Big Number Calculator

What this big number calculator does

Most calculators stop being exact at 9,007,199,254,740,992. That number is 2 to the 53rd power, and it is the largest whole number a standard double precision float can hold without gaps. Add one to it on a phone calculator or in a spreadsheet and the result comes back unchanged, because the next value the format can represent is two higher. A big number calculator sidesteps the limit by storing the digits themselves and doing the arithmetic on those digits.

This page works with whole numbers of any reasonable length. Enter two values, choose add, subtract, multiply, divide or a power, and the answer appears with comma grouping, a digit count, the same value spelled out in English words, and scientific notation for quick comparisons. Division also reports the quotient and the remainder separately, which is what a count needs when it has to split into whole boxes, seats or payments.

Everything runs in the browser. No number you enter is sent anywhere, and the page keeps working once it has loaded, even without a connection.

Why phone calculators lose digits above 9 quadrillion

A double precision float carries 53 bits of mantissa, which is 15 to 17 significant decimal digits. That is plenty for money and measurements, and not enough for a 30 digit product. The failure is quiet: the display still shows a plausible value, but the trailing digits are guesses. Enter 9007199254740993 into a spreadsheet cell and it is stored as 9007199254740992, one less, with no warning.

The difference looks trivial until the number is a balance held in cents, a serial number, or a count that has to match a second system exactly. Rounding in the last digits of an identifier breaks the match. A ledger that stores 9007199254740993 cents and reads it back as 9007199254740992 cents is off by one cent in a way no one will notice until a reconciliation fails.

Integer types fix the problem only up to their own width. A signed 64 bit integer stops at 9,223,372,036,854,775,807, which is 19 digits, so a 20 digit product overflows silently in many languages. Libraries that promise arbitrary precision remove the ceiling by growing the number as needed, and that is the approach used here.

Reading a long number in groups of three

English number names work in blocks of three digits, counted from the right. Take 1,267,650,600,228,229,401,496,703,205,376. Split it every three digits and each block takes a name: 1 nonillion, 267 octillion, 650 septillion, 600 sextillion, 228 quintillion, 229 quadrillion, 401 trillion, 496 billion, 703 million, 205 thousand, 376. Read left to right and the spoken name follows the digits.

Two habits keep this straight. Count blocks, not digits, starting at the right edge, and remember that a scale name belongs to a block, not to a digit position. A 31 digit number starts in the nonillion block because 10^30 is the largest power of ten at or below it.

The same method works for shorter values. A 16 digit number such as 4,500,000,000,000,000 has six blocks, so the leading block is quadrillion: four quadrillion five hundred trillion. A 7 digit number has three blocks and starts at million. Once you can count blocks, any length becomes readable in a few seconds, and the words row in the calculator is a check on your own reading.

Leading zeros never change a value, and a block of three zeros still occupies a place. That is why 1,000,001 reads one million one, with an empty thousand block in the middle that is skipped when spoken.

Scale names from thousand to decillion

The names below are the short scale, the one used in the United States: each new name is a thousand times the one before it. The long scale used in much of Europe steps by a million instead, so a German or French reader may take billion to mean 10^12. In digits the two scales agree; in words they do not, which is worth remembering when a contract crosses a language.

The same naming shows up in measurement. The NIST table of metric prefixes pairs 10^27 with ronna and 10^30 with quetta, and lists octillion and nonillion as their short scale names. The SI Brochure sets those prefixes internationally.

Worked example: 2 to the 100th power

Doubling a dollar a hundred times is the classic compound growth puzzle. The exact answer is 2^100 = 1,267,650,600,228,229,401,496,703,205,376. That is a 31 digit number, and the digit count can be predicted: the base ten logarithm of 2 is 0.301, and 0.301 x 100 = 30.1, so the value sits just past 10^30 and needs 31 digits.

In words it reads one nonillion two hundred sixty-seven octillion six hundred fifty septillion six hundred sextillion two hundred twenty-eight quintillion two hundred twenty-nine quadrillion four hundred one trillion four hundred ninety-six billion seven hundred three million two hundred five thousand three hundred seventy-six. Rounded for comparison it is about 1.27 nonillion, or 1.2677 x 10^30.

A phone calculator shows the same quantity as 1.2676506002282294e+30. That keeps 17 significant digits and drops the other 14, so it answers the question of scale and nothing else. For counting money or matching an identifier, the full string is the answer.

The doubling example also explains why compound growth surprises people. A penny doubled every day for 30 days is 2^30 cents, which is 10,737,418.24 dollars, and the same penny doubled for 100 days passes a nonillion cents. The digits grow by roughly 0.301 per doubling, so 100 doublings land just past 10^30.

Division that keeps the remainder

Dividing 10^20 by 7 gives 14,285,714,285,714,285,714 with a remainder of 2. The check is 7 x 14,285,714,285,714,285,714 + 2 = 100,000,000,000,000,000,000, and it comes out exact. Ordinary calculators hide the remainder behind decimal places, which is fine for measurements and wrong for counts: you cannot ship seven tenths of a crate.

The same two numbers under every operation look like this.

The cube row is worth a pause: a 13 digit number cubed is a 37 digit number, past the decillion mark and far beyond what a phone calculator shows without switching to scientific notation.

Long division by hand and integer division in code follow the same rule. Take the digits from the left, carry what is left over into the next digit, and record one quotient digit at a time. A calculator that stores digit strings does exactly that, which is why the quotient and the remainder are both exact and why the remainder is always smaller than the divisor.

Big numbers you meet in a normal week

Scale matters more than digit count. One million seconds is 11.6 days, one billion seconds is 31.7 years, and one trillion seconds is 31,688 years. Each step of a thousand in the name multiplies the time by a thousand too, which is why a billion dollars and a trillion dollars are different worlds rather than two sizes of the same thing.

Live 14 digit numbers are easy to find. The Treasury Debt to the Penny dataset publishes the total public debt every business day, a figure in the tens of trillions that moves by billions within a week. Population counts, data center storage and national budgets all live in the 9 to 15 digit range, comfortably inside what a double can hold, but long enough that comma grouping and word names start to earn their keep.

Storage is the other place where digits pile up. A 1 TB drive sold with decimal units holds 1,000,000,000,000 bytes, but the operating system divides by 1,024 to display the figure, so it reports 931 GB. Both numbers are correct, and the gap between them is why a 1 TB drive never seems to hold a full terabyte in a file manager.

Astronomy uses bigger values still. The distance from Earth to the Sun is about 149,600,000,000 metres, a 12 digit figure, and a light year is about 9,460,730,472,580,800 metres, 16 digits. Neither needs more precision than a double can give, but both read far more easily with commas and a scale name.

Entering numbers and reading the output

Type digits with or without commas; spaces, underscores and commas are stripped before the arithmetic. A leading minus sign works, and a decimal point is dropped with a note, because the page computes in whole numbers only. Leave the second box empty for an error rather than a wrong answer.

The output rows are fixed. The result row shows every digit with commas. The digit count tells you which scale name leads. The words row spells the value out, which is how the number would be read aloud or written on a check. Scientific notation rounds to six significant digits for comparisons. For division there is a quotient row, a remainder row, and a check row that multiplies the quotient by the divisor and adds the remainder back, so you can see the identity close.

Powers are capped at an exponent of 10,000, and any result that would run past 12,000 digits is refused rather than truncated, so a typing slip cannot freeze the tab. For numbers with decimals, use a scientific notation calculator or a standard calculator instead, then come back here when the value is a whole count.