Mars Climate Orbiter · 1999
Mars Climate Orbiter: a factor of 4.45, filed hundreds of times
The software aboard the spacecraft was correct. A program on the ground wrote pound-seconds where the specification required newton-seconds, and every thruster firing entered the navigation model 4.45 times too weak.
The carrier signal was last seen at 09:04:52 UTC on Thursday, 23 September 1999(Verified) . The spacecraft had launched on 11 December 1998(Verified) and flown for 286 days.
The interesting part of this case is where the error is not.
The spacecraft was right
The Mars Climate Orbiter carried software that handled angular momentum desaturation — the small thruster firings a spacecraft uses to stop itself slowly twisting. That software computed in metric units and it was correct(Verified) . The investigation board says so directly. Most retellings of this story get that detail backwards.
The error sat in a program on the ground, named SM_FORCES. Its job was narrow: record how hard each of those thruster pushes had been, and hand the figure to the navigation team.
A written specification governed the handover. It required newton-seconds. The program wrote pound-seconds.
Newton-seconds. Metric, as required by the project software interface specification.
Pound-seconds. Every value arrived at the navigation model a factor of 4.45 too small.
1 lbf = 4.45 N
One pound of force is 4.45 newtons. Reading a pound-second value as a newton-second value does not produce a small error. It produces the same event, filed at less than a quarter of its true strength.
Source: NASA, Mishap Investigation Board Phase I Report
The navigation software underestimated the effect of the thruster firings by a factor of 4.45(Verified) . One pound of force is 4.45 newtons.
Why one wrong number became hundreds
A single mis-filed figure is recoverable. This spacecraft did not produce a single figure.
Its solar panel was mounted asymmetrically with respect to the body(Verified) , unlike the Mars Global Surveyor that had flown before it. Sunlight pressed on the lopsided panel, the craft slowly rotated, and it fired thrusters to hold its attitude. The navigation team had planned for that. It happened ten to fourteen times more often than they expected(Verified) .
The board is careful about this, and it is worth quoting the shape of its finding rather than paraphrasing it. The cause was the increased frequency of those desaturation events coupled with the fact that the data arrived in English rather than metric units. Not the units on their own.(Verified)
So the error was not one wrong number. It was a wrong number, filed hundreds of times, by a spacecraft that could not stop correcting itself.
The drift was visible
Across nine months the navigation team steered using those figures. The orbiter drifted. The drift was in their data. They discussed it and kept flying.
On 29 September 1999, six days after the loss(Verified) , an engineer lined up the two files and found that the reported velocity changes were low by a factor of 4.45.
Afterwards the board re-ran the navigation with corrected values. The orbiter’s lowest point of approach came out at 57 kilometres(Verified) . The report’s phrase for that altitude is “too low for spacecraft survival.”
The board’s root cause is one sentence: the failure to use metric units in the coding of a ground software file.
The figure this article does not use
The 226 figure is the kind of number that survives by usefulness. It makes the contrast legible: a target, a miss, a gap you can picture. That is exactly why it needs a document behind it, and it does not have one here.
One document
This article rests on a single source: the Mishap Investigation Board’s Phase I report. Under the method used here that is enough, because the report is the primary record of the event it investigates, not an organisation summarising its own programme. It is still one document, and that is worth saying plainly rather than implying a consensus that was never assembled.
Everything above is in it. Nothing above is from anywhere else.
A smaller version of the same mistake stopped a bridge in mid-air four years later, and that one was known about in advance.
Cut register
Figures that circulate and are not published here
- 226 km UNVERIFIED Repeated in most retellings and absent from the investigation report. No intended altitude is stated here.
- $125 million DISPUTED $125M, $327.6M and other totals circulate. The report contains no cost figure, so this site uses none.
Sources
1 entry · 1 organisation
-
S01
NASA PRIMARY
Mars Climate Orbiter Mishap Investigation Board, Phase I Report
The official investigation, 44 pages, chaired by Arthur G. Stephenson. Retrieved from a University of Glasgow mirror; the document itself is NASA's.
Accessed 11 September 2026