01. The Worn Pages — How a Library Book Gave It Away

The Astronomer and the Logarithm Tables

In 1881, the American astronomer and mathematician Simon Newcomb made a peculiar discovery while working with logarithm tables in a library. He observed that the pages at the beginning of the book were significantly more worn and tattered than the pages located toward the end. This physical deterioration revealed a mathematical truth: the ten digits do not occur with equal frequency as the leading digits of numbers. Newcomb noted that the digit 1 appears as the first significant digit far more often than the digit 9 in natural numbers. He published this observation in the American Journal of Mathematics, suggesting that the mantissae of logarithms are equally probable in their distribution. This foundational insight remained largely unremarked and unexplained for 57 years, existing in the literature as an evident but unexamined phenomenon.

The Physicist and the Law of Anomalous Numbers

In 1938, the physicist Frank Benford, working at the GE Research Laboratory in Schenectady NY, provided the empirical proof that solidified this observation. Benford compiled 20,229 first-digit observations across 20 disparate datasets, including river areas, population counts, physical constants, and even street addresses. His measurements demonstrated that the frequency of digits follows a logarithmic law, starting at 30.6% for the digit 1 and declining to 4.7% for the digit 9. These observed frequencies closely matched the predicted logarithmic values, which range from 30.1% for the digit 1 down to 4.6% for the digit 9. This distribution is deeply surprising to everyone who encounters it, as human intuition falsely suggests that each digit should appear with equal frequency. Benford’s comprehensive study transformed an obscure observation into a recognized law, detailing the specific frequencies for all digits from 1 through 9.

The Question of Theoretical Derivation

Benford’s empirical compilation of these 20,229 observations is entirely uncontested and remains a standard in the field of statistical analysis. However, his own original theoretical derivation of the law, which relied on a geometric-series argument, was later superseded by more rigorous work. Specifically, the theoretical foundations he proposed were replaced by the mathematical proofs developed by Pinkham in 1961 and Hill in 1995.

Example: Expecting digits to appear with equal frequency is like assuming every shelf in a library holds an identical number of books regardless of its location.

Verdict. The leading digits of numbers in natural datasets are not evenly distributed, with the digit 1 appearing significantly more frequently than the digit 9.

Overturned if: A comprehensive replication of the 20,229 observations fails to show the logarithmic decline from 30.1% to 4.6% across diverse datasets.

Sources. 1881 · Simon Newcomb, "Note on the Frequency of Use of the Different Digits in Natural Numbers" — PDF · 1938 · Frank Benford, "The Law of Anomalous Numbers" — PDF