Next Contents

1. A FEW STATISTICS

When my brother and I were pursuing the theme that the occasional "tails" and similar odd appendages of galaxies represent tidal damage of a violent and fairly recent sort, we were struck (Toomre and Toomre 1972, p. 662) by photographs often showing the main hulks of the presumed partners to have remained so close together that they can now be distinguished only with difficulty, if indeed at all. Such behavior was certainly not exhibited by our disks of test particles or their respective central masses, though together they had seemed adequate to mimic the tails themselves. Dilemmas like that always tend to broaden one's outlook - and indeed it was not too long before we had plucked up enough courage to guess that those actual galaxies must have suffered a surprisingly fierce orbital decay or braking in addition to the indignity of having their outer disks ripped almost literally into shreds.

The surprise here was not the idea that even galaxies composed entirely of stars might behave in a somewhat inelastic manner during close encounters. (1) What was startling was only (a) the severity of these apparent decelerations, and (b) the number of old merged relics - i.e., piles of stars without any identifying transient tails - that extrapolations into the past from a dozen or so recent NGC examples seemed to imply. In fact, as Section II will stress anew, even item (a) proved far from a total surprise: The impulsive-tide estimates of Alladin (1965), though aimed at faster passages, had already hinted much more explicitly than Chandrasekhar's formulas that "the increase in internal [and the loss of orbital] energy of a galaxy due to an encounter with a second galaxy can easily be of the same order as its initial internal energy", when the speed of approach is roughly parabolic and the impact takes place nearly head-on. But item (b) on the past statistics may still seem unfamiliar. Thus perhaps we should now review at once, starting with Table 1, just why those extrapolated numbers demand some respect.

TABLE 1: Eleven NGC Prospects for Ongoing Mergers

Name Croos-ID RA (1950) DEC V (km/s)

NGC 4038 / 9 VV 245 = Arp 244 11 59.3 -18 36 1,650
NGC 4676 VV 224 = Arp 242 12 43.7 +31 00 6,600
NGC 7592 23 15.8 -04 41 7,300
NGC 7764A 23 50.8 -41 05
NGC 6621 / 2 VV 247 = Arp 081 18 13.2 +68 21 6,250
NGC 3509 VV 075 = Arp 335 11 01.8 +05 06 7,650
NGC 0520 VV 231 = Arp 157 01 22.0 +03 32 2,150
NGC 2623 VV 079 = Arp 243 08 35.4 +25 56 5,450
NGC 3256 VV 065 10 25.7 -43 39 2,900
NGC 3921 VV 031 = Arp 224 11 48.5 +55 21 6,000
NGC 7252 Arp 226 22 18.0 -24 56 4,750

Exclusively from among the 4000-odd NGC galaxies or close pairs, Table 1 identifies the foremost 11 examples known to me to exhibit major tails plus main bodies that appear to be nearly in contact or perhaps not even separable. They are listed here in rough order of completeness of the imagined mergers. As noted, most of these peculiar galaxies were known already to Vorontsov-Velyaminov (1959) if not to Zwicky. Also for most, Arp's (1966) Atlas still provides the best published photographs. Nine of these examples were cited in the present context (2) in our 1972 paper (= TT), where we even concocted some frictionless (but e appeq 0.5 and therefore much too short-period!) models for the first two. NGC3256 and NGC7764A were kindly pointed out to me by de Vaucouleurs in 1973; a recent photograph by Schweizer using the CTIO 4-m telescope affirms that the faint tails of NGC3256 resemble those of NGC2623.

Since appearances are often deceptive, beware that only for three of the above cases has the tidal idea yet been checked spectroscopically:

(1) For the Antennae, NGC4038 / 9, Huchtmeier and Bohnenstengel (1975; also van der Hulst 1976) detected 21-cm neutral hydrogen with a plausibly narrow spread of velocities, especially in the southern tail - though its sense of recession proved opposite to that guessed by TT. Moreover, for an H II region near that southern tip, Schweizer (1977, private communication) has recently managed to determine a similar optical redshift after much effort.

(2) For the Mice, NGC4676, Stockton (1974) reobserved the relative optical velocities of the two main bodies, and also of the bright northern tail. He judged those hulks to be moving in the sense demanded by the tidal hypothesis (contrary to some decade-old measurements), but his tail velocities still seem worrisomely high (if less so than first reported by Theys et al 1972).

(3) For NGC0520, Stockton (1975, private communication) measured the senses of rotation of its two relatively edge-on hulks. Indeed both northwest ends seem to be approaching, consistent with the simple view that the main dustlane there belongs to the northwestward tail.

That seems to have been about the extent of the checking, apart from the display by Schweizer (1976) that the tails of NGC4038 / 9 thicken markedly (as they ought to, assuming larger random velocities for old disk stars) in very deep photographs, and also apart from the demonstration by Larson and Tinsley (1977) that the unusual UBV colors of NGC2623 and NGC7252 are consistent with a major burst of star formation roughly 2 x 108 years ago.

Limited though it is, this evidence at least does not upset the idea that Table 1 refers mostly to galaxy pairs in fairly advanced throes of merger. Moreover, it seems unlikely that the apparent proximity of those main bodies stems from lucky projections: I remain aware of only one well-separated NGC binary, NGC2992 / 3 = Arp 245, that boasts of a pair of major tails. Finally, speaking of double tails, notice such duplicity in all but two examples chosen for Table 1. This itself warns that our little list is apt to have underestimated the number of serious recent interactors from the ranks of the NGC. Remember that double tails seem theoretically feasible only when the original disks are at least vaguely comparable in mass, size, and angular orientation. And besides, much deeper and more splattering interpenetrations should yield no decent tails at all.

Ten examples or so from among 4000+ may not seem like much - until we notice that the median age of this existing lot of transient shapes can hardly be much greater than 5 x 108 years. Admittedly, in the absence of specific data, this estimate merely supposes "normal" rotational velocities of order 200 km/s, but at least for NGC4676 at an assumed 100 Mpc and for NGC4038 / 9 at a probably excessive 20 Mpc (vs. 9 ± 3 Mpc recommended by Rubin et al 1970), the elapsed times since closest approach seem to be about 2 and 7 x 108 years, respectively. Simply on that basis, we should expect to find roughly 250 old relics of mergers among the NGC systems alone, provided the present rate of those intense encounters is at all typical of the 10-15 billion years that galaxies have existed.

In fact, as TT remarked and as Figure 1 again tries to underscore, such a rate now is probably only a fraction of what it used to be. Briefly put, the difficulties of meeting, distorting, and stopping all argue against hyperbolic fly-bys. The only alternative still seems to be that we are here dealing with loosely bound pairs of galaxies that had separated to great distances in the general cosmic expansion, and that have only lately fallen (almost) together again in comet-like plunging ellipses.

Figure 1

Figure 1. Estimates of numbers of past remnants.

Such "loitering" in long, skinny orbits with periods of the order of 1010 years does not, of course, sound statistically probable - just as we would not normally expect any terrestrial rocket, after vertical launch with nearly escape speed, to wait till next week to fall back down. But that unlikelihood remains exactly my point; If we observe a certain number of such returns now, there must have been considerably more of them during similar 5 x 108 year intervals in the past.

To estimate that excess number, I still suspect that even the rate of returns when the Universe was one-half its present age was less than the mean rate reckoned over the entire 10-15 billion years. That midway rate should have been roughly three times larger than now - as would indeed be true if the binding energies of distinct galaxy pairs had a flat distribution consistent with the t-5/3 time-behavior drawn in Figure 1. The resulting naive total then comes to no fewer than 750 probably spheroidal heaps of stars! It is to be compared with the fact that the Second Reference Catalogue (= 2RC) of de Vaucouleurs, de Vaucouleurs, and Corwin (1976) classifies only about 11 per cent of its (approximately 2900) NGC entries as cE, E types -6, -5, or -4, and only another 10 per cent as SO- type -3.

This near-coincidence has, I admit, been something of a hobbyhorse of mine ever since Juri and I first noticed it. As usual, it is difficult to be sufficiently self-critical, but I really do believe its main weakness to be just one of observational selection favoring the unusual. This is not to say that the late 19th-century NGC, heterogeneous though it was in its choices, erred consciously in favor of faint features like tails - the earliest of which, in NGC4038 / 9, was not discovered until 1917 by Lampland. Yet facts like a close binary appearance, or simply a doubling (or more) of the total light, must have favorably influenced the selectors. Such doubling would of course increase 2sqrt2-fold the physical volume out to some magnitude cutoff. And any fourfold rise in the total luminosity, such as Tinsley (1977, private communication) reckons plausible for NGC2623 and NGC7252 from recent star formation induced presumably by the mechanical violence, would make the entries in Table 1 unfairly numerous by a whole factor of eight.

My own guess is that this selection bias actually exaggerates the number of expected old relics by about the same factor of three as was contributed, more legitimately, by the time-evolution. That impression rests mainly on the fact that the median value of the ten (heliocentric) redshifts copied into Table 1 from the 2RC is 5700 km s-1, whereas the median for all measured NGC galaxies barely exceeds 3000, judging from every 10th entry in the 2RC. The latter estimate refers only to about 35 per cent of all NGC galaxies, however; it seems very likely to rise past 4000 km s-1 as the redshift incompleteness for fainter objects is gradually remedied. Clearly (5.7/4.0)3 appeq 3, and this lowers our crucial total back down to 250 or thereabouts. Remember again, though, that Table 1 includes only the more clearcut examples. And also note, for what it may be worth, that the median of the known redshifts for the NGC ellipticals listed in the 2RC is not 3000 km s-1 but already 5000.

All in all, if several hundred merged remnants of former disks do not now constitute many of the ellipticals, where else have they possibly gone?


1 At least in the context of creating loose pairs from stray field or cluster galaxies, such awareness goes well back to Holmberg (1941), Zwicky (1959), and Rood (1965). And already Lindblad (1926) wrote that "sharp encounters between nebulae ... must probably be considered as highly `unelastic' and must tend to convert translational into rotational kinetic energy. An encounter of this kind may even lead to a fusion of the respective bodies." However, it seems he was thinking mostly of gaseous impacts. Back.

2 TT also cited NGC0455, NGC7727, and "perhaps even" NGC7603. On further thought, these may simply be instances of recent major infall of extragalactic gas, roughly as envisaged by Oort, Shklovsky, and Larson among others. Back.

Next Contents