Let V be a finite dimensional simple g-module, ω its highest weight, X its set of weights. Choose a W-invariant non-degenerate positive symmetric bilinear form (⋅∣⋅) on hR∗, and let λ→∥λ∥=(λ∣λ)1/2be the corresponding norm.
(i) X is the smallest R-saturated subset of P containing ω.
(ii) The R-extremal elements of X are the W-transforms of ω.
(iii) If μ∈X, we have ∥μ∥≤∥ω∥. If, in addition, μ=ω, we have ∥μ+ρ∥<∥ω+ρ∥. If μ is not R-extremal in X, then ∥μ∥<∥ω∥.
(iv) We have X=W.(X∩P++). An element λ of P++ belongs to X∩P++ if and only if ω−λ∈Q+.
(i) Let X′ be the smallest R-saturated subset of P containing ω. We have X′⊂X (Prop. 4). Assume that X=X′. Let λ be a maximal element of X -- X′. Since λ=ω, there exists α∈B such that λ+α∈X. Introduce p and q as in Prop. 3. Since λ is maximal in X -- X′,λ+pα∈X′. By Prop. 3 (ii), λ−qα∈X′ since X′ is stable under W. Hence λ+uα∈X′ for every integer u in the interval −q,p. This contradicts λ∈/X′ and proves (i).
(ii) It is clear that ω is an R-extremal element of X; its W-transforms are therefore also R-extremal in X. Let λ be an R-extremal element of X; we shall prove that λ∈W.ω. Since there exists w∈W such that wλ∈P++ (Chap. VI, §1, no. 10), we can assume that λ∈P++. Let α∈B; introduce p and q as in Prop. 3. Since λ is R-extremal, either p=0 or q=0. Since
q−p=λ(Hα)≥0
we cannot have p>0. Hence p=0 and λ=ω.
(iii) Let μ∈X∩P++. Then ω+μ∈P++ and ω−μ∈Q+(§6, no. 1, Prop. 1), so 0≤(ω−μ∣ω+μ)=(ω∣ω)−(μ∣μ); hence, (μ∣μ)≤(ω∣ω), and this extends to all μ∈X by using the Weyl group. If μ∈X -- {ω},
(μ+ρ∣μ+ρ)=(μ∣μ)+2(μ∣ρ)+(ρ∣ρ)≤(ω∣ω)+2(μ∣ρ)+(ρ∣ρ)
=(ω+ρ∣ω+ρ)−2(ω−μ∣ρ)
Now ω−μ=∑α∈Bnαα with integers nα≥0 not all zero, so (ω−μ∣ρ)>0 since (ρ∣α)>0 for all α∈B (Chap. VI, §1, no. 10, Prop. 29 (iii)). If μ is not R-extremal in X, there exists α∈R such that μ+α∈X and μ−α∈X; then
∥μ∥< sup(∥μ+α∥,∥μ−α∥)≤ supλ∈X∥λ∥
and this last upper bound is ∥ω∥ by the preceding.
(iv) We have X=W.(X∩P++) by Chap. VI, §1, no. 10. If λ∈X, then ω−λ∈Q+(§6, no. 1, Prop. 1). If λ∈P++ and ω−λ∈Q+, then λ∈X (Cor. 2 of Prop. 3).
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.