Few mistakes are as humbling as the unit conversion error. The arithmetic is usually perfect; the disaster comes from everyone agreeing on the numbers while silently disagreeing on what the numbers mean. Pounds versus kilograms, inches versus centimeters, one small assumption, and the results range from expensive to catastrophic.
These are true stories. Each one was investigated, each one was preventable, and each one carries the same lesson for anyone who works with measurements, which is nearly everyone.
The $125 million typo: Mars Climate Orbiter
In September 1999, NASA's Mars Climate Orbiter reached Mars and was supposed to settle into orbit as a weather and communications relay. Instead it flew too low, broke apart in the Martian atmosphere, and was lost. The mission had cost about $125 million.
The investigation found a mismatch hiding in plain sight. One engineering team, at contractor Lockheed Martin, had been producing thruster performance data in English units of pound-seconds. The navigation team at NASA's Jet Propulsion Laboratory expected metric newton-seconds. The numbers flowed between the teams for months, everyone assuming the units matched, and nobody checking.
Each thruster firing nudged the spacecraft slightly off its planned trajectory. The errors accumulated over the cruise to Mars until the orbiter arrived roughly 170 kilometers lower than intended, deep enough that the atmosphere destroyed it. The failure review board's phrasing was blunt: the root cause was the failure to use metric units in the coding of a ground software file. A conversion nobody performed, on data everyone trusted.
The Gimli Glider: running on empty at 41,000 feet
On July 23, 1983, Air Canada Flight 143, a Boeing 767, ran out of fuel midway through its flight and glided to an emergency landing on a decommissioned airstrip in Gimli, Manitoba, that was being used that day for drag racing. All 69 people aboard survived, and the aircraft earned its nickname: the Gimli Glider.
The chain of errors began on the ground in Montreal. Canada was converting to the metric system, and the brand-new 767 was the first Air Canada aircraft to measure fuel in kilograms. The fuel gauges were not working, so the crew had to calculate the fuel load by hand from dipstick measurements.
Working from the volume of fuel, they needed to convert to mass. The correct factor was 1.77 pounds per liter... no, kilograms per liter. They used 1.77 as pounds per liter out of habit, which meant the aircraft took off with roughly half the fuel the crew believed it had. About halfway to Edmonton, both engines went quiet.
Captain Robert Pearson, an experienced glider pilot, flew the powerless jet in a steep sideslip maneuver to lose altitude without gaining speed, and touched down on the old runway. It remains one of the most remarkable emergency landings in aviation history, and it began with a single wrong conversion factor written down by tired people at midnight.
Medicine: where milligrams matter most
Aviation and spaceflight get the dramatic retellings, but conversion errors in medicine are arguably the most consequential category, because the margin between a dose and an overdose can be small. Health agencies have documented cases where confusion between units, between milligrams and micrograms, or between different concentration expressions contributed to serious dosing errors.
The pattern is always the same: the calculation is correct and the units are wrong. A microgram is one-thousandth of a milligram, so mistaking one for the other is a thousand-fold error hiding behind nearly identical abbreviations. This is why modern prescribing standards push for explicit units on every order, for leading zeros (0.5 mg, never .5 mg), and for banning trailing zeros (5 mg, never 5.0 mg, where the decimal point can be missed).
The lesson for everyone else: in any high-stakes calculation, write the units at every step, not just the numbers. If the units do not work out in the final expression, the answer is wrong no matter how clean the arithmetic looks.
Construction and engineering mix-ups
Large structures have their own catalog of unit mishaps. Perhaps the most cited involved a new office tower where drawings mixed metric and imperial dimensions, a confusion discovered only after expensive rework. Similar stories surface in manufacturing whenever a supplier works in inches and the buyer works in millimeters: parts arrive, parts do not fit, and the finger-pointing begins.
These are rarely single-person errors. They are interface errors, occurring where two teams, two companies, or two countries meet. Each side does its own work correctly in its own system, and the failure lives in the boundary nobody owned. The Mars orbiter story is the purest example: both teams were competent, both were careful, and the spacecraft still burned up because the handoff between them carried an unstated assumption.
Why it keeps happening
If these errors are so well documented, why do they recur? Partly because unit systems are cultural. The United States still uses customary units in daily life while science and most of the world use metric, so every international project is a translation exercise, and translations fail.
Partly it is cognitive. Once a number looks reasonable, the brain stops interrogating it. The orbiter's trajectory numbers looked plausible month after month. The fuel calculation in Montreal produced a number the crew accepted. Sanity-checking requires the deliberate habit of asking whether the magnitude makes sense, and habits are exactly what erode under time pressure.
And partly it is that unit errors are invisible to the usual quality checks. Spell-checkers catch typos; nothing in a spreadsheet flags that a column labeled in pounds is being added to one in kilograms. The error passes every review that does not specifically look for it.
How to protect your own work
You will probably never navigate a spacecraft, but the same defenses scale down to homework, home renovation, and business spreadsheets. They are cheap and they work.
Make the units explicit everywhere: in column headers, in variable names, in comments, in the email where you send the numbers. Convert at the boundary the moment data changes hands, and state the conversion you used. Finally, sanity-check the magnitude: if a room measures 4,000 square feet instead of 400, or a trip needs 40 liters instead of 400, the units are lying to you even when the arithmetic is not.
- State units in every header, label, and variable name, never just in your head
- Convert explicitly at every handoff between people or systems, and write down the factor used
- Sanity-check magnitudes: does the answer make physical sense?
- In shared documents, agree on one unit system up front and enforce it
- For critical work, have a second person independently redo the conversion
Frequently asked questions
What exactly was the Mars Climate Orbiter error?
Thruster data produced in pound-seconds by one team was used as if it were newton-seconds by the navigation team. The uncorrected trajectory errors accumulated until the $125 million spacecraft entered the Martian atmosphere too low and broke apart in 1999.
What is the Gimli Glider?
Air Canada Flight 143 in 1983, a Boeing 767 that ran out of fuel after a pounds-versus-kilograms conversion error during manual fuel calculation. The crew glided it to a safe landing on a decommissioned airstrip in Gimli, Manitoba, with no fatalities.
How common are unit conversion errors?
They are a classic category of engineering and medical error, recurring wherever two unit systems meet. They persist because they are interface errors: each side's work is correct in its own system, and the failure hides in the unstated assumption at the boundary.
What is the simplest way to avoid them?
Write units at every step, convert explicitly at every handoff, agree on one system up front, and sanity-check whether the final magnitude makes physical sense. Most disasters on this list would have been caught by any one of those habits.
Conclusion
Every story in this article ends the same way: the math was right and the meaning was wrong. Units are the quiet half of every measurement, and they demand the same attention as the numbers they accompany. State them, convert them explicitly, and check that the answer makes sense in the physical world, and you will have immunized yourself against one of the oldest failure modes in technical work.