The Number That Broke Benford

Pick a random real-world number from a list — populations of cities, sizes of rivers, corporate expense reports. What's the chance the first digit is 1? Common sense says 1 in 9, about 11%. Reality says 30.1%. This is Benford's Law, and it's one of the strangest, most useful facts in mathematics.

1. The pattern

Across thousands of datasets — tax returns, stock prices, river lengths, population counts, electricity bills, the distances of stars — the leading digit follows a specific skewed distribution:

• 1 appears as the first digit about 30.1% of the time
• 2 about 17.6%
• 3 about 12.5%
• Each successive digit appears less often
• 9 appears as the first digit only 4.6% of the time

This holds even though the numbers come from completely unrelated sources. The pattern is a property of magnitudes, not the data itself.

2. Why it happens

Real-world quantities span many orders of magnitude. A river might be 10 km or 1000 km. A company might have 50 employees or 50,000. To go from "starts with 1" (say, 100) to "starts with 2" (200), a value has to double — a 100% increase. To go from "starts with 8" (800) to "starts with 9" (900), it only has to grow by 12.5%.

In logarithmic terms, numbers spend a lot of "space" with leading digit 1 (100 → 199) before jumping to 2. Each subsequent leading digit occupies a smaller logarithmic interval. That's the entire mechanism.

3. When it works

Benford's Law applies when:

• The data spans multiple orders of magnitude (10s, 100s, 1000s — not all the same scale).
• The data is naturally occurring — measured from the world, not constrained.
• There is no imposed minimum or maximum (not "people aged 18-25").

It applies to: tax returns, expense accounts, street addresses, populations, lengths of rivers, daily stock prices, electricity bills, the half-lives of radioactive elements, the sizes of files on your computer.

4. When it doesn't

It fails on:

• Assigned numbers — phone numbers, ID codes, ZIP codes (these are labels, not quantities).
• Normally distributed data — heights of adult humans, IQ scores (these cluster too tightly around a single value).
• Constrained data — "people between 5' and 6' tall."

If the data doesn't span multiple orders of magnitude, Benford doesn't apply.

5. The fraud-detector

Here's the punchline. Humans, when making up numbers, distribute leading digits roughly evenly — they pick 1, 2, 3, 4 … 9 with about equal frequency, because that feels "random."

Real-world financial data doesn't. It follows Benford's Law.

So tax authorities and forensic accountants run Benford analysis on large datasets. If an expense report's leading digits don't match the predicted distribution — too many 7s, 8s, and 9s — that's a flag. The numbers may be fabricated.

Several major fraud cases, including embezzlement at the Arizona state treasury (1993), the Diana Ross tax case, and various Enron-related investigations, used Benford's Law to detect anomalous data. The IRS uses it. So does the SEC.

The strange truth

Most numbers in the world begin with 1. The number you make up to fake a financial report probably begins with 7. A tiny piece of math — the spacing of digits on a logarithmic scale — quietly catches thousands of frauds every year. The universe has a preference, and criminals don't notice.