IN my presidential address to Section A of the British Association, reprinted in NATURE (September 16), I stated that 137 was the least value of n for which the prime or composite character of 2n–1 was still undecided. Mr. W. W. Rouse Ball has pointed out to me that this is incorrect, as 2137–1 has been shown to be composite by M. A. Gerardin (Comptes rendus du Congres des Societes Savantes, 1920, pp. 53–55). The result is quoted in The American Mathematical Monthly, vol. 28, 1921, p. 380. The number 139 should therefore be substituted for 137 wherever it occurs in my address.