312 CHAPTER 12. THE CONSTRUCTION OF MEASURES

=∞

∑i=1

µ(A∩A′i

)ν(B∩B′i

)+µ

(A′i \A

)ν(B′i)+µ

(A∩A′i

)ν(B′i \B

)=

∑i=1

µ(A∩A′i

)ν(B′i)+µ

(A′i \A

)ν(B′i)=

∑i=1

µ(A′i)

ν(B′i)= ρ (P) .

This proves the claim.Now continuing to verify 12.9.39, without loss of generality, (µ×ν)(S) can be as-

sumed finite. Let P⊇ S for P ∈R and

(µ×ν)(S)+ ε > ρ (P) .

Then from the claim,

(µ×ν)(S)+ ε > ρ (P) = ρ (P∩ (A×B))+ρ (P\ (A×B))

≥ (µ×ν)(S∩ (A×B))+(µ×ν)(S\ (A×B)) .

Since ε > 0 this shows A×B is µ×ν measurable as claimed.

Lemma 12.9.9 Let R1 be defined as the set of all countable intersections of sets of R.Then if S ⊆ X ×Y, there exists R ∈ R1 for which it makes sense to write ρ (R) because12.9.32 and 12.9.33 hold such that

(µ×ν)(S) = ρ (R) . (12.9.40)

Also, every element of R1 is µ×ν measurable.

Proof: Consider 12.9.40. Let S⊆X×Y. If (µ×ν)(S)=∞, let R=X×Y and it followsρ (X×Y ) = ∞ = (µ×ν)(S) . Assume then that (µ×ν)(S)< ∞.

Therefore, there exists Pn ∈R such that Pn ⊇ S and

(µ×ν)(S)≤ ρ (Pn)< (µ×ν)(S)+1/n. (12.9.41)

Let Qn = ∩ni=1Pi ∈R. Define

P≡ ∩∞i=1Qi ⊇ S.

Then 12.9.41 holds with Qn in place of Pn. It is clear that

x→XP (x,y) is µ measurable

because this function is the pointwise limit of functions for which this is so. It remains toconsider whether y→

∫XP (x,y)dµ is ν measurable. First observe Qn ⊇ Qn+1, XQi ≤

XPi , and

ρ (Q1) = ρ (P1) =∫ ∫

XP1 (x,y)dµdν < ∞. (12.9.42)

Therefore, there exists a set of ν measure 0, N, such that if y /∈ N, then∫XP1 (x,y)dµ < ∞.

312 CHAPTER 12. THE CONSTRUCTION OF MEASURESImM:llunH(ANAs) v (BN Bi) +H (4; \A) v (Bj) +H (ANA}) v (B;\ B)co(474%) v (Bt) +4 (4}\A) v (BE) = Ya (AY) v (Bt) = 9 (P).i=lIMsThis proves the claim.Now continuing to verify 12.9.39, without loss of generality, (u x Vv) (S) can be as-sumed finite. Let P D S for P € & and(uxv)(S)+e>p(P).Then from the claim,(XV) (S)+e > p(P)=p(PA(AxB))+p(P\(AxB))> (HRV) (SN(A xB) + (WV) (S\ (Ax B)).Since € > 0 this shows A x B is fu x V measurable as claimed.Lemma 12.9.9 Let #, be defined as the set of all countable intersections of sets of &.Then if S CX x Y, there exists R € &, for which it makes sense to write p (R) because12.9.32 and 12.9.33 hold such that(ux v)(S)=p(R). (12.9.40)Also, every element of &, is L x V measurable.Proof: Consider 12.9.40. Let SCX x Y. If (u x v) (S) =c0, let R= X x Y and it followsp(X x Y) =co= (ux v)(S). Assume then that (ux V) (S) < 0%.Therefore, there exists P, € & such that P, D S and(HX) (8) <p (Pr) < (@XV)(S) + 1/n. (12.9.41)Let On = Ni_,P; € Z. DefineThen 12.9.41 holds with Q, in place of P,. It is clear thatx— Xp(x,y) is W measurablebecause this function is the pointwise limit of functions for which this is so. It remains toconsider whether y > { 2p(x,y)dp is v measurable. First observe Qn D Qn+1, %Q; <Xp,, andp(Q1)=p(P\)= / / Xp, (x,y) dudv <oo, (12.9.42)Therefore, there exists a set of v measure 0, N, such that if y ¢ N, then[% (x,y) du <0,