it was frank benford in 1938 - it's called benford's law in his honour


it applies to a fairly arbitrary piece of real data - the lengths of various rivers; the individual payments received in a month by a mortgage lender; the population of a collection of towns; or similar


the only restriction is that it should be neither systematic, nor deliberately random: no lists of telephone numbers, or of prime numbers, or the output from a computer's random number generator


whatever the source of the data focus on the first significant digit - whether the data value is 81 or 0.81 or 0.081 that digit is 8


the probability of that digit being 1 is 30.1%