Let G1, G2, G3 be graphs. Then
(G3∘G2)∘G1=G3∘(G2∘G1).
The relation (x,t)∈(G3∘G2)∘G1 is equivalent to the relation
(∃y)((x,y)∈G1 and ∃z((y,z)∈G2 and (z,t)∈G3))
and therefore (by C33 (Chapter I, § 4, no. 3)) to the relation
(1)
(∃y)(∃z)((x,y)∈G1 and (y,z)∈G2 and (z,t)∈G3).
Similarly, the relation (x,t)∈G3∘(G2∘G1) is equivalent to
(2)
(∃z)(∃y)((x,y)∈G1 and (y,z)∈G2 and (z,t)∈G3).
But the relations (1) and (2) are equivalent; hence the result.
¶ The graph G3∘(G2∘G1) is denoted by G3∘G2∘G1. Similarly, if G1, G2, G3, G4 are graphs, we put
G4∘(G3∘G2∘G1)=G4∘G3∘G2∘G1;
and so on.
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.
Nothing.