Stirling's Theorem

Abstract

IN starting from dn = 1 and then making dn infinitesimal, Dr. Satterly's demonstration in NATURE of February 17, p. 220, is scarcely convincing, and the error introduced by this step is represented in his answer by the absence of the factor 1/n or the term ½ log n, neither of which is entirely negligible when n is large. I suggest the following adaptation of his proof, which avoids, I think, the inconsistency referred to above.