(7) Λ(f)=δiαi(f)δi−1 for i∈ Orb(G) and f∈Ga(i),
(8) Λ(j)=δφ0(j)τ0(j)δψ0(j)−1 for j∈ Orb(H).
The homomorphism Λ is surjective; its kernel is the smallest normal subgroup of (∗iGa(i))∗ F(Orb(H)) containing the elements j of Fl(T) and the elements φj(f)jψj(f)−1j−1, for j∈ Orb(H) and f∈Hb(j).
The existence and uniqueness of the homomorphism Λ result from the universal property of free products and free groups (A, I, p. 85, prop. 8).
Let A denote the set of the a(i) for i∈ Orb(G) and GA the full subgroupoid of G whose set of vertices is A. For every x∈ Som(G), let x denote the orbit of x in G and choose an arrow dx of G joining x to a(x). The pair v formed by the mapping x↦a(x) of Som(G) into A and the mapping which to f∈ Flx,y(G) associates the element dx−1fdy of Fla(x),a(y)(GA) is a groupoid morphism. It follows from Proposition 1 of II, p. 182 that v is a homotopism. Let φ′=v∘φ and ψ′=v∘ψ, and then let w be the canonical morphism of Coh(φ,ψ ) into Coh(φ′,ψ′) ; it is a homotopism (II, p. 187, Theorem 1).
The orbits of GA are the sets {a}, for a∈A, and the injection GA→G induces a bijection of Orb(GA) onto Orb(G), by means of which we shall identify these two sets. One defines a basic equipment (a′,b′,β1′,β2′,T′,i0) of the pair (φ′,ψ′) by setting a′(i)=a(i) for i∈ Orb(G),b′(j)=b(j),β1′(j)=v(c1(j)),β2′(j)=v(c2(j)) for j∈ Orb(H) and T′=T. The group homomorphisms φj′ and ψj′ (for j∈ Orb(H)), the quiver morphism τ0′, the arrows δi′ (for i∈ Orb(G)), and therefore the group homomorphism Λ′, deduced from this basic equipment, are the composites with w of the corresponding homomorphisms φj,ψj, of the quiver morphism τ0, of the corresponding arrows δi and of the homomorphism Λ.
Let B be the set of the b(j) for j∈ Orb(H), let HB be the full subgroupoid of H with set of vertices B; let us denote by u:HB→H the canonical injection; put φ′′=φ′∘u and ψ′′=ψ′∘u. The morphism u induces a bijection B → Orb(H) by which we shall identify these two sets. From theorem 1 of II, p. 187 one further deduces a canonical homotopism w′: Coh(φ′′,ψ′′)→ Coh(φ′,ψ′). Moreover, the pair (φ′′,ψ′′) is endowed with a base equipment (a′′,b′′,β1′′,β2′′,T′′,i0), so that Λ′,φj′,ψj′,τ0′,δi′(i,i′∈ Orb(G), j∈ Orb(H)) are the composites with w′ of Λ′′,φj′′,ψj′′,τ0′′,δi′′.
The various morphisms of groupoids introduced are summarized by the following diagram:
HBψ′′φ′′GAα′′ Coh(φ′′,ψ′′)
w′
(9) H φ′GAα′ Coh(φ′′,ψ′′)
ψ′
vw
H ψφ G α Coh(φ,ψ ) .
To prove the proposition, we may therefore suppose that A = Som(G) and B = Som(H), in other words that the canonical mappings Som(G) → Orb(G) and Som(H) → Orb(H) are bijective, hypotheses under which we shall place ourselves in the rest of the proof.
The quiver G1 then has for set of vertices A and for arrows the sum set of the sets Ga,a∈A, and of the set B. The arrows of Ga are loops at a; if b∈B, the arrow b joins φ(b) to ψ(b), the arrows c1(b) and c2(b) are loops respectively at φ(b) and ψ(b). The quiver T will be identified with an oriented tree of G1; it is a maximal oriented tree since the set of its vertices is equal to the set of vertices of G1(II, p. 157, prop. 1). We put a0=a(i0). The set of arrows of the framework Γ0 of the pair (φ,ψ) being identified with B, the morphism of quivers τ1:Γ0→ Grp(G1) associates to the arrow b the class of paths c1(b)−1bc2(b) in the graph G1.
Recall that θ1 denotes the canonical morphism of quivers of G1 into Grp(G1). Let
λ:a∗∈AF(Ga)∗F(B)→ Grp(G1)a0
be the unique group homomorphism such that one has
λ(f)=τ1(da)θ1(f)τ1(da)−1 if a∈A and f∈Ga;
λ(b)=τ1(dφ(b))τ1(b)τ1(dψ(b))−1 if b∈B.
We thus have λ=λ′∘ε, where λ′ denotes the canonical group homomorphism from a∗∈AF(Ga)∗F(B) into Grp(G1)a0 defined by the maximal oriented tree T (II, p. 179, prop. 9), and where ε is the unique automorphism of the group a∗∈AF(Ga)∗F(B) such that ε(f)=f for a∈A and f∈Ga, and ε(b)=c1(b)−1h1(b)c2(b) for b∈B. According to remark 2 of II, p. 179, the homomorphism λ is surjective and its kernel is the smallest normal subgroup of a∗∈AF(Ga)∗F(B) containing the arrows of T.
Let us denote by π: Grp(G1)→ Coh(φ,ψ ) the canonical groupoid morphism. According to II, p. 177, cor. 1 of prop. 8, the group homomorphism πa0 from Grp(G1)a0 into Coh(φ,ψ )a0 is surjective, and its kernel is the smallest normal subgroup of Grp(G1)a0 containing the loops Int(τ1(da))(α1(f)α1(g)α1(fg)−1), for a∈A and f,g∈Ga, and the loops Int(τ1(dφ(b)))(φ(f)bψ(f)−1b−1), for b∈B and f∈Hb.
If p:F(⋃Ga∪B)→(a∗∈AGa)∗F(B) denotes the canonical surjective homomorphism, we thus have Λ∘p=πa0∘λ. This formula implies that the homomorphism Λ is surjective; it remains to determine its kernel.
For a∈A and f∈Ga, let us denote by [f] the image of f∈F(Ga) in the group a∗∈AF(Ga)∗F(B). For a∈A,f,g∈Ga, we then have
Consequently, the kernel of the homomorphism πa0∘λ is the smallest normal subgroup of a∗∈AF(Ga)∗F(B) which contains the elements [f][g][fg]−1 for a∈A and f,g∈Ga, the elements φb(f)[b]ψb(f)−1[b]−1 for b∈B and f∈Hb, and the elements [b], for b∈ Fl(T).
Finally, the kernel of the homomorphism Λ is the smallest normal subgroup of (a∗∈AGa)∗F(B) which contains the images under p of the preceding elements, in other words, the elements [b], for b∈ Fl(T), and the elements φb(f)[b]ψb(f)[b]−1, for b∈B and f∈Hb. The proposition is thus proved.
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.