example, some galaxies classified as (r)-variety by de Vaucouleurs (1963) are classified as (s)-variety by Sandage and Tammann (1981; ST) because the inner ring appears to be made of very tightly wrapped spiral arms. Hubble's revised classification system, as applied by ST, has very broad categories along the spiral sequence that encompass a wide range of forms, yet the ring category they envision is extremely conservative (as is also the bar category). I believe that overconservatism leads to unrealistically sharp edges to the morphological "cells" that may be misleading; the cell boundaries are in reality rather fuzzy. The advantage of de Vaucouleurs's revision is that it views galaxy structure in the broadest possible way, that is, as a continuum of forms in all possible dimensions.
Surprisingly, many astronomers today still persist in adhering to the original Hubble revision, perhaps because it is simpler to understand and apply in practice, and because it serves their needs. However, modern extragalactic astronomy demands a system with greater versatility and applicability, which can serve as a basis for the type of research that can now be done with the sophisticated telescopes and instruments currently in use.