Not long after he founded the Mirror Laboratory at the University of Arizona's Steward Observatory in 1984, astronomer Roger Angel threw some Pyrex custard cups into a backyard kiln to get a feel for how borosilicate glass melts. He has been playing with fire ever since. On a sweltering July day in Tucson, his seventh giant mirror takes shape in a 200-ton rotating oven underneath the campus football stadium. After a week of heating, the borosilicate blocks in the oven have reached 1,170 degrees Celsius and begun to liquefy, sending molten glass seeping over hexagonal pillars to form a 21-ton honeycomb that is 8.4 meters (28 feet) in diameter, roughly a meter high, but nowhere thicker than three centimeters.

Usually Angel would be in the control room, nervously watching the hot, red carousel spin five times a minute, just fast enough to pull the solid top of the glass honeycomb into the desired gentle curve. To Angel, after all, this process represents the birth both of a new era of astronomy and of the Giant Magellan Telescope (GMT), which, by banding together seven huge mirrors with sophisticated control mechanisms, would outdo the Hubble Space Telescope. And the first time his team made a mirror of this size, the mold leaked two tons of glass, causing months of delay. This first GMT mirror is trickier still, Angel explains, because "it is wickedly curved into an extreme shape that no one has ever made before at this scale."

But Angel had to leave the control center this morning to zip over to a hotel, where he is trying to persuade some of the high rollers of science funding that the GMT is worth its $400-million price. Wendy Freedman, director of the Carnegie Observatories, which is a GMT partner, says she has full confidence in Angel and his mirror-casting abilities. As a cosmologist, Freedman is keen to use the GMT to solve mysteries about dark energy and the "dark ages" before stars and galaxies formed. But those are not the questions that most excite Angel.

"I got into this because we have been searching for another Earth for the past two millennia, and for the first time it seems possible to find planets like ours around other stars," he says in a British accent tempered by almost four decades in the U.S. "You can now sit down and plan seriously what kind of telescopes you would need to image an Earth-like planet and to take its spectra," to look for the chemical signatures of liquid water and alien life.

Angel did those calculations before he came up with the basic design of the GMT, and he figures that with a bit of luck and some technical wizardry, it may be able to pick evidence of an Earth-like world out of the glare of its star. His characteristically unconventional solution is to arrange the primary collectors like the petals of a daisy, with one symmetric mirror in the center surrounded by six off-axis partners, each distinctly lopsided in shape. The assembly will gather 4.6 times as much starlight as does one of the California Institute of Technology's two 10-meter Keck Telescopes, currently the world's largest.

To snap legible photographs of such planets, GMT will use advanced adaptive optics to make images that are 10 times as sharp as those from Hubble. Each primary mirror will focus its photons onto a separate, 1.1-meter secondary reflector, a thin membrane attached to 672 actuators. Computers will then cancel out much of the atmospheric blurring by making 1,000 subtle adjustments to the shape of each secondary mirror every second. This past summer Angel and his co-workers installed a similar adaptive optics system in the Multiple Mirror Telescope and demonstrated that it boosted the resolving power of the instrument to its theoretical maximum.

"When we started describing our plans for the GMT, lots of people told us it would never work," says Peter A. Strittmatter, director of Steward Observatory. Angel faced similar skepticism 20 years ago when he and his colleague Nick Woolf first proposed making large mirrors with honeycomb backs, a strategy that cuts the weight by four fifths and enables the mirror to settle to the temperature of the night air within half an hour.

But the idea worked: as evidence, Freedman points to the twin Magellan Telescopes, precursors of the GMT equipped with two Angel-baked mirrors cast a decade ago and now operating on an Andean mountaintop in Chile. "These are the best natural imaging telescopes in the world, bar none," she avers. Making such mirrors is sufficiently complicated, however, that the Arizona mirror lab is still the only outfit in the world that does it at large sizes.

Although the GMT has high-profile partners backing it--including the Carnegie Institution, the Harvard-Smithsonian Center for Astrophysics and five major universities--it competes with several other mega-telescopes for funding. Two of the competing proposals would scale up the Keck's segmented mirror design. Caltech and others want to build one 30 meters in diameter. The European Southern Observatory is working out plans for a 1.2-billion-euro, 100-meter observatory that it has dubbed the OWL, for Overwhelmingly Large Telescope.

"It may come to a shootout, with only one getting funded," Angel says. To get a foot in the door, GMT proponents opted to spend nearly all their $20 million start-up money on this first off-axis mirror. "All the other groups who are thinking about building really big telescopes are still writing talking points," he observes. "We're the first to do something about it."

With this hurdle nearly cleared, Angel is already turning his attention elsewhere. "By the time things get into the planning stage, Roger has already moved on to the next exciting thing," Freedman notes. "He's just full of ideas that have revolutionized how telescopes are being built." Those ideas now include designs for large telescopes in deep space and at the poles of Earth and the moon.

In recent papers, Angel has touted a plateau in central Antarctica at an altitude of 3,300 meters, called Dome C, as a near-ideal observing site. "At Dome C, every quality factor important for ground-based astronomy is better by a factor of two or more than at any other spot on the planet," he enthuses. The polar base completed there last year has on average the least wind, the lowest temperatures, and the driest air of any human outpost. Although the National Science Foundation recently rejected a proposal to construct a two-meter telescope in Antarctica, Angel continues to push the idea. "A 100-meter telescope there could rival the best you could do in space," he argues.

Of course, space has its allure as well. Angel has analyzed the advantages of a 20-meter lunar telescope built in a crater near the moon's south pole by using a spinning pool of reflective liquid as a mirror. "The idea of installing something on the moon makes most astronomers nervous, because it means being the tail on a very large dog," he acknowledges. But by staring for long periods at the same patch of sky above it, such a telescope could peer to the very edges of the visible universe.

Even if the Giant Magellan Telescope finds funding, it is unlikely to see its first light before the middle of the next decade. At age 64, Angel has little hope of living to see any of his more ambitious designs completed. But that doesn't bother him. "It's like building cathedrals," he says. "You can't set the criterion of wanting to see it finished in your lifetime."