April 24, 1990 NASA launched a telescope worth $1.5 billion into space — and for five weeks concealed that it was practically blind. A defect of 2.2 micrometers on the primary mirror 2.4 meters in diameter turned the most ambitious scientific project of the decade into a national disgrace. The cause — a 1.3 millimeter displacement of an instrument due to a removed cap from a washer. It could have been fixed for $6000 and a few hours of ground testing. Instead, it took three years, a heroic shuttle mission, and a miracle of engineering ingenuity to turn catastrophe into legend.
Hubble's primary mirror weighed 825 kilograms — a solid disk of ultra-low expansion glass, polished to perfection. Or nearly. Perkin-Elmer ground it to a precision of 1/65 the wavelength of light — but the shape turned out wrong. The mirror's edge was too flat, the defect depth was about 2.2 micrometers — the thickness of a spider's web, but enough for light to focus not at a single point but to smear into a halo.
The instrument for checking the shape — a reflective null corrector — should have guaranteed a perfect parabola. It's an optical system of lenses and mirrors that compares reflected light against a reference standard. But the corrector itself was assembled incorrectly: one of the lenses ended up displaced by 1.3 millimeters from its required position. A technician removed the protective cap from a calibration washer to check the distance with a laser measurer — and forgot to account for the cap's thickness during assembly.
Perkin-Elmer conducted two independent tests using other methods. Both showed discrepancies. Engineers brushed them off: the primary instrument seemed more reliable, the backup methods were deemed less accurate. The mirror was sent to assembly. NASA received a telescope with built-in spherical aberration — a defect that turns point-like stars into blurry blobs.
The spacecraft, weighing 11 tons, was loaded into the cargo bay of the shuttle Discovery. Five astronauts put it into orbit at an altitude of 547 kilometers. Engineers on the ground began calibrating the instruments. The first weeks were routine system checks — solar panels, gyroscopes, antennas. Science was planned for later.
June 27, 1990 — 64 days after launch — Hubble pointed its camera at a test star in the cluster NGC 3532. Instead of a bright point, a fuzzy spot appeared on screens, surrounded by concentric rings. Engineers checked the focus. Then checked again. Then began running through possibilities: launch vibrations, thermal deformations, secondary mirror displacement.
NASA stayed silent. The press office issued vague statements about "calibration" and "fine-tuning." Astronomers who'd been promised a revolution in observations were getting images worse than ground-based telescopes. Rumors leaked to Science and Nature by mid-summer, but there was no official confirmation.
July 2 NASA convened an investigation board chaired by Lew Allen — a retired Air Force general and former director of the Jet Propulsion Laboratory. The investigation proceeded in parallel with desperate attempts by engineers to find a software solution. Maybe it's the pointing software? Maybe the defect can be compensated mathematically?
By late summer it became clear: the problem was in the optics, and it was fundamental. Analysis revealed the characteristic pattern of spherical aberration — sharp image center, blurry edges, a halo of scattered light around bright objects. They compared it with archived ground-testing data of the mirror. Pulled up Perkin-Elmer reports. Found records of that very reflective null corrector. Found the rejected results of two backup tests.
The Allen Commission completed its work in November. The 400-page report established: the mirror defect arose from an assembly error in the measuring instrument, which could have been detected by simple interferometric verification. Cost of such verification — about $6000. Time — a few hours. But the culture at Perkin-Elmer encouraged trust in the primary instrument rather than cross-checking with multiple methods.
The physics of the aberration proved brutal: 85% of light from a star was smeared into a halo with a radius up to 2 arc seconds, only 15% made it to the central core. For comparison: ground-based telescopes of that era achieved 70-80% light concentration in the center, despite atmospheric distortions. Hubble in orbit, far from air turbulence, turned out worse than instruments on Earth.
The U.S. Congress held hearings. Senators demanded heads. The press dubbed the project "Hubble trouble" and "techno-turkey" — jargon for failed military programs. Cartoonists drew the telescope wearing glasses. The scientific community split: some demanded a repair mission at any cost, others proposed writing off the project and directing funds to ground-based astronomy.
NASA faced a choice: admit defeat or risk another shuttle, another $700 million, and the agency's reputation. The decision came quickly — fix it. But how do you repair a mirror that's 559 kilometers of vacuum and weightlessness away?
Replacing the primary mirror was impossible — Hubble wasn't designed for such a repair. The mirror is integrated into the structure; dismounting it would destroy half the telescope's systems. Engineers proposed something different: don't fix the defect, compensate for it with additional optics. Contact lenses for a cosmic eye, essentially.
COSTAR — Corrective Optics Space Telescope Axial Replacement — a system of 10 small mirrors the size of coins that intercept the distorted light before it reaches the scientific instruments. Each mirror has a shape inverse to the primary mirror's defect: it introduces a counter-distortion, neutralizing the spherical aberration. Light passes through COSTAR and becomes what it should have been originally.
Development took two years. Engineers at Ball Aerospace created the optics in clean rooms, tested them on the ground with telescope models, rehearsed installation in a giant pool with a full-scale Hubble mockup. COSTAR weighed 27 kilograms and occupied the space of one instrument bay — they had to sacrifice the High Speed Photometer, one of the science cameras.
Mission STS-61 launched December 2, 1993 on the shuttle Endeavour. Seven astronauts, 11 days in orbit, 5 spacewalks totaling 35 hours. They replaced the solar arrays, installed COSTAR, upgraded the Wide Field and Planetary Camera to a version with its own corrective optics, swapped out gyroscopes and electronics.
December 13 the shuttle undocked. Engineers began tests. The first image with corrected optics arrived a week later — star M15 in a globular cluster, a sharp point without a halo. Then nebula M100 — spiral arms emerged clearly, better than any ground-based image.
January 1994 — NASA holds a press conference, shows "before" and "after" images. The difference is dramatic: blurry blobs transformed into crystal-sharp images of galaxies, nebulae, planets. The press, which had been mocking the project for three years, pivots 180 degrees. Hubble becomes a symbol not of failure but of triumph over error.
The following years bring discoveries one after another. Hubble measures the expansion rate of the Universe with a precision that settles half-century-long debates. It photographs colliding galaxies, star birth in nebulae, exoplanets in the atmospheres of other stars. Deep Field — a 1995 image where in a patch of sky the size of a pinhead there are 3000 galaxies at distances up to 12 billion light-years.
COSTAR operated until May 2009, when the final repair mission STS-125 installed new instruments with their own corrective optics and removed the prosthesis. It was sent to the National Air and Space Museum in Washington — physical proof that 1.3 millimeters of error can be fixed in orbit.
Over 34 years of operation, Hubble made more than 1.5 million observations, collected data for 20 thousand scientific papers, changed our understanding of the age, size, and fate of the Universe. The most expensive mistake in spaceflight turned into the most productive scientific instrument in history. But this transformation cost three years of national humiliation, heroic engineering, and acknowledgment of a simple truth: trust, but verify — especially when the stakes are measured in billions of dollars and the reputation of an entire generation of scientists.