There has been a veritable feeding frenzy going on with the first JWST data. This is to be expected. Also to be expected is that some of these early results will ultimately prove to have been premature. So – caveat emptor! That said, I want to highlight one important aspect of these early results, there being too many to do all them all justice.

The basic theme is that people are finding very faint yet surprisingly bright galaxies that are consistent with being at redshift 9 and above. The universe has expanded by a factor of ten since then, when it was barely half a billion years old. That’s a long time to you and me, and even to a geologist, but it is a relatively short time for a universe that is now over 13 billion years old, and it isn’t a lot of time for objects as large as galaxies to form.

In the standard LCDM cosmogony, we expect large galaxies to build up from the merger of many smaller galaxies. These smaller galaxies form first, and many of the stars that end up in big galaxies may have formed in these smaller galaxies prior to merging. So when we look to high redshift, we expect to catch this formation-by-merging process in action. We should see lots of small, actively star forming protogalactic fragments (Searle-Zinn fragments in Old School speak) before they’ve had time to assemble into the large galaxies we see relatively nearby to us at low redshift.

So what are we seeing? Here is one example from Labbe et al.:

JWST images of a candidate galaxy at z~10 in different filters, ordered by increasing wavelength from optical light (left) to the mid-infrared (right). Image credit: Labbe et al.

Not much to look at, is it? But really it is pretty awesome for light that has been traveling 13 billion years to get to us and had its wavelength stretched by a factor of ten. Measuring the brightness in these various passbands enables us to estimate both its redshift and stellar mass:

The JWST data plotted as a spectrum (points) with template stellar population models (lines) that indicate a mass of nearly 85 billion suns at z=9.92. Image credit: Labbe et al.

Eighty five billion solar masses is a lot of stars. It’s a bit bigger than the Milky Way, which has had the full 13+ billion years to make its complement of roughly 60 billion solar masses of stars. Object 19424 is a big galaxy, and it grew up fast.

In LCDM, it is not particularly hard to build a model that forms a lot of stars early on. What is challenging is assembling this many into a single object. We should see lots of much smaller fragments (and may yet still) but we shouldn’t see many really big objects like this already in place. How many there are is a critical question.

Labbe et al. make an estimate of the stellar mass density in massive high redshift galaxies, and find it to be rather a lot. This is a fraught exercise in the best of circumstances when one has excellent data for thousands of galaxies. Here we have only a handful. We must also assume that the small region surveyed is typical, which it may not be. Moreover, the photometric redshift method illustrated above is fraught. It looks convincing. It is convincing. It also gives me the heebie-jeebies. Many times I have seen photometric redshifts turn out to be wrong when good spectroscopic data are obtained. But usually the method works, and it’s what we got so far, so let’s see where this ride takes us.

A short paper that nicely illustrates the prime issue is provided by Prof. Boylan-Kolchin. His key figure:

The integrated mass density of stars as a function of the stellar mass of individual galaxies, or equivalently, the baryons available to form stars in their dark matter halos. The data of Labbe et al. reside in the forbidden region (shaded) where there are more stars than there is normal matter from which to make them. Image credit: Boylan-Kolchin.

The basic issue is that there are too many stars in these big galaxies. There are many astrophysical uncertainties about how stars form: how fast, how efficiently, with what mass distribution, etc., etc. – much of the literature is obsessed with these issues. In contrast, once the parameters of cosmology are known, as we think them to be, it is relatively straightforward to calculate the number density of dark matter halos as a function of mass at a given redshift. This is the dark skeleton on which large scale structure depends; getting this right is absolutely fundamental to the cold dark matter picture.

Every dark matter halo should host a universal fraction of normal matter. The baryon fraction (fb) is known to be very close to 16% in LCDM. Prof. Boylan-Kolchin points out that this sets an important upper limit on how many stars could possibly form. The shaded region in the figure above is excluded: there simply isn’t enough normal matter to make that many stars. The data of Labbe et al. fall in this region, which should be impossible.

The data only fall a little way into the excluded region, so maybe it doesn’t look that bad, but the real situation is more dire. Star formation is very inefficient, but the shaded region assumes that all the available material has been converted into stars. A more realistic expectation is closer to the gray line (ε = 0.1), not the hard limit where all the available material has been magically turned into stars with a cosmic snap of the fingers.

Indeed, I would argue that the real efficiency ε is likely lower than 0.1 as it is locally. This runs into problems with precursors of the JWST result, so we’ve already been under pressure to tweak this free parameter upwards. Turning it up to eleven is just the inevitable consequence of needing to get more stars to form in the first big halos to appear sooner than the theory naturally predicts.

So, does this spell doom for LCDM? I doubt it. There are too many uncertainties at present. It is an intriguing result, but it will take a lot of follow-up work to sort out. I expect some of these candidate high redshift galaxies will fall by the wayside, and turn out to be objects at lower redshift. How many, and how that impacts the basic result, remains to be determined.

After years of testing LCDM, it would be ironic if it could be falsified by this one simple (expensive, technologically amazing) observation. Still, it is something important to watch, as it is at least conceivable that we could measure a stellar mass density that is impossibly high. Wither then?

These are early days.

13 thoughts on “An early result from JWST

  1. Are there any current data that disfavor a Mond hypothesis that could see a revision in the JWST results?


  2. Some notable new paper it would be nice to hear about:

    E. Asencio, I. Banik, S. Mieske, A. Venhola, P. Kroupa, H. Zhao, “The distribution and morphologies of Fornax Cluster dwarf galaxies suggest they lack dark matter” arXiv:2208.02265 (August 3, 2022) (accepted for publication in MNRAS);

    Mariia Khelashvili, Anton Rudakovskyi, Sabine Hossenfelder, “Dark matter profiles of SPARC galaxies: a challenge to fuzzy dark matter” arXiv:2207.14165 (July 28, 2022).

    Also FWIW here are lots of impossible early galaxies papers from JWST: arXiv:2207.11558 arXiv:2207.12338 arXiv:2207.12356 and

    This JWST is at lower z but also makes some pretty interesting observations about earlyish galaxy types:


    1. Regarding our work on the Fornax Cluster dwarfs that you mentioned, you can find a press release of it here:

      For a longer explanation, check out this video by the lead author:

      We also explained it in this blog post, which shows some of the main figures:

      Thanks for your interest in our work!

      Fuzzy dark matter is not really going to work when observations rule out substantial amounts of dark matter on galaxy scales. So this at least is not very surprising to me. The evidence is reviewed in more detail here:

      As for the JWST observations, it was predicted a long time ago that Milgromian galaxies should form much faster than in LCDM. So this at least did not come as a surprise to some of us.


  3. Data from the JWST are excitingly informative. However, the evaluation of these within the frame of standard big bang cosmology is paradoxical irrespective of any data: In the mentioned example, we see light from a galaxy at redshift 9. When this light was emitted, the expanding universe was barely half a billion years old and so less than one billion lightyears in diameter. However, the light was on its way for over 12.5 billion years and so must have come from a distance of over 12.5 billion lightyears. This is far outside a prototypical big bang universe.

    Cosmologists simply switch to an expanding view model for calculating line-of sight distances, while still using an expanding universe model for other aspects. This results in a paradox that becomes most drastic when the CMB is considered.

    In Liddle (2015) Introduction to Modern Cosmology, 3rd ed., p. 82, one can read the following: “Since decoupling happened when the Universe was only about one thousandth of its present size, and the photons have been travelling uninterrupted since then, they come from a considerable distance away. Indeed, a distance close to the size of the observable Universe.”

    The first part of this quotation, “Since decoupling happened when the Universe was only about one thousandth of its present size” presupposes an expanding universe, while the remainder “and the photons … come from … a distance close to the size of the observable Universe” presupposes a transcendentally expanding view.

    By placing an expanding view model over an expanding universe model, it is, in fact, taught that the universe was at least as large as it is now, or even infinite, when it was much younger and smaller than now, or even arose out of a point-like singularity.


    1. There is nothing in cosmology that says that the universe started at a point. Indeed, LCDM is compatible with an infinitely large, near-infinitely dense universe at the origin. Moreover, the observable universe is much, much smaller than the actual universe — observations show the universe is at least 125,000 times as big as the part we can observe.


  4. Scott P., your statements neither involve a paradox (as long as one does not consider the universe that is revealed by the observations to be the “actual” one) nor do they resolve any. In the paradoxes I mentioned, a specific galaxy or a certain surface of last scattering is (in different contexts) considered to be at different distances.


Comments are closed.