Let (Xι)ι∈I be a family of sets, and let f be a mapping of a set K onto I. Then
ϰ∈K⋃Xf(ϰ)=ι∈I⋃Xι,
and, if I=∅,
ϰ∈K⋂Xf(ϰ)=ι∈I⋂Xι.
Let x be an element of ι∈I⋂Xι. There exists an index ι∈I such that x∈Xι. Since f⟨K⟩=I, there exists an index ϰ∈K such that ι=f(ϰ), whence x∈Xf(ϰ) and consequently
x∈ϰ∈K⋃Xf(ϰ).
Conversely, if x∈ϰ∈K⋃Xf(ϰ), there exists ϰ∈K such that x∈Xf(ϰ), and therefore, since f(ϰ)∈I, x∈ι∈I⋃Xι. Hence
ϰ∈K⋃Xf(ϰ)=ι∈I⋃Xι.
¶ Now suppose that I=∅, and let x be an element of ι∈I⋂Xι. For each element ϰ of K we have f(ϰ)∈I, hence x∈Xf(ϰ), and therefore
x∈ϰ∈K⋂Xf(ϰ).
Conversely, let x be an element of ϰ∈K⋂Xf(ϰ). If ι is any element of I, there exists an element ϰ of K such that ι=f(ϰ), whence x∈Xι and consequently x∈⋂ι∈IXι. Hence
ϰ∈K⋂Xf(ϰ)=ι∈I⋂Xι.
For families of subsets of a given set, it is clear that the second part of Proposition 1 remains valid without the restriction I=∅.
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.