4. The Proof of the Main Result
Proof of Theorem 1. As k = pa1 1 pa22 · · · pas s we can write k in the form k = lp,
where p is any of the primes pj, j = 1, 2, . . . , s. Thus without loss of generality
we have
Mp
j i(k) = Mpi(lp) = (lp + 1)pi − 1
lp
and the assertion can be rewritten as M
pi(lp) ≡ 0 (mod lpi+1) or equivalently
(lp + 1)pi ≡ 1 (mod lpi+1). (6)
To prove the last congruence we use the binomial theorem
(lp + 1)pi =
pi
Xm
=0
mpi (lp)m = 1 + Xpi
m=1
mpi (lp)m.
Now congruence (6) follows from Lemma 5.