If one asks the man in the street (or even the average reader of Word Ways) for a listing of letters by frequency of occurrence, he is likely to come up with the linotyper's nonsense-phrase ETAOIN SHRDLU. More serious students of letter-frequencies, such as solvers of monoalphabetic substitution ciphers, are likely to point out in addition that the frequencies of certain letters are so nearly alike (for example, A and O, or N and R) that the ordering depends upon the sample of letters tabulated. Although most people will readily understand that the listing is based on a sample of letters from some larger population, few will include a careful specification of this sample and population as part of their answer. It is the purpose of this article to show how important such a specification can be; we exhibit a number of startlingly different letter-orderings corresponding to different populations.
Card, Leslie E. and Eckler, A. Ross
"A Survey of Letter-Frequencies,"
Word Ways: Vol. 8
, Article 6.
Available at: http://digitalcommons.butler.edu/wordways/vol8/iss2/6