To give a little background, one game that astronomers like to play is “Find the oldest galaxy around at some time”. There are a number of reasons this game has so many eager practitioners. First, the oldest galaxies at any time are probably the locations of the some of the very first collapsed structures in the universe, which make them cool. Second, the most massive galaxies we see today seem to be those that that host the oldest stellar populations — turning this around, if you hunt for the oldest things at a given time, you’re hopefully finding the most massive things at that time. Finally, it’s very tough to figure out how the population of galaxies you see at one time is related to the one you see at the present day (i.e., just because a galaxy is blue and distorted 10 Gigayears (Gyr) ago, doesn’t mean it’s not red and symmetric today). However, if you stick to studying massive galaxies that already looked old 10 Gyr ago, it’s probably a reasonably good guess that they’re comparable to the predecessors of the most massive old galaxies today, giving you one of the few cases where you can match up the two populations with reasonable confidence.
Which brings us to the latest paper by Pieter van Dokkum and his collaborators, on the spectroscopic and HST/NICMOS follow-up of 9 massive red galaxies from the MUSYC NIR imaging survey. These galaxies are at redshifts of 2.5-ish, corresponding to about 2-3 Gyr after the Big Bang for WMAP cosmology. They are very red, and have spectra consistent with a dormant stellar population (i.e. one that stopped forming stars roughly a Gyr previously). An earlier paper by Kriek et al estimated masses for these galaxies (assuming reasonable models of the kinds of stars and dust that are in the galaxies), and found that not only are the galaxies dormant, they’re also extremely massive.
So, now you’ve got yourself a bunch of massive galaxies, that in spite of being no more than 2-3 Gyr old, already finished all their star formation a gigayear previously. Pretty nifty, and what you’d hope to find as the precursors of the old massive galaxies we see today. That’s all well and good, but the problem is that the galaxies are just too dang small. They’re tiny. The most massive galaxies today are typically much larger than the Milky Way, but the galaxies van Dokkum is reporting on are much much smaller. (On the plot at right, the solid scale bar shows a size of 10 kpc. On the plot below, the big solid dots are van Dokkum’s sample, and the little dots are the galaxies seen today in the Sloan Digital Sky Survey.)
Ok fine, you might say. A lot can happen in the intervening 10 Gyr. However, not that much can happen, since these galaxies are already pushing the mass limit of the most massive galaxies we see today. You therefore can’t grow the galaxies by padding them out with new stars from accreted satellite galaxies, without taking them over the limit (i.e. looking at the plot on the upper left, accreting new galaxies would take the galaxies up, but would also move them to the right). Even if you could pull this off, the centers would still be too dense. You also can’t magically puff the galaxies up over time. You could imagine perhaps driving a bunch of mass out of the center of the galaxy, maybe through stellar winds, but it would take an absurd amount of mass loss, even if you believed you could get all that mass out of such a dense honking galaxy. You could also imagine trying to disrupt the galaxies somehow by puffing them up kinematically, maybe through violent mergers. However, the problem here is that the galaxies are waaaaaay too dense. If the galaxies really are as dense as the plot above suggests, then they’re going to be almost impossible to disrupt. It would be like trying to disrupt an iron ball by dropping it into a bowl of pudding.
My take on this is that some of the assumptions that went into making the plot above are wrong, because it’s very hard to imagine hiding the descendants of incredibly massive, incredibly dense galaxies somewhere in the local universe. The authors understand this, and argue that it might be a combination of issues related to measuring the radius (loss of diffuse light at large radii, radial gradients in the conversion of light to stellar mass), and possibly the initial mass function (the great “here be dragons” of all extragalactic astronomy).
I actually think the errant step might wind up being the assumed conversion of light to mass. The standard lore is that when you observe galaxies in the NIR, the light is dominated by old red giant branch stars, giving you a reasonably robust conversion from light into stellar mass. However, at a redshift of 2.5, there’s no way that any star in the galaxy is older than 3 Gyr. In this case, most of the red light from the galaxy will be coming from asymptotic giant branch stars, which are notoriously difficult to model. Thus, the calibration of light to mass can easily be off. AGB stars can also give you a nice red spectrum like those observed, and can be potentially centrally concentrated in response to a central burst of star formation. I think the authors have done the best that they could with their analysis, but I suspect we may have run into some of the real limits of our current modeling.