Let f∈A(G). Then F(f)∈L1(G) and, for every x∈G, one has
f(x)=∫G⟨x,x⟩F(f)(x)dx(25)
In other words, for f∈A(G), one has
f=FG(FG(f))∘η(26)
where η denotes the canonical mapping of G into the bidual group G.
By Lemma 4 of II, p. 213 and Proposition 9 of II, p. 214, one has F(f)∈L1(G) for every function f∈A(G). By the Plancherel formula (24), for f and g in L2(G), one has
(f∗g)(e)=∫GF(f)(χ)F(g)(χ)dx(χ)(27)
Let f and g be in L1(G)∩L2(G) and let h=f∗g∈A(G). Since the Fourier transform is an involutive morphism, formula (27) is assertion (25) for the function h at the point x=e. By linearity, one deduces that formula (25) is valid at the point x=e for every function h∈A(G).
Let x∈G and h∈A(G). Let h1=εx−1∗h. Then h1∈A(G) and h1(e)=h(x). Since moreover F(h1)(χ)=⟨χ,x⟩F(f)(χ) for every χ∈G(cf. formula (11) of II, p. 208), formula (25) for the function h1 at the point e implies formula (25) for h at the point x.
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.