Benford's Law: Why the Number 1 Leads More Often Than You'd Think

Open a random book of financial statements, a table of river lengths, or a spreadsheet of physical constants, and count how often each leading digit — the very first number you see, ignoring decimals — shows up. You might expect each digit from 1 to 9 to appear about 11% of the time, since there are nine of them. Instead, in an enormous range of real-world datasets, the digit 1 leads about 30% of the time, 2 leads about 18% of the time, and by the time you reach 9, it's leading barely 4.6% of the time. This isn't a coincidence, a trick, or a quirk of one dataset. It's a mathematical law, and it's strange enough that it has been used to catch tax cheats, expose a national government's cooked books, and win convictions in criminal fraud trials.

It's called Benford's Law, and once you see it, you can't stop noticing it.

The Concept

Benford's Law (more properly the Newcomb-Benford Law) predicts the distribution of leading digits in many kinds of naturally occurring numerical data. Instead of a uniform 1-in-9 chance for each first digit, the expected frequency follows a logarithmic curve:

P(d) = log₁₀(1 + 1/d)

Plug in each digit and you get:

1 → 30.1%
2 → 17.6%
3 → 12.5%
4 → 9.7%
5 → 7.9%
6 → 6.7%
7 → 5.8%
8 → 5.1%
9 → 4.6%

Notice the pattern: the probabilities shrink as the digit grows, and they shrink fast. A number starting with 1 is nearly seven times more likely to appear than one starting with 9. This holds for stock prices, populations of cities, electricity bills, physical constants, the areas of countries, and the line items in a company's general ledger — basically any collection of numbers that spans multiple orders of magnitude and isn't artificially constrained (like phone numbers or zip codes, which are assigned rather than measured, and don't follow the law at all).

Why It Matters

The law was first noticed in 1881 by Simon Newcomb, a Canadian-American astronomer, who observed something odd in the library: the pages of logarithm tables covering numbers starting with 1 were far dirtier and more worn than the pages for numbers starting with 9. People were looking up numbers beginning with small digits much more often. Newcomb wrote a short paper about it in the American Journal of Mathematics and the observation was essentially forgotten for nearly 60 years.

In 1938, physicist Frank Benford independently noticed the exact same thing on his own well-thumbed log tables. Rather than stopping at an anecdote, he tested it rigorously, compiling 20,229 observations across 20 wildly different datasets — river drainage areas, atomic weights, baseball statistics, street addresses, numbers pulled from newspapers, death rates, and more. Nearly all of them matched the logarithmic pattern. His paper, "The Law of Anomalous Numbers," gave the phenomenon the name it carries today, even though Newcomb got there first. (A rigorous mathematical proof of why the law holds so broadly wasn't published until 1995, by mathematician Theodore Hill.)

The practical payoff came decades later, largely through the work of accounting professor Mark Nigrini in the 1990s. Nigrini showed that real financial data — expense reports, tax filings, invoice amounts — reliably follows Benford's Law, while fabricated numbers usually don't. People who invent numbers, whether padding an expense report or forging a balance sheet, unconsciously favor certain digits (many of us like round numbers, or "safe-looking" middle digits) and avoid the natural skew toward 1 and 2. That mismatch is detectable with basic statistics, and it's now a standard tool in forensic accounting: Benford's Law tests have been used to flag irregularities in tax returns, vendor payment records, payroll data, and audits, and the technique is legally admissible as evidence in U.S. fraud cases at the federal, state, and local level.

The Details

Why would something so simple — the first digit of a number — obey a logarithmic law at all? The intuitive answer is scale invariance. Think about what happens if you take a dataset that follows Benford's Law and convert its units — say, a table of country populations converted from a raw headcount to population expressed in "thousands," or river lengths converted from kilometers to miles. Multiplying every number by a constant shouldn't change whether the data "looks natural" or not — the law should hold regardless of units. It turns out that the only distribution of leading digits that survives this kind of rescaling unchanged is the logarithmic one described by Benford's formula. Uniformly distributed leading digits (the naive 11%-for-everyone guess) would NOT survive a unit change — multiply a "uniform" dataset by 2, and its digit distribution shifts. Benford's distribution is the unique fixed point.

A cleaner way to see it: take the base-10 logarithm of every number in a dataset. A number "starts with digit d" exactly when the fractional part of its logarithm falls in a specific sub-interval of [0, 1]. For data spanning several orders of magnitude — city populations ranging from hundreds to millions, say — those fractional parts end up spread roughly uniformly across [0, 1]. Since the interval corresponding to leading digit 1 (from log₁₀(1) to log₁₀(2)) is wider than the interval for leading digit 9 (from log₁₀(9) to log₁₀(10)), a uniform spread over the log scale translates into exactly the lopsided digit frequencies Benford observed.

This is also why the law fails for some datasets: numbers confined to a narrow range (adult human heights in centimeters, say — almost all start with 1) don't span enough orders of magnitude for the log-uniform effect to kick in. Assigned numbers like phone numbers, ID numbers, or lottery picks aren't "natural" measurements at all, so there's no reason to expect the pattern.

Real-world sleuthing has put the law to dramatic use. Greece's government-reported macroeconomic data submitted to the European Union before it joined the eurozone was later shown, using Benford's Law, to deviate suspiciously from the expected distribution — a red flag, in hindsight, for the fraudulent bookkeeping that helped trigger the Greek debt crisis. Forensic accountants have applied it to Ponzi schemes and corporate collapses, checking whether transaction records look "natural." But the law isn't magic, and it has real limits: applying it to election vote counts has proven far more controversial. While it was cited as one piece of evidence questioning irregularities in Iran's disputed 2009 election, researchers have since shown that Benford's Law is a poor and unreliable tool for detecting ballot fraud specifically — simulations of both fair and rigged elections can each conform to, or deviate from, the expected digit distribution with no consistent pattern, and its use to allege fraud in the 2020 U.S. presidential election was a well-documented misapplication of the statistic. The lesson: Benford's Law is a genuine and useful anomaly detector for naturally scaled, multi-order-of-magnitude data like accounting ledgers — but it's not a universal fraud detector, and applying it outside its proper domain can generate false confidence in either direction.

Takeaways

Benford's Law predicts that in many real-world numerical datasets, the leading digit 1 appears about 30% of the time, dropping logarithmically down to about 4.6% for the digit 9 — not the naive uniform 11% you'd expect.
It was first spotted by astronomer Simon Newcomb in 1881 from worn logarithm-table pages, forgotten for almost 60 years, then independently rediscovered and rigorously tested by physicist Frank Benford in 1938 across more than 20,000 data points.
The underlying reason is scale invariance: for data spanning multiple orders of magnitude, the logarithm's fractional part becomes evenly spread, and the numerically "wider" logarithmic interval belonging to digit 1 makes it the most common leading digit.
It's a real forensic accounting tool — used to flag fabricated numbers in tax returns, expense reports, and financial statements, and admissible as court evidence — but it doesn't apply to data that's artificially constrained (phone numbers, zip codes) or confined to a narrow range.
Its use outside finance, especially in election-fraud analysis, is contested and has been shown to produce false positives and false negatives alike — a reminder that even elegant mathematical patterns have a proper domain of application.

Resources: - Frank Benford, "The Law of Anomalous Numbers", Proceedings of the American Philosophical Society (1938) - Mark Nigrini, Benford's Law: Applications for Forensic Accounting, Auditing, and Fraud Detection (Wiley, 2012)