The clue was grime. In 1881, the astronomer Simon Newcomb noticed something odd about the books of logarithm tables that every scientist of his era relied on for calculation: the early pages — the ones used for numbers beginning with 1 — were far more worn and dirty than the later ones. People, it seemed, were looking up numbers that started with 1 much more often than numbers that started with 8 or 9. Newcomb wrote a short paper proposing a formula for how often each leading digit should appear, and the world almost completely ignored it.
Fifty-seven years later, a physicist at General Electric named Frank Benford rediscovered the same pattern — reportedly from the same clue of worn logarithm pages — and did what Newcomb had not: he tested it exhaustively. Benford compiled more than 20,000 numbers from wildly unrelated sources: the areas of rivers, the populations of towns, physical constants, death rates, street addresses, numbers pulled from magazine articles, even baseball statistics. Again and again, the same lopsided distribution appeared. The 1938 paper made the pattern famous, and by one of science's small injustices, it carries Benford's name rather than Newcomb's.
The lopsided law
Here is the pattern. Intuition says that in a big messy pile of real-world numbers, each leading digit from 1 to 9 should appear about one time in nine — roughly 11 percent each. Instead, the leading digit 1 shows up about 30.1 percent of the time. The digit 2 leads about 17.6 percent of numbers. The frequencies keep falling from there, until poor 9 leads a mere 4.6 percent. The precise rule: the probability that a number's first digit is d equals log10(1 + 1/d).
Why on earth should this be true? The cleanest intuition comes from growth. Imagine a quantity that grows by a steady percentage — an investment, a population, a bacterial colony. Going from 1,000 to 2,000 requires it to double: a 100 percent increase. Going from 8,000 to 9,000 requires only 12.5 percent growth. So anything growing multiplicatively spends far longer with a 1 out front than with an 8 or a 9. Numbers that spread across several orders of magnitude, built from multiplied causes, drift naturally into Benford's pattern.
The law has an elegant signature: scale invariance. Convert a Benford-distributed dataset of river lengths from miles to kilometres, or prices from dollars to yen, and the distribution of leading digits stays the same. In fact, it is the only leading-digit distribution with that property — which is a strong hint that any universal law of first digits had to be this one.
The fraudster's blind spot
Benford's law would have remained a mathematical curiosity but for one very human weakness: people who invent numbers do it badly. When someone fabricates expenses, cooks sales figures or launders transactions, they tend to sprinkle digits around evenly, or favour middling digits that "look random". Genuine financial data, built from real quantities multiplying and compounding, follows Benford. Faked data usually does not.
The accountant Mark Nigrini pioneered turning this into a forensic tool in the 1990s, showing that digit analysis could flag suspicious tax returns and corporate ledgers for closer inspection. The technique is now a standard instrument in forensic accounting, and Benford-based analysis has featured in fraud investigations and court proceedings. It has been aimed at governments, too: a 2011 academic study found that Greece's reported macroeconomic statistics in the years before its debt crisis deviated from Benford's law more than those of any other eurozone member — an after-the-fact echo of the data problems that later came to light.
One important caveat: a Benford deviation is a smoke detector, not a conviction. Plenty of honest datasets fail the law for innocent reasons. It does not apply to numbers that are assigned rather than measured (phone numbers, lottery draws), to data trapped in a narrow range (adult heights, where nearly everything starts with 1 in centimetres anyway), or to figures shaped by human rules like $9.99 pricing. Analysts have also tried to use it to detect election fraud — it was invoked in disputes over Iran's 2009 presidential vote — but statisticians remain divided on whether vote counts should follow the law at all, which makes it a shaky tool there.
Still, within its proper domain, Benford's law is a genuinely eerie piece of mathematics: a hidden regularity threading through river basins, stock prices, city populations and electricity bills. The universe, it turns out, has a favourite number to start with — and it is 1.
Quiz nuggets
- In naturally occurring data, about 30.1 percent of numbers begin with the digit 1, but only 4.6 percent begin with 9.
- The law was first noticed in 1881 by astronomer Simon Newcomb, from the worn early pages of logarithm books.
- It is named after Frank Benford, a General Electric physicist who confirmed it across 20,000+ numbers in 1938.
- Forensic accountants use Benford's law to flag fabricated figures, because invented numbers rarely match the pattern.
- The exact rule: the chance a number starts with digit d is log10(1 + 1/d).