ALGEBRAIC
APPROACH
TO STEENROD
OPERATIONS
by J. Peter May
i.
Introduction.
Since the introduction of the Steenrod operations in the cohoclear that similar operations exist in there are Steenrod operations in the in the c o h o m o l o g y of c o c o m 
m o l o g y of topological spaces, it has b e c o m e a variety of other situations. cohomology F o r example,
of simplicial restricted Lie algebras,
mutative Hopf algebras, w e r e introduced m o d
and in the h o m o l o g y of infinite loop spaces (where they m o d p, p>2, by D y e r and Lashof
2 by Araki and K u d o [3] and
[6]).
The purpose of this expository paper is to develop a general algebraic setting in which all such operations can be studied simultaneously. This approach allows
a single proof, applicable to all of the above examples, the operations, including the A d e m relations.
of the basic properties of
In contrast to categorical treatments
of Steenrod operations, simplify s o m e w h a t
the elegant proofs developed by Steenrod [2530] actually Further, even the m o s t general exist
in our algebraic setting.
ing categorical study of Steenrod operations, to iterated loop spaces.
that of Epstein [7], cannot be applied
W e e m p h a s i z e that this is an expository paper.
Although a n u m b e r
of n e w re
sults and n e w proofs of old results are scattered throughout, the only real claim to originality lies in the basic context. results since a large n u m b e r W e have chosen to give complete proofs of all allows a
of m i n o r simplifications in the arguments
substantial simplification of the theory as a whole.
W e have also included a n u m b e r
of topological results which should be wellknown but appear not to be in the litera

154

ture.
In particular, in section I0, w e give a quick complete calculation of the and s h o w that Serre's
m o d p c o h o m o l o g y Bockstein spectral sequence of K(w,n)'s simple proof [Z3] of the axiomatization of the m o d with only slight modifications to the case p > Z.
2 Steenrod operations applies
T h e general theory is presented in the first five sections.
Most of the proofs
in sections i, 2, and 4 are based on those of Steenrod [Z530], and those of section 3 are simplifications of arguments of D y e r and Lashof [6 ]. Via acyclic models and a l e m m a due to Dold [5], the theory is applied to several simplicial categories The standard properties of the
and to topological spaces in sections 7 and 8. Steenrod operations in spaces, except and
po = I, drop out of the algebraic theory, In contrast, w e prove that The theory is
1 = I is s h o w n to follow f r o m these properties. m°
po = 0 on the c o h o m o l o g y of simplicial restricted Lie algebras. applied to the c o h o m o l o g y of c o c o m m u t a t i v e
Hopf algebras in section II; the opera
tions here are important in the study of the c o h o m o l o g y of the Steenrod algebra [13, 18]. The present analysis arose out of w o r k on iterated loop spaces, but this application will appear elsewhere. The material of sections 6 and 9, which is is presented here with a view
peripheral to the study of Steenrod operations, towards this application.

155

1.
Algebraic preliminaries; Let A be a c o m m u t a t i v e
equivariant
homology By a Acomplex, we u n d e r s t a n d a
(ungraded) ring.
Zgraded
differential Amodule,
graded by subscripts,
with d i f f e r e n t i a l of d e g r e e q = 0 for q < 0 and
minus one. n e g a t i v e if
We s a y t h a t a A  c o m p l e x K q = 0 for q > 0. We u s e
K i s p o s i t i v e if K Zgraded
c o m p l e x e s in o r d e r t h a t o u r The exposition and the nota
t h e o r y c a n be a p p l i e d e q u a l l y w e l l to h o m o l o g y a n d to c o h o m o l o g y . w i l l be g e a r e d to h o m o l o g y , w h e r e t h e n o t a t i o n is s l i g h t l y s i m p l e r , tions appropriate mentary f o r c o h o m o l o g y w i l l b e g i v e n in s e c t i o n 5.
We g i v e s o m e e l e 
homological lemmas
in t h i s s e c t i o n ; t h e s e e x t r a c t t h e s l i g h t a m o u n t of
i n f o r m a t i o n a b o u t t h e h o m o l o g y o f g r o u p s t h a t is n e e d e d f o r t h e d e v e l o p m e n t of t h e Steenrod operations. If ~ is a g r o u p , w e l e t s p e a k of A v  m o r p h i s m s s p e a k of ~  m o r p h i s m s on is r A letters, and let AT denote its group ring o v e r A. We s h a l l g e n e r a l l y and we s h a l l group
rather than ~equivariant Amorphisms, when ~r C ~ A is understood. r . For q e Z, l e t Let ~ A(q) r
denote the s y m m e t r i c
denote the ATrmodule which by let 0k = (  1 ) q s ( ° ' ) k , w h e r e M(q) denote rheAs).
as a A  m o d u l e
and has the A T  a c t i o n d e t e r m i n e d ~e ~r If M is a A w  m o d u l e , 0(re@k) = r Kr factors
(  1 ) s(°) i s t h e s i g n of module If M@A(q)
with the diagonal action let K r = K~...~K,
0m@ 0k ( w h e r e @ = ~ A K.
K is a A  c o m p l e x ,
V i a p e r m u t a t i o n of
factors,
with the standard sign convention, Kr(q) is defined.
becomes
a A~rcomplex for
w C E r, a n d Let
I denote the Afree Acomplex one basis element A and e
w h i c h has two b a s i s e l e m e n t s
e
o
and
e1
of degree zero,
of d e g r e e
o n e , a n d d i f f e r e n t i a l d(e) = e I  eo. Fmodule structure,
If F i s a H o p f a l g e b r a o v e r ya = 8(y)a f, g: K~,L Fmorphism for y ~F and
I is given the t r i v i a l
a • I, t h e n t h e n o t i o n of a F  h o m o t o p y of F  c o m p l e x e s , such that
h : f ~ g, w h e r e
are morphisms H:I~K'PL
is e q u i v a l e n t to t h e n o t i o n of a and H ( e o @ k ) = g(k), where
H ( e l @ k ) = f(k)

156

I~K
is given the diagonal Faction.
In fact, H
determines
h by h(k) = H ( e ~ k )
and conversely. With these notations, w e have the following l e m m a . diagonally on tensor products. Lemma (i) I.I. Let TrC ~
r
In all parts, A T
acts
and let V
be a positive ATrfree complex. r suchthat
There exists a A ~ r  m o r p h i s m and h( e l ~ v) = v @ e
h:I~V> V ~ I r I
h(eo@V) = V@eo r (ii) i~fr (iii) to H(K) (iv) a,b e N q
for all v e V . of A  c o m p l e x e s , then
If f,g: K   > L i~ gr:v~K If A and r>
are A  h o m o t o p i c m o r p h i s m s
V ~ L r are A ~  h o m o t o p i c m o r p h i s m s K is a Acomplex, then
of A T  c o m p l e x e s . equivalent
is a field and V@K r is
K is Ahomotopy V@H(K) r,
AThomotopy
e q u i v a l e n t to
Let v ~ V
satisfy d ( v ~ I) = 0 in V ~
T
A; let K and vl~b r
be a A  c o m p l e x
and let
be h o m o l o g o u s c y c l e s .
Then
v~a
r
are homologous cycles
of v ®
T
Kr(q).
(i) Let 8 : I ~> A be t h e a u g m e n t a t i o n Define k:V>V@J g ( e o ) : 1 = ~.~ ( e l ) , by k ( v ) = v @ ( e s:V>V@J and let
Proof. J = Ker(~r), Since
8r : ir ~ A r = A . 0.
r r i  eo). from ktothe
H(J)= 0, H ( V @ J ) =
Define a A T  h o m o t o p y Let
zero m a p by induction on degree as follows.
s I = 01 given Let {xj} be a si_ldi(x ) h is
*
Si.l:Vi_ I > (V~J).,wel find easily that di(ki si.idi) = 0. A~rbasis for and extend s.
1
V.;
1
for
x ~ {xj}, choose
1
si(x ) such that di+lsi(x)=ki(x)The d e s i r e d
to a l l of V. h(e~v) L
by Tequivariance.  s(v) for v e V.
ATrmorphism
obtained byletting
(ii) Let t : l ~ K   >
determine a A  h o m o t o p y
from
f to g.
T h e n the followl~fr to
ing composite is a A T  m o r p h i s m
which determines aA~rhomotopy f r o m
l®g r :
I@v@Kr where u: I r ~ h~l>v@ir~Kr l~u >V~(I@K)r l @ tr > V ~ L r
s
K r > ( I ~ K) r
is the evident shuffling isomorphism.

157

(iii)
Define
f: H(K) > 14 b y s e n d i n g e a c h e l e m e n t of a b a s i s f o r
H(K) to a
chosen representative cycle.
K ~ I m f (9 Coker f as a A  c o m p l e x and Coker f is is a field. The first half follows and
acyclic and therefore contractible since A implies the second half by (ii). (iv) Define a m o r p h i s m of A  c o m p l e x e s f(eo) = b, and f(e)= (l)qc, where df(e) = (l)qfd(e)). Let F: I @ V   > l~V h
f:l> K, of degree
q, by f(ei) = a,
d(c)= ab in K V@I<r(q) Kr(q).
(the sign ensures that
be the composite
> v~Ir
l~f r
> V~
A check of signs shows that fr is a qr. By
A:rmorphism, hence that F (i), w e find that
is a m o r p h i s m of A,rcomplexes of degree
F(ei®v) : (i ® f r ) ( v ® e ~ ) :
(i)qrdeg V v ® f(ei)r , i = 0 or I.
A, w e have that
Since w operates trivially on I and d ( v ® i ) = 0 in V ®
d(e®v)=
(e i  e o ) @ V
i n 1 ® V.
iT
Thus, in V ~
1T
e
Kr(q),

d F ( e @ v ) = (l)qrF(el @ v and this proves the result.
@v)= (i)qr(degv+l)(v@a r
0
v@br),
W e n o w consider the cyclic group w of prime order tion of the standard A~free resolution W = W(p,A) Definition i. 2. Let W.
1
p.
We r e c a l l the d e f i n i 
of A. p with generator o~. and
Let
w be the cyclic group of prime order e., i > 0.
1
be Awfree on one generator
Let N = i+ ~ + ... + apI
T = c~i in Aw. by the formulas
Define a differential d, augmentation ~ , and coproduct ~ on W
(I)
d(ezi+l) = Tezi
and
d(ezi ) = Nezi_l ;
~(o~Je o) = I ;
and
•(ezi+l) = ~ e2j@eZk+ i + ~ e2j+l@aeZk j+k=i j+k= i
*(ez)ezj®eZk + j+k= iI j+k=i
Then W
0< r <s <p
J
ezj+l ~ °zSezk+l
is a differentiaiAwcoalgebra and a A w  f r e e resolution of A.
When
necessary
f o r c l a r i t y , we s h a l l w r i t e
W(p,A)
for
W.
Of c o u r s e ,

158

W ( p , A ) = W(p, Z) ~ A .
The structure
of W(p, Zp)
shows that ~3 and c o p r o 
H.(~r; Zp) = H(W(p, Z p ) ~ r r Z p ) duct (2) ~b, b y the f o r m u l a Ho.(~r;Zp).,. h a s Zpbasis
is given, with its B o c k s t e i n o p e r a t i o n
{ell i>0}_
suchthat
~3(e2i)= e z i _ l
and
if p > Z .
¢(e i) = E e.j ~ e k if p = 2 o r i is odd, j+k = i We e m b e d ~ in E P b y o~(1. . . . . p) = (p, 1 . . . . .
J2(e2i ) = E e2j(~e2k j+k = i pl), where Z P
a c t s on {1 . . . . . p } .
We then h a v e the following l e m m a . E W. be the n  s k e l e t o n of W = W(p, Z ). L e t G 1 p inn be a n y s e t of i e f t c o s e t r e p r e s e n t a t i v e s f o r ~r in E . L e t K be a Z  m o d u l e with P P totally ordered BCK basis Lemma 1.3. L e t W 'n'=
{xjl j
e J}
Let
AC Kp have basis < 'Jp' Jl < Jp}"
{ x j P l j c J } a n d let Then
p have basis
{x.~(jp). I "~ ~ G, jl .<. . . .
n
H(w(n)~
K p) = ( (9 e . ~ A ) 1 i=O
~B(eo~B ) (B(Kerdn(~B)
,
d :W
n
n
  > Wn_ 1.
Proof.
It is easy to see that where
K p is isomorphic as a Z Trmodule to P A and acts on Tr p Z Tr ~ B P by its left
A (9 (Z w ~ B ) , P
w acts trivially on ~
a c t i o n on
Z w,
p
Since
H ( w ( n ) ( ~ A ) " = H(w(n)(~" Z ) @ A "
and
H(W (n) ~ Z p ~ r ( ~ B) = H(W (n)) ~ B, the r e s u l t f o l l o w s . R e c a l l t h a t if ~r is a n y s u b g r o u p of E
in E, then conjugation by any AEmodule M.
and if W e N(Tr), the n o r m a l i z e r
of 7r
for
"~ defines a h o m o m o r p h i s m
x/,:H(w; M) > H,(,rl M)
~/;:. is the m a p induced on h o m o l o g y f r o m where W is a n y A w  f r e e resolution of A and for
~# ~ y : W ~ M     > ~# : W > W 0 ( *r and
W~M
is any m o r p h i s m w c W.
of A  c o m p l e x e s
i such that ~# (0w) = yoy y#(w)
(It is easy to verify that ~# exists and that ?,:. is independent and of " / # . ) so that Clearly ~;:, 1 if ~ ~ w = since w e m a y then
of the choice of W define ~/#(w) = ~ v
(~#~/)(w~m)
= W@~/m
=w~m
,
we
W and m ~ M.

159

If ~r C P C ~' a n d
y e N(w) t% N ( p ) , t h e n t h e f o l l o w i n g d i a g r a m J,
commutes:
H,(~r; M)
, > H,(p ; M)
J,
H , (~r; M) In fact, if W is anyApfree
> H,(p ; M)
W is also aAwfree resolution
r e s o l u t i o n of A , t h e n
of A,
W~ rr
and the above
M> W~ p M.
diagram
results from
the observation
that
j,
is induced from
In p a r t i c u l a r , Let
j , = j~.y, if y ¢ O. . order p > 2 and let q e Z. Consider
Lemma
1.4.
~ b e c y c l i c of p r i m e Then
j,: H: ¢ (Tr; Z p (q))
(i) (ii) If q If q Proof.
> H~(Zp; Zp(q)). "
is even, j,(ei) = 0 unless is odd, j,(ei) = 0 unless Let Z P in N(w). Define k
i = 2t(pl)
or
i = 2t(pl)  I. or i = (Zt+l)(pi) I. Let Y is an
i = (2t+l)(pi)
generate the multiplicative and define W e ~ P y# : W > W .ki = k~ E by
subgroup of Zp, k p = I. Then yo~ l = ~ k and
operate on P odd permutation
y(i) = ki. by
~/#(ezi ) = k i e g i , .
7#(eii+i)
j =0
a)ezi+l, .
y#(~ei) = y0~
iY#1ei),
(l) q on
0~c ~r
Then y ~ Ep Z P
y # d = dy#
and
y#~y
induces the conjugation
y~< on
H,(~;Zp(q)).
Since
, j,y, = j~,..,hence '
j~,.(e  Y , ei)= 0 for all i. y operates by i ,
(q) and therefore y,(ezi) = (l)qkiezi and y,(eZi+l) = (l)qki+lezi+l p and .
Thus
j,(ezi ) = 0 unless p.
1  (l)qk i = 0 m o d Clearly k i=
j,(eZi+l ) = 0 unless for s o m e
I  (l)qk i+l = 0 m o d t and k iI
i rood p
if and only if i = t(pl) for s o m e t.
rood p if and only if Zi = (2t+l)(pl)
The result
follows easily.

1 6 0

2.
The d e f i n i t i o n a n d e l e m e n t a r y p r o p e r t i e s of the o p e r a t i o n s We n o w d e f i n e a l a r g e a l g e b r a i c c a t e g o r y ~ (p,n) on which the S t e e n r o d
operations will be defined.
Steenrod operations will be obtained for particular
~ ' ( p , n). T h e i n t e r e s t of the i n t e g e r
c a t e g o r i e s of i n t e r e s t b y o b t a i n i n g f u n c t o r s to n
i n t h e f o l l o w i n g d e f i n i t i o n l i e s s o l e i y in the a p p l i c a t i o n s to i t e r a t e d l o o p s p a c e s . n = o® is r e l e v a n t . r be a n i n t e g e r , a n d l e t V b e a AN r
F o r a l l o t h e r k n o w n a p p l i c a t i o n s , o n l y the c a s e Definitions 2.1. ~r b e a s u b g r o u p of ~
r
Let .
A Let
be a c o m m u t a t i v e r i n g , l e t
W be a A ~ r  f r e e r e s o l u t i o n of A, l e t j:W>V
f r e e r e s o l u t i o n of A, a n d i e t Assume that W
O
b e a m o r p h i s m of A ~ r  c o m p l e x e s o v e r A. e .
O
= AT w i t h g e n e r a t o r W a n d V. are pairs
Let
0 Kn<oo

and let ~(~r,n,A)
W (n) a n d V (n) as follows.
d e n o t e the n  s k e l e t o n s of T h e o b j e c t s of ~ ( r r , n , A ) differential Aalgebra and such that
(i)
Define a category (K,@), w h e r e
K is a h o m o t o p y a s s o c i a t i v e
O:w(n)(~K r>
K is a m o r p h i s m of A ~ r  c o m p l e x e s
T h e r e s t r i c t i o n of
@ to
e ~K r o
is
A  h o r n ® t o p i c to the i t e r a t e d p r o d u c t and ~®1 >v(n)®Kr ¢ > K,
K r   > K, a s s o c i a t e d i n ~ m e f i x e d o r d e r ,
(ii)
@ i s A ~ r  h o m o t o p i c to a c o m p o s i t e
w(n)~K r
where
A morphism
¢ is a m o r p h i s m of A N
r
complexes.
~(w,n,A) is a m o r p h i s m of A  c o m p l e x e s
f.(K, 0)> (K',0')
in
f, K > K' such that the diagram
w(n)
® Kr
I ®fr
,
0
....
> K
w ( n ) ® ( K , ) r ........
I
[f 8,
> K,
is A~rhomotopy commutative.
A morphism
f is said to be perfect if
~(I@ fr) = f@, with no homotopy required, and @ ( 7 , n,A) denotes the subc a t e g o r y of ~(%n,A) having the same objects (K, 0) a n d a l l p e r f e c t m o r p h i s m s
between them.
A i s i t s e l f a n o b j e c t of
~ (7, n, A), w i t h

161

@= e @ l : W ( n ) ( ~ A
r
> A r = A , and an o b j e c t
(K,@)e
~(w,n,A)
is s a i d t o b e
unital if K has a t w o  s i d e d h o m o t o p y i d e n t i t y
e s u c h that ~]:A > K, n(1) = e,
is a m o r p h i s m in ~'~(w,n,A). The tensor product of objects (K, @) and (L, O') in ~(~,n,A) is the pair ( K e L , ~ ) , where ~ is the composite 1 @ T @ 1> w(n)@i<r@w(n) @ L r
W (n) @ (K(~ L) r ~ ) U > w ( n ) @ w(n) ~) K r ( ~ L r
e®e, > K ( ~ L
Here U is the evident shuffling i s o m o r p h i s m , >W(~W T ( x ~ y ) = (1) deg x deg y y ~ x (i) and (ii) ,
and ~ : W
is any fixed A ~ r  m o r p h i s m o v e r A; conditions
are clearly satisfied by the pair ( K ~ L , ~ ) .
A n object (K,O) e ~(~r,n,A) is said is a m o r p h i s m in ~(w,n,A). to be the explicit ~ ( p , n)
to be a Caftan object if the product K ~ K > K When
w is cyclic of prime order p, w e agree to choose W W(p,A) of D e f i n i t i o n 1. Z, and we abbreviate
resolution
and
~(1T, n, Zp) to
(~(~,n, Zp) to
~O(p,n). Anobject (K,e)~ C(p,n)
is said to be reduced
rood p if (K,O) is obtained by reduction m o d p from an object (K,~) ~ ~(w,n, Z) such that ~ is a flat Zmodule. W e can now define the Steenrod operations in the homology (K,@) ( ~(p,n). H(K) of an object
Observe that if x e H(K) and 0 < i< n, then e . ~ x p is a well1 K P) ~ ( w ( n ) ~ H(K) p) by L e m m a s H
W W
defined element of H ( w ( n ) ~ (iv) of L e m m a
I
I. 1 and 1.3; here
I. i applies, since
w C E
P
contains only even permutations, and K p for any representative
W
shows that e . ~ x p is represented by e i ~ ape w ( n ) ~ cycle a of x. Definitions Z. 2. L e t (K,O) ( ~ ( p , n ) and let
x e H (K). q
For
0 <_i<_n, If p = Z,
define Di(x ) e Hpq+i(K ) by Di(x ) = O~.,~(ei®x @,:H(W(n)(~wK p) > H(K). ~, define % : Hq(K) > Hq+s(K ) for s <__q+n by the formula (i) Ps(x) = 0 if s < q ; P s ( X ) = D (x) if s _ > q .
sq
If p > Z, define Ps:Hq(K)> Hq+2s,p_l,(I<)~) for Zs(pl)_< q(p1)+n and define

162

~Ps:Hq(K) (ii)
> Hq+Zs(p_l )_ I(K) for 2s(pl) <_q(pl)+n+l if 2 s < q ; 2s < q ; Ps(X)=(1)Sv(q)O(Zs_q)(p_l)(X)
by the formulas if Z s > q a n d if Zs > q, 1 ~(p1), (ml) q
Ps(X)=0
~Ps(S) = 0 i f where or,
~Ps(X) = (  1 ) S v ( q ) D ( z s . q ) ( p _ t ) _ l ( X ) ~= 0or 1, m =
v ( Z j + e ) = (_ 1)J(m: )g , j a n y i n t e g e r ,
equivalently since
( m ' ) 2  (  1 ) m + l m o d p , v(q) = (  1 ) q ( q  t ) m / 2 Ps a n d , if p > Z, t h e ~Ps are defined for all
Observe that, integers
if n = co, t h e ~Ps The
s and that
is a single symbol which is not a priori P
s
r e l a t e d to a n y
Bockstein operation. to h o m o l o g y ;
and
~P
s
are appropriately
defined for applications
a s s h o w n in s e c t i o n 5, t h e a p p r o p r i a t e
formulation for cohomology is
o b t a i n e d b y a s i m p l e c h a n g e of n o t a t i o n . T h e f o l l o w i n g p r o p o s i t i o n c o n t a i n s m o s t of t h e e l e m e n t a r y D.,i Ps' and ~Ps" In particular, if p > Z, it shows that the Ps D. and that i ~P s properties and of t h e
~Ps account s and the
for all nontrivial operations Bockstein
is the composition of P
~ p r o v i d e d that (K,O) Let
i s r e d u c e d m o d p. Di: Hq(K) > Hpq+i(K ).
Proposition 2.3. (i) (ii) (iii) If f : K   > K'
(K, 0) ( ~(p, n) and consider
is a m o r p h i s m
in ~(p,n), then Dif;:.= f~D i .
If i < n, then D. is a h o m o m o r p h i s m . I D o is the pth power operation in the algebra H(K) and if (K, 0) is unital,
then Di(e ) = 0 for i ~ 0, w h e r e (iv) If p > Z and
e e H o ( K ) is the identity.
i < n, then D. = 0 unless either i i = 2t(pl) or i = (2t+l)(p1) i = gt(pl)I or for s o m e t, or for some t.
(a) q is even and (b) q i s o d d a n d (v)
i = (Zt+l)(p1)i then
If (K, 8) i s r e d u c e d m o d p a n d (a) ~Dzi = D2i_l (b) ~D2i+l = Dzi
~ is the Bockstein,
if either p > 2 or q is even, and if p = 2 and q is odd, and
2i < n
2i+i < n.

1 6 3

Proof.
Part (i) is immediate f r o m the definitions and f r o m Z e m m a
I. I (iv),
and p a r t (iii) is i m m e d i a t e f r o m the definitions.
(ii) Let a,b ¢ K q be cycles and define A(a,b) = (a+b) p  ap  b p ~ K p. involving both a's and b's, and ~(a,b) is
a s u m of m o n o m i a l s freely. Let
~r permutes such m o n o m i a l s w give
c c K p be a s u m of m o n o m i a l s w h o s e permutations under of A(a, b) exactly once. andif Then A(a,b) = Nc.
each m o n o m i a l
If i is odd, then = e.•Ncl in and
d(ei+l~C)__ = ei~Nc__ w(n)~ K p, i < n , ~r D.
I
i is even, then d ( T P  Z e i + l @ C ) Thus e.~ZX(a,b)
since TPI = N in Z w. p i < n.
is a b o u n d a r y
therefore (iv)
is a h o m o m o r p h i s m ,
In the notations of Definition/~. i, w e have that O is homotopic to a composite
v ( n ) ~ E Kp ~ > K. Since nothing is changed P b y t e n s o r i n g with two copies of Z (q), this c o m p o s i t e can equally well be w r i t t e n as P
w(n)(q)%KP(q) Let a ( K be a cycle. j~ I >v(n)(q)~E P q Then, bythe definition of KP(q), Therefore, ap is a basis for a Kp(q ) ~ > K.
w(n)~Kp
1 >v(n)~Kp_..>
trivial E subcomplex of KP(q). P v(n)~ For E
if _ J(ei) = d(f) in
Z (q)= v(n)(q)~ E Z , then d ( f ~ a p) = J(ei)(~aP in v(n)(q)(~ E KP(q). p p P P P i < n, j induces j.: Hi(w; Zp(q)) ~ > Hi(Ep; Zp(q)), and the desired conclusion I. 4.
n o w follows immediately f r o m L e m m a
(v)
Let (K, O) be the mod p r e d u c t i o n of (K,~). up
,
Let a ~ ~
q
s a t i s f y d(a) = pb.
A n easy calculation demonstrates that, in K d(a p) = p N b a p1 d(a 2) = 2Tab
if p > 2 or q is even, q is odd.
if p = 2 and
In the f o r m e r case, if 2i< n, then, in W(p, z)(n)~Ir~P , d(e2i ~ a p ) = ezii ~ N a p + P e z i ~ NbaPI Z  p[ezi.l~aP since TPl= N rood p. + d(TPgezi+1~baPl)] mod p , In the latter case, if 2i+I <n, then  2 [ e z i ~ a 2  d(egi+z~ab)] rood 4.
d ( e z i + l ~ a 2) = e z i ~ r a Z  2 e Z i + l ~ T a b

164

T h u s , in H(W(p, Zp)(n)~TrKT,.,~t if K is the rood p r e d u c t i o n of a, t h e n [3{egi@~P } = { e Z i _ l ~ g P } Since incase (a)and p{eZi+l@g2} 2} in c a s e (b). = {ezi@a > K , ~8,.. = @,.~, , ,
@ is the rood p r e d u c t i o n of the m a p @: W(p, z ) ( n ) o ~rK p
and the r e s u l t f o l l o w s . Of c o u r s e , (i) and (ii) i m p l y t h a t the P
s
and ~P
s
are natural homomorphisms
(except, if n < co, for the last operation). corollary of part (iii). Corollary 2.4. Let (K,O) e ~'~(p,n).
A check of constants gives the following
Then
P q ( x ) = x p if p = 2 and X e Hq(K)
or if p > 2 and x e HZq(K ). If (K, 8) is unital, then Ps(e) = 0 for s J 0. T h e i m p l i c a t i o n s of (iv) and (v) f o r the Ps and ~Ps a r e c l e a r if p > 2. If
p = 2, w e have the following corollary of (v). C0,,r,ollary 2.5. If (K,@) e ~'(2,CO) is r e d u c e d rood 2, t h e n ~Ps+l = SPs.
The following result is the external Cartan formula.
Proposition 2.6.
x e H (K) 9 and
Let
(K,@) and Consider
(L,@')
be o b j e c t s of
~(p,n),
Let
y • H (L). r
x@y • H(K)@H(L):
H(K@L).
(i) (ii)
If p = g, t h e n
Di(x@y) =
~ Dj(x)@Dk(Y ) j+k= i
for i<_n .
If p > 2, then D z i ( x ~ y ) = (I) m q r
~ D2j(x)~ngk(y ) j+k= i
for 2i<_n, and
D Dgi+l(x@y) = (1) m q rj+k=i~(gj~rl(x)~Dzk(Y) + (1)qDgj(x)(~D2k+l(Y)) 2i+I <__n.
for
Proof. Since
By L e m m a s
1. 1 and 1 . 3 , we m a y w o r k in w(n)(~Tr[H(K)(~H(L)]P. O , ( e i ~, x ~ y ) P ) , ( , by m e a n s The re
~r operates trivially on ( x ~ y ) p, w e m a y compute
of the induced coproduct on w ( n ) ~ Z p , sult follows by direct calculation f r o m
as given in (Z) of Definition I. 2.
0,(%® (x® y)P) = (0,®%)(1®
T®
1)(¢®U)(ei®(x®y)P).

165

A trivial verification of constants, together with part (iv) of Proposition 2.3,
yields the following corollary.
Corollary g.7. and y ~ Hr(L ).
Let (K, 0) and (L, 0') be objects of
~(p,n).
Let x e H q (K)
Then Ps(X(~y)= ~ Pi(x)(~Pj(y) and, if p > Z, i+j =s
~P+I(xOY) = ~ (OPi+l(X) OPj(Y) +(1)qPi(x)O~Pj+l(Y)) • i+j= s Of course, if (K, 0) is a Cartan object in ~ ( p , n ) , then the corollary and the
naturality of the operations imply that the P
S
and, if p > 2, the
~P
S
on
H(l~)
satisfy the internal Cartan formulas
:
>_5
i+j : s
Pi(x)pj(y)
and
(1)
~Ps+l (xy) = ~ (~Pi+I(X)Pj (y) + (1)degxp'(x)~P:+l I ( y )2 ) i+j= s

1 6 6

3.
Chain
tevel
operations,
suspension,
and spectral
sequences
In t h i s s e c t i o n , w e d e f i n e c h a i n l e v e l S t e e n r o d o p e r a t i o n s a n d u s e t h e m to p r o v e that the homology operations c o m m u t e wSth s u s p e n s i o n . The chain level operations in s p e c t r a l sequences
c a n a l s o b e u s e d to s t u d y t h e b e h a v i o r of S t e e n r o d o p e r a t i o n s a n d , in p a r t i c u l a r , theorem. Theorem 3.1. Let (K, O) e ~ ( p , co).
w e s h a l l p r o v e a g e n e r a l v e r s i o n of t h e K u d o t r a n s g r e s s i o n
Then there exist functions
P :K >K s q q+s
if p=
2 and
P :K   > K s q q+Zs(pl)
and
~ P s : K q > 
Kq+zs(pl)  1
if p > 2 w h i c h s a t i s f y t h e f o l l o w i n g p r o p e r t i e s .
(i) (ii)
dPs = P d s
and
d~Ps=~Ps
d x e H(K), then Ps(a) and ~Ps(X). in ~3(p, co), so that fO = O'(l®fP), ~Ps(a) are
If a i s a cycle which represents cycles which represent Ps(x) and
(iii)
If f:(K,O)> (K',O') is a m o r p h i s m
then Proof. fP
S
= P f and
S
fBP = ~Pf.
~S
Let
a ( Kq a n d w r i t e
b = d(a) e Kq_ 1.
In t h e c a s e
p = 2, d e f i n e
(1)
P s ( a ) = 0(c), w h e r e
c = es_q+l•b@a
+ es_qOaOa
e WOlf KZ" that p > Z. K p. Let (a,b)
The verification
of (i), ( i i ) , a n d ( i i i )
is t r i v i a l . a and
Thus assume b, s o t h a t s(b) = a. 1 on
denote the subcomplex of K with basis s:(a,b)> (a,b). Let ( a , b ) , of d e g r e e o n e , b y S = 1Pl~s on ( a , b ) p.
(a,b)Pc Then
Define
s(a) = 0 and Then
ds + sd = 1 on S is given
dS + S d =
(a,b) p and
explicitly by S(ea) = 0 and
S(eb) = (I) deg eea
for e e (a,b) pI.
Define t.e (a,b)p i
for 0 <_ i <_ p by the inductive formula
(Z) t
o =
be; t 1
bp  1 = S(¢~1 = a', tZk
t2k_l t2k_l ) ; tZk+l = S(Nt2k ) 
Since (3)
dS + S d = I, an easy calculation demonstrates that d(tl) = to , d(tzk) = (~I 1)tZk_1 and d(t2k+1) = Ntzk , l<__k<_m.
A straightforward induction, which uses the explicit formula for S , yields

I 6 7

(4)
tZk = E I
(  1 ) k q ( k  1 ) ' b i l a Z b i Z a Z ' ' " bikaZ , 1 _< k_< m ,
summed
over all
ktuples
(5)
I = (iI ..... ik) such that
•
E
o
ij = pZk ; and
a, 0 < _ k < m, s u m m e d o v e r ij = p  l  Z k . Now let following ali
tZk+l = E (  1 ) k q k l b I (k+l)tuples I = (i 1 .
it
a
2
..
, hlkaZblk+l
. . . .
ik+l)
such that ~
In p a r t i c u l a r ,
tp = t Z m + l = (1)mq'm~ a p ( s i n c e e a c h i.j = 0). and d e f i n e c h a i n s
1
j = (Zsq+l)(p1) formulas
c and
c'
in W ~
rr
Np b y t h e
(where, by convention,
m
e. = 0 if i < 0):
m
(6)
c = k=OE (I
ITI
)kej_Zk~
tZk+l
 E
k=l
m
(1)k ej+l_Zk~(a1

1)P'Z
tZk;
(7)
c' = E(1) k=0
k
ej_l.ZkOtZk+l
+ E (1)k ej_zk~tZk k=l 1 . 2 and (3), g i v e s
T h e n an e a s y c o m p u t a t i o n ,
which uses Definitim d(c') = ej_l~)bP
(8)
d(c) = e p@Jb .
and
(j = (ZS  q + l ) ( p  1 ) ) are that Nt = 0, that and t h a t
In c a l c u l a t i n g ae.~t
1 1
d(c), the s a l i e n t o b s e r v a t i o n s
P
= e.~alt
f o r t e K p b y the v e r y d e f i n i t i o n of a t e n s o r p r o d u c t , Z w. F i n a l l y , P and define ~ P s ( a ) = (  1 ) S u ( q  1 ) 0 ( e ') . i < p and t
(a  1  1) p1 = N in (9) If
Ps(a) = (  1 ) S v ( q  1 ) O ( e ) a is a c y c l e , b = 0, t h e n
t. = 0 f o r 1
p
= (  1 ) m q m ' a p, h e n c e
(i0)
c= (l)m(q+l)m'e(zs_q)(p_l)@a
p and
c' = (1)m(cl+l)m.'e(zs_q)(p_l)_l@a p
It is easy to verify that
w(q) = (l)m(q+l)m' v(ql) and now (ii) is obvious from
be K ql"
(9) a n d (10) a n d (i) f o l l o w s f r o m (8), (9), and (10), a p p l i e d to the c y c l e P a r t (iii) is i m m e d i a t e The remaining proof. f r o m 49). of the t h e o r e m
r e s u l t s of this s e c t i o n a r e c o r o l l a r i e s
and its
We f i r s t define and s t u d y a v e r y g e n e r a l notion of s u s p e n s i o n .

168

Definition. Let such that where
that o
f:K'> K
and
g:K> K"
be m o r p h i s m s
of A  c o m p l e x e s ~{b'} = {g(a)},
gf = O.
Define
0: K e r f,.~> C o k e r g. by the formula d(a) = f(b') in K.
b'
represents
{b'} ~ K e r f,~ and w e call
that the P
S
It is trivial to verify
is welldefined, and
We can now prove
o ~
the suspension.
commute with suspension. We remark that
if n = oo, the hypotheses of the next t h e o r e m simplify to the requirement that and g be m o r p h i s m s in ~(p, oo) s u c h that gf = O. For n < oo, the stated
f
hypotheses arise in practice in the study of iterated loop spaces. Theorem3.3. K
Let
(K',8') E #(p,n+l)
K and
and let (K", @") e
K"
C(p,n).
Let
be a Z  c o m p l e x and let f:K' m > P gf = O.
g:K> "~
be m o r p h i s m s
of c o m 
plexes such that
Define a s u b c o m p l e x
K of W
(n+l)
~
Kp
by
= w ( n + l ) ~ f(K,)p + ~ ( n + l ) ~
f(K,)p1 ~ K + w ( n ) ~ ~P ,

w h e r e ~(n+l) = w(n)~D Zpen+ 1 (that is, en+l e w ( n + l ) i <_ i < p).
but ~ en+ 1 ¢
i
~ (n+l)
for
Suppose g i v e n a T r  m o r p h i s m 0: ~   > K ( w h e r e , by c o n v e n t i o n , Tr does
not a c t on en+ 1 ~ f(K') p 1 0 K) such that the following d i a g r a m is c o m m u t a t i v e :
w(n+l) ~ (x,)P l(~f p > ~
i (~ gP
> w(n) (~ (K,,)p
K'
f
> K
g
>
K"
(Here closed Coker
gf = under g,.
0 ensures the P
S
that and
(l~gP)(~) C w(n)~(K") p.) Observe that Ker f, is
S
~P
f,.
and that there are welldefined induced
oP(x) = Pc(x)
P
S
on
Let x
e Ker
Then
and
c~Ps(X)=~PsC(X)
when
ever
P(x)
and
are
defined.
Proof.
Let
deg(x)= qI
and let b' E I<' represent
x.
Let
b = f(b') and
let
d(a)
= b in K, so that
or
g(a)
represents
g(x).
The h y p o t h e s i s g u a r a n t e e s that
Ps(a)
if S iS sueh that Ps(X) or ~Ps(a)
~Ps(X)
is defined, then the chain level operation
constructed in the previous proof is also defined.
Of course, this is

169

clear if n = co; if n <co, w e need only verify that all elements involved in the definition of Ps(a) or operation ~Ps(a) are present in K. F o r example, if p = 2, the last
Ps(x) occurs for s = q + n a + e ~ a ( ~ a, and
n
and then Ps(a)= @(c), w h e r e
~v
c = en+ l ~ b ~
c is indeed in K.
N o w our d i a g r a m ensures
that fPs(b')= Pf(b'), hence
fPs(b') = dPs(a), and that gPs(a)= Psg(a). (r~pS(x) =
OPs(X) = Ps0(x) follows f r o m the definition of o% and the proof that ~Ps~(X) is equally simple.
Note that if p > 2 and all objects are reduced rood p, then which is consistent with the theorem. that 0~Ps(x) = 0 if p > 2 and
0~ = ~o,
The t h e o r e m implies that 0(x = 0 and p) is reduced rood p, the
deg(x) = 2sl; if (K",O") The operation
latter statement b e c o m e s
~0(x) = 0. p
~Ps(x), deg(x) = 2s  i, plays
a special role in m a n y applications; the following very useful technical result about about this operation is k n o w n as the Kudo transgression theorem. It applies to the
D y e r  L a s h o f operations in the h o m o l o g y Serre spectral sequence of the pathspace fibration ~ n x > IzhqnIx > ~nIx, to the Steenrod operations in the c o h o m o l o g y Serre spectral sequence of a fibration F > E > B (with K' > K > K" being
C":"(B) > C;:'(E)> C '(F), graded by subscripts) and to the spectral sequence of Adams [I, p. Zl0] for c o c o m m u t a t i v e Hopf algebras. Theorem K 3.4. Assume, in addition to the hypotheses of T h e o r e m that Ho(K' ) = Z
' p
3.3, that
has an increasing titration {FiK } of complexes u: K ~
= H
o
(K"), and that there is
a morphism
f(I<')> K
such that either if i>__0,
(i)
K',K, and K" f(K') C F o K
are positively graded, F.I< = 0 if i < 0, F.K. = X
i 1 1
, ~ ( F i K ~ f(K')) C F i K
, and f and g induce i s o m o r p h i s m s w induces a m o r p h i s m
EZf:H.(K')>EZ.K 3 oj Z 2
and EZg:K> Hi(K" ) and Z
EZ'rr: Eij K ~ Eok K > El, j+k K such that the c o m p o s i t e m o r p h i s m EZrr[(EZg) 1 ~ EZf]: Hi(K")(~ Hj(K') > E..2 K is an i s o m o r p h i s m  1j or
 170

(ii)
K',K,
a n d K"
are negatively graded,
F . K = K if i > 0, F i _ I K i = 0 i f i < 0, 1 and f and g induce isomorphisms
f(K~) C F i K ,
=(FiK®f(Kj))CF i + j K ,
2
oj
Ezf:Hi(K,)__> ~z K and ~ : ~
i, o
EZw:
K> H (K") and . induces a morphism
J
EZ.. K ~ Eko K > E Z j 2 lJ i+k, .K such that the composite m o r p h i s m
EZw[(EEg) I 8 EZf]: H / K " ) e Hi(K' ) > EZijE is an isomorphism. Let 7 be the transgression, ~ = dt:EttoK> E t0,t_lK m = d l.t:Elt K _ >
ot
in (i) and
E 1t
tl,o
K
(t < 0) in (ii). Then m is the inverse additive
relationto cr, and if y ~ Hq(K") transgresses to x • Hq_l(K'), then Ps(y) and if p > Z, ~Ps(y ) transgresses to Ps(X) and ~P(x), whenever the operations transgresses to
are defined. ~P(x)
Moreover, if p > 2 and q = 2s, then ypi ~ x
(that is, dq(p_l)(ypi ~ x ) = ~Ps(x) in case (i) and
dl_q(p_l)(ypl~x) = ~Ps(X) in case (ii) provided that
(iii) (iv)
if a.j c F i . K , t h e n J The restriction
0(ek(~a I ~...
~ap)
¢ FiK , i = ~
ij+k,
and
of @ to
;:'~
e ~K pl(~f(K') o
induces a morphism
~O
E20:(E~oK)PI~E 2 K> E2K
• 0
in (i)and E20:(EoZ.K)PI~E2 K> E Z K
in (ii) such t h a t
EZo =
Eaw[(EZglI@(EZg)pI~ w h e r e I],
By the definition of the differentials in the
~:H(K") pI >H(E") is the iterated product. Proof. Let y ~ Hq(K").
spectral sequence of a filtered complex, T(y) is defined if and only if y is
represented by g(a) for some a • K q such that d(a) = f ( b ' ) for some cycle
b' • K' ql' and then x = T(y) = {b'}. Thus the first statement follows from the properties of the chains Ps(a) and ~Ps(a). F o r the second statement, consider as above, it suffices to
~%(a), q = 2s. Since d~P(a) = ~Psf(b'), a and b' prove that ~Ps(a) represents ypI e x in EZK.
~Ps(a) =  m v.o(c') by (9) of the v(Zsl) = (l)S'im~.
proof of T h e o r e m 3. 1 and the observation that v(ql) = In the definition (7) of c', the term with k = m
in the first s u m involves e_l (since

171

q = 2s
implies j = pl) and is therefore zero.
The term with k = m
in the
m  1 aZiba2(m.i ) m second s u m is (I) e o ~ t p _ I where, by (4), tp. 1 = ( m  t ) : ~ i=O b = d(a). It is easy to see that mI m ziI ~.a2iba 2(m'i) = P(a)aPlb , where i=0 i=l a n d d i r e c t c a l c u l a t i o n s h o w s t h a t P(~) = m + Q(a), in Z w, w h e r e m P Q(~) = ~ j ( a 2j + aZJ+l). Let c" = (l)m(ml) ' e l ~ Q ( a ) a p  l ( ~ b ¢ W @ K p. Then j=l e'  d(e") = (l)mm:e ~ a P  l b plus a linear combination of terms e i ~ g such o that g has i+l factors b and pi1 factors a. Condition (iii)ensures that
each O(ei~ g) has lower filtrationthan does 0(eo~ aP'Ib) and condition (iv) ensures that 8 ( e o ~ a P  l b ) represents y p  l ~ x m: e(c'  d(c")) = 8(eo~ aP'ib) represent is c o m p l e t e . The following proposition gives a general prescription for the study of Steenrod operations in spectral sequences; it will be useful in the study of the cohomology of the Steenrod algebra in [18]. In the applications, the determination of the function f is often quite difficult and depends on h o w the given O was constructed. Proposition 3.5. Let (K, O) be an increasingly filtered object of ~ (p, co) e EZK. Since ~Ps(a) and
the same element of E2K, the proof
Suppose given a function f(i,j,k) such that (i) If ate FitKit+Jt , where O(ek@al@... (ii) (iii) Let ~ i = i and ~ t Ji = j' then
@ap) e Ff(i,j,k)Ki+j+ k ; and r >_ i. y) e and
f(i,j,k) > f(ir,j+r I, k+ I), r>_l;
f(i,j,k)> r + f(ipr,j+p(rl),k+pl),
r y e E.K. tJ
Then there exist elements such that dtPs(g)= psdr(g)
Et k
K and, if p > 2 , where
~Ps(y) e Eti,~,
dt ' ~ P s ( y ) =  ~ P s d r ( y ) ,

172

(iv)
If p = 2, t h e n
k = f(Zi, 2j, s  i  j ) ,
g = i+j+sk,
and
t = kf(2i2r, 2j+2(rl),s+ (v) If p > 2, then
lij) , g = i+j+2s(pl)k , and
k = f(pi, pj,(Zsij)(pl)),
t = k  f ( p i  p r , pj+p(rl), (vi) If p > 2, then
(2s+l ij)(pl)). g' = i + j + 2 s ( p  1 )  l  k , and
k' = f(pi, pj,(Zsij)(p1)l),
t' = k'  f(pipr, pj+p(rl),
( 2 s + I ij)(p1) I).
Proof.
Let
a ~ FiKi+ j represent %(a) constructed enGaP
y
and let b = d(a) ~ Fi_rKi+j_ 1 3.1. B y (ii), all s u m m a n d s n) have lower of
C o n s i d e r the chain Ps(a)
in T h e o r e m
other than that involving k
(for the appropriate
filtration than
and, by (i), O(e n O a p) h a s f i l t r a t i o n
k.
Since
dPs(b), w h e r e P(y) follows.
Ps(b) ~ F k _ t K b y (i)
T h e proof for
and
kt
~ r by (iii), the statement about
~Ps(y ) is similar.
4.
The Aden
relations relations are valid for the Steenrod operations The general algeare able to exploit a relations, and this since
W e here s h o w that the A d e n in H(K) if (K,O) ~
~ ( p , co) satisfies certain hypotheses. in the proof. We
braic context is distinctly advantageous trick (Lemma 4.3) used by A d e n
to prove the classical A d e n
trick w o u l d not be available in a categorical approach to Steenrod operations it depends on the usage of objects of C(P,°°) which are not present in m a n y
categories of interest, such as infinite loop spaces and c o c o m m u t a t i v e bras.
Hopf alge
We require
proof.
s o m e n o t a t i o n s a n d d e f i n i t i o n s b e f o r e we c a n p r o c e e d to t h e
2 act as permutations on {(i,j) I l ~ i < _ p , l<__j<__p}. P in ~n 2 by letting ~(i,j) = (i+ 1 j). Define ~. ~ Z 1 <i < p, by ' 1 P pZ'  Let ~
Embed

173

ai(i,j) = (i,j+l) 13(i,j) = ( i , j + l ) .
and Then
ozi(k,j) = ( k , j ) f o r k / i ,
and let
~ = ~l..C~p
so that
(I) Let Let
ozozi = ~i+l ~ ; ~iozj = ajoti ; a n d oz. generate
I
c~ = ~oz . v are cyclic of o r d e r p.
~.
I
and
~
generate v, so that ~. a n d i
0 = ~v
and let
T be g e n e r a t e d
by the
oe. a n d 1
o~. T h e n 0  C T , T by
is a rr.
p  S y l o w s u b g r o u p of 2~p2,
and
T i s a s p l i t e x t e n s i o n of ~ l . . . ~ p respectively,
Let
W 1 = W and
Z P . Let
W 2 = W regarded, v
a s Trfree a n d v  f r e e
resolutions of andlet of Z. P If M o
operate trivially on W1, let
WI~W Z. Then WI~W
w o p e r a t e t r i v i a l l y on WZ,
g is a 0  f r e e resolution
operate diagonally o n
is a n y v  m o d u l e ,
let T operate o n
M p by letting ~ operate b y cyclic M as does ~. Let ~. i
p e r m u t a t i o n and b y letting ~i operate on the operate trivially on W I. on Then
ith factor
T operates on W l and w e let T operate diagonally
W 1 ( ~ M p. Let
In p a r t i c u l a r ,
W 1 ~W?
is then a We l e t
K be any Zpcomplex.
Tfree resolution of Z . P 2 2~pZ o p e r a t e on K t~ b y p e r m u t a t i o n s
w i t h t h e ( i , j )  t h f a c t o r K b e i n g t h e j  t h f a c t o r K in t h e i  t h f a c t o r K p o f 2 Kp = (KP)P. We l e t v o p e r a t e in t h e s t a n d a r d f a s h i o n on W Z ( ~ K p (13 a c t i n g as cyclic permutation of T on on KP)o By the previous paragraph, this fixes an operation
W I~(wZ~Kp)p. Y be any over 2~ _  f r e e pZ Z . P resoiutionof Z p and let w:W I~W P z >Y be
Let any
of w
Tmorphism
are
w exists since
Y is acyclic,
a n d a n y two c h o i c e s
Thomotopic.
W i t h these notations, w e h a v e the following definition.
Definition 4. I. object if there exists a ing d i a g r a m
L e t (K,@) e ~ ( p , n ) .
~ P 2_morphism commutative:
We s a y t h a t ( K , @ )
> K
i s an A d e m
~ : y ( n ) ~ K p2
such that the follow
is T  h o m o t o p y

1 7 4

(WI @ W } ) ( n ) @ Kp 2
w@l
> y ( n ) @ Kp z K
Wl(n )
(wJn)~
Kp)p
11~ e p
> Wl(n)l ~ Kp
Here
U
is the evident shuffle map, and is clearly a T  m o r p h i s m (~. p2
acts
t r i v i a l l y on K and
a.1 a c t s t r i v i a l l y on W l ( n ) ~ KP).
The relations obtained ~(p,n), pro
F o r clarity, w e only treat the case n = co below. will be valid f o r o p e r a t i o n s
on H (K), (K, 0) an A d e m q
object of
vided that n is sufficiently large relative to q.
We f i r s t show t h a t the t e n s o r p r o d u c t of A d e m o b j e c t s is an A d e m o b j e c t and t h e n u s e this f a c t to s h o w t h a t a n y r e l a t i o n s v a l i d on H objects q. Lemma 4.2. If (I<,0) and (L, 0') a r e A d e m o b j e c t s of ~ ( p , co), then (K, 0) and s u i t a b l e qi will n e c e s s a r i l y (K) f o r all A d e m qi be v a l i d on H (K) f o r a r b i t r a r y q
( @L , ~ ) is an A d e m object of K
Proof. and 6'
¢ ( p , oo). By h y p o t h e s i s , we a r e g i v e n
"~ is as d e f i n e d in D e f i n i t i o n s 2. 1. (K,~) and (K,B')
such that
a r e o b j e c t s of
~
m a y define
a s in D e f i n i t i o n s 2. 1 so t h a t
(K~
L, 6)
~'(]E Z, co, Zp), hence we P (~]p2' co, Zp). We must Thomotopy commutative,
s h o w t h a t the d i a g r a m of D e f i n i t i o n 4 . 1 , f o r
K ~ L, is
and this follows easily f r o m a simple chase of a large diagram and the definition
of ~ and g.
and Y ~ Y V
The crueial point is the observation that since W, ® W 2
is T free
is acyclic, the following diagram is Thomotopy commutative where ~ :Y >Y ~ Y is any given ~p2Coproduct:
is the evident shuffle and

175

w
iw®w
.... ~ > Y@Y denote the free associative algebra generated by {PslS ¢ Z} Let JpC Fp d e n o t e the t w o  s i d e d i d e a l c o n s i s t i n g x • H(K) and all Adem objects a e F P such that ax = 0 f o r a l l
y Let F
P
a n d , if p > 2, {~P_ls s e Z}. of a l l e l e m e n t s
(K, 8) e ~ (p, oo). Let Bp = Fp/Jp.
Bp
is a universal Steenrod algebra.
Both
the classical Steenrod algebra and the DyerLashof algebra [17] are quotients of B . P
L e m m a 4. 3. s e q u e n c e of i n t e g e r s . objects Let a • F . p Let { q i i i > 0} be a s t r i c t l y d e c r e a s i n g i > 0, a n d a l l A d e m
Suppose that Then
a x = 0 f o r a l l x e Hq (K),
t
(K, 0) ¢ ~ (p, oo). Proof. Let
a • J . P
K be an A d e m object in Choose ( L r , Or) • r < 0
~(p, co) a n d l e t
q + r = q.
1
x•
H (K). q i.
We There
must prove that
ax = 0.
such that
for some
exists an Adem object
~ (p, co) a n d a c l a s s
y • Hr(Lr)
such that
Po(y) = y, Ps(y) = 0 for s ~ 0, and ~Ps(y) = 0 for all s. Such an object can easily be constructed explicitly, but it is quicker to appeal to the results of section 8, which show that the singular cochains of a (r)sphere, graded by nonpositive subscripts, provide such an object. N o w by the previous l e m m a .
a(x@y) = ax@y.
(K~Lr,~)
is an A d e m object of ~(p,oo)
B y the external Cartan formula, Corollary 2.7,
, a(x@y) = 0 and therefore a x = 0,
Since x@y e Hqi(K~Lr)
a s w a s to be s h o w n .
The A d e m relations will be proven by chosing the diagram of Definition 4.1,
a n d we s h a l l n e e d s o m e i n f o r m a t i o n a b o u t the h o m o l o g y of ~, 7, a n d ~:W I~W 2> WI~W 7 be a 0   m o r p h i s m o v e r 2 :E ~. Let
Zp.
Define N ( Ep2
by
~(i,j) = (j,i). Observe that ~
= 1 and ~ a =
~'~. For
q • Z, conjugation by

176

gives a commutative diagram
~#
W~
G~(~; Z p ( q ) ) >
H.(.;, z p (q))
> H,~,(z z " z p(q))
H.(¢; Z p ( q ) )
Thus
> H.(T; Z p (q)) ,
>H=(G Z; Z p (q)) ," P
Nz~ and ~ . Note that
w:,(~, @,N.~) = 0. The following l e m m a s compute
H,(T; Zp(q)) = H,(T; Zp) since T contains only even p e r m u t a t i o n s .
Lemma Proof. 4.4. ~.,. is given on Ii,~(¢;, Zp(q)) by ~(ei~, ej) = (l)ij+mqe.~e..3 l W l~V Z by the formula
Define "~# : W I ~ W Z  >
"f#(~kei@~f ej) = (_l)iJ~ej~ ~kei. Then d'f# = N#d and X#(~x)= (Np'fl)f#(x) for ~ ~ 0 and xE W I ~ W 2. Thus
,~#~:
(Wl@Wz)~Zp(q)>
(W I ~ W Z ) ~ Z p ( q )
induces
N,.
Sincethe signor
"f is (I)m, N.I = (I)m q Before computing
in Zp(q), and the result follows. @~=. we fix notations concerning binomial coefficients.
Notations 4.5.
j >__0
Let i and j be i n t e g e r s .
Define (i,j) = ( i + j ) ' / i ' j '
if i k 0 and
(0' = I) and define (i,j) = 0 if i< 0 or j < 0. Recall that if i >__0 and i= ~ akP k and j = ~ b k, then kp
j > 0 have padic expansions
(i,j)  U (ak,b k) rood p° Clearly (ak, bk) ~ 0 rood p if and only if a k + b k < p, k k hence (i,j) ~ 0 rood p if and only if ~ ( a k + bk)p is the padic expansion
of i + j . L e m m a 4.6. H,~.,~.(T; Zp) = H (~r;Hl,(v,Zp) p) and ~ , : H , ( ~ ; Z p )   > H~.,..(T;Zp)
is given by the following formulas (with sums taken over the integers). (i) If p
=
Z, ~,(er~es)
=
~(k,s_Zk)er+zk_s k
~ es_ k , . Z and
(ii)
If p > Z, ~ , = . ( e r ~ e s ) =
~.(1)kv(s)(k,[s/Z]pk) k
er~2pk_s)(p_i)~esP_2k(p_l)
6(r)8(s I) .~(l)k~sl)(k, IS'i/z]  pk)e r+p +(Zpk  s)(p l)~ePzk(p I)i ' k

1 7 7

where
v(Zj +E) = (1)J(m~)~
Proof. Let W Z = W 2 %
and 6(Zj+~) =E , j any integer,
Z
~= 0 or 1.
P
= H•(w ; Z ). By the definition of the action of P
•
on
and the f i r s t part follows. ~,
Of c o u r s e ,
P P HI~(T; Zp) is now computed by L e m m a 1.3. Let
Z
could be computed d i r e c t l y , but it is s i m p l e r to use topology.
K(Zp, 1) = E/v , w h e r e v o p e r a t e s p r o p e r l y o n t h e acyctic space E, so that, by C . ( E ) = Z v ~ C , ( E / v ) , with Z coefficients. Let D:E~'~ E p be ,~ p P the iterated diagonal. Then 1 ~ D : W I ~ C,(E) > W I ~C:~(E p) is a 0morphism,
and we shall obtain a ~  m o r p h i s m Let d = ~: W 1 ~ C.(E p) >W 1 ~ C~(E) p in L e m m a v  m o r p h i s m over Z p . Since 7. i.
[14, IV l l ] ,
~(I ~)D) and let f:W Z ~ > C,(E) be a ~
W 1 ~ W Z is 0~free and W I ~ C ~ ; ( E ) p is acyclic (and Tfree, with the evident
raction), the following diagram is ~homotopy commutative.
W1]W2
lef W 1 (~ C , ( E ) Therefore
¢
> WliWPz
l~f p
d
> W1 @ C , ( E ) p
~, = d~: H:~(cr;Zp) > H,(T; Zp), and this can clearly be computed from d : W l ( ~ r C ;.(E/v) > W l ~ rG:..(E/v)P. We shall prove the
the quotient m a p
following f o r m u l a s in P r o p o s i t i o n 9 . 1 .
(a) (b) If p = Z, d,(er~ e s) = ~ er+2k_s ~P,k(es)Z • and ~,. ;,., k If p > Z, d;~(er~es ) = ~(l)kv(s)er+(Zpk_sl(p_l)~P,k(es )p k

6(r) ~~.(l)kv(sI )er+p+(Zpk_sl(p.l) e p.k ~(eslP . k
Here the p~:~k are the duals to the Steenrod operations in H * (K( Zp, i); Zp) = H (v; Zp). The latter operations satisfy pO = I and the internal Cartan formula, hence, by (1.2), if w t is dual to et, then
(c)
If p = Z, pk(wt) = (k,tk)Wk+ t, hence
pk(es),:, = (k, s2k)es_ k ; and

1 7 8

(d)
If p > Z,
pk(wt) : (k, [t/Z]  k)wt+zk(p_l)
, hence
P;k(es) = (k, [s/Z]  pk)es_zk(p_l )
C o m b i n i n g (a) and (c), w e o b t a i n (i). we o b t a i n (ii). Theorem 4.7.
Combining (b),(d),
and
~(ei) = 6 ( i  1 ) e i _ l ,
T h e f o l l o w i n g r e l a t i o n s a m o n g the P ~ ( p , oo).
S
and
~P
are valid
on all h o m o l o g y c l a s s e s of all A d e m o b j e c t s in
(i) (ii)
If p
=
z
and
a > Zb,
PaPb = Z(Zia,abil)Pa+b_iPi i i
p > 2 and
a > pb, PaPb = ~ (  1 ) a + i ( p i  a , a  ( p  1 ) b  i  1 ) P a + b _ i P i
and
~PaPb = E (  1 ) a + i ( p i  a , a  ( p  1 ) b  i  1 ) ~ P a + b _ i P i i
Pa~Pb = E(l)a+i(pia,a i  (pl)bi)~Pa+b_iPi
(iii)
If p > 2 and a > pb,
 E(1)a+i(pi
1
 al, a(p1)bi)Pa+b_i~Pi
and
~Pa~Pb =  E (  l ) a + i ( p i i  al,a  (pl)bi)~Pa+b_i~Pi
Proof.
Note f i r s t t h a t the s e c o n d r e l a t i o n s of (ii) and (iii) a r e i m p l i e d by
the f i r s t f o r o b j e c t s w h i c h a r e r e d u c e d m o d p, but a r e l o g i c a l l y i n d e p e n d e n t in o u r general setting. Let (K,O) be an A d e m o b j e c t in C ( p , oo) and l e t x ¢ H (K). q
D e f i n i t i o n 4. 1 i m p l i e s that we h a v e a Z  h o m o t o p y c o m m u t a t i v e P (WI~W7)~ KPZ(q) w~l > Y~E KPZ(q)
diagram
I® u
P
K(q)
W ®=(Wz® KP)P(q) l®0P WI® KP(q)/~e
I Z Since xp is E pZ Z invariant in K p (q), w e have, for all Z Z r and s,
(a)
~,:,(w®l)*(er ® esP®XP ) = ~*(w,(er® asp)®xp
)"

1 7 9

In the other direction,
U
introduces the sign (i) m q s Z
and w e have
(b) Since (c)
0*(l~oP)*(l{~U)~(er~eP~xPs w,#, = w , # , ~/,, L e m m a
) = (1)mqsDrDs(X)
"
4.4 gives the formula er).
w,#;:.(er ~ es) = (1) rs +mqw,#,(es~
Combining formulas
(a) a n d ( c ) . w e o b t a i n t h e f o r m u l a 2 Z
(d)
~,(w®l),(L(er®es)®xP ) =(1)rs +mq<:,(w® 1),(#.:.(es®er)®xP ).
and these relations are
In v i e w of (b), (d) g i v e s r e l a t i o n s on i t e r a t e d o p e r a t i o n s , explicit since In a l l p a r t s , ~, is known.
W e p r o v e t h e t h r e e p a r t s of t h e t h e o r e m
successiveIy
the statements
about binomial coefficients are verified by writing out in
t h e p  a d i c e x p a n s i o n s o f t h e r e l e v a n t i n t e g e r s a n d a p p e a l i n g to t h e r e m a r k s Notations 4.5. (i) {e) By (b) a n d (d), L e m m a 4.6 implies the formula = ~(~,rZl)Ds+zl_rDr_~(x) ~ r = aZq and and . s = bq
~(k,sZk)Dr+Zk_sDs_k(X) k (e) i s v a l i d f o r a l l
Formula fixed
r a n d s, a n d w e s e t to
for our
a>Zb.
If w e t h e n c h a n g e v a r i a b l e s
j=bk
i = aqI
and apply
Definition 2. 2, w e obtain (f) ~(bj,2jbq)Pa+b_jPj(x) j guarantees = ~(aqi, i that the s a m e Zia)Pa+b_iPi(x). t e r m s do not appear with nonzero q = b2 t + 1 for s o m e t > 0.
T h e condition a > 2b
coefficients on both sides of (f). N o w Then, if j ~ b, (bj, Zjb c 0 =
suppose that
(bj, ZtlZ(bj))
=
0. while if
O n the right side of (f), Pa+b_iPi(x) = 0 unless Zt > 2ia, then (aqi, Zia) Thus Lemma
Z tI = b  q ~ 2ia,
= (Zia,abil+Z t) = (gia,abi1). q = bZt+l for s o m e t > 0. By
(f) reduces to the desired relation (i) w h e n 4. 3, it follows that (i) holds for all q.

180

(ii) Observe that ~w, = w , ~ the Bockstein operation where
and that, by the proof of part (v) of Proposition 2. 3, ~(ek@eP ) = ~(ek)~eP ~@g., so does
~ in H,(q Zp) is given by
~(ek) = 5(kl)ek_ I. N o w
since (c) holds with ~.:.. replaced by
(d); that is,
(d')
~g.(w~ 1),(~¢,(er~ e s ) ~ X p2) = (1)
rs + m q
2
~ , ( w ~ 1),(~¢,(es~ e r ) ~ x P ).
Replace r and s by 2r and 2s in (d) and (d') and let 8 = 0 or 1; then, by (b)
and L e m m a respectively. 4.6, (d) and (d') imply the following formula for ~ = 0 and ~ = i,
(g)
~ (  1 ) k v ( g s ) ( k , spk)Dzr+(gpk_gs)(p_ I)_E Dgsgk(p t)(x) k
=
~ (l)e+mqv(2r)(e, rpC
and
)Des +(2p~ 2r)(pl)£ D z r  Z~ (pl) (x)"
In (g), set r = a ( p  l )  p q m and i = amqi. Let
s = b(pl)qm
I
and change variables to j = bk
~°P s = Ps and
~ Ps = ~Ps' by abuse; then, by Definitions
Z. 2 and a check of constants, w e obtain (h) ~(1)b+J(bj, p j  b  m q ) ~ Pa+b_j~(x)
J
= ~ i Again, a > pb (1)a+i(a  m q  i , p i a)~ Pa+b_iq(x)
ensures that the s a m e terms do not appear on both sides of (h). q = 2b 2(I+p +... + pt1), t > 0. Then (bj, pjbmq) = 0
N o w suppose that unless I +...
j = b and, on the other side of (h), ~E Pa+b_iPi(x ) = 0 unless + ptI = bq/2 > pi  a, w h e n pt > pi  a implies
(a  mqi, pia) =
(ela, a(pl)b + pt_ 1i) = (pia, a(pl)bi1). q = Zb2(I +... + pt1) for
Thus (h) reduces to the desired relations (ii) w h e n some (iii) t > 0. Replace By Lemma
4. 3, it follows that (ii) holds for all q. 2si in (d) and (d'); then, by (b) and L e m m a ~ = 0 and ~ = I, respectively.
r and s by Zr and
4.6, (d) and (d') imply the following formula for

181

(i)
~(l)k+mqk
~Zs1)(k, slpk)D2r+{Zpk_2s+1)(p_l)_ED2s_l_Zk(p_1)(x) (I)~ +mqv(2r)(~ , rp~ )Des_1+(Zp ~ _2r)(p_ l)DZr.2~ (p_1)(x)
= (i  a) ~
~
and i = amq~.
(i) ~ v(Zrl)(i,rlp~)D2s+{2p ~_2r+~(p_l)_~D2r_l_2~(p_l) (x)" and s = b(pl)qm and change variables to j = bk
In (i), set r = a(pl)pqm
B y Definitions 2. 2, w e obtain
(J)
~(1)b+J(bj, pjbmqt)~ ~ Pa+b_j~Pj(x)
J
= (18) ~ i ~(i (l)a+i(amqi, pia)~Pa+b_iPi(x)
l)a+i(amqi, pia I)~~ Pa+b_i~Pi(x)
Again, a >__pb ensures that the s a m e terms do not occur on both sides of the equation. j = b. Now suppose that q = 2b2p t, t > 0. Then (bj, pjbmql) = 0 unless t p > pia, w h e n
O n the other side of (j), (amqi, pia)
~Pa+b_iPi(x) = 0 unless
= (pia,a(pl)bi + (p1)p t) = (pia,a(pl)bi), pt > pia1, w h e n . t < 0,
and
~E Pa+b_i~Pi(x) = 0 unless
(amqi, pia])
= (pial, a(pl)b  i +(p1)p t) = (piaI, a(pl)bi)
Thus (j) reduces to the desired relations (iii) w h e n and L e m m a Remark 4. 3 implies that (iii) holds for all q. It should be observed,
q = 2bZp t for s o m e
4. 8.
for use in section 9, that the relations (f), a and b
(h), and (j) derived in the proof above are valid for arbitrary integers (without the restrictions a > pb or
a > pb). Indeed, these conditions on a and
b w e r e only required in order to obtain disjoint nontrivial terms on the two sides of the cited equations.

182

5.
Re i n d e x i n g f o r c o h o m o l o g y : We h a v e g e a r e d o u r d i s c u s s i o n to h o m o l o g y , but the reformulation approThus If
p r i a t e to c o h o m o l o g y is o b t a i n e d b y a m i n o r a n d s t a n d a r d c h a n g e of n o t a t i o n . let K be a Z  c o m p l e x P IK b y Zgraded by superscripts, with d of d e g r e e p l u s o n e . sections applies. and reformulate
we regrade Equivalently,
the theory. with IK and
tK q = K q, t h e n t h e t h e o r y o f t h e p r e v i o u s W by nonpositive superscripts
we can regrade
Obviously, W
this in no w a y changes the proofs. and let x ¢ Hq(K), define ~pS(x)
Let Then
(K,@) ~ ~ ( p , oo)~
graded by superscripts, i>0,
Di(x ) = @ , ( e  i ~ x p) ~ HPqi(K), p > 2, f3pS(x) = ~Ps(X).
and w e m a y pS(x) and
pS(x) = P s ( X )
and, if
Explicitly,
are defined by the
formulas
(i)
If p = 2, p S ( x ) = Dq_s(X ) ~ H q + s ( K ) , w h e r e
If p > Z, pS(x) =(l)Sv(q)D(q_2s)(p_l)(X)
D.I = 0 f o r
i < O; a n d
and
(2)
~ Hq+Zs(pI)(K)
~ p S ( x ) = (  1 ) s v (  q ) D ( q _ Z s ) ( p _ l ) _ l(X) ~ H q + Z s ( p ' l ) + l ( K ) , D. = 0 f o r
1
where
i < 0 a n d if q = Z j  ~ , E = 0 o r 1, t h e n
S
v(q) = (1)J(m!) ¢ .
Of course,
if p = 2, w e should write
p S = Sq
in order to c o n f o r m pS. In this way, the
to
standard notations, but w e prefer to retain the notation C a r t a n f o r m u l a and A d e m p>2. The integers (3) (4) s. ps and ~pS are natural h o m o m o r p h i s m s
relations are formally the s a m e
in the cases p = 2 and
and are defined for all
If (K,O) ~
C ( p , oo) a n d
x ~ Hq(K), t h e n pq
(x) = x
If p = 2, If p > 2,
pS(x) = 0 if s > q and
Z
; and
pS(x) = 0 if 2s > q, ~pS(x) = 0 if 2s ~ q, and pS(x) = x p if 2s = q. pS(x) = 0 if s < 0 or that is unital, then p O = i; these formulas s ~ 0. If (IK, @)
Note that w e do not claim that are not true in general. is reduced m o d p, then
If (IK, @)
pS(e)  0 for

183

(5)
~pSI = spS if p > 2 .
if p = 2 and
~ps
is the composition of pS
and the Bockstein
The external Caftan formula n o w reads
(6)
pS(x~)y)
=
~ Pi(x)(~PJ(y) i+j = s = ~
i+j =s
and, if p > 2,
~pS+l(x @y)
(~pi+l(x)~)Pj(y)+(_l)degxPi(x)~Pj+l(y)).
W e have
~pS = p S
and
0 ~ P s = ~pS0
and of course the Kudo transgression
t h e o r e m takes on a m o r e familiar f o r m with grading by superscripts in case (ii). The A d e m relations, reformulated in terms of the pS, take on the f o r m given in
the following corollary. Corollary 5. I. The following relations a m o n g the pS objects in ~(p, oo) and ~pS are
valid on all c o h o m o l o g y classes of all A d e m (i) If p >__2, a < pb, and papb
~ = 0 or 1 if p > Z, 6 = 0 if p = Z, then
: ~(1)a+i(a
i
 pi,(p1)b  a+i1)~ ~ pa+bipi
(ii)
If p > 2, a <_pb, and ~pa~pb
a = 0 or i, then
= (I ~) ~ (  1 ) a + i ( a  p i , ( p  l ) b  a + i  l ) ~ p a + b  i p i i ~(I
l)a+i(api 1, (p l)ba+i)~ ~ pa+bi~pi ~OpS = ps and ips = ~pS), 4.7 and the
(where, by abuse of notation,
While the two forms of the Adern relations given in T h e o r e m
corollary are completely equivalent, they w o r k out quite differently in practice. The relations of T h e o r e m a,b >_0; but a,b >_ 0 4.7 apply to positive complexes, in homology, with in the corollary,
in T h e o r e m
4.7 corresponds to a,b <_ 0
which is designed for use in c o h o m o l o g y with a, b _> 0.
F o r this reason, the Dyer
Lashof algebra [ 17 ], which operates on the h o m o l o g y of infinite loop spaces, is a very different algebraic object than the classical Steenrod algebra.

1 8 4

6.
Cupi products, B r o w d e r operations, and higher Bocksteins. W e here discuss v.products and certain h o m o l o g y operations of two
i
variables, which w e r e first studied by B r o w d e r in [4]; these operations occur in the presence of a Vnproduct and the absence of a ~n+lproduct and are central We shall also obtain a In section
to the study of the h o m o l o g y of (n+l)fold loop spaces.
very useful result, Proposition 6.8, on higher Bockstein operations.
i0, w e shall show that this result suffices to give a complete computation of the rood p c o h o m o l o g y Bockstein spectral sequence of K(~r,n) for any Abelian group and any prime p. /i is a commutative ring, ir is the cyclic group of W is the canonicalA~rfree resolutionof A. Z~iei_I for i > I. If (K,O) ~ ~(~,n,A),
O
Throughout this section, order Z with generator i = o~ + (  1 ) i ~ Aw, so ~, and
Let then
that d(ei) =
w e m a y a s s u m e that the restriction of 0 to e ~ K ~ K product on K by (i) of Definition g. i. Let (K,@) ~ ~(Tr, n,A)
agrees with the given
Definition 6. I.
0 < i < n, d e f i n e a n d if i > 0, t h e n to
and let x ~ K
Then U
O
q
and y ~ K
r
. For
K
! i(i+l ) x •. y = (1) 2 @ ( e i. ~ ) x Q y ) _
1
i s the p r o d u c t o n i from •i1
w:K~K1
  > K is a c h a i n h o m o t o p y of d e g r e e
(  1 ) i  l ~ i _ 1 e: ; t h a t i s ,
(i)
d(x~.1y) = (l)id(x)wi y +(l)i+qx r i d ( Y ) + x ~ i  i
Y +('l)l+qry vii x.
If A = Z2
and x ~ K q
is a cycle, then Pi+q{X}
= Di{x} = {xw.1 x}, We now
which, in cohomology, w a s Steenrod's first definition [Z5] of the squares.
d e f i n e the B r o w d e r o p e r a t i o n s f o r D e f i n i t i o n 6. Z. y c H r ( K ). Let (K, 0) ~ ~(~r,n,A). n <oo, andlet xe Hq(K)
(K,O) a ~ ( ~ , n , A ) ,
and
O b s e r v e t h a t if a a n d b
are representative
c y c l e s f o r x a n d y, t h e n
An+len~a~b
is a cycle in w ( n ) ~ K y. Define
Z w h o s e h o m o l o g y class ~ n + l e n ~ X ~ y kn(x,y ) ~ H q + r + n ( K ) by
depends only on x and

185

kn(X,y) = (  1 ) n q + 1 0 , ( A n + l e n O X O y ) . equivariant homology;
Note that we have c h o s e n not to pass to K z,
of c o u r s e , we can do so, and, in w ( n ) ~
(l)nq+IAn+lenOaOb = (l)nq+nenO a O b  (1)nq+qren(~boa" nq+n + in(n+l) Thus kn(X,y) is represented by (  I ) (a~.~nb(l)n+qrb%2n a). The following propositioncontains many of the elementary properties of the kn; its proof is i m m e d i a t e from the definition.
P r o p o s i t i o n 6.3. L e t (K, 0) > Hq+r+n (K). kn: H q ( K ) O H r ( I K )   > Hq+r+n(K) ~(~r,n,A), then Xn(f, O f ~ ) = f , k n thenk = 0
n
~'(lr, n , A ) , n < c o , and c o n s i d e r
Xn:H q (Z) X Hr(K)
(i) (ii) (iii) (iv) (v)
kn i n d u c e s a h o m o m o r p h i s m If f : K   > I K ' If O i s t h e
is a m o r p h i s m in r e s t r i c t i o n to
w ( n ) @ K 2 of O ' : w ( n + I ) ~ ) K Z   > K ,
If n = 0, t h e n k ( x , y ) = x y  (  t ) q r y x
O
If (K, e) is unital and the r e s t r i c t i o n of e to w ( n ) o ( e O K h o m o t o p i c to ~ O~,
+ K O e) is
~ the p r o d u c t , then kn(X , e) = 0 = kn(e , y). kn(X,X)
=
(vi)
kn(X,y )
:
(1) q r + l + n ( q + r + l ) k n ( Y , X ) and, if 2 = 0 i n / i ,
0
(Note that the f i r s t p a r t i m p l i e s The k
n
Zkn(X , x) = 0 if n+q is even. )
s a t i s f y the following a n a l o g of the e x t e r n a l G a r t a n f o r m u l a . L e t (K,O) and (L, 0') be in ~(lr, n , A ) , n <oo and Then
P r o p o s i t i o n 6.4. A a field.
L e t x e Hq(K), x' e Hr(K), y e H s ( L ) and y' e Ht(L ).
kn(xOy, x ' O y ' )
= (1)r(s+n)xx'Okn(y,y')+(1)s(r+t+n)kn(X,X')Oy'y.
Let
Proof. Let a,a',b,b' represent x,x',y ,y' r e s p e c t i v e l y .
c = (1)n(q+s)+l ~ ( A n + l e n @ a @ b @ a ' ( ~ b ' ) , By (1) of Definition 1.2,
so that c r e p r e s e n t s kn(x(~y , x ' ~ y').

186

n
5 ~(e n) = "=
eJ®aJenj
in W, and the definition of ~ shows that
c = E
n
(_ I) rs+ n(1 +r+s)_j(q+r)e(ej
ea~a'l®8'(~en_j®b®b')
j=O
 ~
n
n
(1)rs+n(r+s)J(q÷r)e(~e.®a®a')®e'(~+le
.1 nj
.®h®b')
jO
Let
e = ~
(1) r s + n ( r + s )  ( J + l ) ( q + r ) e ( e j calculation demonstrates r( sPn)+sn+n
a @ a ' ) ® 8'(alien+ l_j ® b ® b ' ) . that
Then
j=l a straightforward
c +d(e)= (1)
e(eo® a®a')®e'(~n+len®b®b')
+ ( l)s(r+n)+qn+%(Z~n+len ® a® a') @ e'(n+leo ® y® y').
Since L is h o m o t o p y c o m m u t a t i v e for n > 0, 6 ' ( ~ n + l e o ®b®b') represents
(1)Sty'y
f o r a n y n, and the r e s u l t f o l l o w s .
commute
n
W e next prove that the k Proposition6.5. Let K be a A  c o m p l e x Let
with suspension. and g : K   > K"
(K',O')(C(w,n+l,A)
and
(K",O")~
~(w,n,A). of
and let f:K' > K Define
be m o r p h i s m s
complexes such that gf = 0.
= w(n+l)@f(K,) 2 + w(n+I)@f(K')@K
where ~(n+l) = w(n) + A e n + 1 (aen+l / ~(n+l)).
+ w(n)@K 2 ,
Suppose given a w  m o r p h i s m
O: K   > K s u c h that t h e f o l l o w i n g d i a g r a m is c o m m u t a t i v e : w(n+I)oK,@K ,
i®f®f
>~
iQg~g
> W ' n ' • K" (~ K" 0" g
> K"
f ~
io,
K'
Let x, y e M e t f,. Proof. Then
f
io
>K
Okn+l(X , y) = kn(OX,~y) e C o k e r g,. x and y respectively. Let
Let a' e K' and b' ~ K' represent q r
a = f(a') and h = f(b') and choose
u ~ Kq+ 1 and v e Kr+ 1 such that d(u) = a

187

and d(v) = b. Define c ~ ~ by c = (  l ) q A n + l e n ~ U ( ~ V  (l)n+qen+l(~ a ~ v Then a straightforward calculation demonstrates that d(c) = e n + l @ a ~ b Thus  (l)r(q+l)en+ 1 ( ~ b O u .
 n+qr + (i) en+l@b@a
n = (i) ~ n + z e n + l ~ a @ b
(1) (n+ 1)q + ] f O ' ( A n + 2 e n + l ( ~ a' O b') = ( l) (n+l)(q+l)dO(c) (l)(n+l)(q+l)go(c)
and
= (1)n(q+1)+Io"(An+lenOg(u)Og(v)), a n d this p r o v e s the r e s u l t .
by our commutative diagram,
T h e a n a l o g f o r the B r o w d e r o p e r a t i o n s of the A d e m r e l a t i o n s is the f o l l o w ing Jacobi identity: let x ~ Hq(K), y E Hr(L), and z ~ Hs(K); then, under appropriate hypotheses, ( i~(q+~)(s+~)× % X
j n ~
l
n k, Y
z)) + (1)(r+:n)(q+n)kn(Y, kn(Z ,x)) + (_l)(s+n)(r+n)kn(Z, kn(X , Y))
= 0,
and, if 3 = 0 in A
and q+n is odd,
kn(X ,kn(x ,x))=0. W e omit the proof as an
easier geometric argument can be obtained for the homology of (n+l)fold loop spaces. This identity, and the identity of (vi) of Proposition 6.3, lead to a notion There is In
of k algebra which generalizes that of Lie algebra (or k algebra). n o
also a notion of restricted k algebra which is important for the applications. n
the case A = Z2, the restriction is already present in our algebraic context; it is the last Steenrod operation for an object (K, 8) ~ ~ (2, n). The following addendum to Proposition Z.3 gives s o m e properties of this operation that are needed in the study of (n+l)fold loop spaces. Proposition 6" 5. Then Let (K, 8)
~(Z,n). Let ~n = Pq+~:Hq(K)>HZq+~(K)"
(i)
~n(X+y) : ~n(X) + ~n(y) + ×n(X,y), and

1 8 8

(ii)
~n(X) Proof.
: (q+nl)Pq+n_l(X) + kn(X , ~x) if F o r (i), if a and b represent
(K,O)
is reduced m o d
2
x and y, then, in K Z, An+len~ a~b yields the stated
(a+b) z = a Z + b 2 + ~n+lab, and the error t e r m
deviation f r o m additivity of ~n" (v) of Proposition 2.3. W e n o w relate the k
n
Part (ii) follows f r o m a glance at the proof of
to the Bockstein operations on H(K)
when
(K, 8) c
C(w,n, Zp) is reduced m o d p.
In contrast to the Steenrod operations,
the higher Bocksteins are all of interest. Proposition 6.7. x,y ~ H(K), deg(x) = q. Let (K, 8) ~ ~(~, n, Zp) be reduced m o d p. Assume that ~r(X) and ~r(y ) are defined. Let
Then
~rkn(X, y) is defined and, modulo indeterminacy,
~rkn(X, y) Proof. Let
:
kn(~rX ,y) + (l)n÷qkn(X , ~ry ). Let x and y. a, b e K be such that their
(K,O) = ( ~ Z p , ~ Z p ) . ~ , ~ ~ I< represent
rood p reductions and and
We may
a s s u m e that d(a) = pra' ~r(X)
d(b)= prb'; the rood p reduction a' and ~' of a' and b' represent ~r(y ). In W ( n ) ~ g d(An+len~aOb)
= (l) n p rAn+le n o ( a ' ~ b + (  l ) q a ~ b ' ) 
B y reduction m o d p and a check of signs, this implies the result. Surprisingly, the following fundamental result appears not to be in the literature, although it is presumably wellknown. It allows complete calculation for any
of the m o d p h o m o l o g y Bockstein spectral sequence of Q X = lim ~ n s n x space X
and, as w e shall show later, the m o d p cohomology Bockstein spectral Together with the previous result, it also suffices for the
sequence of K(w,n).
computation of the rood p h o m o l o g y Bockstein spectral sequence of ~nsnx, n >__I. Proposition 6.8. Let K Let be a Zgraded associative differential ring K have a u lproduct such that
which is flat as a Zmodule. (a) d(a Vl b) :d(a) v I b

(i)deg aa ~i d(b) + ab  (i) dega d e g b b a

1 8 9

and, f o r c a s e (ii), s u c h t h a t the H i r s c h f o r m u l a (b) h o l d s
(b)
Let
abv I c = (1)degaa(bUl c) + (l)degbdegc(at~ 1 c)b.
~r d e n o t e the r  t h m o d p B o c k s t e i n on H ( K ~ Zp), ~1 = ~" Let ~r(y p) is de
y ~ Hzq(K~) Zp) and assume that ~r_l(y) is defined, r >2. fined and, modulo indeterminacy,
(i) (ii) If p = 2 and r = 2, ~2(y2) = ~(y) y + Pzq~(y )
m
Then
If p > 2 and
r = 2, ~2(y p) = ~ ( y ) y p  1 + E j=l
JXI({3(Y)yjI'~(Y)YPJ1)
(iii) If p > 2 and r > 3 ,
Proof. Let
~r(y p) = ~r_l(y)y pI.
~ represents y.
b ~ K2q be s u c h t h a t its m o d p r e d u c t i o n d(b) = p r1 a, and t h e n
We m a y a s s u m e t h a t
a is a c y c l e w h o s e rood p r e d u c t i o n
a represents
~r_l(y ). Clearly we have d(b p) = pri ~ bii ab p1 , i=l
and
d(abPiul b il) = ab pI  bilab pi m o d p ri , 2 K i < p
Therefore d(bP + p r  1 i=2 If r >_ 3, t h e n 2r2 > r and p a r t (iii) f o l l o w s . arising from d(b) = pa T h u s let abP'l~l
"
.
bil)

p r a b p1
rood p 2 r  2
r = 2; we m u s t now t a k e If p = 2, t h e n
into a c c o u n t the t e r m s
in d ( a b P  i v l b i  1 ) .
d(b 2 + 2a~l b) = 4ab + 4 a v 1 a. Since the m o d 2 reduction of assume that p > 2. Then d(b p + p ~ abPi~l b il) = p2ab pI + p2c + p2c' , i=2 c =~ ~ i abJ_labP_i_j ~i bi.l i=2 j=l and c' = ~ l<_j <i<_p where a ~I a represents P2q~(y), this proves (i). Thus
abP1 bJlab1"J'l ~I
By the Hirsch formula, and a separate reindexing of the two resulting sums, we
find t h a t

190

c =
~ Z<j<i<p
[(abijl~l bJl)abPi  abPi(abijI ~i bJ1)].
Therefore if e = 2<j<i<p
abPi~l (abiji ~i bJl), t h e n
d(e) m  c +
~ a b P  l•• l 2<j<i< p
( a b l~°
.  bJlab 1jl)
rood p .
Comparing
d(e) to
c ' , we e a s i l y find t h a t
P pi bi1 d(bp + p ~ ab ~i i=2
+ pZe) =pZabP
1
+
p2 ~
" i2 (il)abPl~l ab
i=2
2 m " " 3 = pZabPI + P ~ j(abPJI Ul ab31  ab31 ~I abPJl) rood p j=l By Definition 6.2 , this implies (ii) and so completes the proof.
Of c o u r s e , •2product. the t e r m s involving k 1 in (ii) a r e z e r o if K a d m i t s a The Hirsch
T h e g e n e r a l r e s u l t is n e e d e d f o r s e c o n d l o o p s p a c e s .
f o r m u l a is v a l i d f o r the c o c h a i n s of a s p a c e [8, 16], f o r the c h a i n s of a s e c o n d
loop space [I0], and for the dual of the bar construction of a cocommutative Hopf algebra [i 6]. In connection with this formula, we m a k e the following remarks. R e m a r k s 6.9. Let K be an associative differential Z algebra, p > 2, with a P Define
v lproduct which satisfies the Hirsch formula.
< > P : H z s _ I ( K )>H2sp_2(K ) as follows. Let a I represent x e H2s_I(K ) and 1 iI define ai = ~ ai. lul al for 2 < i< p. Then d(ai) = j ~ a.a. and pi "= J 1j = ~ a.a is a cycle. A computation demonstrates that if {a~l i < i< p} j=l J PJ is a n y set of e l e m e n t s of K s u c h t h a t a~ r e p r e s e n t s x and
d(a~) = a[ a~ j 1j for 2 Ki< p, t h e n ~ ' = ~ a!a' is h o m o l o g o u s t o ~ . j=l J pj T h u s the
class of ~ depends only on x, and we define < x > p = {~}.
is o f t e n the case t h a t if (K,@) e ~(p,p2),
In the applications, it
K r a i n e s [11]
t h e n < x > p = ~Ps(X).
h a s p r o v e n t h i s r e s u l t f o r the c o h o m o l o g y of s p a c e s ,
w h e r e it r e a d s

191

<x> p
=
~pS(x)
for
x e
H Zs+l
(K), a n d K o c h m a n [10] h a s p r o v e n it f o r t h e h o m o A general proof within our algebraic context should
logy of iterated loop spaces. be possible, but appears
to b e d i f f i c u l t .
7. The category of simplicial Amodules W e here develop some machinery that will allow us to apply the theory of
the previous s e c t i o n s to a l a r g e s i m p l i c i a l of i n t e r e s t category ~.. We s h a l l s p e c i a l i z e familiarity with
to s p e c i f i c c a t e g o r i e s
in t h e n e x t s e c t i o n .
We a s s u m e
t h e b a s i c d e f i n i t i o n s o f t h e t h e o r y of s i m p l i c i a l o b j e c t s a n d of a c y c l i c m o d e l s
(see, e.g., [15, §1,2,28,29]). and ~ d e n o t e
Let A
be a commutative ring, andlet ~ ,
~~
the categories of (ungraded) Amodules, positively graded Let C: /~> ~0 be the normalized
i s t h e q u o t i e n t o f K, r e g a r d e d as a
Acomplexes, and simplicial Amodules.
chain complex functor (for chain complex with simplices). d = ~ K eJ~ , C(K)
(  1 ) i d i , b y the s u b c o m p l e x H(C(K)) and
generated by the degenerate
Define H , ( K ) =
H * ( K ) = H(C*(K)), w h e r e 6(f)(k) = (  1 ) q + l f ( d k ) for
C*(K) = H o m A ( C ( K ) , A ) i s g i v e n t h e d i f f e r e n t i a l f • Cq(K) Dold[ 5 ]. Lemma r e s o l u t i o n of A K 1....
'
and
k e C q + I ( K ).
The following key lemma
i s b a s e d on i d e a s of
7.1.
L e t w b e a s u b g r o u p of W
O
~
r
and let e .
O
W be aAwfree Let
such that
= Aw w i t h A w  g e n e r a t o r
K
r
e J~
" then there exists a morphism "W®C(KI~
1
of A  c o m p l e x e s @ C ( K r)
. . . . ~ K r)
> W@C(K1)•...
w h i c h i s n a t u r a l i n t he K. a n d s a t i s f i e s t h e f o l l o w i n g p r o p e r t i e s :

192

(i)
For
c~ • ~r, t h e following d i a g r a m W@C(K lx
... X K r)
is c o m m u t a t i v e : > WI~C(KI)@...
@ C ( K r)
W@C(K0~I) (ii) (iii)
X...
X K (r)) on
> W®C(K (1))® ... ® C(K~(r)).
W(~Co(K 1 X ... ~kr) X Kr).
~ is the i d e n t i t y h o m o m o r p h i s m ¢(eo~kl(~... where
map.
@kr) = eo@ X Kr)>
~(kl®...
if k i • K i is a j  s i m p l e x ,
~'C(K 1 × ...
C(K1)®...
( ~ C ( K r ) is the A l e x a n d e r  W h i t n e y
(iv)
e(w®cj(~ 1 x...
x Kr)) C >2 W®[C(K1)®'~C(Kr)]k"
k < rj
Moreover,
any two such • are naturally equivariantly homotopic Since (K X L)j = K . @ L . formulas (ii) and (iii) m a k e sense. J J'
Proof.
Write
A.j = Cj(K 1 X . . .
i j
X Mr)
and
B.j = [C(K1) ~ . . .
@C(Kr)]j.
We c o n s t r u c t Formula for (ii)
on W. ~ A . defines •
b y i n d u c t i o n on i and f o r f i x e d j = 0 a n d all
i b y i n d u c t i o n on j. •
for
i and f o r m u l a s (i) a n d ( i l l ) define
i = 0 and
all j. Thus let i k l
and j > i and a s s u m e that • is defined for i' < i and for Awbasis { w k} for W..1 It suffices to • can then be uniquely
the given i and j' < j. Choose a define # on w ~ x for w
• { w k } and x • Aj, since
extended to all of W i ~ Aj by (i). Let AA[j]
denote the free simplicial [15,p. 14]. Then the
A  m o d u l e generated by the standard simplicial jsimplex funetor w O
A. is represented by the rfold Cartesian product AA[j] r, and J Therefore ~(w~.~j ... ~ A j ) c a n b e definedby
W ~ B ( A A [ j ] r) is acyclic. choosing
a chain whose boundary is r
~ d ( w ~ Aj ~ ... ~ Aj), and • can be carried by representability. N o w (i),(ii), and (iii) are
over to a r b i t r a r y w ~ k I ~ . . . ~ k
clearly satisfied and (iv) follows from the fact that Ck(AA[j]) = 0 for k > j. The proof that • is unique up to natural equivariant homotopy is equally simple.

193

Remarks = I~] Since
7. Z.
Define k g : W ~ C ( K I ) ~ 9 . . .
~ C ( K r)> W ~ C ( K X ... X K r )
1 X ... × Kr) by is the shuffle m a p .
where
I]:C(KI)~... ~C(Kr)>C(KI •
~] , unlike [ , is commutative, and ~
is equivariant.
B y an easy acyclic
models proof, ~ maps.
are equivarlantly homotopic to the respective identity
W e shall only be interested in the case • :W~ C ( K r) >W ~
K 1 = ... = Kr; here The
C(K) r is a natural rnorphism of A T  c o m p l e x e s .
general case w a s required in the proof in order to have the representability of the functors A.. J Starting with objects of the following category ~ , w e shall use
to obtain diagonal approximations and so to pass to the category p ( w, oo, A) < C(w, co, A) defined in Definitions Z.I. Let where ~ denote the following category. and D:K> K X K The objects of ~]
Definitions 7.3. ~ are pairs (K,D)
K a ~
is a m o r p h i s m i n
suchthat
( D × 1) = ( I X D ) D
and tD = D, w h e r e are those m o r p h i s m s
t(x~y) = y~x. f: K > K'
The rnorphisms such that
f: (K, D) >(K', D') in ~ (f X f)D = D'f. Each I< ¢ 7a
in : ~
admits the natural diagonal D(k) = k e k, and ~d~ is thereHowever, an object K ~ . ~ may
by e m b e d d e d as a full subcategory of ~2, admit other interesting diagonals.
F o r example, if I< is a simplicial cocorn~ :K > K X I< is a
rnutative coassociative Acoalgebra, then the coproduct permissible diagonal; that is,
(K, ~ ) ~ ~.b . The following r e m a r k s will be of
use in the study of relative and reduced cohornology. Remarks 7.4. (i) If L C K in ~ define H.(K,L) = H(C(K/L)) and D(L) and
H • (K, L) = H(C*(K/L)). then K / L K D >KX
If (K, D) E ~i,
is contained in L X K + K X L,
admits the diagonal ~ K w× w > K/LX in ~  . K/L,
induced f r o m the composite where w:K> K/L is the projection, and
then w is a m o r p h i s m

194
(ii) Let ~ : hAlO] ~ / ~
and C ( ~ ) : C (~k) : A .
O
; thus ~ Give X
n
= A for n > O, each d. and s. is the identity,
1 1
t h e natural d i a g o n a l .
We s a y t h a t (K, D) e and an epimorphism
i s u n i t a l if w e a r e g i v e n a m o n o m o r p h i s m
v : A > K i n ~ L
6:K>~
in ~
such that ev = i and (~.X I)D = (I X £ )D (where
If ( K , D ) is u n i t a l and IK = K e r ~ , t h e n K = v
" X K = K = KXA).
(X)~) IK
Clearly,
and
for k c IK, D(k) = k ®
(IK,~)
v(1) + v ( 1 ) ® k
(K/v(~Y),D)
+D(k), in ~0 .
where
D(k) ~ I K X IK.
i s i s o m o r p h i c to
If ( K , D ) e
~L
, then C (K) is an associative differential Aalgebra, with
c u p p r o d u c t d e f i n e d as t h e c o m p o s i t e
(i) ~:C*(K)®C*(K)
Here a
~ >[C(K)®C(K)]*
>C*(K×K)
) = (i) d e g y d e g k
,
D
* >c (K).
~ is
is the natural map, c~(x~ y)(k®~ map. If (K, D)
x(k)y(g ), and
the AlexanderWhitney
and augmented
is unital, then C*(K)
is unital (via v*)
(via ~;:"). F: ~H*(K) > for ~9(Tr, oo, A) a n d t h e n s h o w h o w to u s e
We n o w d e f i n e a f u n c t o r F to a p p l y o u r g e n e r a l t h e o r y to
(K,D) e k065 in the c a s e A D
= Z . P
Definitions 7.5. K   > K r. Define Let ~ C E
r
Let (K, D) e J~[b and write and let W
for the iterated diagonal with W
o
be aA~rfree resolution of A
= ATr.
A:W@C(K)
> C(K) r to he the composite
(z)
Let
~:w®c(K)
I®~W®C(K
r)
~
> weC(K)
r
~®l
>C(K) r
¢~:C~(K)r> [C(K) r] ~ be the natural m a p
and define a AIIrnorphism
8:W®C*(K)
(3) Since
r  > C*(K)
by the formula
w ~ W, x e C*(K) r , k e C(K). j : W > V a s i n
0(wex)(k)
= (1)degwdegx=(x)(Z~(wek)),
w= ~
r
0 may be defined for
and then factored through
Definition 2.1,
a n d t h e r e s u l t i n g c o m p o s i t e i s A ~  h o m o t o p i c to t h e o r i g i n a l 0 d e s a t i s f i e s c o n d i t i o n (ii) of D e f i n i t i o n Z. 1. By L e m m a 7 . 1 ,
f i n e d i n t e r m s of W, 0
f o r m u l a (3) s p e c i a l i z e s to give
(4)
0 ( e o ® X ) = D* ~*a(x) for any x ~ C*(K) r, and

1 9 5

(5)
O(w(~9x) = $(w)D$e~(x)
if x • C°(K) r
a n d w e W. Since O is n a t u r a l on
B y (1) and (4), O s a t i s f i e s c o n d i t i o n (i) of D e f i n i t i o n Z. 1. morphisms by letting if ( K , D ) in ~)6u , we thus o b t a i n a c o n t r a v a r i a n t
functor
F : ~96L  > ~9 (% oo, A) B y (5),
F(K,D) = (C"(K),0) is u n i t a t in gg0~ t h e n
on o b j e c t s and F(f) = C*(f) F(K,D)
on m o r p h i s m s . IfA
=
is u n i t a l in ~ ( % c o , A ) . Zp) w h e r e ~
Zp, ~r is
c y c l i c of o r d e r p, and Zmodule, then
(K,D) = (~
Zp,5~
is a Z  f r e e s i m p l i c i a l ~ ; is
we a g r e e to c h o o s e is r e d u c e d m o d p
0 for (since
K to be the m o d p r e d u c t i o n of 0 f o r C(~) is Z  f r e e a n d t h e r e f o r e C*(~)
F(K,D)
Z  f l a t , a s r e q u i r e d by D e f i n i t i o n 2 . 1 ) . Observe that, by Definition
any (K,D) e ~(L . W h e n
6.1,
we now h a v e
~.products
1
in C"(K)
for
A = Zp, the results of Proposition 2.3 will clearly apply pS defined on the c o h o m o l o g y of objects (K,D) ¢ ~0 •
to the Steenrod operations
If (K, D) and ( L , D ' )
a r e o b j e c t s of kgCL , t h e n
K X L a d m i t s the d i a g o n a l
Thus
D = (i X t X I)(DX D'); if D a n d D '
are the natural diagonals, then so is D.
(C*(K X L), O) is d e f i n e d in
¢ (w, oo, A).
The f o l l o w i n g l e m m a c o m p a r e s
(C*(K X L), 0) to (C*(K) ~ C*(L), ~) and will imply the applicability of the external Cartan formula to H * ( K X L) w h e n gemma 7.6 . F o r any objects A = Z . P (L,D') in k~0~ , the following
(I{,D) and
d i a g r a m is A w  h o m o t o p y
commutative
>
W~ C*(K X L)r l®(~*)r l[ l®(£~r
W @ [C*(K) ~ C*(L)] r
That is, * ~* N and are morphisms Proof.
to show
c*(~
x
L)
~* i n*
> C"~(K)@C*(L)
in the c a t e g o r y ~(~r, c o , A ) . 1)(~ U), it s u f f i c e s
B y the d e f i n i t i o n s of O a n d @= ( 0 ~ O)(1 ~ T ~
is Awhomotopy
that the following diagram
commutative:

196

W @ C ( K × L)
 (e® 1){~ ® 5 )
> C(K X L) r
1~}111 11@6 W @ C(K) ~ C(L) U(A@A)(I® T® 1)(~®1®1) > [G(K)@C(L)] r
Since we let
(I~D~
I@D')(I@T@
1)(~
1(~1) = ( I @ T ~ I ) ( ~ @ I @
1)(I~D~D'),
if
~ = (E@ 1)~
and let u : K r × L r   ~ ( K
X L) be the evident shuffle, so that
= u(D × D ' ) , t h e n this d i a g r a m b e c o m e s
w@c(~x L)
I@DX
D ' > W @ C ( K r X L r) ! @ U > w ~ C ( [ K X
L]r)
~ > C ( K X L) r
io[Io
W@C(K)@C(L) I@D~D'>
l 111o
W@C(Er)@C(L r) U ( @ @ @ ~ I ~ T @ I ) ( ~ @ I @ I ) )
rI[
[C(IQ @ C ( L )]r
The lefthand square c o m m u t e s by the naturality of ~] and 6 Since the diagonals are not involved inthe righthand square, w e can prove that it c o m m u t e s up to A w  h o m o t o p y by an acyclic models argument, with K r and L r replaced by
K 1 × ...
× Kr
and L 1 × . . .
× Lr
so as to h a v e d o m a i n s g i v e n b y r e p r e s e n t a b l e
f u n c t o r s f o r fixed w e W. w e W and on e o
On z e r o s i m p l i c e s , the d i a g r a m c o m m u t e s f o r any
{ W, as the s i m p l i c e s v a r y , the d i a g r a m is A  h o m o t o p y c o m T h i s s t a r t s the inductive c o n 
m u t a t i v e by a s t a n d a r d a c y c l i c m o d e l s a r g u m e n t .
s t r u c t i o n of the d e s i r e d h o m o t o p i e s , and the p r o o f is c o m p l e t e d p r e c i s e l y as w a s the p r o o f of L e m m a 7 . 1 . C o r o l l a r y T ~ 7. If (K,D) { ~(k , then F ( K , D ) is a C a r t a n o b j e c t of
(% oo,A).
Proof. Since D: K >K X K is commutative and associative, it is a
m o r p h i ~ m in @0 . Therefore the cup product (1) is a m o r p h i s m in
as required.
C (%oo,A),
197
Lemma 7.8.
of
If
(K,D)~ ~9C6 , A : Zp, then F(K,D) is an Adem object
~(p, co).
Proof. In the notations of Definition 4. I (with Yo = ZpEpZ), it suffices
to
prove that the following diagram is Thomotopy commutative: Z
w~,
Z
Wl~ Wff~ C*(K)P
I I@U
> y ~ C*(K)P
~ C*(K)
Wl ~ (Wz 0 C*(K)P)P
All m a p s
looP
> W, ~ C * ( K ) P ~
0 are as defined in Definition 7.5; by dualization, it suffices to prove
that the following diagram is Thomotopy commutative Z >Y 0 C(K) > C(K) p
W1 ® W ? O C(K)
TO'
w@l
W ? ~ W , OC(K)
Let @=
> W? ~ C(K) p
a = @(wO,):W,OW?~C(K
U
> (Wz 0 C(K))P Z 2
(~,O1)@
anddefine
p
)
> C(K) p
Since
Z
A = ¢(I~D), A ( w O , ) = a(IO I O D), D:C(K)> C(KP ). By the naturality of ~, the following diagram is commutative:
IOWP c(KIT®D>WfO ,OcIKP> lliOD I 'OD I
W,OW Z ~C(K p )
I p 2 T®I ¥ 2
>Writ<KIP I'Op
"
> (Wz®C(K))P
(l@D) p U > ( W Z ® c(KP)) p
> Wf®W,~C(K p ) ' ® ~ > Wf~c(KP) p
Z Let ~ = ~Pu(1x~)(TX1)'W1 X W ? X C(KP )
APu(I X A)(T X I) = 6(' X i X D). "rhomotopic.
Z >C(K) p . Bythe diagram above,
Thus it suffices to prove that ~ and ~ are
Since a and ~ do not involve the diagonal, this can easily be shown
by acyclic models precisely as in our previous proofs.

198

The following t h e o r e m s u m m a r i z e s
properties of the Steenrod operations ~ , A = Z . Of course, P we
that are valid for arbitrary objects of the category
use the notations of section 5 since w e are dealing with cohomology. Theorem homomorphisms 7.9, pS Let (K,D) c k961. , A = Z P and, if p > 2, ~pS defined on each Hq(K); degree (pS) = s . T h e n there exist natural
if p = Z and d e g ( ~ £ P s) = 2s(pl) +6 operations on (i) ~pS ~
, ~ = 0 or i, if p > g.
These c o h o m o l o g y
satisfy the properties (6 = 0) and s > q or if p > 2 and 2s+E > q.
= 0 if s < 0 or if p = 2
(ii)
pS(x) = x Z if p = 2 and s = q; pS(x) = x p if p > Z and 2 s = q
if p = 2
(iii) If ( K , D ) = (K(~ Z p , ~ O Zp), where K is Z  f r e e , then ~pSI = spS and ~pS is the composition of ~ and pS if p > 2.
(iv)
pS=
~
piopsi
and ~pS = ~ ( ~ p i o p s  i + p i o ~ p S  i
) on H * ( K X
L);
the internal Cartan f o r m u l a is s a t i s f i e d in H*(K)
(v)
then
If f:K' > K
:
and
g:K> K"
, where
are m o r p h i s m s
>
in [9~
such that
gf = 0,
H q  I ( K ') is the suspension
associated with (vi) If L ~ i
C*(K") > C*(K) > C*(K'). and D(L)C LX L, then 6~£pS = (l)£~PS 6 w h e r e
6 : H q ( L )   > Hq+l(K, L) is the connecting h o m o m o r p h i s m .
(vii) The ~[pS satisfy the A d e m relations as stated in Corollary 5. I. ~pS = 0 for s < 0 (the rest is the
Proof. convention
F o r (i), w e m u s t prove that
e. = 0 for i < 0), by formulas (5. I) and (5.2), it suffices to s h o w that 1
D.(x) = 0 for i > (pl)q, deg (x) = q. B y (3) of Definition 7.5, it suffices to s h o w
1
that A ( e i O k ) = 0 for k ~ Cpq_i(K ).
Lemma 7. I, if i > (pl)q, then ~(ei~
Now A = ( £ @ I ) @ ( I e D ) and, b y ( i v ) of
D(k)) ~ ~
j < pq
PqJ
W
~[C(K)]~
C
Ker ( { ~ l).
(ii) and (iii) follow f r o m P r o p o s i t i o n 2.3; (iv) follows f r o m C o r o l l a r y 2.7, L e m m a 7 . 6 , and C o r o l l a r y 7.7; (v) and (vi) follow f r o m T h e o r e m 3.3, noting for

1 9 9

(vi) that the s u s p e n s i o n a s s o c i a t e d with C a ( K / L ) > Ca(K) > Ca(L) i n v e r s e additive r e l a t i o n to the c o n n e c t i n g h o m o m o r p h i s m 5 ; T h e o r e m 4.7 and L e m m a 7 . 8 .
is the
(vii) follows f r o m
By T h e o r e m 3 . 4 , the K u d o t r a n s g r e s s i o n t h e o r e m applies to a p p r o p r i a t e s p e c t r a l s e q u e n c e s involving o b j e c t s of ~0~ and, u n d e r the h y p o t h e s e s of (iii) of the t h e o r e m , P r o p o s i t i o n 6 . 8 applies to c o m p u t e the h i g h e r B o c k s t e i n s on p  t h
powers of elements of Ha(K), (K,D) ( ~)(3L. In the next section, w e shall show h o w to eompoute pO for arbitrary objects (K, D) ~ ~(L and shall give nontrivial
examples to show that p o / I in general.
8.
Simpli_cial sets and s i m p l i c i a l r e s t r i c t e d Lie a l g e b r a s We shall h e r e obtain the S t e e n r o d o p e r a t i o n s on the c o h o m o l o g y of
topological spaces, simplicial sets, and simplicial restricted Lie algebras, and shall consider the evaluation of po Let $ on Ha(K) for any (K,D) ~ ~ K ~ ~ , A = Zp, ,let ~ denote
denote the category of simplicial sets. F o r Let A
the free simplicial Abelian group generated by K. and define a functor A: ~ > ~CL
be a commutative ring with its natural dia
by letting A(K) = ~ @ A
gonal D; here 190~ is as defined in Definition 7. 3 and D diagonal k> (k,k) on K. Composing A with F E
r
is induced f r o m the
of Definition 7.5, w e obtain ~ denote the category
a functor FA: ,~ ~ ~ (w, oo,A) for any ~ C of topological spaces and let S: T ~ ~
. Let
be the total singular complex functor. If (K, L) is a simplicial pair, define 2Z of simplicial pairs Since the normaliz
Then FAS: ~   > A(K,L) = K/L(~A.
~(Tr, oo,A) Then FA
is defined.
is defined on the category
and FAS is defined on the c a t e t o g y
ed cochains with coefficients in A pair (X,Y) m a y bedefinedas
YZ of t o p o l o g i c a l pairs.
of a simplicial pair (K, L) and of a topological and
C'(K,L)= ca(~/~@A)
# C*(X, Y) = C (SX, SY), the results of the previous section apply to the cohomology
 200 
of simplicial and topological pairs. F o r the r e m a i n d e r of this section, w e take A = Z P of order on p. Via FA: ~Z> ~(p' oo), w e have Steenrod operations 7 z , hence on H (X,Y) Sqs. pS, s > 0, Of and w e let ~ be cyclic
H (K,L)
for all (E,L) E
for all (X,Y) ~ 7 Z .
course, if p = Z, pS
is usually dencfed by pS except
Theorem
7.9 gives all of the
standard properties of the fromthe
pO = I.
W e n o w s h o w that pO = 1 follows mS.
previously obtained properties of the Proposition 8. I. pO
is the identity operation and, if p = Z, pl is the
Bockstein operation on the c o h o m o l o g y of simplicial (or topological) pairs. Proof. (K, L) ~ complex. Since ~mO = pl then if p = Z, it suffices to prove that H (K, L) = H'~(K/L, P), w h e r e P pO = 1. If
, L nonempty,
is a point
Thus it suffices to prove that L
P°(x) = x for x ~ ~n(K) = Hn(l<, P), If E (n) is the nskeleton of that such
since the result for K, then
e m p t y will follow trivially. is a m o n o m o r p h i s m ,
~n(l<) > ~n(K(n))
and w e m a y thus a s s u m e f: K > Sn
K = K (n). Then, by the Hopf T h e o r e m that f (in) = x, w h e r e nsphere. i~"ne ~n(sn)
[Z4, p. 431], there exists
is the fundamental class of the simplicial P (in) = i n N o w for any K, the
It therefore suffices to prove that S;:": Hq+l(sK)
suspension i s o m o r p h i s m Hq+I(sK) > H q + I ( c K , K) hence S* commutes
> ~{q(K) m a y be defined as the composite CI~ is the simplicial cone of K,
* = in_ 1 for n_> 1 and
> H (K), w h e r e pS. Since
with the (where i o
s*(* "n)
P°(i:~ = (i P = i o We
generates
S° ) =
), this proves the result.
n o w use the fact that pO = 1 on the c o h o m o l o g y of simplicial sets to pO on H ~~ (K) for any object T. 1 w h e n (K,D) ~ ~0~ . In fact, w e have
s h o w h o w to compute the following a d d e n d u m tion of Z P Lemma Then, for any 8. Z. .
to L e m m a
9f is the canonical Z wfree resoluP
Let
K. e ~0~ l
, 1 <__i <__p, and let k. be a qsimplex of K.. i 1 ~... ~C(Kp) which satisfies
@:W~C(K
I × ... X K p )   > W ~ C ( K I )

201

satisfies the conclusions of L e m m a
7. I, ® k p ) = (1) m q v (  q ) _ikl ~ . . . ~)kp ,
(£®1)~(eq(p_l)®kl®...
where v (  q ) = 1 if p = Z and
v (  a j + ~ ) = (  1 ) J ( m ')~" , a= 0 o r 1, if p > Z.
Proof.
Let
A
q
be the fundamental qsimplex of AA[q].
Since ~
q
is a
Z basis for Cq(AZk[q]), we clearly have that P (i) ¢ ( e q ( p _ l ) ® ~ q ® ... ® A q ) = Y e o ® A q ® . . .
~)Ziq rood Eer(F~® I), y e Zp~.
By the n a t u r a l i t y of • ( o r b y the p r o o f of L e m m a 7 . 1 ) , (i) i m p l i e s
(ii)
(e× l)~(eq(p.1)®k I ®. "" ®kp) = ~(~)kl®... ® k p for any k i ~ (Ki) q
"
To evaluate
6(y), let iq e Cq(S q) represent the fundamental class of H q (sq); we
m a y take S q = z~[q]/~[q] so that i is a basis for Cq(Sq) and i~ e Cq(S q) is q q welldefined. • By (ii) and D(iq) = iq®... ®lq, Definition 7.5 gives
(iii) 0(eq(p_ I) ® i~qP)(iq = =(iq P)[(e® 1)• (eq(p_ 1)® i$)] = ( 1)mq ~(y). ) Since
,q} = ~_q)Dq(p.l){ i: } = iq , v(q)@(eq(p_l) ®(i : P) = iq * * pO{.~
Thus
(i)m q v(o~(y) = i and the result is proven.
Corollary 8.3. Let (K,D) e ~ . W r i t e D(k) = E k(1)®'''®k(P)e Cq (Kp)
for k ~ Cq(K), and regard each k (i) as an element of Cq(K). acocycle. y(k) = E Then P°{x}
Let x ( C ~ K ) suchthat
be
is represented by that cocycle yE Cq(K)
for each k e C (K). in particular, if P q D(k) = N i e K p f o r e a c h k ( K, whe re N = E ai e Z w, then p o = 0 on H':"(K). P Proof. po {x} By formulas (5.1) and(5. Z), y = v(q)O(eq(p_l)®xP) represents
x(k(1))''"x(k(P)) e Z
, and the r esult follows by an easy computation from Definition 7.5 and
the lemma. W e now give a useful application of the theory for which the Steenrod operations satisfy the results of T h e o r e m 7.9 and pO = 0. Let ~ and ~ denote
the categories of restricted Lie algebras and of primitively generated Hopf algebras

202

over V: ~
Z . Let P > ~
F: (~_ > ~
denote the free restricted Lie algebra functor, let
denote the universal enveloping algebra functor, and let P : ~   > ~ H c 24 its restricted Lie algebra of primitive 6.1~], PV(L) = L for
denote the functor which assigns to elements. L e~ and
B y a result of Milnor and M o o r e [19, T h e o r e m VPH = H for H e 94 By T h e o r e m s
of Witt and Friedrich [9, if E ~ F K = PT(K) , J~ and
Theorems and T(K) in ~ ~9~
7 and 9, p.168170],
extended to restricted Lie algebras, in ~ and
is the tensor algebra of K, then V ( F K ) = T(K)
These statements clearly remain valid for the categories ~ of simplicial objects in 0, ~ , and ~
[see 15, DefinitionZ.l]. W e
shall need
the following algebraic l e m m a . Lemma 8.4. Let L be a restricted Lie algebra and let IV(L) = K e r ~, Let Then, for
: V(L) > Z P ~:V(L) each
, be its augmentation ideal.
> V(L) p = V(L) @ ... @ V ( L ) y ~ V(L) p
denote the iterated coproduct. such that L ~(x) = Ny.
x e IV(L), there exists Proof. Let w:FK~> g
represent
as a quotient of a free restricted is an e p i m o r p h i s m of Hopf algebras, that K is a finite
Lie algebra. and w e m a y
Then
V(w) = V ( F K ) > V(L) L = FK. Since
a s s u m e that
Clearly w e m a y T(K)
also a s s u m e
dimensional Z module. P
admits a grading under which it is connected, ~ is zero on the
[19, Proposition 4.20] implies that the pth p o w e r operation augmentation ideal of the dual Hopf algebra implies that, for x e IT(K), (x) = N y zi = 0 + ~ z ~... i ~(x) T(K)*.
The cocommutativity of T(K)
can be written in the f o r m B y the triviality of ~ on IT(K)*, each
~ z.l in T(K) p.
and the result follows. for the study of h o m o t o p y invariants '~2 M ~2as follows. The
W e n o w sketch a definitional f r a m e w o r k of simplicial restricted Lie algebras. objects of ideal of L ~2 are pairs (L,M)
Define a category L e ~ in and
such that
is a restricted Lie are m o r p h i s m s
and the
morphisms
f: (L, M) > (L',M')
 203 
f : L > L '
in ~ ~
s u c h that
f(M) C M'.
Two s u c h m o r p h i s m s ,
f and g, a r e
s a i d to be L i e h o m o t o p i c if t h e r e e x i s t m o r p h i s m s of r e s t r i c t e d Lie a l g e b r a s h.s L q >L' q+l ' 0 _< i_< q, such that hi(Mq) C M'q+1and the identities (i)  (iii)
of [15, Definition 5. I] are satisfied. Define the homotopy, homology, and eohomology groups of (L, M) ~ (i) (Z)
The
~Z by
and and H*(L, M) = H*(IV(L/M)) are as
~;.(L, M) = H.(L/M)
H. (L, M) = H.,.(IV(L/M)) homology and eohomology
groupson the right sides of these equations L/M and IV(L/M) 5.3] shows
defined at the start of section 7, with Z modules. P m o r p h i s m s in cohomology. The ~Z
regarded simplicial as that Lie homotopic
argument of [15, Proposition
induce the same m o r p h i s m s on homotopy, homology, and and H,(L, M) are, respectively,
B y [15, T h e o r e m ZZ. l], ~r,(L,M) and of IV(L/M)
the homotopy groups of L / M Hurewicz h o m o m o r p h i s m
regarded as simplicial sets. The m a y thus be defined as the m a p Since
h: ~r,(L,M) > H,(L, M)
induced on homotopy from the inclusion L/Ivl>IV(L/M).
IV(L/M) = IV(L)/IV(M), we have natural long exact homotopy, homology, and cohomology sequences on pairs (L, M) E ~ 2 ' formation of long exact sequences. of the Hopf algebra H,(V(L/M)) and h defines a natural trans
Note that H ~(L, M) is the augmentation ideal is of finite type. Consider It can be proven that
if H,(V(L/M))
F A S n = F(~ n ~ Zp),where Sn is the simplicial nsphere.
Wn(L) is the Z module of Lie homotopy equivalence classes of m o r p h i s m s P F A S n  > L for L ~ ~ and that H . ( F A S n)    ~;,..(, Sn+l) is the augmentation ~ ideal of the free commutative algebra on one primitive generator of degree n. Our theory immediately yields Steenrod operations on H*(L,M). T h e o r e m 8.5. ~pS There exist natural h o m o m o r p h i s m s pS and ,if p > 2,
defined on H#(L, M) for (L, M) ~ ~ Z "
These operations satisfy the con
 204 
clusions of T h e o r e m 7.9 (except that the hypothesis of (iii) is not satisfied in general) and, in addition, the operation pO
Proof. diagonal Theorem We may regard
is identically zero.
as an object of ~,,, 7.4(ii). with Thus
G,(L, M) = C(IV(L/M))
D = ~b , t h e r e d u c e d c o p r o d u c t a s d e f i n e d in R e m a r k s 7.9 applies directly, and
pO = 0 f o l l o w s f r o m t h e p r e v i o u s l e m m a a n d
corollary. In [22], Priddy has given a different definition of H, (g) and H~'~(L) be the functor from simplicial Z algebras to ~(G P is a simplicial Z algebra, then % ( A ) p q > 0, as
[15, p. 87]. Let W
defined by M o o r e [20]. If A o'
= Zp
and Wq(A) = A q i (~" ~A ""
Z modules. P
For L ~ t~
The face and degeneracy operators are as defined in
, WV(L) is a simplicial cocommutative coalgebra with
coproduct ~ given by
!
~(aq_l~... ~ao) = ~ where ai ~ Vi(L ) satisfies ~ (ai) = ~
(aq_l®... ~ a ' o ) @ ( a " a _ l ~ . . . a ! ~ a!' . Priddy defines
1 1
~gao)'
H~,¢(L) = II~(IWV(L)) and H~T'(L) = H;:~'(IWV(L)), where simplicial Z module. P
IWV(L)
is regarded as a
For spectra, Priddy's definition and ours clearly differ
only by a shift of degree; with his definition, H ~ ( F A S n) ~'~(Sn+l) = Z . By P Definition 7.3 and R e m a r k s 7.4(ii), (C(IWV(L),~) H$(L) • L@0~ and therefore Priddy's
alos admits Steenrod operations which satisfy the conclusions of T h e o r e m
7.9 (except, in general, for (iii)) and pO = O.
9. The dual homology operations; Nishida's theorem For applications to loop spaces and to obtain a result used in the proof of the Adem relations, w e shall discuss the homology operations P~
S
whose duals are the
Steenrod operations on the rood p cohomology of a space X.
H#(X)" = H;:,(X) 'l' : Horn Z (H:.(X), Zp) "
Of course,
a n d , if H:~(X) i s of f i n i t e t y p e , H~(X) = H (X).
Define
P~¢ on H~(X)
s
by
P
pS
s ~ " pS = (P.#) ; # is clearly welldefined if H:~(X) is of
 205 
finite type, and either a direct limit argument or the next proposition imply that pS ~ . is welldefined in general, : if p > 2. Our results on the pS W e shall write the operations Adem P/ p ~ s lowers degrees by s if p = 2 and by 2s(pl) immediately yield the dual results for the Pf. on the left; the order of composition in the dual is a left module over the
relations must thus be reversed (that is, H~(X)
opposite algebra of the Steenrod algebra). the proof of L e m m a tion [30] of the D..
1
The following proposition was used in
4.6.
F o r m u l a (2) of the proof is essentially Steenrod's defini
Proposition W % C g , ( X ) p.
9. t . Consider
Let
X be a s p a c e a n d l e t
d = ~(t@D):W@wCt(X Let
)> Then
d . : H , ( ~ ; H ~ ( X ) ) , .   > H'(w;H,',(x)P)'.,. . o ,
x ~ Hs(X).
(i) (ii)
If p = 2, d l~(er~X ) = ~ e r + z k _ s ~ P k ( x ) ~ p . k ( x . ~ , ) k If p > Z, dg,(er@X ) = v ( s ) ~ (  l ) k er+(Zpk_s)(p_l)@Pgk(x)P k  6 (r) ~ s  l ) ~ k where Proof. v(2j+E) = ( t) j (m : )e and 6(2j+e) =e , g= 0 or i. is of finite type. W e shall compute (l)ker+p+(Zpk_s)(p_l)~ p k ~(x)p.
W e m a y a s s u m e that H~(X) H*(w)~H;I"(X) Z w~B P
d g':H(w;* H *:'(x) )  > p Lemma 1.3,
and then dualize.
In the notations of
H,(x)P~ A@
as a wmodule, and
H,(~; H,(X) P) ~Hg,(~) @ A (~)B. W e claim first that d~'~(B = 0. ¢) from B
It follows that H*(~ ; Hg'~(X) ~ H~(w) @ A* (~) B*. p) To see this, w e m a k e explicit the isomorphism Z n~B)*. p For y ~ B , define
to the homology of ( W ~ Z ~@B)*
p
~c ( W @
by
= ~(w)y(b) for w ~ W, 0 K i < p, b ~ B.
~(w@ aiOb)
Then
v:Z P
y is a cocylce and y>~
w> Z P by v(t) = I and
induces the desired isomorphism.
v ( a i ) = 0, t < i < p, a n d d e f i n e
Define
P w@B)*
y~ (W@Z
for y ~ B
by
y ( w ~ c ~ l ~•b ) = 6(w) v ( i ) y ( b ) for wc W, 0 < _ i < p, b ~ B.

206

Clearly
~(w@aiOb)
= V(N[wO~i@b]).
Therefore = ~d,(Nw~x) = 0
d*(~)(wOx) =~d,(w~x) f o r w ~ W and x ~ H,(X) since Nw~x
is a b o u n d a r y in W~H.~..(X)
(because
d(ezi ) = ne2i_l and d(TPZezi+l) = Ne2i in W). W e next compute d on H (~r)~A.
This p r o v e s that d""(B*) = 0.
.(X). By Let y ~ Hq(x) and x c H pq:
Definition 7.5, we have the formula (:) Di(Y)(X) = @.(ei@ yP)(x) = (:)iqd~:~(e@ l)*(yP)(ei~)x) from the tensor product of duals to the dual of tensor
(where the isomorphism
products has been omitted from the notation). * p Let w.1 be dual to e..: Then (E~l) (y) = w o ~ y p and therefore
d (WoO yP)(eiO x) = (0iqDi(Y)(x).
F o r any z ~ H P q  i ( x ) , the sign in the definition of a gives (w i O z)(e i O x) : (  t ) i ( p q  i ) z ( x ) . C o m p a r i n g t h e s e f o r m u l a s , we see that (2) d*(w o O y p )
= ~
(l)iwi@Di(Y)
To compute
d (wjOyP)
for j > 0, observe that if p " W   > W
= Zp~wW
is the
natural epimorphism, then we have the commutative diagram: d#
(W@W) @~r H,(X)
> t@d,
(W @ W) ~ H . , ( X ) p
p@l@l I W@ (W ~ . H . t (X))
$ p@i@l > ~ @ (W % H,(X)p) Dually, d is
(The upper rectangle requires an easy acyclic models argument.) a morphism of H (w)modules.
wjw i = w i + j if p = 2 o r e i t h e r
Now
wj(w O @ yP) = w j ~ y P
and, in H (w),
i o r j is even and ~ w i = 0 if p > 2 and i and j
are odd by formula (1.2). Therefore (2) implies the formulas
 207 
(3) (4)
if p = 2, d ( w j ~ y P ) = i ~ wi+ j ~ D i ( y )
and ® DZi+t(Y).
if p > 2, d (wjl~y p) = . ~ W z i + j ( ~ D z i ( y )  6(j+t) E w z i + t + j
1
By formulas (5. t) and (5. Z) and a reindexing, (3) and (4) become (5) if p = 2, d (wjO yP) = k~Wj+q_k@ pk(y) if p > Z, d (wj@yP) = v(q) t E ( k t)k
k
and pk(y) @ ~pk(y).
wj+(q Zk)(Pi)I
(6)
wj+(q Zk)(P 1) ~
 6 (j+1) v(q) i E (  I ) k
We now dualize,
d~..(H..(~)@H,,(X))C H,.(w)~A
since d'~:"(B = 0o F o r xe H (X), '1')
w e m a y therefore write
d.(er~X) = E Let y ~ Hq(X). (7)
e r + s _ p q ~ E q r ( x ) P ' Eqr(X) ~ Hq(X) .
Using the K r o n e c k e r pairing < , >, we have = (  t ) ( r + s ' q + m ) q < y , Eqr(X)> .
<Wr+s_pq@yP , d , ( e r @ X ) >
Since < p k ( y ) , x >
= <y, lmk(x)> , (5) implies that if p = 2, then
(8)
Thus
<d ( W r + s _ Z q ~ y Z ) , e r ~ X >
E qr (x) = P~, .
= <Wr@PSq(y),er@X>
= <Y'P*'sq(x)>"
sq
(x) if p = 2 and, with k = sq, this implies (i). N o w a s s u m e
that 'p > 2. By (6),
q = sZk(pi)E, For
d * ( W r + s _ p q ~ y p) has a summand involving Wr only if
and E= 0 or i, hence E qr (x) = 0 for other values of q.
k>_0
q = sZk(pi),
(6) gives
(9)
< d (Wr+(Zpk_s)(p_t)~yP), k+mq
er•X>= t
v (  q )  t (  t ) k + r q < y , plk(x)>. k
if q = sZk(pl); since
B y (7) and (9), Eqr(X) = (I) (I) m q v(q) I = that < ~ y , x > v(q) =
v(q) P,(x)
v(s), this yields the first s u m of (ii). Observe next by the chain and cochain definitions of the 6 (f) = (1)degf+Ifd used in defining C~(X).
= (l)q+l<y, ~x>
Bockstein and the sign convention N o w for q = s2k(p1)i, (6) gives
 208 
(10)
<d ( W r + p + ( Z p k _ s ) ( p _ t ) ~ y P ) , e r ( ~ X > = 6 ( r ) w (  q )  t (  t ) k+r(q+l)+q<y,pk~B(x)>;r.
since
B y (7) a n d ( t 0 ) , E q r ( X ) : (1)k+r+mqw(q)16 (r)p:k~(x) if q = s  Z k ( p  t )  l ; 6 (r) = 0 if r is e v e n and (1)mqw(q) "t = w(q) = v ( s  t ) ,
t h i s y i e l d s the s e c o n d
s u m of (ii) and so c o m p l e t e s the proof. Remark 9.2. The p r o o f a b o v e u s e d no p r o p e r t i e s s p e c i f i c to t o p o l o g i c a l s p a c e s
a n d so a p p l i e s to c o m p u t e d , = <~,(1 O D ) , : H,(w ; H , ( K ) )   > H,(w ; H , ( K ) p)
in terms of
,
and ~P,
s
(where ~P:: is defined by
(.6
<s), =~3P s
if no
Bockstein is present) for arbitrary objects (IK, e ~ D)
We now give a n e w p r o o f of a r e s u l t due to N i s h i d a [Zt], w h i c h is e s s e n t i a l to the c o m p u t a t i o n of S t e e n r o d o p e r a t i o n s in i t e r a t e d l o o p s p a c e s . K(Zp, 1) : E/w w h e r e C , ( E ) = Z p rr@ C , ( E / r r ) . be a w  m o r p h i s m o v e r w o p e r a t e s p r o p e r l y on the a c y c l i c s p a c e Let Z . p Let
E; by [14, I V l l ] ,
f: W > C,(E) ;,"
or: E   > E/Tr be the p r o j e c t i o n and let If W = Z ~ W, t h e n f i n d u c e s p ~r
f: W   > C~.(E/w), and
fp
is homotopic to
~ f, 9 :W > W.
By Remarks 7. 2, if I] is the shuffle
X:
m a p , t h e n we h a v e the f o l l o w i n g h o m o t o p y c o m m u t a t i v e d i a g r a m f o r a n y s p a c e
wif<xl
w%
f® 1
>
c,(E)® c,(x)
> c,(E x
T[
x)
I I®D
>
I IXD
c,(x)P
,1 ¢ ® t
%C,(X) p

Ce.(E) {~,w C,(xP)
"
> C,(E x X p)
~ DX 1
rl > C...(EX E'X X p)
~ DOl
W e W
,n(f® f)® n> c,(E x E ) ~ c , ( x p)
o x 1 ® 1
~" I
oX
I X1T l
W'O W
%C.(X) p
n(7®f)® ~ >c,(s/~
"
x E) @c,(x p)
"p) "
n >C,(E/~xEx
Tr
X p)
hori
Let
1¢, = ""I . 9 ~ I ~ I ) , [ j ~ @ I ) ; : . ' H , [ w ; " , [ X ) " " H
> H ,i w ;w ) ,@x ),P ) . " " I H" ~ H
"
" " The
zontal arrows are homology isomorphisms and we therefore have Steenrod operations p hes
O n
H.(w,H.(X)), H:~(w;H,=(X) ' and H.,.(w)~H, (w; H,,(X)p) such that d,~ p) , , .~. ~,¢ ,¢ 4
 209 
and },g. c o m m u t e w i t h the
evaluate the p,.S on
pS
The following theorem uses
d~ and bt,. to
H~(~r; H~(x)P).
O u r result differs f r o m Nishida's by a sign;
the reason for this is that our formulas (2) and (6) in the proof above differ f r o m the corresponding formulas in [30,p. I03 and p. li9]. W e w e r e pedantic about W e shall need the
signs in the preceding proof because of this disagreement.
following identity on binomial coefficients in the proof of the theorem. Lemma Proof. 9.3.
~Q(i,ai)(ni,i+bn) i
= (n,a+bn)
for
a > _ 0 , b > _ 0 , and n > _ 0 .
gives the only nonzero
The result is obvious if b = 0, w h e n
i= n
s u m m a n d on the left.
the triples (a,bl,n)
Using
and
(ci,d)
+ (c,dt)
= ( c , d ) , we find that the r e s u l t f o r
(a,b,n).
(a,bl,nl)
implies the result for the triple
Theorem9.4. (i)
(ii)
L e t X be a s p a c e , x ~ Hq(X).
S
T h e n , in H.,.(~r;H,(x)P),.,. :'; P..,}(x)z .
If p = 2,
Ps,c ( e r @ x2) = ~  ( s  Z i , r + q  Z s + Z i ) e r _ s + 2 i ~ i
S
If p > 2, P:= (er ~ x p) = ~
(spi,[~ ] + q m  p s
r
+pi)er+z(pi_s)(p.l) ~ P..jI(x)P
+ 6(r)~(q) ~ ( s  p i  l ,
[ r+l l Lj J + q m  p s + p i ) e r + p + Z ( p i . s ) ( p _ i )
~ P,~ ~(x) p I.
i
where
~(q)
=
v(q)lv(ql)
=
(l)mqm. ~
and
6(2j+a)
=E
,
~.= 0 o r
Proof. then eo ~ x p
W e a s s u m e that
s > 0, since the result is trivial for s = 0. If r = 0, In H,.(E,.X xP),
is in the image of H~(E. X X p) > H,(E,..X xP).
P,,S(e ° ~ x p) = ~ .
such that ~
ii e ° ~ P.:~ ( x ) ~ . . .
The s u m
i ~ P:,..,P(x) s u m m e d o v e r all p  t u p l e s ( i l , ,.. , ip)
i. / ik lies in e •NH.,.(X) p 3 o ,s = pt, w h e n
i = s. p
of all terms with any
S
and is thus zero in H,..(E XTrXp)"
Therefore P:. ( e o ~ x P ) = 0 unless
P;:i(eo e x p) = e° ~ P~¢ (x)p, which is in agreement with (i) and (ii). Recall that, by Definition i. 2 and the proof of L e m m a 4.6, w e have the following relations in
H :..(Tr). (a) i If p = 2, % (%)  ( i , j  Z i ) e j _ i and , ( % )  ~ ej®er_ j

2 1 0

(b)
i If p > Z, P~(ej)= ( i , [ j / Z ]  p i ) e j _ z i ( p _ l ) where 6(r,j) = i unless
and ~(e r) = ~ 6 ( r , j ) e j O e r .
j
r is even and j is odd, when 6 ( r , j ) = O.
If q = 0, then d ~ ( e r ® X ) = e r ~ x p and t h e r e f o r e
p,S(er @ xp ) : d~.,(P~:(er) ® x) s
s = P~, (er) ~ xp ;
b y (a) and (b), the r e s u l t holds in this c a s e .
and for fixed q by induction on our fixed q and r' < r. Let r.
We now p r o c e e d by i n d u c t i o n on q
Thus a s s u m e the result for q' < q and for
z = P:S(er(~.. x) and let z' denote the right side of z  z' = ~ e l ® Yi e H#(~; H.~.(x)P). W e shall
the equation to be proven. first prove that
Write
~,(zz'). = e® ~ (zz'); this will imply that Yi = 0 for all i > 0 such that Yi ~ 0, then e.x® e® ~ Yi clearly occurs as a Y® = 0 by explicit c o m 
since if i is m a x i m a l nonzero s u m m a n d
of }*.:..(zz'). W e
shall then prove that
putation and so complete the proof. p = Z and p> Z.
S
W e give the details separately in the cases
(i) p = Z. Since F~:,(z) = 1D~ ~ # ( e r ~ X 2 ) ,
we find by (a), the C a r t a n formula in
r, that
H , ( ~ ) ® H#(~;I~.(x)Z), and the induction hypothesis on
(c)
~,(z) = P : ( ~ e j ~ e r _ j ® x Z
J
) =~(i,jZi)ej.i®P:,
1, J
si
(er_j~xZ),
where if j > 0.
psSi(er_ j ® x Z ) = ~ k ( s ' i ' Z k ' r  j + q  Z s + Z i + Z ~ e r  j  s + i + z k ~ P " k ( x ) Z " The t e r m s with j = i > 0 a r e z e r o s i n c e
(i,j) such that ji = i>0, with
(i, i) = O. Applying the l e m m a to t h o s e
n = sZk, w e see that
a = ~, b = ri +qs, and
(c) r e d u c e s to the f o r m u l a
(d)
~,(~) = eo®Z + ~
k,~ >0
(sZk,r+qZs+Zk)e~ ®er.s+Zk_~ ® P,k(x)Z
A glance at the right side of (i) shows that ~t~.=(zz') = e® ® (zz'), hence Yi = 0 fo r i > 0. To c o m p u t e Yo' observe that p® = 1 and Proposition 9. I imply
(e)
er ®
x2 : d # ( e r + q O X ) + ~
k>0
e r + Z k ~ P  ~k(x)Z "

211

.q
Pg:d,.. = d , P : : , the C a f t a n f o r m u l a on I{,(w)~ H , ( X ) , and P r o p o s i t i o n evaluate pS d , ( e r + q ~ ,
9. i
x), and the induction hypothesis on q evaluates k > 0. Carrying out t h e s e c o m p u t a t i o n s , we find t h a t
pS k Z , ( e r + z k ( ~ P,,~..(x) ) f o r (e) i m p l i e s the f o r m u l a
(f)
z
=
r+qZs+Z
k,£ +
)er.s+Zk+Z
p,~ (x) z
~ ( s  2 ~ , r + q + k  Z s + Z ~ ) e r _ s + Z k + Z ~ ® P,~,,.. e,.k(x) z,0 k>0,~ but o u r a r g u m e n t w i t h e o, w i t h Z(k+l ) = s  r . ~ , s h o w s t h a t we Let t = k+l s
In p r i n c i p l e ,
(f) m u s t i m p l y (i) d i r e c t l y , involving
need only consider those terms and c = qkl;
then these terms
become
(g)
~ (kt,t+cZk)eo~ , k>O
~
+ ~ < s+t
s,
By formula (f) of the proof of T h e o r e m 4. 7, rephrased as in section 5 and dualized (with the o r d e r of c o m p o s i t i o n operations p,s reversed u n d e r d u a l i z a t i o n s i n c e we a r e w r i t i n g the 4. 8, (g) w o u l d be z e r o if I = s+t
on the left), and by R e m a r k s
w e r e a l l o w e d in the s e c o n d s u m ; thus (g) r e d u c e s to (h) (c+t,sZt)eo~: `
_ s+t,
(x) = (qs,r)eo@
,g
P 1/2 ( s  r )
:=
(x)2.
=
Since (h) is equal to the s u m m a n d of z' involving e , it follows that Yo
0
O.
(ii)
p > Z.
For brevity of notation, write d = g(pJ). As in the case p = Z, we r that si ~ (er_j@xP) , where, if j > 0,
find b y (b) a n d i n d u c t i o n on (i) ~,(z) =~
5(r,j)(i,[j/Zlpi)ej_di@P.
ij
pssi(er_ J.~xp) = ~] (s "iPk' [ r~" k z + q m  PS +pi+pk)e r J+d(pk s +i)~ P"k(x) P "
+ 5(rj)a(q) ~(sipki, [r~z+l + qmps+pi+pk)er_j+p+d(pk_s+i )~ p k ~(x)p ]
k The terms with j = di > 0 a r e z e r o . B y the l e m m a , a p p l i e d to t h o s e (i, j) s u c h
that jdi=~>O, with a = [~/Z], b = [ ~ ]
+qms(p1), and n = spk forthe

212

f i r s t s u m and a =
[~/2],
b
= [r~t 
Z
]+ q m  s ( p  t ) ,
and n = s  p k  i
for the second
s u m , (i) r e d u c e s to (j) ~,(z) = e ~ z o + Z 6 ( r , g ) ( s ' p k , [ ~ ] + q m  p s + p k ) e i ~ e r + d ( p k _ s ) _ I ~P~k(x)P
r
k,~>O
~
+ ~ 6(r,~ )5(r~ )a(q)(spkl, [~]+qmps+pk)eg~ er+p+d(pk_s)_ g ¢9 P:~(x)p . k, i >0
Now 6 ( r , l )5(r£ ) = 5 ( r ) 6 ( r + t , l ), and it follows f r o m a glance at (ii) that Thus Yi = 0 for i > 0, To compute Yo' o b s e r v e that
~(zz'), = eo ~ ( z  z ' ) "
Proposition 9. 1 implies that (k) e ~x p= u(q)ld,(er+q(p_l)~x )  ~
r k>O (t)ker+dpk~l='~k(x)P '
+ 5(r)~(q) ~(1)ker+p+dpk~ P~k~(x)P . k Precisely as in the case p = 2, we can compute Pj on the right side of (k);
c a r r y i n g out this computation, we find
(J[)
z = ~(i)k(si,[r/Z]+qmps+p~)er+d(pk+p~_s)~P~ k P~/(x)p
k,~  6(r)~(q)~(i)k(s~, [r/g] + qmps+pt )er+p+d(pk+p~ s) ~ PJ ~P~,f(x)p k,~
 ~ (  l ) k ( s  P ~, [ r / Z ] + k ( p  l ) + q m  p s + p ~ ) e r + d ( p k + p ~ _ s ) ~ P~ Plk(x) p r+l
k>O,l 6(r)~(q) ~ (  t ) k ( s  p ~  i , [   ~ ] +k(p1)+qmps+p~ )er+p+d(pk+p ~ _ s ) @ ~ ~ k (x)p k>O,f + 6(r)~(q)~(1)k(sp~
k,~ C o n s i d e r the f i r s t and t h i r d s u m s , with r+d(pk+p~s) = 0.
c = qd(k+f).
,
[~]+k(pl)+qmps+p~)er+p+d(pk+pi_s)(~ ,
p~
p k~(~)p
Let t = k+~s and
Then these two sums b e c o m e
(m)
k
(_l)k(k.t,t+mc _pk)eo~ p:,k +tk.(x) p,s "p

~
(  i ) s+t+~ (~ + m c  s , s  p ~ ) e o ~ P~: p~?+t~ (x)p r is odd, hence
~<s+t
C o n s i d e r the r e m a i n i n g sums of (~), with r+p+d(pk+p~s) = 0; 6(r) = t. Let t be as above and let c' = c  l .
Then these t h r e e sums become

213

(n)
~(q) ~ k
(l)k(kt,t+mc'pkt)eo
~
pk pf , ~ , (x)p +tf(x) p
 (~(q) ~ (  t ) s + t + f (f + m c '  s , s  p f  t ) e o ~ P . : ~ B P x ? ~<s+t
+~(q) 2 (i)s+t+~(:~+mc's'spl)eo~Pfp.:+ti(B(x)p
By formulas (h) and (j) of the p r o o f of T h e o r e m 4. 7 (with e = 0), r e p h r a s e d as in
section 5 and dualized,
we s e e t h a t (m) and (n) w o u l d be z e r o if ~ = s + t w e r e Therefore, by an e a s y v e r i f i c a t i o n , (m) a n d (n) r e i = k+i = s+t.
a l l o w e d in t h e s e c o n d s u m s .
duce to the following expressions, w h e r e (o) (spi, ~ + qm  ps+pi)eo~P (x)p
with dpi = dsr, and with dpi = dsrp. in its first s u m and (p) is
0
(p)
~(q)(spit,
r+l + q m  ps + p i ) e o ~ P!:~(x)p z
C l e a r l y (o) is e q u a l to the s u m m a n d of z' e q u a l to the s u m m a n d of z'
the proof is complete.
involving e
involving
e
O
in its second sum.
Thus
Yo
=
0 and
10.
The c o h o m o l o g y O f K ( % n) and the axiomatization of the pS
We r e c a l l the s t r u c t u r e
of H * ( K ( % n ) ; Z p )
= H*(Tr, n, Zp)
and c o m p u t e c o m 
p l e t e l y the rood p c o h o m o l o g y B o c k s t e i n s p e c t r a l
section.
s e q u e n c e of K(Tr, n) in t h i s
W e also s h o w (as should be wellknown) that Serre's proof [Z3 ] of the K(Z2, n) can be simply modified so as to apply w = Z pt '
axiomatization of the Sq I using in the case of oddprimes.
W e shall consider only the cyclic groups pco
I <_ t <__co, where, by convention, Z monomials relative to t. (a) p = 2. sk ~ l . For
= Z. W e first fix conventions on admissible
Notations i0. i. s i>_ 2si+ l and
I = (s I ..... Sk) , w e say that
I is admissible if
The length, degree, and excess of I are defined by and, if I = (s,J), e(I)= sd(/). Define
~(I) = k, d(I)= ~ s j ,

2 1 4

pI
t
= P
sI
.
""
pSk_ l
P
t
sk
'
where
P
t
s
=
pS
if s > 2,

ptI
= ~t if t < co, and
P
I
co
= 0",
thus,
if t = co, we a g r e e t h a t a d m i s s i b i l i t y
requires
s k>_ 2.
The empty sequence
I is admissible,
with length, degree, and excess zero, and
I determines the
identity o pe ration. (b) p > 2. For I= (El,s I ..... &k, Sk, £k+l), and 6.1 = 0 or I, w e say that I is I = (~). Define
admissible if si>PSi+l + &i+l
~(I) = k, d(I) = ~ £j + ~
sk >__i or if k = 0, w h e n
a n d , if I  ( £ , s , J ) ,
2sj(pt),
e(I) = 2 s + e  d ( J ) .
Define
pl g s £ k Sk ek+l t = ~ I p l... ~ P ~t , where
o i ~t = I for all t, [5 = ~t for t < co, t ~ k+l =0.
and ~1 = 0; thus, if t = co, w e agree that admissibility requires co
We now givea quick calculation of H*(Z pt' n, Zp).
Lemma
10. 2.
H (Z,l,Zp) = E(il)
and
H*(Z,2, Zp) = P(i2).
If t < co, then if
* H (Zgt ,I, Z2) = P(il), with p>2. Proof.
2 ~t(il) = iI , and
H*(Zpt, l, Zp) = E(il)@P(~t(il))
K(Z,I) = S I and
K(Z,2) = CPco,
so the first statement is obvious.
F o r the second statement,
H*(Zpt, l,Zp) = H (Zpt; Zp), and w e can define a precisely as in Definition 1.2 (with ring A. T h e result follows by an
A Z p t  f r e e resolution of A, with coproduct, p there replaced by easy computation. Theorem H*(Z
t ' n , Zp)
pt) for any c o m m u t a t i v e
10. 3.
If n >
2 (or if n = I and either
algebra
p > Z or t < co), then
is the free commutative and e(I) < n o r ,
generated by the following set: t = t}. More
I p {Pt i n l I is a d m i s s i b l e over, H*(Z Proof. n = 2. Assume pt'
if p > 2, e(I) = n a n d IIopf a l g e b r a . and
n, Z ) i s a p r i m i t i v e l y p The lemma the result
generated
gives the result for for nl. Of course,
t < co
n = 1 and for spectral
t = co a n d
the Serre
sequence
{Er}
of I<(Zpt,nl ) > E   >
E Z = H*(Z
K(Z
pt'
n), E
acyclic
satisfies
and E
pt' n, Zp)
~H*(Z
pt' ni , Zp)
co
: PZ

215

First,
let
p = 2; t h e n ,
r e g a r d i n g s q u a r e s as S t e e n r o d o p e r a t i o n s , H*(Z2t, n  t , Z2), a d d i t i v e l y ,
we s e e t h a t we as the e x t e r i o r
m a y r e w r i t e the p o l y n o m i a l a l g e b r a algebra E(S), w h e r e
S = { P:in_ I I I is admissible and By Theorem 3. 4, P:in. I transgresses to P:i n.
e(1) < n }.
Define an abstract spectral where 7(S)
sequence of differential algebras, {E'r}
'
by letting E' = P ( T S ) ~ E ( S ) ,
2
is a copy of S with degrees augmented by one, and by requiring dress to TS e T(S). Clearly f :E' >E r r r by f2 = g ~ l ' E' = ZZ. co where Define a m o r p h i s m zt'
s ~ S
to trans
of spectral sequences is the m o r p h i s m
g'.P(TS)> H*(Z g(TPtlin_l)
n, Z2)
of algebras defined on generators by with the differentials determines morphisms,
= Pt lin; clearly commutation o, fz and f are isoco
f for r > 2. Since r
*o f2 = g is an i s o m o r p h i s m m a y rewrite
by the comparison t h e o r e m [14, p. 355]. where
N o w let p > Z. W e
H*(Zpt, n, Zp), additively, as E ( S ) ~ Q ( T ) , (xp = 0 for x ~ T) and where e(1)< ni, d ( 1 ) + n
Q denotes a truncated polynomial algebra
s =
in_i I I is admissible,
even},
T = {Pt I" I I I is admissible, In_ (Note that e(1)  d(1) n o d Z, hence By Theorem d(1)+n even and
e(1) < n, d(1)+n odd}. e(1) = nI is impossible.) and, if Define an Let
3.4, Ptlin_i transgresses
to (l)d(1)Ptli n
d(1)+n= 2q+l, pli n ~(P:in_i) pi
transgresses to (i) n ~ P q P : i n.
abstract spectral sequence of differential algebras , {Er} , as follows. E 'Z = [P(~S) @ E ( T ) • (the b r a c k e t e d e x p r e s s i o n s TT P ( ~ T ) ] @ [ E ( S ) @ Q(T)] respectively). ~T Here
a r e the b a s e and f i b r e ,
TS and
are copies of S and T with degrees augmented by one and p and then augmented by two.
is a copy of T
with degrees multiplied by {E'} r
The differentials in to trans
are specified by requiring Tt ~ TT, and Ttl~lt p  t
s ~ S to transgress to ~s E TS, t ~ T to transgress to ~t e ~T.
g r e s s to
A n easy computation
demonstrates that E' = Z . Define a m o r p h i s m co p
of spectral sequences

216

!
f :E' r r
> Er
by
f2 = g @ l ,
where
g:P(TS)@E(TT)~P(~T)>
H * ( Z p t , n, Zp)
is the m o r p h i s m
g(,PtIin.t)
of algebras defined on generators by
g(~P[in_l)=~ p q p I i n if d ( I ) + n = by commutation 2q+t. with the The last
= (t)d(I)PtIi n and f r for
As in
t h e c a s e p = 2, t h e differentials, and
r > 2 are determined
g is an isomorphism
by the comparison
theorem.
statement follows since, by the external Cartan formula, if X
x~ H~'(X) satisfies ~b(x) = ~ x'e x", then
is an Hspace and
~PS(x) = ~
i+j = s Thus, since i
n
•pi(x,)®PJ(x,,)
and ~t(in)
and,~(x) = ~~(~(x')®x"+(t) degx' x' ® ~(x")).
s o a r e a l l of t h e pI t in spectral sequence
are primitive,
We c a n now c o m p u t e t h e
rood p cohomology Bockstein
{ E r ) of K(Zpt, n ).
algebras such that I3r for
Recall that {Er} ~¢ t E t = H (p , n , Zp)
r > t. pr Since
is a spectral sequence of differential
and Er+ t i s t h e h o m o l o g y of E r with
r e s p e c t to
H (p , n , Z)
i s a d i r e c t s u m of c y c l i c g r o u p s w i t h Ira(13. __ C E r r)
OO
one g e n e r a t o r generator of
of o r d e r
f o r e a c h b a s i s e l e m e n t of
and
one
of i n f i n i t e o r d e r f o r e a c h b a s i s e l e m e n t o f E determined, E l = E t and additively, by Et+ t = E oo
, the integral cohomology If t < oo a n d n = t,
K(Zpt, n ) is completely t0, Z implies that
{Er}°
Lemma
; Z , hence we need only conp
sider the case Theorem
n ~ 2. i0.4. Let n ~ 2. Define a subset S of the set of generators for 10.3 by ~(I) > 0}.
El = H~(Zpt' n, Zp) (a) (b) For
given in T h e o r e m
If p = Z, S = {Ptlin I sI and If p > 2,
d(1) + n are even and
S = {ptlin I ~I = 0, d(I) + n z(y) = ~(y)y + pZq~(y)
is even, and ~(I) > 0}. if p = Z and degree (y) = Zq and define Ar(n,t ) by
y c S, define
z(y) = ~(y)yp1 (c)
if p > Z. Define an algebra
Ar(Zn ,00) = P(in}
and Ar(Zn+l ,cO) = E{in}if r < t < o o and
(d) Ar(Zn,t)= P{iZn} ®E{~t(Zn)}

217

rt
rt
Ar(Zn, t ) = P { i p 2n
where
} @E{m(izn)iP n 2n .
 P} if r > t , t = t
z(izn ) = ~(iZn)iZn+ P
~(12n ) if p = 2 and
and
(e) A r(2n+l,t)
z(i2n ) =
12n)1
if e i t h e r
if
p> 2 or
t > t
= E( iZn+t} @ P { ~t(iZn+~ )}
r <_t < co and
A (2n+J,t) = Z
r
p
if r > t. r r I yeS} ~ E{z(y)y p "
T h e n , if r _ > t ,
r
Er+i = P{yP
r
Pl
y e s} @ A r + i ( n , t ) ,
~ r + t ( y p ) = z(y)y p  P
Proof.
for
y ~ S, and
r r ~ r + t ( i P n ) = z(izn)iP n P. p > 2. Let
W e first compute E 2 separately in the cases p = 2 and T and U of the set of generators of E l by ~(1) > 0}
p = 2 and define subsets
T = {PIinl sI is even, d(I)+n is odd, e(1) < ni, and U = {P[inl d(1)+n Recall that ~pSi = spS d(J) + n = 2q+l, and is odd, e(1)= nl, and ~(1)> 0}.
and observe that if Ptlin e U, then I = (2q, I), w h e r e be the (additive) s u b c o m p l e x of E l
~ID lin = (P ff in) Z . Let C t
which is the tensor product of the following collections of subcomplexes:
(i)
P{~(y)} @ E { y }
E{y}
for
y e T, and
deg(z) = 2q+l. Then H(IC) = 0 under 6, and
(ii) P{z 2 } ~
for y = PZqz e U,
Let IC be the positive degree elements of C. therefore E 2 is isomorphic to H(EI/IC). C
If y e T U U, then p Z q y e U, is a
deg y = 2q+l, and therefore
is actually a subalgebra of E l and EI/IC
quotient differential algebra of E l . It is easy to see that
Et/IC = P{ytye where
S} ~
E{~(y) l y e
S} •
A~(n,t),
,
A~(n, t) is the q u o t i e n t of the p o l y n o m i a l a l g e b r a g e n e r a t e d by i and, if n
.2 t < oo, ~t(in) by the i d e a l g e n e r a t e d by in2 if n is odd o r b y ~t (in ) if n is even. Therefore

2 1 8

H(Et/IC) = P{yZ [ y~ S} O E{~(y)yl y~ S} ~ A'2 (n,t),
where A~(n,t) A~(n,t)= Az(n,t ) unless n is evenand For y e S or t = t, when y = i n if n is even and t = 1,
= P { i n 2} ( ~ E { ~ ( i n ) i n } .
z ( y ) = ~(y)y + p 2 q ~ ( y ) , d e g y = 2q, i s a c y c l e in E t ~(y)y i n Et/IC. Since Next, z(y) 2 b o u n d s in let Et,
w h i c h p r o j e c t s to t h e c y c l e E 2 has the stated
it f o l l o w s t h a t
f o r m i f p = 2. of E l by T =
p > 2 and define a subset
T of t h e s e t of g e n e r a t o r s
{ Ptlin I 5
= O, d(1)+n
is odd, and
~(I) > 0 }.
Then, as a differential algebra, E l breaks up into the tensor product of the following collection of subalgebras: (iii) (iv) (v) P { y } ~ E{~(y)} E{y}~ P{~(y)} for y ~ S for y e T
The free commutative algebra generated by in and, if t < oo, ~t(in). P{yP} ~ E{z(y)}, those of (iv) are acyclic, hence E 2 is as stated. Now a s s u m e ~r+t that and
The algebras in (iii) have h o m o l o g y a n d t h a t o f (v) h a s h o m o l o g y Er+ 1 is as stated, r >_ I A2(n,t), p.
and any
Then Proposition
6.8 computes
E r + 1 b r e a k s up i n t o t h e t e n s o r P{x}• E{~r+l(X)} , where
p r o d u c t o f A r + l ( n , t) w i t h s u b a l g e b r a s r x = yP , y e S. T h i s p r o v e s t h e r e s u l t .
of the form
Finally, w e prove the axiomatization of the pS
on topological spaces. commute with
Recall first that the Caftan formula and p O = I imply that the pS
suspension [28, 30] and that w e have shown in Proposition 8.1 that pO = i is implied by the c o m m u t a t i o n of pO with S ~ and the fact that pO a zero dimensional class. are in fact redundant. is the pth power on
Thus the axioms w e choose (for convenience of proof)

2 1 9

Theorem
I0.5.
T h e r e exists a unique family
{ P S I s .>_0}
of natural h o m o 
morphisms
H (X; Zp) > H (X; Zp) if p > Z, and
s u c h that
d e g ( P s) = s if p = Z,
d e g ( P s)  Z s ( p  t ) (i)
(ii) (iii)
pO is the i d e n t i t y h o m o m o r p h i s m
pS(x) = x p if p = 2 and s = deg x s > deg x or or p > 2 and 2s = deg x
pS(x) = 0 if p = Z and
p > Z and
2s > deg x
(iv)
(v)
pS(xey)=
~ pi(x) @ PJ(y) f o r x @ y e i+j=s
~ { Rsl
H*(XXY)
at"pS = pScr , w h e r e Proof. S u p p o s e given
is the suspension of a fibration. s > 0} w h i c h also satisfy the axioms. If
x ~ Hn(x, Zp), then x  f*(in) for s o m e
that pS(in) = RS(i ). n n = Z.
f: X > K(Zp, n), hence it suffices to prove
T h e result is obvious f r o m (i), (ii), and (iii) if n = I or if that pS(i n_1) = RS(in_l) for all s and consider B y (v), 0"(y) = 0,
p > Z and
Assume
y = pS(in)  RS(in ), 0 < s < n if p = 2 and w h e re
0 < Zs < n if p > Z.
• *: H*(Zp, n, Zp) > H~"(Zp, nl, Z ). P If p = Z, oLet p > Z. is an i s o m o r p h i s m in degrees less than Zn and therefore y = 0.
A s s h o w n in the proof of T h e o r e m
i0.3, (i) and (iv) i m p l y that both
pS(in)
and
RS(in ) a r e p r i m i t i v e .
By T h e o r e m
t 0 . 3 , we s e e that e(I) <
{pIi n I I admissible,
n}
T h e o n l y e l e m e n t s of this /3Pq(y),
is a b a s i s f o r the p r i m i t i v e e l e m e n t s of H*(Zp, n, Zp).
s e t w h i c h a r e in N e t ~* a r e p  t h p o w e r s and e l e m e n t s of the f o r m deg y = 2q+t, w h i c h h a v e d e g r e e and is thus z e r o . at l e a s t 2pq + 2. If n is odd, t h e n
y has odd d e g r e e
If n is e v e n , t h e n all p r i m i t i v e e l e m e n t s in K e r ~* have d e g r e e y = 0.
pn, which is g r e a t e r t h a n the d e g r e e of y, and a g a i n
 220 
11.
Cocommutative
Hopf algebras ~ . The objects of Hopf algebra
In this section, w e consider the following category are triples C = (E,A,F) ring where A,E Thus E A
is a (Zgraded) cocommutative F
over a c o m m u t a t i v e
is a right and and F
is a left (Zgraded) c o c o m and c o c o m m u t a t i v e ~ are co
mutative Acoalgebra.
are A  m o d u l e s
algebras (not necessarily unital or augmented), morphisms of A  m o d u l e s . We say that C
and their coproducts
is unital if E and F are m o r p h i s m s
are unital and A is a
a u g m e n t e d and their units and augmentations morphism morphism morphisms a e A, and ~ : C   > C' in ~is atriple
of A  m o d u l e s . k:A> are A'
~ = (~,k ,~), w h e r e and ~:F > F' and
of Hopf algebras and of coalgebras; thus f e F. We
~ : E > E'
k equivariant for e e E,
~(ea)=~(e)k(a)
~(af) =k(a)~(f)
say that "f is unital if c~ and ~ are m o r p h i s m s For C and C' in ~ , define C~C'
of unital
a u g m e n t e d coalgebras0 F~F') e ~
= (E~E',A~A',
and observe that ~ (E@E, is unital if C ~" by
st
~b = (~,~,~):C = (E,A,F) is a m o r p h i s m in ~ ; clearly ~
A@A,
F~F)
= C@C
is unital. Define h o m o l o g y and
c o h o m o l o g y functors on the category (I) We Hst(C) = Tor(A'A)(E,F) st and
Hst(c) = Ext(A,A)(E,
F*)
m
shall define and study Steenrod operations on
H%'(C) w h e n
71 = Z P
. The
results here generalize w o r k of Liulevicius [13]. In the following definitions, w e recall the description of H*(C), with its product, in t e r m s of the bar construction. Definitions il.l. Let IA For C = (E,A,F) e ~, let C'= (A,A,F) e ¢ Define the bar construction where T(JA) B(C)
be the cokernel of the unit A ~ A. B(C) = E~) T ( J A ) ~ F JA.
as follows. algebra on
as a A  m o d u l e ,
is the tensor
Write elements of B(C)
in the f o r m
e[all.., las]f; such an t = deg e + ~ d e g ai + deg f,
element has homological degree
s, internal degree

221

a n d t o t a l d e g r e e s+t,. Define
e.:B(C)>E ~
A
F and d:B
,(C) by s, . ( C )   > Bs_I,
(2)
~(e[ ]f) = e~f,
~(e[all... ]as]f) = O, and
(3)
d(e[a l l    l as]f) = @al[aZl.. l as]f si
 ~ ~[~11...
i=i ~[:il... l:s_l]asf , w h e r e : = (i)i+deg x
dS + S d = t  ~ ,
I ~i_ll~iai+llai+zl""
laslf
If E = A, then d is a m o r p h i s m of left A  m o d u l e s and
where ¢: F ~B(~) and S: Bs,.,..,.(~) Bs+l, ,,.(~),.are defined by the formulas ~
(4) Clearly ¢(f) = [ ]f and S ( a [ a l l . . . d : l~Ad las]f ) = [ a t a l l . . . tas]f
.
on B ( C ) : E ~ A B ( ~ ).
By adjoint a s s o c i a t i v i t y ,
HomA(B(U), E*) ~ B*(C) =~ HomA(B(E , A, A), F$). T h e r e f o r e (1) admits the equivalent r e f o r m u l a t i o n (5) H,(C) = H(B(C)) Definitions 11. Z. Whitney map and H*(C) = H(B;'(C)) = Ext(A,A)(F, E'*). be objects of ~ . Define the A l e x a n d e r 
Let C and C'
~:B(C~C')
~ B(C)(~B(C') and the shuffle map by the formulas
:B(C)®B(C')~ B(C@C')
(6)
~ ( e ® e ' [ a l ® a~ I . . . I as ~ a's]f® f')
= ~s(  I )~(k)e[a 1 1* * * k=O
where
l aklak+ 1. • .a S f®e'a~...a~[a~+ll k
**t
la']f' sj
~(k) = deg e'(k+deg a l . . . a s f ) + ~ deg a: ( k  i + d e g a i + t . . . a s f ) i=t
S
+
~ (I +deg aj)degaj+l...asf j =k+l
l as+t If')
,
and
(7)
~](e[atl... las]f®e'[as+il..,
=
where
1
~ (t) v ( " ) e ® e ' [ a ( ~ ) l . . . l % ( s ~ ) ] f ® f '
7r 1
,
a. e A if i<__ s, a. e A' lr (see [ t 5 , p .
if i > s, the s u m is taken o v e r all t7]), and
(s,t)shuffles
 222 
( t + deg a i ) ( t + deg a s + j ) . (~) = ~(i) > ~ ( s + j ) The u n n o r m a l i z e d b a r c o n s t r u c t i o n E (~ T ( A ) ~ F graded Amodule u n d e r which ~ and a d m i t s a s t r u c t u r e of s i m p l i c i a l
~ a r e in fact the c l a s s i c a l n o r m a l i z e d
Define D = ~ B ( ~ ):B(C)> B ( C ) @ B ( C ) .
A l e x a n d e r  W h i t n e y and shuffle m a p s . Then D gives
B(C) a s t r u c t u r e of c o a s s o c i a t i v e c o a l g e b r a ; if C is unital, t h e n If E = A, t h e n D c o i n c i d e s with the m o r p h i s m
B(C) is unital and a u g m e n t e d .
of left A  m o d u l e s
defined inductively by
(8)
(9)
D([ ]f) =
~
[ ]f'@[
S = S@
]f"
if
~(f) = ~
on
f'~f",
and
DS = SD , where
I + fie @ S
B(~)eB(~).
Clearly
D on B(C) = E ~ A B ( ~ ) is the c o m p o s i t e E~AB(~)~ B(C) ~ ~ T ~ t> E ~ A B ~ ) ~ E ~ A B ( C  ).
E~AB(~) ¢@ D >E~
We (I0)
define the cup product on
IB~:=(C) to be the composite >[B(C)eB(C)]* D * > B*(C).
u:B*(C)~gB*(C)'~
We h a v e the following analog of L e m m a 7 . t ;
of
a m o r e p r e c i s e analog (in t e r m s
~ ) could also be proven, and an alternative proof by semisimplicial rather
than homological techniques is available. Lemma of A W~B(C) bigraded 11.3 . Let o =Aw Ir be a subgroup of with A~r generator = ~ WiGBjt(C i+j=s ). ~ and let W be a A w  f r e e . Bigrade of resolution
r
such that W by
e . Let o
C e ~
[W~B(C)]st
T h e n there exists a m o r p h i s m
Awcomplexes
A: W ~ B(C) > B(C) r which is natural in C
and satis
fie s the following prope rtie s :
(i)
~(w~b)
~(eo@b
= 0 if b e B
) = D(b) A
o,
,(C)
and w e W. f o r i > 0 1
D is the iterated coproduct where A operates
(ii) (iii)
if b e B(C), w h e r e is a m o r p h i s m
If E = A, then
of left A  m o d u l e s ,
 223 
on
WOB(G)
by
a(wOb)
= (t) deg w deg a WOab.
(iv)
A(WiOBst(G))
= 0 if i > ( r  t ) s . Z~ a r e n a t u r a l l y A~rhomotopic. of A e n s u r e s that (iii) is
Moreover, Proof.
a n y two s u c h
Observe first that the cocommutativity of A
compatible with the =equivariance the result when
O b s e r v e n e x t t h a t i t s u f f i c e s to p r o v e A on WeB(G)= EOAWOB('C~ )
E = A, s i n c e w e c a n t h e n d e f i n e
to b e t h e c o m p o s i t e EOAWOB(~ where ) ,OZk> ErOAB(~)r.. We d e f i n e
s. Formula defines
U
> [E ®AB(~)]r
WiG B s t ( ~ ) by induction on d~ for i
U is t h e e v i d e n t s h u f f l e .
i by induction on
A on
and for fixed
(i) defines ix for i= 0
s = 0 and all i> 0 Let i ~ 1
and f o r m u l a (ii) and wequivariance and s > i and a s s u m e Let { Wk} that A
and all s.
has b e e n defined for W.. 1
i' < i and for our given i and it suffices
s' < s.
be a diwbasis for for w ~ {Wk}
B y (iii) a n d wequivariance, si, ~:~). Let
to define A ( w O S ( y ) ) r s = ~ (~)i®s O i=1
and
y e B
~ri1 on
B(~) r.
Then
dS + Sd = i  (~6
)r.
We
define
(v)
d~(w@ S(y)~) :
(  1 ) d o g w S A ( w O y) + S a ( d ( w ) O S(y)).
w = e o and that (v) is welldefined by the
O b s e r v e that
(v) is equivalent to (ii) on T o verify that Then:
induction hypothesis.
d & = d~d, write (v) in the f o r m
A ( I ~ ) S ) = SZ~(I ~)i + d O S ) .
dA(tOS)
= dSA(IO t + dOS) = [a  Sa(dOi
= (1  S d ) A ( I O
1 + dOS)
+ lOd)](iO
I + dOS)
= A + A(d O S)  S A ( d O t )  S A ( i O d) + S A ( d O dS)
= a + a(d®S)
 S a ( d O t )  S a ( t O d) + S A ( d O t)  S a ( d O Sd)
= a + a ( d O S)  a ( i O dS) = a + a ( d O S)  a + a ( t O dS) = a(dOi + iOd)(iOS)
( w h e r e no t e r m s
involving
c E a r e r e l e v a n t b y (i) a n d a n e a s y v e r i f i c a t i o n ) .
Thus
 224 
(i), ( i i ) , ( i i i ) , a n d (v), t o g e t h e r w i t h ~ r  e q u i v a r i a n c e , of a natural morphism A of A~rcomplexes.
provide an explicit construction observe that
To s e e t h a t (iv) h o l d s ,
if w ~ {Wk} C W
i and y = al[a21.., las]f, then ~(w~S(y))
i
is a linear combina
tion in B(~) r of terms involving precisely the factors a.(j) and f(J) in the B(~), where ,(ai) = ~ a~i)~... ~a'(r)1 and ,(f) = ~ f ( 1 ) ~ . . . Thus no s u m m a n d of A(w~S(y)) ~f(r) give
the iterated coproducts. degree greater than rs.
can have homological
Since A ( w ~ S(y)) has homological degree i+s > rs if
i > (rl)s, this proves (iv). The uniqueness of A up to A1rhomotopy follows easily by use of the contracting homotopy W e now pass to the category Definition ii.4. Let C ~ ~ S on B(~) r.
~D(Tr, oo,A) of Defitions 2. i. Let ~:B=~(C) r  > [B(C)r] ~" be the natural
m a p and define a A ~  m o r p h i s m
(tt) Since
@:W~B*(C)>
B~"(C) by the formula r
w e W, x e B"(C) r, k e B(C). j: W ~ V a s in
@(w@x)(k) = (1) degw degx~(x)A(w@k), @ m a y be d e f i n e d f o r T~= X
r
and then factored through
D e f i n i t i o n 2. t , a n d t h e r e s u l t i n g c o m p o s i t e i s n a t u r a l l y A ~ r  h o m o t o p i c to t h e original @ defined in terms of W, @ s a t i s f i e s to g i v e c o n d i t i o n (ii) o f D e f i n i t i o n Z. 1. By
the lemma, (12) (t3)
f o r m u l a (11) s p e c i a l i z e s for any
@(eo@X ) = D*~(x)
x e B*(C) r
and and w ~ W. Since @ is natural by setting is
@(w(~ x) =
~(w)D * ~(x) f o r a n y x e
B ° ' *(C) r
B y ( t 0 ) a n d (12), @ s a t i s f i e s on morphisms r(c)= in ~,
c o n d i t i o n (i) of D e f i n i t i o n 2. t . F: ~ >
we thus obtain a functor and
~O(~r,o o , A )
(B*(C),@) ~ , then
onobjects F(C) P
] ? ( ~ ) = B*(,~) o n m o r p h i s m s . ~ (%co,A).
B y (13), i f C
unital in
is unital in
If A = Z p , ~r i s c y c l i c of
order
p, and C = ~ ® Z
where
~ is Z  f r e e
(that is, ~ , 2 ,
and ~
are Zfree),
C; I~(C)
t h e n w e a g r e e to c h o o s e w i l l t h u s be r e d u c e d
@ for
C to b e t h e
r o o d p r e d u c t i o n of @ f o r
r o o d p.
N o t e t h a t if x e B * ( C )
h a s b i d e g r e e ( s , t), t h e n
@(w@x) has b i d e g r e e
(s  degw, t).
W ( ~ B * ( C ) r and B*(C) should be thought of
 225 
as regarded by total degree in defining the functor Observe that, by Definition 6. i, w e n o w have C e ~ . When
F. viproducts in B*(C) for any
A = Zp, the results of Proposition 2.3 will clearly apply to the (C), and the following l e m m a s will imply the applicability
Steenrod operations in H
of the e x t e r n a l C a f t a n f o r m u l a and the Adem r e l a t i o n s .
Lemma i i. 5. F o r any objects C and C' in ~ , the following d i a g r a m
is A •  h o m o t o p y
commutative.
w ® B*(C ®C,) r
e
>
B*(C ® C,)
W @ [B*(C) (~ B'*(C')] r
Proof.
~
> B*(C) @ B*'(C') commutativity of the d i a g r a m
It suffices to prove the A ~  h o m o t o p y
W ® B(C)(~ B(C') u(z~® z~)(1@ T @ i)(~® l® I) > [B(C)® B(C,)]r I®~ [ i®n
...........
W®B(C®C'),,
a
C and C'
> B(C ® c')r
replaced by C and C'. the
and this d i a g r a m need only be studied with Since B ~ ' ) r and
[B(~)~B(C')] r have obvious contracting homotopies,
result follows by an easy double induction like that in the proof of L e m m a Corollary 11.6. Lemma Proof. it.7. If C e ~ If C e ~ , then F(C) is a Cartan object of F(C) is an A d e m
i i. 3.
~(%co,A)
, A = Z , then P
object of ~(p, oo).
Precisely as inthe proof of L e m m a
7.8, it suffices to prove the
"rhomotopy commutativity of the following diagram:

226

2 Wi@W:@B(C) ' w@i
>Y~B(C)
....... A
> B(C) p AP
[
W?~Wi
H*(C)
r~l
~B(C)
I (~ &mw?~
B(C) p
U
> ( w z ® B(C)) p
Z We need only consider this diagram with C replaced by "C, and, since B(~) p
has a contracting homotopy, the result then holds by another easy double induction. the properties of the pS and ~ps on
T h e following t h e o r e m s u m m a r i z e s for C ~ ~ , A = Z P . We
shall be very precise as to grading since there is W e are thinking of H*(C) A n alternative
considerable confusion on this point in the literature.
as regraded by total degree in applying our general theory.
formulation Theorem
isms pi
that is sometimes II.8. Let C
~pl
convenient will be given after the theorem. • ~ , A = Z Then . P there exist natural homomorph
and, if p > 2,
defined on
H (C), with
(a) (b)
p i : H s,t(C ) p i HSt(c) ~Pi: HSt(c)
> Hs + i ' t ' z t ( C )
if p = 2 ; and if p > 2 .
> H s ~ Z i  t ) ( p  i ) ' Pt(c) > HS +i +(2it)(P i)' Pt(c)
T h e s e operations satisfy the following properties: (i) pi = 0 if p = Z and either i < t or 2i<t or i > s+t 2 i > s+t
pi
= 0 if p > Z and either
~p1 = 0 if p > 2 and either (ii) (iii) pi(x) = x p if p = 2 and if C = ~ Z p , where ~
2i< t or 2i >__s + t 2i = s+t
i = s +t or if p > Z and is Zfree, then pi if p > 2.
~pii = ipi if p = Z and ~pi
is the composition of ~ and
(iv)
pj= ~
pi~)pji
and ~ P J = ~
(~pi@pji+pi@~pji)
or H*(C@C');
the internal Cartan formula is satisfied in H"(C).
 227 
(v)
If 7:C' > C
and
¢:C> C"
are unital m o r p h i s m s
in
~
such that
~N = 0 on the cokernels of the units, then where (vi) The Proof.
uP i = pi0 and
u~Pi = _ ~pi0_'
u:HSt(c '')> HSl't(c ') is the suspension. pi and ~pi satisfy the A d e n relations as stated in Corollary 5. I. The pi and
If x c Hst(c), then Di(x ) = 8:¢(ei~xP ) e HPSi'pt(c).
~p1 are defined by formulas (5. i) and (5.2), with x having its total degree q = s +t ; thus (a) and (b) are valid. The vanishing of pi(x) for i < t if p = 2 ll.3. The
and of ~ pi(x) for 2i< t if p > 2 follows f r o m part (iv) of L e m m a
remainder of the t h e o r e m follows immediately f r o m our general theory and the previous l e m m a s . BXC(C '') B~(~) F o r (v), note that the composite
> B (C)
*
B*(~)>B":"(C,) is zero on the kernel of the augmentation
B (C") > Z . A n alternative formulation of (v) in the nonunital case can easily P be obtained. In addition to (v), the I~udo transgression theorem, appropriate spectral sequences involving objects of is seldom satisfied in practice, and ~pl ~ Theorem 3.4, applies to
. The hypothesis of (iii)
is generally an independent operation There is an alternative definition of the
having nothing to do with any Bockstein.
operations, which amounts to the following regrading of our operations . Define (c)
(d)
~i
" =Sq I= pi+t: HSt(c) ~ Hs+i, Zt(c) if p = Z ; ~i = pi+t: H s, 2t(c ) ' Hs+2i(pi)' 2pt(c) and ~i= ~pi+t:HS, Zt(c ) ~ H'+Zi(Pi)+i'2Pt(c) if p >2.
This regrading is reasonable if p = 2, but has the effect of eliminating all operations on Hst(c) for
t odd if p > 2; of course, these operations are nontrivial
is nontrivial in ~P ;
since, if s and t are odd, the pth power operation on HSt(c) general.
The results of the t h e o r e m can easily be transcribed for the P~ and
for example, the A d e n pi and ~pi
relations are still correct precisely as stated but with all and ~i The motivation for the reindexing is just This operation is of
replaced by ~i ~o
the desire to m a k e
the first nontrivial operation.
 228 
particular i m p o r t a n c e Definition I i. 9.
in the applications, Let
and w e n o w evaluate it. , where E, A, and F are positively
C = (E, A, F) ~ ~ Then B#(C) may
g r a d e d and of finite type. E ~T(IA ~')~F . Define
be identified with by
k :BSt(c)
> B s'pt(C)
>~(e[++i ]. . .
Then k commutes
I%]@)
= eP[=IPl.l+s
p]~p "
k~ : H s t ( c )   > H s'pt(C). Of
with the differential and induces
course, if p > z ,
and thus k~.. = 0
then
×(~[~ll...f%]~)=0
if E,~,I or ~ has odd degree
if t is odd. Let C = (E,A,F) Let ( ~ , where E,A, andF are
Proposition 11.10.
positively g r a d e d and of finite type. Then
x ~ M s , t(c) w h e r e
t is even if p > 2.
T°(x)= Proof.
k~,(x). Let y = e[a I ]... that (i) ms v(s)l(e'[a~l... summands ]a's]f' ) + N z , e'~... ~e',a!~... 1 ~ a 1. , ' ] a s ] f ( B s , pt(C). A straightforward, but t e d i o u s ,
calculation demonstrates A(es(p_l)~y
) = ~
w h e r e the s u m is t a k e n o v e r the s y m m e t r i c and f'~... ~f' of the i t e r a t e d c o p r o d u c t s . G is a g r o u p ,
(A m o m e n t ' s
r e f l e c t i o n on the c a s e
C = (Zp, ZpG, Zp), w h e r e
and a g l a n c e at L e m m a 8 . 2 s h o u l d c o n T h e r e s u l t now f o l l o w s
v i n c e the r e a d e r of the p l a u s i b i l i t y of this s t a t e m e n t . ) e a s i l y f r o m the d e f i n i t i o n s . Remarks ~°(x) 1 1, l i .
If p > 2, t h e n it c a n be s h o w n b y a t e d i o u s c a l c u l a t i o n t h a t where C = ( Z p , A, Zp) e ~

= < x > p for x ~ H l ' 2 t ( C ) ,
6.9
•
and < x > p is as
2t(c) '
defined in R e m a r k s and any C c ~
It is possible that
~S(x)
= < x > p for x ~ H 2s+i
, but this appears to be difficult to prove.
 229 
Bibliography
I. J.F. A d a m s , Comment.
O n the structure and applications of the Steenrod algebra.
Math. Helv., i958. The relations on Steenrod powers of cohomology classes. A symposium in honor of S. Lefschetz, 1957.
2. J. A d e m ,
Algebraic geometry and topology. 3. Mem.
4.
S. Araki and T. Kudo,
Topology of H spaces and Hsquaring operations. n
Fac. Sci. Kyusyu Univ. Ser. A, 1956. W. Browder, H o m o l o g y operations and loop spaces. Illinois J. Math. ,
1960. 5. A Dold, Uber die Steenrodschen Kohomologieoperationen. i961. 6. E. D y e r and R.K. 1962. Epstein, Steenrod operations in homological algebra. Invent. Lashof, H o m o l o g y of iterated loop spaces. Amer. Ann. of Math. ,
J. Math., 7.
D.B.A.
Math. I, 1966. 8. Sci. G. Hirsch, Quelques proprietes des produits de Steenrod. C.R. Acad.
Paris, 1955. 9. N. Jacobson, S. K o c h m a n , D. K r a i n e s , T. Kudo, Lie Algebras. Ph.D. Thesis. Interscience Publishers. University of Chicago, Trans. A m e r . 1962.
I0. I I. 12.
1970. Math. Soc. i966.
M a s s e y higher products.
A transgression theorem.
Mere. Fac. Sci. Kyusyu Univ.
Ser. A , 1956. 13. A. Liulevicius, The factorization of cyclic reduced powers by secondary Mere. A m e r . Math. Soc., 1962. Press, i963.
cohomology operations. 14. S. M a c L a n e ,
Homology.
Academic

230

15.
J.P. May,
Simplicial objects in algebraic topology°
D. V a n Nostrand
npany, 1967. 16. The structure and applications of the EilenbergMoore
ctral sequences (to appear). 17. 18. 19. , H o m o l o g y operations in iterated loop spaces (to appear), , The cohomology of the Steenrod algebra (to appear). J. Milnor and J.C. Moore, O n the structure of Hopf algebras. Ann. of
h., 1965. 20. J.C. Moore, Constructions sur les complexes d'anneaux. Seminaire
ri Cartan, 1954/55. 21. G. Nishida, C o h o m o l o g y operations in iterated loop spaces. Proc.
m Acad., i968. 22. S. Priddy, P r i m a r y cohomology operations for simplicial Lie algebras.
ppear in Ill. J. Math. 23. J.P. Serre, Cohomologie modulo 2 des complexes d'EilenbergMacLane.
ment.
Math. Helv., 1953. Algebraic Topology. McGraw Hill B o o k C o m p a n y , 1966.
~4. E. Spanier,
~5. N.E. Steenrod, of Math., 1947. 6.
Products of cocycles and extensions of mappings.
, Reduced powers of cohomology classes.
Ann. of Math,,
7.
, H o m o l o g y groups of symmetric groups and reduced power Proc. Nat. Acad. Sci., U.S.A., 1953. Proc.
;ions.
3.
, Cyclic reduced powers of cohomology classes. i953.
%cad. Sci., U.S.A., ).
, C o h o m o l o g y operations derived from the symmetric group.
lent. Math. Helv., i957.

231

30. N. ]E0 Steenrod, 1962.
Cohomology operations.
Princeton University Press,