An arithmetic theorem and its demonstration

Abstract

Translated from the Latin original, "Theorema arithmeticum eiusque demonstratio", Commentationes arithmeticae collectae 2 (1849), 588-592. E794 in the Enestroem index. For m distinct numbers a,b,c,d,...,\upsilon,x this paper evaluates \[ \frac{a^n}{(a-b)(a-c)(a-d)...(a-x)}+\frac{b^n}{(b-a)(b-c)(b-d)...(b-x)} +...+\frac{x^n}{(x-a)(x-b)(x-c)...(x-\upsilon)}. \] When $n \leq m-2$, the sum is 0, which Euler had already shown in sect. 1169 of his Institutiones calculi integralis, vol. I, E366. When $n=m-1$ the sum is $=1$. When $n \geq m$, then using Newton's identities Euler gets expressions in terms of sums of powers of the numbers a,b,c,d,...,\upsilon,x. The notation used in this paper is confus(ing/ed), and indeed the editors of the Opera omnia made several corrections to the original publication. Near the end of the paper, "... for each of them one searches for what the character will be..." is the only reading of the sentence I can come up with. Examples of a work that refers to this theorem is J.-P. Serre, Corps locaux, 2nd ed., ch. III, sect. 6 (for $n \leq m-1$), and D. E. Knuth, The Art of Computer Programming, vol. 1, sect. 1.2.3, exercise 33.