12.8. GENERAL TECHNIQUES FOR Ln(2) ON THE NATURAL MODULE 851
F.8.1 is satisfied for each H E Hz, while the remaining conditions are easily ver-
ified; for example, 12.8.4.4 says ker 0 HCV)(H) = QH, giving (c). Thus (3) implies
(4). Finally (4) implies (1) by F.8.5.2, and (4) and (5) are equivalent by Remark
F.9.9. D
REMARK 12.8.7. Notice that if one of the equivalent conditions in 12.8.6 holds,
then from condition (5), (VH) = (Vf). Thus the subgroups denoted by "UH" and
"VH" in section F.9 both coincide with the group denoted by UH in this section.
When UH is abelian for all H E Hz, by parts (4) and (5) of 12.8.6, we can
apply lemmas from sections F.8 and F.9 to analyze the amalgam defined by LT
and H. Notice in particular by F.8.5 and F.9:11 that in this case the amalgam
parameter "b" of those sections is odd and at least 3. On the other hand when UH
is nonabelian, we normally specialize to the case H = G 1 , and apply methods from
the theory of large extraspecial 2-subgroups, which are developed further in the
following subsection.
12.8.2. (VG^1 ) nonabelian almost extraspecial subgroups. In this sub-
section, we consider the case where (VG^1 ) is nonabelian. The analysis in the sub-
section continues to develop the theory of almost extraspecial 2-subgroups U (i.e.,
U is nonabelian and l<I>(U)I = 2) begun in section G.2 of Volume I. The theory is
a variant of the theory of large extraspecial 2-subgroups appearing in the original
classification literature.
In the remainder of the section we take H := G1 and assume that U :=UH=
(VH) is nonabelian.
As U is nonabelian, 12.8.4.3 says that: ,
(U) =Vi.
Set fl:= H/Z(U) and iI := H/CH(U).
LEMMA 12.8.8. (1) U = U 0 Z(U), with Uo an extraspecial 2-group and <I>(Uo) =
{2) Regard Vi as F2. Then the map
(ili, ii2) := [u1, u2]
defines a symmetric bilinear form on fj with radical Z(ii) preserved by H*, which
induces an H-invariant symplectic form on U. If <I>(Z(U)) = 1, then
q(u) := u^2
defines an H -invariant quadratic form on U with bilinear form ( , ) , which induces
an H -invariant orthogonal space structure on f).
{3) V n Z(U) = V1.
(4) Assume n:::; 3, let I:= (UL), and S := 02(I). Then L:::; I and S has the
I -chief series ·
1 =: So :::; S1 :::; · · · :::; Sn+I := S
described in G. 2. 3 or G. 2. 5, for n = 2 or 3, respectively.
(5) If L1 :SI H then n:::; 3, and when n = 3 the chief series in {3) becomes
1 =: So < Si < S3 = S