Let A and B be Krull domains such that B contains A and is aflat A-module. Then: (a) condition (PDE) of Proposition 14 is fulfilled; (b) for every divisorial ideal a∈A, Ba is the divisorial ideal of B which corresponds to the divisor i(divA(a)).
To show (a), suppose that there exists a prime ideal p of B of height 1 such that p∩A is neither 0 nor of height 1. Take an element x=0 in p∩A. The ideals pi of A of height 1 which contain x are finite in number and none contains p∩A; there therefore exists an element y of p∩A such that y∈/pi for all i (Chapter II, § 2, no. 1, Proposition 2). Thus divA(x) and divA(y) are relatively prime elements of the ordered group P(A), so that sup(divA(x),divA(y))=divA(x)+divA(y)=divA(xy); as the ideals Ax∩Ay and Axy are divisorial, we deduce that Ax∩Ay=Axy. Since B is a flat A-module, Bx∩By=Bxy (Chapter I, § 2, no. 6, Proposition 6). This implies that sup(vp(x),vp(y))=vp(xy)=vp(x)+vp(y), which contradicts the inequalities vp(x)>0,vp(y)>0 (which hold since x and y are in p). Thus (a) has been proved by reductio ad absurdum.
We now show (b). If a is a divisorial ideal of A, it is the intersection of two fractional principal ideals (no. 5, Corollary 2 to Proposition 9), say
a=d−1(Aa∩Ab)
where a,b, dare in A∗; as B is flat over A, Ba=d−1(Ba∩Bb) (Chapter I, § 2, no. 6, Proposition 6), which shows that Ba is divisorial. This shows also that divB(Ba)=sup(divB(a),divB(b))−divB(d); using Proposition 14 (a) and (b), it is seen that
divB(Ba)=sup(i(divA(a)),i(divA(b)))−i(divA(d))=i(sup(divA(a),divA(b)))−i(divA(d))=i(divA(Aa∩Ab))−i(divA(d))=i(divA(d−1(Aa∩Ab)))=i(divA(a)).
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.