Machine or mystery
Hubble's mirror was ground to the wrong shape. How do you fix a telescope in orbit?
Hubble's primary mirror was polished to an exquisitely precise version of the wrong curve, because the instrument used to check it during manufacture was itself assembled incorrectly. The fix was not to replace the mirror but to put corrective optics in front of the instruments — spectacles for a telescope — installed by astronauts three and a half years after launch.
- Revision: 1
- Published: 2026-08-14
- Updated: 2026-08-14
- By: Space Now, edited by J. S. Nova
It is the reason serviceability stopped being treated as a luxury. Hubble was designed to be visited, and that single decision converted a career-ending manufacturing error into an expensive delay.
An error measured in millimetres, then in years
The mirror's outer edge was too flat. Not by a visible amount — by roughly ten times the tolerance the design allowed, which is still far finer than anything the eye could detect. The consequence was spherical aberration: light striking the outer edge of the mirror came to a different focus than light striking the middle, so no single position produced a sharp image. NASA announced the flaw on 27 June 1990, two months after launch.
The instrument that lied
The mirror was not measured carelessly. It was measured with a device called a null corrector, built specifically to verify this shape, and the mirror was polished until it matched what that device reported as perfect. The device was wrong: a spacing inside it was off by 1.3 millimetres. Every subsequent measurement inherited the error and confirmed it, because they were all made with the same faulty reference. The mirror is an almost flawless example of the wrong surface.
Spectacles, not a new mirror
Replacing a primary mirror in orbit was never realistic. Instead, engineers characterised the error precisely — an advantage of a flaw that is consistent rather than random — and built optics that introduced exactly the opposite distortion. In December 1993 astronauts installed two things: a redesigned camera, WFPC2, with the correction built into it, and COSTAR, a package of corrective mirrors that acted, in NASA's description, like a pair of eyeglasses for the remaining instruments.
Chronology
Generated from dated relations, with the precision of each date stated.
- day precisionNASA announces that the telescope's primary mirror suffers from spherical aberration.
- month precisionServicing Mission 1 installs WFPC2 and COSTAR, correcting the optics.
Related records
compared with
What do you do about a flaw you cannot reach?
Confidence: high · 1993-12-02 (day precision)
relates to
Two orbital repairs, twenty years apart.
Confidence: high · 1990-04-24 (day precision)
relates to
Serviceability, before and after it was designed in.
Confidence: high · 1990-04-24 (day precision)
Inspect the record2 assertions, each with its source
Every assertion, its source class, and any conflict or uncertainty in the record.
The mirror's outer edge was too flat by about ten times the specified tolerance, causing spherical aberration; the flaw was announced on 27 June 1990 and traced to a null corrector whose internal spacing was off by 1.3 millimetres.
Source class: space agency recordNASA Science — Hubble's Mirror FlawView the source
The correction was installed during the first servicing mission in December 1993: the redesigned WFPC2 camera with internal corrective optics, and COSTAR, described by NASA as functioning like a pair of eyeglasses for Hubble's remaining instruments.
Source class: space agency recordNASA Science — Hubble's Mirror FlawView the source
CorrectionsNone recorded
No corrections recordedThis record has not been corrected since it was published.
