The books of numbers: quantifying historical trends in numeracy (Q316842)

The Google Ngrams corpus, based on over 15 million scanned and digitized books, records the frequency with which every short phrase has occurred in books published each year. The authors use the Ngrams data to study several aspects of numerical usage: 1) representation of numbers as text or digit strings, 2) the over-representation of rounded five-digit numbers, 3) range of numbers used for physical quantities, 4) verification of Benford's law: the probability that a number begins with digit \(d\) is \(\log_{10}(1+\frac1d)\).