If you tried to fake a giant spreadsheet by spreading the leading digits evenly, a forensic auditor could prove you were lying. In large naturally occurring data sets, the number one leads about 30 percent of the time while nine leads less than 5 percent. This episode traces the discovery from astronomer Simon Newcomb noticing grimy front pages in logarithm tables in 1881, to physicist Frank Benford testing river areas, town populations, molecular weights and Reader’s Digest numbers in 1938, and explains the logarithmic scale that makes the pattern inevitable.

We walk through why multiplicative growth, like a bank account compounding at 10 percent, spends most of its time in the ones bracket, why the law survives converting feet to meters, and Ted Hill’s 1995 proof that mixing unrelated data sets produces the curve. Then we cover the boundaries: human heights, sequential invoice numbers, and psychological 9.99 pricing all break the law. Finally we look at how it is used to flag fraud in Greek economic data, scientific papers and elections, including the 2009 Iranian vote and why viral 2020 claims about Chicago and Milwaukee precincts misapplied the test.

• The binary edge case where every number starts with one
• How the 2002 euro changeover briefly pulled European prices toward Benford’s curve
• Kuiper, Leemis and Cho-Gaines tests used as evidence in court
• Walter Mebane’s second-digit test and its limits in the 2000 Florida election
• Why precinct-level vote counts cannot span enough orders of magnitude to be tested