14.7. FINISHING Ls(2) WITH (vGi) ABELIAN 1075
Therefore 11 appears in one of the cases (2)-(13) of F.6.18. Further the subcase
of case (2) of F.6.18 with 02 (1+) ~ 31+^2 is eliminated, since in that case there is
no subnormal subgroup of J+ isomorphic to S 3. Thus if l+ is solvable, then by
F.6.18, I' ~ S3 x S3, so there is a normal subgroup Kt of l+ contained in J(I)+
isomorphic to S3. Then as Y = [Y, T], either y+ = 02 (Kr)+ or y+ centralizes Kj.
Similarly either Lt= 02 (Kr)+ or Lt centralizes Kj. Therefore as y+ does not
act on Lt, we conclude using F.6.6 that 02 (!) = (Y, Ll) centralizes Kt, impossible
as 02 (Kj) t Z(Kj).
Therefore I' is nonsolvable, so as l+ has a subnormal subgroup isomorphic to
S3, L3(2) or A5, it follows from F.6.18 that I'~ L3(2). Thus Kr= 02 (1) = (Y,L1)
and l = KrT. But now E4 ~ E = [E, Y] :<:::; [Vr, Kr], so as LlT centralizes Vi and
Kr= 02 (!), Vr = [Vr, Kr] = (Vj_Kr) is of rank 3 by H.5.5. This is impossible, since
we saw earlier that (Z~^1 ) is of rank 4. D
LEMMA 14.7.71. (1) H has two noncentral 2-chief factors, both isomorphic to
tJ, one on U and one on QH/CH(U).
(2) L 1 has five noncentral 2-chief factors, one in 02 (L1), and four in S.
(3) [Q, L] :<:::; S.
PROOF. The proof is similar to some of the analysis in the second subsection,
but is substantially easier. First Ll has one noncentral chief factor on 02(Li), two
on tJ, and hence also two on QH/CH(U) by the duality in 14.5.21.1. Thus Ll has
at least five noncentral 2-chief factors.
Next as Ua :<:::; S by 14.7.70.4, using 14.7.66 we have
02(Li) = (U~L^1 ) :<:::; S. ()
Set Qx := [QH, K]CH(U). As [Qx/CH(U), SJ= [Qx/Cx(U), 02(Li)] is of corank
2 in Qx, with (S, Qx] :<:::; S, and as m(Qx/CQx(V)) = 2 by the duality in 14.5.21.1,
we conclude
Thus one noncentral 2-chief factor for Ll in Q lies in S*, two lie in U :<:::; S, and by
(**) a fourth factor also lies in S. Now if (1) holds, then Ll has four noncentral
2-chief factors in QH, so £ 1 has exactly five noncentral 2-chief factors by (*). Then
as L 1 has at least four noncentral chief factors on S, (2) holds, and of course (3)
follows from ( 2).
Thus it remains to prove (1), so we must show that [CH(U), K] :<:::; U. But
K = [K, UaJ, so it suffices to show [CH(Ua), Ua] :<:::; U.
Now CH(U) :<:::; CH(Za) :<:::; Na(Ua), so [Ua, CH(U)] :<:::; Cu,JU). We will show
that m(Cu°'(U)/U n Ua) :<:::; 1; then as m([W, Ua]) ~ 2 for any nontrivial H-chief
factor Won CH(U)/U since U~ :<:::; K* by 14.7.66, our proof will be complete.
By 14.7.69, m(U~) = 1, and by 14.7.70.3, m(Ua/U n Ua) = 2. So indeed
m(Cu°' (U)/U n Ua) :<:::; 1, as desired. D
LEMMA 14.7.72. (1) S = 02(L) = (02(L), L].
(2) S/V is the Steinberg module for L/S.
(3) Uo = V1.
(4) V = Z(S) = (S) = (S, SJ.
PROOF. By 14.7.70.4, b = 3. Set R := (U/J), so that V :<:::; R :<:::; S, and
((U 0 V)L) =RV= R. From Theorem 14.7.63 and 14.7.66, [U,Q] = VUo, and from