Some of the most expensive engineering failures in history trace back to surprisingly small measurement mistakes: a misplaced decimal point, a wrong unit conversion, or a testing device off by a fraction of a millimeter. These errors have destroyed spacecraft, grounded submarines, and nearly crashed a commercial airliner. Here are the most notable cases and what went wrong in each one.
Mars Climate Orbiter: Metric vs. Imperial
In 1999, NASA lost a $327.6 million spacecraft because two software systems spoke different measurement languages. The ground software built by Lockheed Martin calculated thruster impulses in pound-force seconds, an imperial unit. But the navigation software at NASA’s Jet Propulsion Laboratory expected those values in newton-seconds, the metric equivalent. Neither team caught the mismatch during the months the orbiter traveled toward Mars.
The result was a trajectory that brought the spacecraft far too close to the Martian atmosphere. Instead of settling into orbit, it burned up. The entire mission was a total loss. What made this failure so striking was its simplicity. The math itself was correct on both sides. The problem was purely that no one verified the two teams were using the same units.
The Gimli Glider: Pounds Instead of Kilograms
In 1983, an Air Canada Boeing 767 ran out of fuel at 41,000 feet over Manitoba. The aircraft was one of the first in the fleet to use metric measurements, and during fueling, the ground crew used the wrong conversion factor. They calculated fuel density at 1.77 pounds per liter (the figure for older aircraft) instead of 0.8 kilograms per liter, which the new 767 required. The plane was loaded with roughly half the fuel it needed to reach Edmonton.
Both engines flamed out mid-flight. The pilots managed to glide the powerless jet to an emergency landing at a decommissioned airfield in Gimli, Manitoba, where a community event happened to be underway on the old runway. Remarkably, no one was killed. The incident became one of aviation’s most famous emergency landings, and the aircraft flew commercially for another 25 years after repairs.
Hubble Space Telescope: 1.3 Millimeters Off
When NASA launched the Hubble Space Telescope in 1990, the first images it sent back were blurry. The primary mirror, which had been ground to an incredibly precise shape, was too flat near its edges. The cause was a flaw in the device used to test the mirror during manufacturing, called a null corrector. A spacing error of just 1.3 millimeters inside this testing instrument meant the mirror was polished to the wrong specifications. Investigators later found the null corrector undisturbed and in the same configuration as when fabricators had tested the mirror nine years earlier.
The mirror itself was flawless in the sense that it perfectly matched the (incorrect) measurements it had been tested against. Corrective optics were installed by astronauts during a 1993 servicing mission, essentially giving Hubble a set of corrective lenses. The fix worked, but the repair mission and years of degraded science represented an enormous cost, all traceable to a millimeter-scale error in a single calibration tool.
The Laufenburg Bridge: Two Sea Levels
The town of Laufenburg sits on the Rhine River, split between Germany and Switzerland. When engineers from both countries built a bridge to connect the two halves, they discovered during construction that the two sides didn’t line up. There was a height discrepancy of 54 centimeters between the outer ends.
The root cause was a reference point mismatch. Switzerland measures elevation from the mean sea level of the Mediterranean. Germany uses a standard called Normal Null, based on the North Sea. The difference between these two baselines is 27 centimeters, and the engineering teams knew this. The plans accounted for it correctly on paper. But during construction, the Swiss side subtracted 27 centimeters from their measurements instead of adding them, doubling the error to 54 centimeters. The German side had to be lowered to compensate.
Spain’s S-80 Submarine: A Misplaced Decimal
Spain’s Navantia shipbuilder was constructing four new S-80 class submarines for the Spanish navy at a cost of 2.2 billion euros when engineers realized the vessel was roughly 100 tons heavier than designed. According to a former director of the Strategic Assessment Office of Spain’s Ministry of Defense, someone had placed a decimal point in the wrong place during the design calculations, and nobody reviewed the figures closely enough to catch it.
A submarine that weighs too much cannot resurface reliably, which is a fairly critical flaw. The original design called for a 71-meter vessel displacing 2,200 tons. The redesign stretched the submarine to 81 meters and pushed displacement to 3,000 tons, a massive change required to restore buoyancy. The program was delayed by years and the costs ballooned well beyond the original budget.
Ariane 5 Rocket: A Number Too Large
In 1996, the European Space Agency’s Ariane 5 rocket exploded 37 seconds after launch on its maiden flight. The failure was caused by a software conversion error. The rocket’s inertial reference system tried to convert a 64-bit number (representing horizontal velocity) into a 16-bit integer variable. Ariane 5 was faster than its predecessor, and the velocity value exceeded the maximum that a 16-bit integer can hold: 32,767. The resulting overflow crashed the navigation computer, which sent erratic steering commands to the rocket’s engines.
The software had been reused from the Ariane 4 rocket, where the velocity values never came close to exceeding the 16-bit limit. Engineers had even deliberately left this particular conversion unprotected to keep the computer’s workload below 80 percent. The assumption that the old measurement range was sufficient for the new, faster rocket destroyed the vehicle and its payload, worth an estimated $370 million.
The Vasa Warship: Asymmetric Construction
In 1628, the Swedish warship Vasa capsized and sank in Stockholm harbor on its maiden voyage, barely 1,300 meters from the dock. Modern analysis of the recovered hull has confirmed that the ship was asymmetric: its port and starboard sides were built to slightly different dimensions, with the weight distribution skewed toward the port side. Archaeologists believe that different teams of builders worked on each side of the hull using different measurement standards. Swedish feet and Amsterdam feet were both in use at the time, and the slight difference between them compounded across the length of the ship.
The asymmetry alone didn’t sink the Vasa. The ship was also top-heavy, carrying too many cannons on its upper decks for its narrow hull. But the uneven construction made an already unstable vessel even less capable of handling the light wind that toppled it minutes after setting sail. The ship sat on the seabed for 333 years before being raised in 1961 and is now preserved in a museum in Stockholm.
Why These Errors Keep Happening
A pattern runs through nearly all of these disasters. The underlying math was usually correct. The engineers were competent. What failed was the handoff: between teams using different units, between design documents and construction crews, or between software written for one system and reused in another. Measurement errors thrive at the boundaries where one group’s assumptions meet another group’s practices.
International standards now exist specifically to prevent these failures. ISO 10012 provides a framework for organizations to design, maintain, and continually improve their measurement processes. It requires that measurement equipment and procedures be systematically controlled and verified, with the goal of ensuring that every measurement used in design, production, and testing is fit for purpose. The standard applies across industries, from aerospace to shipbuilding to construction, and is meant to manage exactly the kind of risk that destroyed the Mars Climate Orbiter or stretched Spain’s submarines by ten meters. Whether organizations follow it rigorously enough is, of course, another question entirely.