Skip to content
SIGNIFICANT FIGURES Everyone measured. It broke anyway.

Hubble mirror · 1990

Hubble: NASA puts the spacing error at 1.3 millimetres

The mirror was tested. The instrument built to test it had a lens in the wrong place, so the polishing machine stopped at the wrong moment and the test it had to pass was the test that was broken.

The Hubble Space Telescope reached orbit in April 1990. In late June it failed a focusing test(Verified) , and NASA announced on 27 June that the telescope was flawed.

The primary mirror had been polished to the wrong curve: too flat away from its centre.(Verified)

This is the worst case on this site, and not because of the size of the error.

They tested it

Nobody launches a mirror of that size without measuring its shape first. The measurement was done. The instrument that did it is called a null corrector, and it exists for one purpose: to tell the polishing machine when the mirror has reached the right figure.

The null corrector was assembled with one of its lenses in the wrong position, and it guided the polishing.(Verified) Both sources below say so, in those terms.

NASA puts the lens spacing error at 1.3 millimetres.(Single source: NASA) The independent source declines to give a figure at all, which is why that sentence carries a name and the ones above it do not.

Fig. 1 — Order of operations
  1. 01

    The null corrector is built. One lens sits slightly out of position.

  2. 02

    The machine polishes until the null corrector says stop.

  3. 03

    It says stop at the wrong moment.

  4. 04

    The mirror comes out perfectly smooth, perfectly uniform, and the wrong shape.

    The check passes, because the check is the thing that is broken.

A verification instrument does not sit beside the process. It drives it. When the reference is wrong, every step that follows is executed correctly against the wrong target, and each one adds confidence.

Source: NASA; BBC Sky at Night Magazine

NASA describes the resulting error as ten times larger than the specified tolerance.(Single source: NASA) Again: one organisation, named.

What a broken check does to confidence

Every other case on this site has a moment where the evidence was available and somebody could have seen it. The orbiter’s navigation drift sat in the data for months. The bridge halves could have been compared against a single datum. The platform sample could have been questioned.

Here the evidence pointed the other way. The mirror had been measured, with the correct instrument, by people who knew how to use it, and the measurement said the mirror was right. A second test would have been run on the same corrector. More care would have produced a more precisely wrong mirror.

The telescope flew out of focus for three and a half years. Astronauts installed corrective optics in December 1993(Verified) — optics ground to cancel an error that nothing on the ground had been able to see.

The mirror is still the wrong shape. Every Hubble photograph reaches us through a correction.

The figure that is everywhere and nowhere

The 2.2 figure is well formed. It has a plausible magnitude, it uses the right unit for the domain, and it fits the sentence people want to write. Those are the properties of a number that spreads, and none of them are the property of being true.

It is also the clearest argument for counting organisations rather than pages. There are many pages. There is no second organisation, and there is no first one either.

What the sources support

Four claims in this article have two independent organisations behind them: the failed focusing test, the wrong curvature, the null corrector as cause, and the December 1993 repair. Two claims have one organisation, and both name it in the sentence.

That division is the whole method. It is not a disclaimer at the bottom of the page — it is visible in how each sentence above is built.

Cut register

Figures that circulate and are not published here

  • 2.2 µm UNVERIFIED Absent from NASA and from the independent source. Found only in generated content farms whose page titles are full sentences.

Sources

2 entries · 2 independent organisations

  1. S01

    NASA PRIMARY

    Hubble's Mirror Flaw

    Mission owner. The only retrieved source that gives the 1.3 mm spacing figure and the ten-times-tolerance comparison.

    Accessed 11 September 2026

  2. S02

    BBC Sky at Night Magazine SECONDARY

    What was wrong with Hubble's mirror?

    Independent of NASA. Confirms the failed focusing test, the wrong curvature, the null corrector as cause and the 1993 repair. Declines to give a spacing figure.

    Accessed 11 September 2026