8.1 Binary Galaxies

Bound pairs of galaxies offer the attractive possibility of obtaining estimates of halo masses at greater distances than are usually accessible to HI rotation curve measurements. In practice, however, such observations have proved rather difficult (Page 1962; Trimble 1987 and references therein). Since the geometry of the orbit of a given binary pair is unknown, it is impossible to translate the projected velocity difference and separation into orbital velocity and physical separation. Statistical methods have therefore usually been applied to a large sample of binary galaxies. Eliminating chance superpositions and brief encounters from the sample of genuine bound pairs is a major difficulty. Despite these problems, some interesting results have been obtained.

White (1981) described the minimum requirements for deriving an indicative mass from a sample of binary galaxies. These criteria were applied by White et al. (1983) who found dark-to-luminous mass ratios higher than those usually attributed to individual galaxies. (This, of course, was not surprising, since the study probed dark halos to greater radii than investigations of individual galaxies. In fact, these authors noted that their results were consistent with extrapolations of observed rotation curves.) Interestingly, White et al. (1983) found that the mass-to-light ratio of their galaxies increased with decreasing luminosity L:

(8.1)

assuming circular orbits and h = 0.5, where r is pair separation and L_* is the knee of the galaxy luminosity function. This is in agreement with the trend for spirals discussed in Section 6 above.

Lake and Schommer (1984) used a similar method to study 9 pairs of dwarf irregular galaxies. They found extremely high M/L values, ranging from 20 to 5000 for separations between 40 and 850 kpc, which fit in fairly well with the relationship given in equation (8.1). However, as Lake and Schommer (1984) acknowledged, applying statistical arguments to such a small sample is hazardous.

Karachentsev (1983, 1985) studied large samples of binary galaxies to derive masses for galaxy pairs. He found little evidence for substantial amounts of DM around these systems. Indeed, he argued that the amount of DM in the halos around these objects was no more than a factor of 2 greater than the dark mass within R₂₅. Karachentsev (1985) noted that this was a rather suprising result, particularly in view of the fact that some of the galaxies in his sample were known to have flat rotation curves.

A different approach was adopted by van Moorsel (1987). He studied 16 pairs of galaxies, using HI measurements to obtain rotation curves for individual galaxies as well as the projected velocity separation of each pair. He concluded that the dark halos of the galaxies in his sample extended well beyond the HI rotation curves, and that there was typically 3 times as much DM present in these systems as could be accounted for by the rotation curve measurements. He also suggested that the data were consistent with a common DM envelope surrounding galaxy pairs.

More recent studies have highlighted the difficulties involved in using binary galaxies for mass determinations. Sharp (1990) concluded that statistical approaches to derive masses from binary pairs were unable to provide reliable results. He illustrated this by comparing previous studies by different workers who obtained dramatically different mass estimates. Sharp (1990) found that the causes of these discrepancies included different methods of analysis, none of which could be shown to be objectively superior. Confusion was also introduced by the fact that many galaxies in pairs are interacting. These gloomy findings were somewhat tempered by Sharp's (1990) suggestion that detailed analysis and modelling of individual pairs could yield useful dynamical masses.

Further concerns were expressed by Schneider and Salpeter (1992). They showed that if the projected separation for including galaxies in a binary sample was fixed at too low a value, then artifacts in the velocity separation distribution could be introduced. The troublesome ``peak'' at 72 km^-1 in many such samples is probably the result of such a selection bias, unless one prefers Tifft's (1976) suggestion that redshifts are quantized. However, in order for such a peak to be produced by selection procedures, the degree of incompleteness in extant binary samples must be high (Schneider and Salpeter 1992). This result casts serious doubts on the validity of mass determinations based on such samples.

Instead of directly estimating masses of binary galaxies, Charlton and Salpeter (1991) used the distribution of projected separations to study the extent of dark halos. In low-density environments, they found an excess of pairs with low velocity separations out to projected spatial separations as great as 1 Mpc. Moreover, the distribution of projected separations was featureless from small distances out to 2 Mpc. This is somewhat surprising, since galaxy mergers are expected to deplete the number of pairs at separations less than the characteristic halo radius. Charlton and Salpeter (1991) suggested that extremely extended halos, with radii around 1 Mpc, were consistent with these observations. Alternatively, they argued that galaxies in low-density environments may be surrounded by a common DM envelope.

One problem with such large halo radii is that, unless the halo density profile does something pathological, the mean density is extremely low. Thus halos with radii around 1 Mpc would not have virialized by the present epoch (Ashman 1991). The common envelope idea may also have problems, since a smooth DM ``superhalo'' would simply raise the background density. In this case, halos associated with individual galaxies would be of a more conventional size, thereby reintroducing the depletion-by-merger problem that Charlton and Salpeter (1991) were attempting to overcome. Numerical simulations may clarify this issue.