7.3. Do ellipticals form from mergers?
The studies of purely stellar dynamical mergers summarized above yield some interesting results and in particular offer a simple explanation of the observed radial density profiles and perhaps also of the rotational and clustering properties of elliptical galaxies. Despite these moderate successes, the merger hypothesis has been criticized on several grounds (Ostriker, 1980; Tremaine, 1981). One difficulty was discussed in some detail in Section 7.2a, namely it seems difficult to escape the conclusion that merging between two spiral discs with rotational velocities of order 300 km sec-1 will result in an elliptical-type galaxy with a velocity dispersion that is smaller than would be expected from observation. This may mean that the most massive spirals (and hence those with the largest vm) had a much higher probability for merging in the past (though this does not seem to resolve the problem for typical L* ellipticals) or that dissipation has played an important role (Negroponte and White, 1983). Perhaps ellipticals formed by the merging of predominantly gaseous systems (Silk, 1978; Silk and Norman, 1981).
The ellipticals satisfy a colour-luminosity relation (Visvanathan and Sandage, 1977) and a metallicity-luminosity relation (Faber, 1977) which, it has been argued, would be difficult to explain on the merger scheme. However, White, 1980 has shown that metallicity gradients, whilst weakened by merging, are not completely erased. Metallicity gradients have been observed in spiral galaxies (see Pagel and Edmunds, 1981) so it may be that metallicity gradients in ellipticals reflect the metallicity gradients of their progenitors. Metallicity gradients could be set up if some of the gas in their progenitors dissipates into the middle of the remnant after merging (Negroponte and White, 1983). The presence of large numbers of globular clusters around elliptical galaxies may also be difficult to explain if they formed by mergers (Harris, 1981).
It has also been argued
(Tremaine, 1981)
that dwarf ellipticals and
spiral bulges look so much like giant ellipticals (i.e. they have
similar surface density profiles) that all should have formed in the
same way. In the case of dwarf ellipticals it is hard to see how they
could have formed by mergers because their relative velocities are too
high whilst spiral bulges may have formed dissipatively during the
collapse of the disc
(Larson, 1976).
However, there is no compelling
reason for insisting on a common formation mechanism for all
spheroidal systems The fact that Hubble-like density profiles result
from the dissipative collapse models of
Larson (1975),
the dissipationless collapse models of
Aarseth and Binney (1978)
and van Albada
(1982)
as well as from merging may indicate that quite
different formation mechanisms can lead to the same end product
(though one should bear in mind a strong selection effect in the
literature which favours the publication of "correct" models). From
the point of view of forming galaxies within the framework of
hierarchical clustering theories, the main argument against the
dissipative formation of giant ellipticals would seem to be their low
observed rotational velocities. A protocloud with an initial spin
parameter 
0.07 acquired by tidal
torques would spin up
roughly as
r-1/2 if it collapsed dissipatively. It
is difficult to see how
dissipational collapse could lead to slowly rotating ellipticals.
In summary, there are several points that are not well understood and further work is necessary before we can judge just how important mergers could have been in determining the morphologies of galaxies. Obtaining more definite answers does not look to be easy. For example, a more accurate discussion of the merging of discs, including a realistic model for the interstellar medium, is beyond the capabilities of present-day computers. Also, the cosmological N-body simulations discussed in Section 7.2 provide only a rough guide to the details of the merging process. A more definitive treatment must necessarily include the dark material in a self-consistent way.