5.1. PRELIMINARY ANALYSIS OF THE L 2 (2n) CASE 633
Also ifs is the involution in (a'), then GA(U) = S(s)D_ and U = (zD+), so U = Vv.
In particular [D_, Z] = 1, so Z :::; Z(X-). If n = 2 mod 4, then T :=:; S(s),
so Z = U is D+-invariant; hence X = (H, DL) :::; Na(Z), contrary to 02 (X) = 1.
Thus n = 0 mod 4. Finally D+ is faithful on Vv, so applying 3.1.6 with TD+, Vv in
the roles of "Mo, V" as before, either 02(X+) of. 1 or q(D+T/0 2 (D+T), Vv) .:=:; 2. In
the latter case, as T/Gr(Vv) is cyclic and m(Vv/Gvn (t)) ~ n/4 fort E T-Gr(Vv),
n = 4 or 8. Thus (4) holds. D
LEMMA 5.1.7. (1) Na(Baum(T)) '.SM.
(2) Let HE H*(T, M) and set K := 02 (H). Assume [Z, H] of. 1. Then
(i) L = [L, J(T)].
(ii) K = [K, J(T)].
(iii) Either 02( (NL(T n L), H)) of. 1, or [V, L] is the S5-module for LT~
S5, and Z(H) = 1.
PROOF. We first prove (1). Let S := Baum(T). If J(T) :::; Gr(V), then
(1) follows from 5.1.2. Thus we may assume J(T) f:. Gr(V), so by 5.1.2, either
V/Gv(L) is the natural module for Lor [V,L] is the A 5 -module. In the former
case, Sn LE Sylz(L) by E.2.3.2, so (1) follows from 4.3.17.
Therefore we may assume that [V, L] is the A5-module. As [V, J(T)] of. 1,
we conclude from E.2.3 that LT~ S 5 , S = J(T) ~ E 4 is generated by the two
transvections in T, and (ZL) = [V,L] EB Gz(L). We may assume V = [V,L].
Assume that Na(S) f:. M; then no nontrivial characteristic subgroup of S is
normal in LT as M = !M(LT). Hence by E.2.3.3, L is an A 5 -block, so V =
02 (L) :::;I M. Let Q := 02 (LS). It follows using C.1.13.b that Q = V x Qc, where
Qc := Gs(L).
For any 1 of. S 0 :::; S normalized by LT, we have Na(So) :::; M = !M(LT), so
Na(S) f:. Na(So) by our assumption. Thus Hypothesis C.6.2 is satisfied with L, S,
T, Na(S) in the roles of "L, R, TH, A". Therefore by C.6.3.1 there is g E Na(S)
with Vg f:. Q. As V :::;I M, g f:. M.
Suppose that Qc f:. Qg. Since [Qc, Vg] n [V, Vg] :S Qc n V = 1, from the
action of S on V and hence on Vg, we conclude that Qc and V induce distinct
transvections on vg. Thus as IS : Qgl = 4, s = QcVQg. Let x E [Qc, Vg]#;
then x E Qc :::; Ga(L), so as M = !M(L) by Theorem 4.3.2, Ga(x) :::; M, so
V :::; 02 (Ca(x)). Since Qc induces a transvection on the A5-module Vg for Lg,
GL9s(x)QcQg/QcQg ~ S3, so V :S 02(GL9s(x)QcQg) = QcQg, contrary to V
and Qc inducing distinct transvections on vg.
Therefore Qc :::; Qg. Hence
(Qc) :S (Qg) = (QbVg) = (Qb),
so <I>(Qc) = <I>(Qb). Thus as <I>(Qc) ~ LT and g f:. M = !M(LT), <I>(Qc) = 1 =
<I>(Q).
Next we claim we can choose g so that S = QQg. If not then Q n Qg is a
hyperplane of Q and Qg centralized by Qg, so Qg induces a transvection on Q
and hence S = QgQgt fort ET - S0 2 (LT). Thus as g E Na(S), S = QQgtg-
1
,
establishing the claim.
As S = QQg with (Q) = 1 and Qc :::; Qg, S = Qc x D1 x D2, where