My research centers on the search for the first galaxies ever to form in the universe. Over the past decade, we have entered the era of precision cosmology, where my cosmologist colleagues can tell me the age and composition of the universe to decimal-level precision. However, despite these amazing advances, we still do not know when the first galaxies formed. To directly observe these galaxies, it would take an enormous space telescope, several times larger than the soon-to-be-launched James Webb Space Telescope (at least, we think so; right now, all we have to guide us are theoretical model-based simulations).

However, these galaxies had an effect on the universe around them that we can see. The energetic radiation emitted from these early galaxies heated the gas in the intergalactic medium—that is, gas between galaxies—in a process known as reionization (severing the electromagnetic link between a hydrogen atom’s proton and its electron). We know that reionization happened, as we can observe that this intergalactic gas is ionized today, and was ionized as far back as a time only one billion years after the Big Bang. If we can study the process of reionization, and constrain when it first started, that will inform us about when the first galaxies appeared in the universe, all without directly observing them!

While the common thinking among astrophysicists is that star formation inside galaxies provided all of the needed photons for reionization, this requires a number of crucial, and previously untested, assumptions. Should these assumptions be incorrect, it would require some other, more exotic source of photons, such as early accreting supermassive black holes, or perhaps even self-annihilating dark matter. To see whether or not enough galaxies were in place to complete reionization seems relatively straightforward—just count up the galaxies!

When we do this, we find that galaxies would appear to fall short of the needed energetics to accomplish reionization. However, we need to take into account our observational limitations—even the deepest Hubble Space Telescope survey can only see galaxies about 1/20th of the brightness of the Milky Way (albeit at a time 13 billion years into the past). Do galaxies exist fainter than the limits of Hubble? Logic says yes, as it would be an enormous conspiracy for galaxies to stop existing just at the limits of what Hubble can see, and simulations of the early universe back this up. Fainter galaxies should be there.

Interestingly, when we look at the galaxies we can see, we find that as we have pushed deeper and deeper into the universe we have found that early on a larger fraction of the total light from galaxies was coming from the faintest sources we can see. If the abundance of faint galaxies keeps climbing even beyond Hubble’s limits, the missing photons for reionization would be accounted for! Studies of reionization have found that for galaxies to provide all of the needed photons, galaxies would need to be ever increasingly abundant to a luminosity 100 times fainter than Hubble can presently see in this epoch. As Hubble lacks the sensitivity to directly observe these galaxies, the assumption that these very faint galaxies exist in large numbers is one that is commonly made, but has never been tested. Until now.

When limited by the size of your telescope, one can always take advantage of nature’s magnifying glasses: gravitational lensing, the phenomenon, first predicted by Einstein, and first observed nearly 40 years ago, where large concentrations of mass bend space-time and magnify more distant galaxies. While the universe is mostly empty space, we know of such large concentrations of mass, giant clusters of galaxies, with total masses of close to one quadrillion times that of our Sun (10 to the 15th power). If a distant galaxy happened to lie behind one of these monsters, its light could be magnified by a factor of anywhere from a few to a few hundred, allowing us to see galaxies much fainter than previously possible.

This phenomenon, and the uncertainty about reionization, led the Hubble Space Telescope to embark on one of its largest programs ever, known as the Hubble Frontier Fields. This program stared at six galaxy clusters around the sky for several hundred hours each, pushing Hubble to its limits. The goal is to use gravitational lensing to make up for the missing factor of 100 in luminosity between what we had previously observed and what is thought to be needed for reionization. This program completed late last year, and has been a resounding success, with dozens of papers published using its data already.

Relevant to the matter at hand, in a paper published last month led by astronomer Dr. Rachael Livermore of the University of Texas at Austin and myself, in collaboration with Dr. Jennifer Lotz of the Space Telescope Science Institute, it was shown that the faintest galaxy observed in the Frontier Fields is on order 100 times fainter than that seen in previous, un-lensed, observations (and an astounding 1500x fainter than the Milky Way!). Importantly, the abundance of these very faint galaxies was calculated to be precisely what one would predict if they extrapolated to these extremely faint luminosities from previous Hubble observations. It would therefore appear that the case is closed!

However, as is often the case with scientific inquiry, these observations have opened yet more questions, as many models predict that at these extreme faint luminosities, the abundances of galaxies should start to fall, yet this has not been seen. Additionally, the number of these extremely faint galaxies we have seen is small, so the result is somewhat uncertain. While we may have pushed Hubble to its limits even with the aid of lensing, the James Webb Space Telescope (launching in October 2018) will have a light-gathering power a factor of seven times that of Hubble. Combined with nature’s telescopes, this exciting facility will have to power to peer yet deeper into the distant universe, illuminating whether more mysteries await us at the end of the dark ages.