Sandbox · exhibit 0045

Three in ten of the powers of two start with a 1

1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024 — take ten thousand of them and count only the leading digit. A 1 comes up 3,010 times and a 9 comes up 458, which is not one ninth of anything. It is not a quirk of two either: the first digit of bⁿ is decided by the fractional part of n·log₁₀b, and when that number is irrational the fractional parts spread out evenly and the digits land in proportion to log₁₀(1 + 1/d). When it is not irrational, they do not — and the powers of ten start with a 1 every single time.

Count them

Nine digits, and the first one is exactly as likely as the last five put together — the bands for 5 to 9 telescope to log₁₀(10/5), which is log₁₀2, which is the band for 1. The red line is one ninth, which is what you would guess if you had not counted.

counted log₁₀(1 + 1/d) one ninth
how many terms ·

Where it comes from

A number starts with the digit d exactly when the fractional part of its logarithm lands between log₁₀d and log₁₀(d+1). So the question is only ever: where do those fractional parts land? For 2ⁿ they are the multiples of 0.30103…, wrapped round — and multiples of an irrational number, wrapped round, fill the interval evenly.

the band for each digit where the first 300 terms land

The two that do not obey, and why not

If the argument is right, it should break in a way you can predict. Powers of ten have log₁₀10 = 1: the fractional part is zero every time, so it never leaves the first band. Powers of √10 have a logarithm of exactly one half, so the fractional parts alternate between two points — and the numbers use two digits, 1 and 3, and no others ever.

·

What this page checks about itself

The powers are computed as exact integers in BigInt, so the leading digit of 210000 — a number with 3,011 digits — is read off the number itself and not off a floating-point logarithm.