How Unit Conversion Actually Works

The core method is: choose one base unit per category and express every other unit as a factor of it. For length, UnitBook uses the metre as the base, so a centimetre is 0.01, a kilometre is 1000, and an inch is 0.0254. To convert 5 kilometres to miles, you first go to the base — 5 × 1000 = 5000 metres — then out of the base into the target — 5000 ÷ 1609.344 ≈ 3.107 miles. Every within-category conversion is just "multiply into the base, divide out to the target".

This is why a converter only needs to store one number per unit, not a factor for every possible pair. With a dozen units in a category, storing every pair would mean over a hundred factors; storing one factor each means just a dozen — and any pair is derived on the fly.

The multiply-and-divide method assumes the two units share a zero point — that zero of one is zero of the other. That holds for length, mass, volume and most categories. It fails for temperature, where the scales start at different places. Temperature is an affine conversion: you scale and then add an offset. That is why UnitBook handles temperature with dedicated formulas instead of a single factor, and why 0 °C is 32 °F rather than 0 °F.

The second subtlety is precision. Computers store numbers with finite accuracy, so a converted value can carry a long tail of digits that are not meaningful. A good converter rounds to a sensible number of significant figures and switches to scientific notation for very large or very small results, so 0.0000001 shows cleanly rather than as a wall of zeros. UnitBook rounds to ten significant figures — precise enough for engineering, tidy enough to read.

There is a mechanical way to get conversions right that never depends on remembering which direction to multiply. Write each conversion factor as a fraction equal to one — 1609.344 m over 1 mile, 1 hour over 3600 s — then multiply, and cancel the unit names as if they were algebraic symbols. If the units that survive are the ones you wanted, the arithmetic is right; if a unit is upside down, you will see it before you reach for a calculator.

Sixty miles per hour becomes 60 × 1609.344 m / 3600 s = 26.82 m/s, and "miles" and "hours" vanish on the way. Compound units always need every component converted, which is what makes them error-prone: a fuel figure in litres per 100 km and one in miles per gallon differ in both parts and are also inverted with respect to each other.

The same habit explains derived units. A newton is a kilogram-metre per second squared, a joule is a newton-metre, and a watt is a joule per second — so if a calculation is meant to produce watts and the units cancel down to joules, something upstream is wrong. Checking dimensions is the cheapest error detector in any calculation.

The Mars Climate Orbiter was lost in September 1999 because one ground-software team produced thruster impulse in pound-force-seconds while the navigation software expected newton-seconds. Nobody made an arithmetic mistake; the numbers were simply carried between systems without a conversion, the trajectory drifted about 170 km low, and a 125-million-dollar spacecraft entered the atmosphere instead of orbit.

Sixteen years earlier, Air Canada Flight 143 was fuelled using a pounds-per-litre density figure on an aircraft whose gauges worked in kilograms. It took off with roughly half the fuel it needed, lost both engines at 41 000 feet and was glided onto a disused runway at Gimli, Manitoba. Both accidents share a shape: a correct number in the wrong unit, passed across a boundary where nobody was checking.

The practical lesson for everyday work is smaller but the same: keep the original value, convert once, and convert from the source rather than from an already-rounded intermediate. Chaining three conversions that each rounded to two decimals can shift a result by more than the tolerance you were trying to hold.