This past July astronomers working with the James Webb Space Telescope (JWST) released the deepest astronomical image ever obtained, leaving the world in awe. Against the background of a galaxy cluster named SMACS 0723, seen as it appeared 4.6 billion years ago, myriad galaxies of different shapes and sizes appear like bright gems in the darkness of the cosmos. Some of these lighthouses were already shining when the universe was just a few hundred million years old. To understand how we reached this remarkable achievement—how astronomers have sailed to galactic islands so remote from us in space and time, collecting photons whose journey started breathtakingly close to the big bang—it helps to know how deep-field observations came to be.

The origin of Webb’s first deep field is best traced to the early 1990s, with the launch of JWST’s predecessor, the Hubble Space Telescope. The concept of deep-field observations was still in its infancy back then. Hubble was primarily designed for targeted observations. Astronomers would point the telescope to a source at a specific spot in the sky and expose (or “integrate”) as needed, depending on the source’s brightness. But Hubble could also be used for deep-field imaging, which is the opposite: astronomers would point the telescope to a sky region devoid of any visible source and use a very long exposure time to observe as many faint sources of light as possible, thereby reaching “deep” into the cosmos. From its perch in low-Earth orbit, above our planet’s starlight-scattering atmosphere, Hubble was the best platform for deep-field imaging astronomers had ever known.

Not everyone thought the approach would prove revolutionary. In a famous article published in Science in 1990, John Bahcall of the Institute for Advanced Study in Princeton, N.J., and his colleagues argued that a deep-field image from Hubble would not reveal significantly more galaxies than ground telescopes. Bahcall, a giant in astrophysics, was widely known for his work on the problem of solar neutrinos and his calculations of the distribution of stars around a massive black hole. He contributed fundamentally to the development of the Hubble Space Telescope from its original concept in the 1970s to its launch. Bahcall thought Hubble deep-field images could be used to study the sizes and shapes of faint galaxies and to take a census of quasars (a rather old-fashioned word for accreting supermassive black holes), but he didn’t believe they would reveal new populations of galaxies. Such tepid expectations tamped down any urgency to try deep-field imaging with Hubble.

The first attempt occurred around the winter holidays of 1995, after a much needed optics repair. The telescope spent 10 days of exposure time pointed at the Ursa Major constellation, staring at a tiny patch of the sky just one-thirteenth the moon’s angular diameter. Weeks later, when astronomers saw the resulting image—known as the Deep Field North—they immediately realized it was a Christmas gift for the ages. Because the Milky Way’s stars are sparse in the target region, Hubble was able to probe the cosmic abyss as if through a peephole. The telescope saw almost 3,000 faint galaxies of different shapes and sizes—many more than expected, some of them as far as 12 billion light-years away. Hubble was not only exploring space. It was also probing time, gathering starlight that had been emitted eons ago, during earlier epochs of the universe. The image quickly became iconic.

A crucial question arose: Was the galaxy-rich region revealed by the Deep Field North the norm throughout the universe, or did astronomers just happen to point the telescope toward a Pantagruelian crowding of galaxies? In 1998 Hubble obtained the Deep Field South. The exposure was similar, but the telescope pointed toward the southern celestial hemisphere, as far as possible from the first spot. This new image confirmed that the universe contained many more galaxies than previously thought, especially at vast distances. In addition to their scientific and inspirational value, these and other Hubble deep-field surveys were a technical triumph, capturing more than 10,000 galaxies in one of astronomy’s first “big data” challenges.

Deep-field imaging is not restricted to the visible realm of the spectrum. At the turn of the millennium, the Chandra X-ray observatory, a revolutionary NASA mission launched in July 1999 and still active today, captured the first high-energy deep field. The Chandra Deep Field South was obtained by integrating for about one million seconds over a piece of the sky that was devoid of hydrogen clouds and dust from the Milky Way. The Chandra Deep Field South uncovered the extreme universe, revealing hundreds of black holes, some very remote. The image wasn’t as visually spectacular as the Hubble photographs, but it was dense with science. Chandra later imaged the same field for a total exposure of about seven million seconds, capturing one of the deepest fields ever obtained in x-ray. In 2003 a new image called Chandra Deep Field North was released, containing data from more than 500 x-ray sources.

In 2006 scientists released the Hubble Ultra Deep Field, which was taken using an instrument called the Advanced Camera for Surveys that was added to the telescope during a servicing mission in 2002. This historic shot contained thousands of galaxies, some that we now know were shining when the universe was less than one billion years old. The Ultra Deep Field showed the history of galaxy formation in unprecedented detail. Distant galaxies conclusively appeared to be smaller and more irregular in shape than closer ones, providing substantial evidence to support galaxy evolution theories.

The technology used for Ultra Deep Field provides essentially the deepest image that can be obtained in optical wavelengths. If a galaxy is too far away, its optical light is shifted outside the visible range and into the infrared regime; this is a consequence of the cosmological redshift, in which the expansion of the universe stretches out the wavelengths of light traveling through enormous expanses of intergalactic space. It would take an infrared camera to look farther in space and time. With the addition of a new near-infrared camera to Hubble, an infrared Ultra Deep Field was obtained in 2009, revealing galaxies shining only 600 million years after the big bang. A decade later, in 2019, a deep field produced with NASA’s Spitzer infrared space telescope was released. Both these images are rich with galaxies at the cosmic dawn.

Hubble’s Frontier Fields campaign, completed in 2017, was the real prologue to JWST’s first deep image. During this observational campaign, Hubble was pointed toward six large concentrations of galaxies. According to Einstein’s theory of general relativity, a substantial density of mass along the line of sight can bend and thus amplify the light incoming from a background source, an effect called gravitational lensing. The Frontier Fields campaign used these galaxy clusters as a magnifying glass to see even farther away. Besides being filled with swarming galaxies, the Frontier Fields images are adorned with strange arcs of light, representing the stretched and amplified images of background galaxies much more distant than the cluster and possibly too faint to be directly observed with Hubble. These shots revealed some of the most distant galaxies and the first gravitationally lensed supernova.

It’s been almost 200 years since the advent of photography, when humanity first managed to directly capture and record photons to make images. Today highly complex cameras onboard a space telescope one million miles away are shaking our knowledge of the universe, opening new windows onto space and time. A relatively short time separates these two events, but they are linked by the same goal: achieving a deeper understanding of nature by looking at what our eyes cannot see.