The mirror, a perfect hexagon of gunmetal gray, stands vertically on a low platform. It is about two inches thick and more than four feet wide, a precisely carved slab of beryllium that gleams in the low light of this optics laboratory near San Francisco Bay. My guide, chief engineer Jay Daniel, watches my footing as I step gingerly in front of the mirror to see my reflection. “It’s like your bathroom mirror,” Daniel says, chuckling.

The other side of this looking glass, though, is nothing like a household vanity. The slab of metal is mostly hollow, drilled out by machinists to leave an intricate triangular scaffold of narrow ribs. It is beautiful in its geometric precision, and I resist the urge to touch one of the knifelike edges. The polished front layer that remains, Daniel says, is a mere 2.5 millimeters thick. From its starting weight of 250 kilograms, the entire mirror now weighs just 21 kilos. That is light enough for a rocket to hoist 18 of them deep into space, where the curved mirrors will join as one to form the heart of the most audacious space observatory ever launched.

That observatory, a $5-billion NASA mission (in partnership with the European and Canadian space agencies) called the James Webb Space Telescope, or JWST, is scheduled to carry on in 2014 as a successor to the iconic Hubble Space Telescope. The Hubble has circled 570 kilometers above Earth since 1990, giving astronomers their sharpest views of galaxies in the distant universe and of the births and deaths of stars closer to home. Like the Hubble, the Webb promises stunning images of the cosmos, but with far more penetrating vision. Astronomers have designed it to stare back toward the beginning of the universe. It may spot the explosions of the first stars that arose after the big bang and reveal the origins of galaxies similar to our Milky Way. It also will look deeply into clouds of gas and dust, the wombs of gestating stars and their families of planets.

To meet these goals, the Webb will be radically different from its predecessor. Its lightweight mirror will span more than 6.5 meters, giving it six times the light-collecting power of the Hubble’s 2.4-meter-wide mirror. Coated with gold, the telescope’s 18 hexagonal panels will act as a uniform surface—a feat requiring them to align within one ten-thousandth the width of a human hair. NASA will hurl this honeycombed eye into a looping orbit far beyond the moon. Along the way, it will unfurl a giant sunshield, casting a frigid shadow in which the mercury will fall below 55 kelvins so the telescope can sense traces of light and warmth that have straggled across the universe for more than 13 billion years.

All this involves unprecedented technical risks. Because of the telescope’s remote perch, no astronaut will be able to fix it if something goes wrong. Unlike with the Hubble, which has had several repairs and upgrades throughout the two decades it has been in operation, there will be no do-overs, no shuttle flight to correct an embarrassing optical flaw, no widget to get that pesky shield unstuck. What’s more, to get to its lonely orbit, the probe must first fold up to fit in the cramped cargo bay of an Ariane 5 rocket. Hitching that ride puts strict limits on the telescope’s weight and dimensions. The observatory must then deploy itself with balletic precision, a tortuous sequence that ends with two folded panels of mirror segments rising into place like the sides of a drop-leaf dining table.

“I think of it as the origami telescope,” says Mark Clampin, the observatory project scientist at the NASA Goddard Space Flight Center in Greenbelt, Md. “We have to unpack it, align it and have it work at the right temperature. It’s not designed to be serviced, so everything has to work on the day.”

The combination of mass, size and temperature constraints—and the mechanical derring-do required to pull it off—has forced NASA to spend far more money on the Webb than astronomers had hoped. A national panel in 2001 ranked the space observatory as astronomy’s top priority and called for a $1-billion budget, but that figure was naive. It did not include costs to launch and operate the telescope, and it grossly missed the mark on how complex and time-consuming the design would become. “The [engineering] challenges are much greater than initially anticipated,” says Webb program scientist Eric P. Smith of NASA headquarters in Washington, D.C.

The project’s costs are not far out of line with those of other pioneering satellites, Smith notes. For instance, the Chandra X-ray Observatory (now orbiting Earth) and the Cassini spacecraft (now touring Saturn and its exotic rings and moons) each cost roughly $4 billion for their complete life cycles, in 2007 dollars. “This is what it costs to build a big flagship mission,” says Alan Dressler of the Carnegie Observatories in Pasadena, Calif., who chaired the first report on the Hubble’s successor in 1995. The Webb is “being built quite effectively, and money is not being wasted,” he says.

Some dismayed researchers think the observatory saps a disproportionate share of NASA’s astronomy budget, shutting out other missions. “The opportunity cost of JWST is very high, and it will be felt throughout the decade,” says astrophysicist Shrinivas Kulkarni of the California Institute of Technology. In particular, he says, advanced probes to explore gravitational waves, the high-energy universe, and the details of possible planets like Earth around other stars must now wait until the 2020s and beyond.

Even the Webb’s supporters are on edge about whether this huge investment will pay off in a mission that works as advertised. “This is a really challenging project,” says Garth Illingworth of the University of California, Santa Cruz, a longtime Hubble user. “Even by the standards of most NASA projects, this is a tough one. For most of the things on JWST, if it doesn’t deploy, it’s dead.”

The Undiscovered Country
Most of the telescope’s shape-changing will occur before it reaches its home in the solar system: a gravitational balancing point called L2, more than one million kilometers in deep space. There the spacecraft will use tiny shots of fuel to follow Earth’s pace in a gentle, undisturbed orbit around the sun. Engineers think it will have enough fuel to last a decade or so until they can no longer steer it. (NASA’s budget covers five years of operations; extending it five more years would add $500 million.)

Astronomers knew they would have to propel the satellite far away when they began planning it 15 years ago. The Hubble basks in the unwelcome glow of Earth, which means it has to operate at close to room temperature. It is therefore blind to faint infrared light from the distant objects astronomers would dearly like to see: the earliest ancestors of today’s galaxies, scattered at the margins of the visible universe. Their light would have been visible to our eyes—and to the Hubble’s cameras—back then. But as the cosmos expanded in the intervening billions of years, the light’s waves have stretched out of the visible spectrum and into the infrared. 

“This is where the new opportunity exists,” says chief project scientist John C. Mather, a Nobel laureate at NASA Goddard. “This is the rock we’ve never turned over, the place we’ve never looked. We don’t know what the first objects were that lit up after the big bang, and we ought to go find out.”

Named for NASA’s administrator during the Apollo era, the James Webb Space Telescope will use infrared cameras and other detectors to sense the first shards of galaxies as they assembled into the kind of majestic bodies we see today. Those embryonic objects probably existed 400 million years or so after the big bang—just 3 percent of the universe’s current age. Its cameras may detect the sparks of even earlier stars, behemoths with hundreds of times the mass of our sun. Such stars would have detonated after short, brilliant lives, casting flares of light that still travel across the cosmos.

“We are going to extraordinary lengths to build a far more challenging telescope than the Hubble, to be able to see back as far as we will ever see,” says Carnegie’s Dressler. “NASA had a strong desire to build the first of the new telescopes, rather than the last of the old.”

Infrared light also opens a portal toward objects closer to us, enabling us to see through the shrouds of dust that hide the nurseries of stars and planets in our galaxy. Right now astronomers detect alien planets mostly using the visible spectrum of light, so they are able to see only objects in systems that have “cleared out” of the disks of gas and rocky debris that created them. But because infrared light can penetrate through dust, the Webb will unveil many stages in those acts of creation, helping us determine whether our solar system is rare or common. Some of the planets will cross in front of their stars, giving the sharp-eyed Webb a chance to detect gases in their atmospheres. It is a long shot, but the telescope just might find a planet with an unstable mixture of gases, such as oxygen, carbon dioxide and meth­ane—­­the first signs of life elsewhere.

Beryllium Does It Better
Other space telescopes have used small mirrors made of beryllium, the second-lightest metal. And the Webb’s 6.5-meter primary mirror is not huge, by astronomy standards: several telescopes on the ground now have mirrors ranging from eight to 10 meters across, with far bigger ones on the drawing board. But creating 18 beryllium segments to form a single, smooth surface in deep space has taxed optical technicians as never before.

It is a challenge embraced by the engineers at Tinsley, an optics company in Richmond, just north of Berkeley, Calif., owned by L-3 Communications. Before my visit—the first time Tinsley has allowed a journalist to see its work on the mirrors—Daniel asks me not to bring a camera and lets me know that some questions are off-limits. Making telescope mirrors is intensely competitive; Tinsley has spent years and millions of dollars perfecting NASA’s prescription for the Webb and its metallic eyes.

When I arrive, I learn in a briefing that beryllium powder is toxic. I must sign a waiver that absolves Tinsley in case of pulmonary distress. Not to worry, Daniel assures me: the lab polishes its mirrors only with wet processes, so there is no dust floating around. My lungs can relax—although at times I wear a surgical mask to keep a sniffle from soiling the beryllium.

In the project’s early days, astronomers assumed they would use ultra­low-expansion glass, which holds its shape when temperatures change. But when opticians made test mirrors and plunged them to the kind of bitter cold the telescope will experience, the glass warped in a way that might have thrown the telescope out of whack. In contrast, beryllium is stiff and well behaved in such conditions.

That change, though, added a year to the mirror’s production schedule, because beryllium takes longer to polish. “It’s extremely hard to make a beryllium mirror without leaving stresses in it,” says optical engineer Bob Brown of Ball Aerospace & Technologies in Boulder, Colo., which oversees Tinsley’s work on the telescope. Carving the surface makes the remaining metal want to bend upward, Brown says. The team must remove that layer of stressed metal by gently etching the mirror with acid or grazing it with a sharp tool. It is a tedious, exacting process.

To view the mirrors, I don booties and a smock to keep stray beryllium off my shoes and clothing. Daniel and Brown escort me onto Tinsley’s factory floor, built expressly for the Webb. Eight polishing machines, each about two stories high, dominate the room. A mirror segment sits on one of the computer-controlled machines. A black bellows, shaped like an accordion, makes a groaning sound as it gently moves a robot back and forth over the mirror. Attached to the tip of the robot is a Frisbee-size polishing head. The computer dictates how long the spinning polisher works on each spot to remove an exact amount of beryllium.

A whitish liquid, looking like diluted milk, lubricates the rotating head and flows off the sides of the mirror in a constant stream. When I ask Daniel what it is, he smiles. “It’s a polishing fluid,” he says after a pause. “It’s a home brew. It’s very specifically stipulated, and it’s proprietary.” Brown points to the edges of the hexagon. Within five millimeters of the border, he says, the mirror is still smooth, a difficult polishing feat never before tried on surfaces this large. If the margin were twice as wide, the telescope would focus 1.5 percent less starlight into a sharp image—a big loss of data from the faintest objects.

Opticians measure the precision of the mirrors’ surface in Tinsley’s metrology lab, an enclosed space with strict controls on temperature and air currents. The technicians use holograms, infrared lasers and other tools to gauge the height of the mirror’s surface at hundreds of thousands of points. A segment goes back and forth between the polishing machines and the metrology lab a couple of dozen times to get the shape and smooth finish that NASA requires.

Next, each segment is flown to Ball Aerospace, where engineers attach it to its flight hardware—a graphite composite structure that latches onto the hexagon’s rear grid work and holds it in place in the telescope. Next, it travels to the NASA Marshall Space Flight Center in Hunts­ville, Ala., for testing in a large vacuum chamber cooled by liquid helium to 25 kelvins. In those conditions, the metal warps in subtle ways, which opticians map in microscopic detail. The segment then returns to California, where Tinsley uses the opticians’ maps to guide additional, subtle polishing that will cancel out whatever warping the beryllium undergoes once it is subjected to the cold of space.

This slow waltz has been going on since December 2009. As of August, one mirror segment was done, and about half a dozen others were in their final polishing stages. Tinsley plans to deliver all 18 segments (plus three spares) to NASA by mid-2011. 

Learning from the Hubble
As Tinsley’s engineers work, the dramatic flaw of the Hubble Space Telescope is very much on their minds. The Hubble’s mirror was polished to the wrong shape, thanks to a measuring er­ror that engineers overlooked. Shuttle astronauts installed corrective mir­rors three years after launch and saved the mission. No such option exists here.

Heeding the Hubble’s lessons, NASA recruited engineers who helped to fix the Hubble to work on the new mission. The same technique used to diagnose the shape of the Hubble’s deformed mirror by studying its blurry images will keep the Webb in sharp focus. “As we move the telescope across the sky, certain thermal gradients set in and the telescope gently drifts out of shape,” says Matt Mountain, director of the Space Telescope Science Institute in Baltimore, which will oversee its operations. But unlike any other space observatory, the Webb telescope will have an active, adjustable mirror to compensate for those changes. 

First, small lenses in the telescope’s instruments will create out-of-focus images like the ones that plagued the Hubble. After analyzing those pictures, mission control will send radio signals to activate seven tiny motors on the back of each mirror segment. Each motor, built at Ball Aerospace, can push or pull on the mirror in increments of fewer than 10 nanometers. That gives astronomers control over each segment’s curvature and its position relative to the neighboring hexagons. Mission control will perform that procedure and recalibrate the mirror every two weeks or so.

Of course, the telescope needs to deploy itself properly in the first place. In particular, two folded “leaves,” each bearing three of the mirror segments, must swing properly to form the entire surface. A single, 75-centimeter-wide secondary mirror also must latch into its perch, on a spidery tripod seven meters above the main mirror, to reflect light back through the center of the primary and to the instruments that record the data.

But the transformation that really makes observers gulp is the opening of the enormous sunshield, which is 11 meters wide and 19 meters long. If it does not work, the sun’s heat will blind the instruments to most of their targets. In a video simulation, the shield spreads out like a stack of five candy wrappers, each with the surface area of a volleyball court. NASA’s prime contractor, North­rop Grumman in Redondo Beach, Calif., has designed satellites with giant, unfurling antennae—as well as so-called black operations in space for the government, according to NASA officials. But the Webb will be the most mechanically complex civilian mission of this type yet attempted.

Adding to the stress is that no cold vacuum chamber is big enough to test the entire sunshield before launch. To keep costs from growing even higher, NASA adopted a riskier procedure that tests critical pieces of the observatory—but never the whole thing. “That’s life in the fast lane,” Mountain says. “We’re going to have to take an extra leap of faith.”

For now the mission’s scientists are focused on building the telescope and its instruments. But they cannot help looking past the 2014 launch, too. “At one level, this is our generation’s contribution to civilization,” says NASA Goddard’s Lee Feinberg, the Webb’s optical telescope element manager. “It won’t last forever, but it will be out there in space. Future generations probably will be able to find it with big telescopes.” And perhaps one day the observatory’s golden mirror—pocked by space dust and weathered by radiation—will be towed back to Earth as a monument to the time when we first grasped our cosmic past.