Let f, g and h be monic polynomials in Q[X] such that f=gh. If f has integer coefficients, the same is true of g and h.
Let a (resp. b) be the least of the integers α≥3 (resp. β≥1) such that αg (resp. βh) has integer coefficients; we put g′=ag and h′=bh and show by reductio ad absurdum that a=b=1. If this were not so, there would be a prime divisor p of ab. If u∈Z[X], let us write uˉ for the polynomial with coefficients in the field Fp obtained by reduction mod p of the coefficients of u. We have g′h′=abf, whence gˉ′hˉ′=0; since the ring Fp[X] is an integral domain (IV, p. 9, Prop. 8) we thus have gˉ′=0 or hˉ′=0. In other words, p divides all the coefficients of g′ or all those if h′ and this contradicts the hypothesis.
We have the relation
Xn−1=d∣n∏Φd(X).
For we have Xn−1=∏ζ∈μn(Q)(X−ζ) and the sets Sd for d dividing n form a partition of μn(Q).
The formula (6) determines Φn(X) when we know Φd(X) for all the divisors d<n of n; since Φ1(X)=X−1, we thus have a recursive procedure for calculating Φn. For example for prime p we have
Xp−1=(X−1)Φp(X),
whence
Φp(X)=Xp−1+Xp−2+⋯+X+1,
and
Φpr+1(X)=Φp(Xpr)forr≥0.
Let us list the values of the polynomials Φn(X) for 1≤n≤12:
Φ1(X)Φ2(X)Φ3(X)Φ4(X)Φ5(X)=X−1=X+1=X2+X+1=X2+1=X4+X3+X2+X+1
Φ6(X)=X2−X+1
Φ7(X)=X6+X5+X4+X3+X2+X+1
Φ8(X)=X4+1
Φ9(X)=X6+X3+1
Φ10(X)=X4−X3+X2−X+1
Φ11(X)=X10+X9+X8+X7+X6+X5+X4+X3+X2+X+1
Φ12(X)=X4−X2+1
The values of Φ1,Φ2,Φ3,Φ4,Φ5,Φ7,Φ8,Φ9 and Φ11 follow directly from (7); now we have Φ1Φ2Φ3Φ6=X6−1 and Φ1Φ2Φ3Φ4Φ6Φ12=X12−1, whence Φ4Φ12=X6−1X12−1−X6+1 and finally Φ12=X2+1X6+1=X4−X2+1. The cases n=6 and n=10 may be treated similarly.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.