In fact, if α∈M1(G) satisfies FG(α)=0, one deduces from (31) that α(FG(f))=0 for every f∈L1(G); since the image of L1(G) under the Fourier transform is dense in C0(G) (corollary to Proposition 5 of II, p. 209), it follows that α=0.
There exist function spaces on G and G, other than L2(G) and L2(G), on which F and F are inverse isomorphisms of one another. The following theorem gives an example of this. We denote by B(G) the vector subspace of L1(G) consisting of the elements f∈L1(G) such that FG(f)∈L1(G). This is a subalgebra of L1(G). In fact, let f and g be in B(G). Then f∗g∈L1(G) and FG(f∗g)=FG(f)FG(g)∈L1(G), since FG(f)∈L1(G) and FG(g)∈C0(G).
Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.