Suppose that the duality between F and G (resp. F1 and G1) is separating in G (resp. F1). Let u be a linear mapping of F in F1 that is continuous for σ(F,G) and σ(F1,G1). Let A be a set in F and A1 a set in F1; then:
(i) We have (u(A))∘=tu−1(A∘).
(ii) We have tu(A1∘)⊂(u−1(A1))∘; further, if A is closed, (for σ(F1,G1)) convex, and contains the origin, then we have tu(A1∘)=(u−1(A1))∘.
Let z1∈G1, the relation z1∈(u(A))∘ is equivalent to ⟨u(y),z1⟩≥−1 for all y∈A, and the relation tu(z1)∈A∘ is equivalent to ⟨y,tu(z1)⟩≥−1 for all y∈A and our assertion (i) follows using (1). Next interchanging u and tu and applying (i) to the set A1∘ of G1 we get
(tu(A1∘))∘=u−1(A1∘∘)⊃u−1(A1)
from which, on taking polars
(tu(A1∘))∘∘⊂(u−1(A1))∘.
We have tu(A1∘)⊂(tu(A1∘))∘∘ by the bipolar theorem (II, p. 44, th. 1); the final statement follows from (2) and the bipolar theorem since then A1∘∘=A1 and tu(A1∘) is convex and contains the origin.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.