Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Fix relatively prime integers a>b>0a>b>0 and an integer n>2n>2. Let P(n)P(n) be the greatest prime factor of nn, and let Φn=Φn(a,b)\Phi_n=\Phi_n(a,b) be the homogeneous nnth cyclotomic factor.

Statement

For n>2n>2, P(n)2∤ΦnP(n)^2\nmid\Phi_n, and each prime factor of Φn\Phi_n other than P(n)P(n) is congruent to 1(modn)1\pmod n.

Printed omission and usable range

The printed text of Lemma 3 reads: "The prime P(n)P(n) can divide Φn\Phi_n to at most the first power. All other prime factors of Φn\Phi_n are congruent to 1(mod n)1 (\mathrm{mod}\, n)." (p. 429). It omits an explicit n>2n>2 qualifier. Taken literally at n=2n=2, its first sentence is false: for a=3a=3 and b=1b=1, one has P(2)=2P(2)=2 and Φ2(3,1)=3+1=4\Phi_2(3,1)=3+1=4, so P(2)P(2) divides Φ2(3,1)\Phi_2(3,1) to the second power. Theorem 1 assumes n>2n>2, and the Section 3 application begins with nn sufficiently large. This page therefore records the lemma only in the usable range n>2n>2. This is a transparent compilation qualification, not an author-issued erratum.

Source and proof pointer

This is Lemma 3 on printed p. 429, the right half of physical p. 2 of the retained published scan. Stewart records it as a consequence of Birkhoff and Vandiver's work and credits a first version, apparently, to Sylvester; the paper gives no separate proof at this point. The lemma is used in Sections 3--4 for Theorems 1--2.

This page records the statement and source attribution only. Neither a proof of the lemma nor the cited predecessor arguments are transcribed or verified here.

Bears on. #977.