The relationships between classes of natural objects rely on models or theories. These relationships are not directly obtained by observations. It is always possible to say that classes have always existed, but this is sterile and unsatisfactory. The question is not really how these classes are composed, but rather why there appeared as such. Evolution seems somehow unavoidable since “appearance” does not occur from nothing, it is a transformation of something else.
When Hubble established his classification, he tried to understand the origin of the different classes by relying on physics. He found that elliptical galaxies should progressively become disky because of the diffusion of the angular momentum. He thus postulated the evolution of the angular momentum as the driver of galaxy diversification. So he built the first evolutionary scenario for galaxies known today as the “Hubble diagram” or the “Tuning-fork diagram”, that is still the only global evolutionary scheme for galaxies even if it does not include all of them and the sense of evolution is not the same (in fact it would take a time longer than the age of the Universe for the process to happen).
It is here interesting to note that the Hubble diagram can be computed as a cladogram with cladistics (see my post) assuming that ellipticity and the presence of arms and bars contain an evolutionary information.
The Hubble classification and the Hubble diagram (also called “Hubble sequence”) should not be confused. The Hubble classification is a synthesis of the observations, while the Hubble diagram is a model applied onto this classification. The morphological classification has not changed (the general morphologies of the galaxies have not changed since Hubble’s work!), but the evolutionary models has become more complicated. However, one can sometimes see in the literature some modeling of the Hubble diagram (a model of a model!).
A good illustration of the confusion between classification and evolutionary scenario is the representation of more sophisticated morphological classifications of galaxies that were devised since the first work by Hubble (see a good review in VAN DEN BERGH 1998). They all try to refine the original classification to follow the progress made with the images but they are essentially all represented on the tuning-fork diagram even though no model explains the relationships between two sub-classes.
May be the reason is to be found in the fact that the elliptical galaxies have nearly no visible structures, while spiral galaxies can have very fine details. The orientation of the Hubble diagram goes from “simple” objects to more “complex”ones and this has been preserved more or less unconsciously in the refined “classifications”. This can appear as a hidden assumption which has not found physical justification. But a diagram convey an idea that a classification does not.
Why the Hubble diagram has so much success in astrophysics? There are several reasons. A diagram is very synthetic, and the Hubble classification is very simple. In addition, the refined classifications has tried to quantify it, like in the de Vaucouleurs scheme that gives a number to each morphological class, increasing along the Hubble diagram. But there are other reasons.
The global shape of a galaxy is determined by the distribution of the orbits of the stars. In an elliptical galaxy, the orbits are distributed in the three-D space, while in a disky galaxy they are distributed in a plane. It is then unsurprising that many physical properties are related to the global morphology, and that each morphology must have a specific origin. For instance, there is very little neutral hydrogen and stellar formation in elliptical galaxies. There is also a gradual evolution of the luminosity and the color along the Hubble diagram. As a result, the morphology, being correlated with several physical properties, looks like the proxy of a multivariate and objective classification.
But no physical quantitative and objective property has been found to reproduce so well the Hubble classification and the tuning-fork diagram. Isn’t it natural since it is difficult to reproduce a morphological classification with something else than … morphology? Could it be because several physical parameters are necessary to provide a global evolutionary scenario, which is less simple that the Hubble sequence?
Hence, for nearly a century, we are “stuck” to the Hubble diagram, based only on visual morphology, despite the explosion of available quantitative and objective parameters and despite the enormous progress of our physical understanding of the formation and evolution of galaxies, mainly thanks to the computer power.
So at least it would be legitimate to expect galaxy diversification to be depicted on a scatter (2D) plot using to physical quantities, somewhat like the Hertzsprung-Rüssel diagram for stars. Aren’t galaxies more complex than stars? Some attempts have been made, without much success (DUTIL 2001). Most results of computations can show an evolutionary trajectory on scatter plots, but rapidly it becomes obvious that a single scatter plot do not suffice since many parameters must be tuned in the computations and it is not easy to decide the most influent ones.
The evolution of galaxies cannot be understood if we do not have the right representation space. And if we do not have the right representation space, we cannot build a relevant classification.
The parameter space is thus crucial to build a classification that allows a clear understanding of evolution. Unfortunately, this parameter space cannot be as simple as we would like since it must reflect the complexity of the objects and their evolution. It is where statistics and graphs have proven their usefulness on other disciplines… Why not in astrophysics?
DUTIL, Y., 2001. Astrophysics and Space Science 277 (Suppl.), 165–168.
VAN DEN BERGH, S., 1998. Galaxy Morphology and Classiﬁcation. Cambridge University Press.