The Spy Agency Had Two Spare Hubbles in a Warehouse

The mirror inside NASA’s newest telescope was not made to look at the sky.

That is the part of the Roman Space Telescope I keep turning over. Every flagship observatory has an origin story about a decade of committees and a decade of grinding glass. Roman has one of those too. It also has a stranger one, which is that a U.S. intelligence agency called NASA in 2011 to mention it had a couple of spare Hubbles lying around, and did anybody want them.

A row of Hubbles under dust covers

It surfaced the way most things do — sideways, at a conference. In January 2011, at the American Astronomical Society meeting, the Princeton theorist David Spergel quietly told the astronomer Alan Dressler that the National Reconnaissance Office had telescopes it no longer needed. NASA’s Michael Moore had heard rumors along the same lines through Air Force contacts back in the 1990s, asked once, and been told no. By 2011 the answer had changed.

An NRO representative made it official in a secure room in Washington, to NASA’s John Grunsfeld: two partially disassembled telescopes with 2.4-meter primary mirrors, plus a third primary and a pile of spare components. Grunsfeld is the astronaut who flew three Hubble servicing missions. He had worked on that telescope with his hands, in a suit, in orbit.

So he flew to Rochester, New York, where the hardware sat in a clean room at what was then ITT Exelis, costing about $100,000 a year just to keep dust off. What he found was a row of pristine mirrors that were near-replicas of the one he had serviced.

“It was beautiful,” he said.

What NASA had before the phone call

The mission that would become Roman was then called WFIRST, and in 2010 the National Research Council’s decadal survey — the once-a-decade document in which American astronomers rank what to build next — had put it first. Top priority, whole field, unanimous enough to matter.

The design it endorsed had a primary mirror just over a meter across. Half of Hubble. Not because anybody wanted a small telescope, but because that was the telescope the budget drew. The James Webb Space Telescope was in the middle of consuming $8.7 billion and a lot of goodwill, and there was a real argument at that same 2011 meeting about whether NASA should be starting anything at all.

Then the mirror doubled in diameter, for free, from a direction nobody in the room had been watching.

Why a spy mirror is better at this than Hubble is

Here is the detail that turns a funny procurement story into an actual scientific windfall. The NRO mirrors are the same size as Hubble’s. They are not the same shape.

Hubble’s optics have a focal length of 57.6 meters — f/24. That is a long lens. It is built to take one small piece of sky and hold it very, very still, which is why the pictures you have on your wall exist. The NRO telescopes run 19.2 meters, f/8: same aperture, a third the focal length. Matt Mountain, then the director of the Space Telescope Science Institute, called them “stubby Hubbles,” and the nickname stuck because it is exactly right.

If you have ever shopped for a camera lens you already know what that trade buys. Same glass diameter, shorter barrel, and suddenly you are looking at a landscape instead of a bird. Roman keeps Hubble’s sharpness and puts about 100 times more sky in the frame than Hubble’s infrared camera can hold, onto a 300-megapixel detector array. One Roman exposure covers what would take Hubble a hundred.

And the reason a reconnaissance telescope is shaped that way is not mysterious. A satellite looking down wants ground coverage — the target moves, the swath matters, the field of view is the mission. Point that same requirement at the sky and you get a survey instrument: not “what is that thing,” but “how many of them are there, and where.” Roman is expected to map more than two billion galaxies and turn up something like a hundred thousand planets. It is very good at the census because it was designed by people who needed a census of something else.

Where the mirrors actually came from, the NRO has never said. They were built somewhere between the late 1990s and the early 2000s, and they are widely reported to be leftovers of the Future Imagery Architecture program, a reconnaissance effort that ran badly over budget and was cut back. Officially they come from nowhere, which is its own kind of answer.

Free hardware is not a free telescope

What arrived: mirror bodies, support struts, payload radiators. What did not: detectors, star trackers, filters, prism wheels, the electronics, and every other thing that turns optics into an observatory. No spacecraft. No instruments. No launch.

NASA accepted in August 2011. The public found out on June 4, 2012, and then very little visibly happened for a while, because a gift you cannot afford to use is a storage bill. The mission was not formally approved until February 17, 2016 — four and a half years after the announcement, five after the offer. L3Harris, in the same Rochester building, then reshaped the donated mirror to Roman’s prescription and built the rest of the optical assembly around it.

The finished surface has an average bump height of 1.2 nanometers, more than twice as smooth as the mission required, on a mirror that weighs 410 pounds. Hubble’s mirror, for the same aperture, weighs about a ton.

The donation saved somewhere between a couple hundred million dollars and an argument about how to count it. What it definitely saved was the design: Roman is a fundamentally more capable mission than the one the decadal survey ranked first, and it got that way because of a phone call.

The second one is still in the box

This is the part I find hardest to let go of. There were two.

NASA flew one. The other is still in storage — fifteen years after the offer, with no funded mission attached to it. Astronomers have floated ideas: put it in orbit around Mars, point it at the solar system, fly it as a second survey instrument. Nothing has been funded. Somewhere there is a Hubble-class telescope in a clean room, complete enough to matter, waiting on a line item.

And then there is the name on it

In May 2020, WFIRST was renamed for Nancy Grace Roman. She was NASA’s first Chief of Astronomy and the first woman to hold an executive position at the agency, and she spent the 1960s and 1970s doing the unglamorous, decisive work of arguing a large telescope into orbit — talking to Congress, to astronomers who thought it was a boondoggle, to anyone who would sit still. People called her the Mother of Hubble, and they were not being cute about it.

So: the telescope that a spy agency built and never used, which turned out to be Hubble’s twin with a wider eye, now carries the name of the woman who got Hubble built. Nobody planned that. It is just how it landed.

Launch morning

Liftoff is targeted for 7:26 a.m. EDT on Sunday, August 30, 2026, from Launch Complex 39A at Kennedy, with a backup at 7:22 a.m. on the 31st. Falcon Heavy’s two side boosters come back to Cape Canaveral, which means sonic booms a few minutes after the rocket is gone — for a lot of people that is the part they remember. If you are anywhere near the Space Coast, our Cape Canaveral viewing guide has the free spots and the timing, and the Roman guide covers what happens after: about 100 days of commissioning, a million miles out to L2, then five years of staring.

And there is a version of this you can do from your own yard. Roman’s planet hunt points at the galactic bulge, the crowded core of the Milky Way in Sagittarius, and on a clear night right now you can find that exact patch of sky yourself — face south, look low, find the teapot. The Milky Way guide has the how, and a pair of binoculars is the cheapest way to see it break into stars. You will not see a microlensing event. You will be looking at the same star fields.

The photograph at the top of this page was taken at Goddard in July 2025. A few people are up on a scissor lift in a dark clean room, running ultraviolet light across the primary mirror looking for contamination; the blue patch on the far wall is the mirror throwing that light back at them. The green is a long exposure picking up an airflow indicator on the left-hand wall.

It is thirty-year-old glass that was ground for one purpose by people who were not allowed to say what it was, and it is going up pointed the other way.