How ConvertPlus calculates every answer
The formulas, the definition-based factors, the parsing rules and the rounding policy — written down, so any result on this site can be reproduced by hand.
A converter is only useful if you can check it. This page documents what happens between the number you type and the number you read: which formula runs, which factor it uses, where that factor comes from, how many digits are kept, and when a value is rejected instead of converted. Nothing here depends on a server, a lookup table of pre-computed answers, or a third-party conversion API. The arithmetic runs in your browser against the constants listed below.
Every figure quoted as a definition is quoted because it was fixed by an international agreement, not because it was measured. The distinction matters: 1 inch is exactly 2.54 centimetres, not approximately, so a conversion built on it introduces no error of its own.
Temperature
Celsius and Fahrenheit are related by a linear function, so one multiplication and one addition are the whole calculation:
- F = C × 9 / 5 + 32 (Celsius to Fahrenheit)
- C = (F − 32) × 5 / 9 (Fahrenheit to Celsius)
- K = C + 273.15 (Celsius to kelvin, and the reverse subtracts it)
The two scales meet at one point, −40, where −40 °C and −40 °F are the same temperature. That crossing is the quickest sanity check available: convert −40 and you should get −40 back. The second check is the pair of water anchors, 0 °C = 32 °F and 100 °C = 212 °F, which any correct converter reproduces exactly. Fahrenheit was fixed with 32 as the freezing point and 212 as the boiling point of water at standard pressure, which is why the interval between them is 180 degrees against Celsius 100 — the 9/5 in the formula is that ratio, not a measured constant.
Kelvin is handled as an offset only, with no change of scale, because one kelvin is the same size as one degree Celsius. Absolute zero sits at 0 K = −273.15 °C, and the converter shows negative Kelvin values as a warning rather than pretending they exist, since temperatures below absolute zero are physically meaningless and usually indicate that Celsius and Fahrenheit were swapped somewhere upstream.
Roman numerals
Encoding uses the standard subtractive set, applied greedily from the largest symbol down: M 1000, CM 900, D 500, CD 400, C 100, XC 90, L 50, XL 40, X 10, IX 9, V 5, IV 4, I 1. Each pair is taken as many times as it fits, so 1994 encodes as MCMXCIV and 2026 as MMXXVI. Numbers outside the range 1 to 3999 are refused instead of being encoded with extended notations such as overlines or the apostrophus, because those forms are not part of the classical system and are not read consistently.
Decoding is stricter than most online tools. A Roman numeral is accepted only if the parse survives a canonical round trip: the value is computed left to right, subtracting a symbol that is followed by a larger one, and the result is then re-encoded. If the re-encoding does not reproduce the original string character for character, the input is treated as invalid rather than guessed at. That rejects constructs such as IIII for four, VX for five, or IC for ninety-nine — strings that a lax parser happily turns into a number but that no Roman writer would have used. Whitespace and lower case are tolerated; anything outside the seven letters is not.
The date converter works on the same rules, splitting a year, month and day into three separate Roman values. For the calendar year it uses the positive year number as given: 2026 becomes MMXXVI. It does not invent an offset for years before the Common Era, because there is no single convention for how those were written.
Length, weight and volume factors
These conversions are single multiplications by fixed factors. The imperial and US customary values below were locked to the metric system by the international yard and pound agreement of 1959, which is why they are exact rather than approximate:
- 1 inch = 2.54 cm exactly; 1 foot = 12 inches = 30.48 cm
- 1 yard = 0.9144 m; 1 mile = 1.609344 km
- 1 pound = 0.45359237 kg (so kg to lb uses 2.20462)
- 1 US fluid ounce = 29.5735 mL (the figure the volume pages use)
Where a page states a factor, that is the factor the page actually multiplies by — the kg to lbs pages use 2.20462, while the reverse pages divide by the pound definition of 0.45359237 kg, and the two agree to the digits shown. Feet and inches stay separate units on purpose: a height of 5 ft 9 in is 69 total inches before it becomes 175.26 cm, and collapsing the feet and inches into one decimal number first is the mistake that produces 5.9 ft for 5 ft 9 in.
Rounding policy
Rounding happens once, at the end, and only for display. The calculation itself carries full floating-point precision; no intermediate step is shortened. The displayed value is produced with a fixed maximum number of fraction digits rounded the usual way, then trailing zeros are dropped, so 104.0 shows as 104 and 176.666 as 176.7. Thousands separators follow English conventions, which is why a long result appears as 1,609.34 rather than 1609.34.
Default precision is one decimal place, because one decimal place is enough to describe most everyday answers and short enough to compare at a glance. Weight pages show two, because a tenth of a pound is about 45 grams and rounding there loses information people actually use. If you need more digits than the default, avoid chaining rounded answers: convert once with the page, or carry the unrounded value into your own arithmetic and round only the final result. Rounding twice is the most common cause of a result that is off in the last digit.
How the numbers are checked
Three checks run whenever a converter or a value page changes. First, anchor values with known answers: 0 °C to 32 °F, 100 °C to 212 °F, −40 in both scales, 2026 to MMXXVI, 5 ft to 152.4 cm, 100 kg to 220.46 lb. Second, round trips: converting a value out and back must return the starting number within the precision shown. Third, agreement with published definitions — the SI brochure for metric prefixes and the NIST guide to the SI for the customary unit definitions. The factors on this site are taken from those definitions, not from a secondary summary.
What the tools deliberately do not do
- Millilitres to ounces is a volume-to-volume conversion. Converting a volume to a weight needs a density, so there is no ml to grams result unless a specific ingredient density is stated.
- There are no currency rates. Exchange rates change by the minute and would be wrong the moment a static page cached one, so this site does not publish them.
- Roman numerals stop at 3999, and invalid strings return an error instead of a best guess, as described above.
- No result is stored. Your input is never uploaded, logged or reused, because the calculation never leaves the page.
Review date
This methodology was last reviewed on 2026-11-16, together with the formulas, factors and rounding rules it describes. When a converter changes in a way that affects results, the change is recorded on the changelog.