A function φ2(n) giving the number of pairs of consecutive integers each less than n and prime to n, was considered first by Schemmel. In applying this function to the enumeration of magic squares, D. N. Lehmer has shown that if one replaces consecutive pairs by pairs of integers having a fixed difference λ prime to n = Πti=1pi αi, then the number of such pairs (mod n) whose elements are both prime to n is also given by
|φ2(n) =||Π piαi–1(pi — 2)|
As is the case for Euler's totient function φ(n), the function φ2(n) obviously enjoys the multiplicative property φ2(m)φ2(n) = φ2(mn), (m,n) = 1, φ2(1) = 1. In what follows we call an integer simple if it contains no square factor >1. For a simple number n we have the following analog of Gauss' theorem:
|Σ φ2(δ) = φ(n),|
where n is simple and where the summation extends over all the divisors of n. Using Dedekind's inversion formula, we can write
|Σ φ(n/δ)μ(δ) = φ2(n),|
where n is simple and where μ(n) is Merten's inversion function.
It is the purpose of this note to develop another property of φ2(n), true only for simple numbers, and apply it to the evaluation of the discriminants and resultants of cyclotomic equations.