You are on page 1of 32

APPLICATIONS OF THE

AND

GENERALIZATIONS THEOREM

APPROXIMATION by ~. P. M a y

In its basic form,

the approximation t h e o r e m referred to provides simple ~ E X, w h e r e X
n n

combinatorial m o d e l s for spaces T h e first such

is a connected based space. is equiva-

result w a s given by J a m e s construction MX.

[26], w h o s h o w e d that ~ E X

lent to the J a m e s

T h e unpublished preprint f o r m of D y e r and = lira ~ n E n x , and M i l g r a m [41]

gashof's paper [25] gave an approximation to Q X gave a cellular m o d e l for ~2nEnx Starting f r o m B o a r d m a n for all finite n.

and Vogt's spaces

[5], Dold and Thorn's treatment

of j-tuples of little n-cubes n,j of the infinite s y m m e t r i c product N X in t e r m s of

~

quasifibrations [24], and the category theorists' c o m p a r i s o n between finitary algebraic theories and m o n a d s
n n

(as for e x a m p l e in B e c k [4]), I gave a n e w a p p r o x i m a -

tion

C X
n

to 12 E X

in [36]. This m o d e l has proven m o s t useful for practical n > I, and it is its applications and generalizations This will be a survey of w o r k by various people, and

calculational purposes w h e n that I wish to discuss here.

I would like to mention that I have also given a survey of other recent developments in iterated loop space theory in [39], updating m y T h e first section will give background, 1976summary [38]. relevant work,

mention miscellaneous

and discuss generalizations, notably C a r u s o and W a n e r ' s recent homotopical approximation to ~ n E n x for general non-connected spaces X [9,11]. T h e second Both are based

and third sections will outline the two m a i n lines of applications.
n

on certain stable splittings of C X, due originally to Snaith [47]. One line, initiated by M a h o w a l d [33] and with other m a j o r contributors B r o w n and Peterson

[7, 8] and Ralph C o h e n [Zl], is primarily concerned with a detailed analysis of the pieces in the resulting splitting of 122Sq and leads to n e w infinite families of elem e n t s in the stable stems. The other line, primarily due to F r e d Cohen, Taylor,

and myself [16-19] but also contributed to by C a r u s o [10] and K o s c h o r k e and Sanderson [30], is based on a detailed analysis of the splitting m a p s and their homotopical implications and leads to an unstable f o r m of the K a h n - P r i d d y theorem, a m o n g various other things. T h e s e lines, and thus sections 2 and 3, are

essentially independent of each other.

39 § 1. B a c k g r o u n d and generalizations Tlie construction of the approximating spaces is naively simple. Suppose

given a collection of ~.-spaces ~. with suitable degeneracy operators J J th cri:~._ij " ~ . ,j i < i< j. (Here Z. is the j-- s y m m e t r i c group.) Given a space X J with (nondegenerate) basepoint *, construct a space C X = I ] ~j )<~.xJ/(~), where the equivalence relation is generated by (c,~xI ..... xj) "~ (c0-i,x ..... Xi_l,Xi+ I ..... xj) I if x.1 = * J

If each ~. = E., J J the resulting space is M X . If each ~. is a point, the resulting space is NX. J W e shall largely be concerned with examples C(Y,X) obtained f r o m the configuration spaces
B(Y,j) Artin's

See [16,§~i, Z] for details, examples, naturality properties, etc.

~j(Y)

= F(Y,j)

of j-tuples of distinct points of Y, and w e let
F(Y,j)/Zj. B ( R Z , j) : K ( B j , 1 ) , where B.j i s [30], we

be the orbit (braid) space g r o u p of j - s t r a n d e d braids.

Following 14oschorke and Sanderson

think of C(Y,X) subset of Y and

as the space of pairs X'L -*X

(L,X), w h e r e

L

is a finite (unordered)

is a function.

Here w e impose the equivalence rela-

tion generated by
(L,)~) ~ (L - {y},X IL{y})

if X(y) = "~ • L

The crucial example

C X
n

m a y be described similarly, with the sets
I
n

t a k e n to h a v e a f f i n e e m b e d d i n g s s e e [36,§ 4]. homotopy

-~ I

n

with disjoint interiors In as R n, b y a b u s e ,

as their elements; and obtain a natural

We t h i n k of t h e i n t e r i o r

of

equivalence g-C
n

X

~" C ( R n , X )

by restriction of little n-cubes to their center points.

(The proof that g is an

equivalence in [36, 4.8] is not quite right; it is corrected in [14, p. 485] and also in

[30].)
Define points s ~ Sn a n : C n X -~ flnEnx = In/01 n by I X(c)^c-l(s) ~n(L,X)(s) : ] if s ~ I m c for c ~ L . k The approximation t h e o r e m [36, 6.1] asserts that
n

by letting an(L,k):S n -* X ^ S n be specified on

if s ~ I m c for c ~ L

is a w e a k equivalence if X is ~ is a group completion n w0f~nEnx is the universal

connected.

Segal [45] and C o h e n [14] later proved that case. This m e a n s that

in the general, non-connected,

40

g r o u p associated to the m o n o i d

~r0CnX

(which is e a s y [36.8.14]) and t h a t H , ~ n ~ n x by localizing at its s u b m o n o i d -i n n n hold for ~ng .'C(R ,X) -~ ~ ~n X.

is obtained f r o m the P o n t r y a g i n ring

H,CnX

~0Cn

X.

Of course, X

the same = S O , MoDuff

conclusions [40] gave

In the case property, between suitable

another

proof of the group homological sphere

completion

viewing it as a special C(M,S 0) manifolds and the space M.

case of a general of sections

relationship bundle of M for

of the tangent

She also gave thought

a construction and

C±(M,S O ) of pairs particles interest

of finite

sets in M, annihilation homotopical

with points properties.

of as positive

negative is of some

with suitable and yields in
fails to

While this construction to the space of sections

a

approximation

of a bundle,

the bundle

question is not the tangent s p h e r e bundle of M . n n be a h o m o t o p i c a l a p p r o x i m a t i o n to f~ S .

In particular, C ± (R n , S 0)

V a r i o u s other people h a v e tried to obtain h o m o t o p i c a l (rather than m e r e l y homological) a p p r o x i m a t i o n s to ~ n ~ n x problem is quite delicate. for general n o n - c o n n e c t e d spaces X. The

It is v e r y e a s y to give intuitive a r g u m e n t s

for plaus-

ible candidates but v e r y h a r d to pin d o w n correct details. h a s recently b e e n obtained by C a r u s o ~X n for ~ n ~ n x and Waner [II].

S u c h an a p p r o x i m a t i o n

T h e y construct a m o d e l n-I ~ In , w h e r e c"'l n-l ~I

by use of partial little c u b e s

c'Xc":[a,b]Xl

c': [a,b] -- I

is a linear m a p

(increasing or decreasing) and

n-I is an

affine little (n-l)-cube.

These

partial little c u b e s are required to a p p e a r in be thought of as piecewise linear m a p s In -~ In

closed configurations, w h i c h m a y

given b y a piecewise linear path in the first coordinate a n d a single affine e m b e d d ing in the r e m a i n i n g coordinates. T h e labels in X of the c o m p o n e n t partial little

c u b e s of a closed configuration are all required to be the s a m e . Caruso theorem [9] has generalized both this result and the original a p p r o x i m a t i o n n n by proving that $] C ( y , ~ n x ) is a g r o u p c o m p l e t i o n of C ( Y X R ,X) a n d ~

obtaining a h o m o t o p i c a l a p p r o x i m a t i o n Y

(Y,X) to ~2nc(y, ~ n x ) . T h e case w h e n n is a point reduces to the earlier a p p r o x i m a t i o n s , a n d the use of general Y Further are in

allows a quick inductive reduction of the entire result to the case n = i. generalizations, b a s e d on a c o m b i n a t i o n of the ideas of C a r u s o the w o r k s .

and McDuff,

Related to these ideas are C o h e n and Taylor's extensive calculations of

H,C(M,X) M = N

for certain X R n for some homology

manifolds positive of C(M,X) n.

M

[Z0].

Their

arguments

work

best

when on im-

In particular, for such M,

they give complete these calculations

information having direct

the rational plications

for Gelfand-Fuks

cohomology.

41 T h e r e is also an e quivariant generalization of the approximation theorem, the present status of which is discussed in [39, § 5]. H o w e v e r , w e shall mainly concentrate on the specific approximations C(Rn, X ) - g C X n n ~n ~ n X

and their applications in the rest of this paper. P e r h a p s I should first explain just w h y these approximations are so useful a O n e reason is that C X has internal structure faithfully reflecting that of n iterated loop spaces. M u c h of this is captured by the assertion that C is a m o n a d n and ~ is a m a p of m o n a d s [36, 5.2]. Less cryptically, the three displayed spaces n allhave actions by the little cubes operad ~ n and g and ~ are b o t h ~ n - m a p s n [16,6.Z and 36,5.Z]. In particular, they are H - m a p s , but this is only a fragment of the full structure preserved. CmXACn Y Further, there are s m a s h and composition products ) and C n X X C n SO -+ C n X tool.

-~ C m + n ( X ^ Y

which are carried by the ~ iterated loop spaces

to the standard s m a s h and composition products for n [36,§8]. There is a similar and compatible s m a s h product on

the C(Rn, X), but there is no precise point set level (as opposed to homotopical) n composition product on the C ( R ,X). B y specialization to X = S O , ~ :C S O -~ ~nsn is a m a p of topological n n monoids. A s C o h e n has recently observed [13], this leads to approximations for the localizations of the classifying spaces BSF(n) for sn-fibrations (with section).

Indeed, C n S 0 = l J- ~ n , / ~ q and ~nsn = I I ~nsnq , w h e r e ~nsn consists of q>__0 q q~ Z q the m a p s of degree q. If p is any prime, then the m a p of classifying spaces m~ n "B( ~ _ ~n,pi/~pi) i>_0 -~ B( [ I i>0 ~ . S n) n pl

is an equivalence, and the universal cover of the target is the localization of BSF(n) a w a y f r o m p. When n -- co, this result is due to T o r n e h a v e and is a

special case of a general p h e n o m e n o n [37,VII § 5]. A key reason for the usefulness of the approximation t h e o r e m is that spaces of the general f o r m CX c o m e with an evident natural filtration. The successive and x [q] q where X [q] denotes the q-fold s m a s h p o w e r of X. C X n are D X = ~+ A n,q Z x [q] q

(and equivalent) quotients of C(Rn, X) Dq(Rn, x) = F(IRn, q)+ A S

and

n,q

'

Just as the simplicial version

42

of the J a m e s

construction admits the splitting ~IVfX -~ V ~X[q] found by Milnor q [42], so K a h n in 197Z proved that Barratt's simplicial m o d e l [Z] for Q X splits stably as the w e d g e of its filtration quotients; K a h n has just recently published a

proof [Z7], a different a r g u m e n t having been given by Barratt and Eccles [3]. W h e n w o r d of Kahn's splitting reached C a m b r i d g e , mation theorem, w h e r e I w a s lecturing on the approxi-

Snaith [47] w o r k e d out a corresponding stable splitting ~°°C X --~ V ~°°D X , n q>_l n,q

where my

Z °o is the stabilization functor f r o m spaces to spectra (denoted Q New proofs of such splittings by Cohen,

earlier papers)°

in all oo Taylor, and myself

[16] are the starting point of the w o r k discussed in section 3.

Incidentally, by w o r k

of Kirley [29], these splittings for n >_Z cannot be realized after any finite n u m b e r of suspensions (see also [16, 5.10]). There are two points of view on these splittings. One can either ignore h o w they w e r e obtained and concentrate on analyzing the pieces or one can concentrate on the splitting m a p s and see what kind of extra information they yield. T h e s e two

viewpoints are taken respectively in the following two sections. T h e crucial reason for the usefulness of the approximation t h e o r e m is that w e have very good homological understanding of the filtered spaces Historical background and complete calculations of H,~..~QX and X = S O , w h e n the latter is ~ H~..BZq) q>_0 complete calculations of H,~n~,nx and [lZ, App] and also in [7].) C Xo n H,~.,~CcoX (including

are in [14,11. C o h e n [14,111] has given H, C n X . (Some m i n o r corrections are in as Hopf algebras over the

H e r e "complete" m e a n s

Steenrod algebra, with full information on all relevant h o m o l o g y operations , Since these operations are nicely related to the geometric filtration, complete calculations of all H.Dn, qX drop out. H e r e h o m o l o g y is understood to be taken rood p for s o m e prime p, but w e also give complete information on the Bockstein spectral sequences of all spaces in sight. A m a j o r d r a w b a c k to these calculations is that they give inductive formulae for the Steenrod operations, but not a global picture. as a m o d u l e over the Steenrod algebra Ao for n = Z and X X n,q The solution to this dualization p r o b l e m One wants to k n o w H D

a sphere is basic to the w o r k of the next section, and w e shall n > Z. Before turning to this, however, I

also say what little is k n o w n w h e n

should mention the related w o r k of Wellington [51].

lie has solved the analogous

dualization p r o b l e m for the algebra structure, giving a precise global description * n of H ~ ~,nX for all c o n n e c t e d X (or allX if p = Z; corrections of [51] are needed

43 when p > Z. ) He has also studied the problem of determining the A-annihilated
n n

primitive elements in H, fl E X.

M o r e is said about this in [39,§4]

(but the

description there should have been restricted to connected X).

§Z. The spaces Dq(Rn, S r) andthe Brown-Gitler spectra
The w o r k discussed in this section began with and w a s inspired by Mahowald's brilli~nt paper [33]. I shall reverse historical order by first discussing the w o r k of B r o w n and Peterson [7,8] and Ralph C o h e n [ZI] on the structure of the spaces R (nr ) --~D sr and then briefly explaining the use of this analysis ,S n,q for the detection of elements of the stable stems. I a m very grateful to C o h e n for lucid explanations of s o m e of this material. (In case anyone has not yet noticed, it

is to be emphasized that there are two different Cohens at w o r k in this area. ) Let ~ be the q-plane bundle n,q F(Rn, q) X E (RI) q ~ q B(R n, q) .

With Thorn spaces of vector bundles defined by one-point compactification of fibres followed by identification of all points at infinity, it is obvious that the Thorn space of ~n,q is precisely Dq(Rn,SI). Replacing R 1 by R r in this construction, the

Thorn space is Dq(Rn, sr). connected. fibration). Lemma

resulting bundle is the r-fold Whitney s u m of ~ with itself and the resulting n,q n r Here DI(R ,S ) --~ S r and Dq(Rn,S r) is (rq-l)• Let Jn,q denote the order of ~n,q (or better, ofits associated S q-

W e have the following evident periodicity (see e.g. [37,111 §I]). Z.l.
D q( R n "

,S r+jn'q~ }is equivalent to E°Dn'qmq(R n, sr). Dq(R n,S r) is to determine what is presently known.

Thus the first problem in analyzing the spaces the n u m b e r s Let • in, q" The following lelcama s u m m a r i z e s

Vp(j) denote the p-order of j (the exponent of p in j). Theorem Z.Z.
z~(n-l)

(i) Jz,q , where

g for all q >_ Z . 9~(n-l) is the vector fields n u m b e r (namely the n u m 0 < i < n).

(ii)

Jn,Z

=

ber of i - 0, I,Z, 4 rood 8 with (iii) Jn, q divides Jn, q+l (iv) F o r an odd prime

(and, trivially, Jn,q divides Jn+1,q ) . q < p and Vp(Jn,p) = [n-l/Z].

p, Vp(Jn,q) = 0 for

44

(v) (vi)

For any prime J4,4 = 12 .

p,

Vp(Jn,q ) = Vp(Jn,pi) if p i < q < pi+l .

N o n e of t h e s e is v e r y h a r d .

P a r t (i) w a s f i r s t p r o v e n by C o h e n , M a h o w a l d ,

a n d M i l g r a m by u s e of v a r i o u s r e s u l t s of m i n e i n i n f i n i t e loop s p a c e t h e o r y , but B r o w n l a t e r found the t r i v i a l t r i v i a l i z a t i o n of 2 ~ Z , q = ~ 2 , 2 q d i s p l a y e d i n [15].
(ii), B(R n, 2) ~ R P n-I are in Y a n g of F.Cohen. and ~n,Z is the canonical bundle

For

~ (g& . Parts (iii)-(v)

[SZ] and w e r e also proven by Kuhn.

Part (vi) is an unpublished result

It remains to determine the n u m b e r s Vp(Jn ' pi) for i >__Z and n >_>3 ,

and this is an interesting and apparently difficult problem. In connection with this, the only general w o r k on the K-theory of spaces n n ~ X for I < n < oo that I a m aware of is the computation by Saitoti [43] and for X a finite torsion-free C W - c o m p l e x . However,

Snaith [48] of K~,(eZE3x;zz)

there is w o r k in progress by Kuhn. As would be expected f r o m the l e m m a , D q ( R n,S r) when n = Z than w h e n n > Z. m u c h m o r e is k n o w n about the spaces Before restricting to n = 2, however,

we summarize

the results of B r o w n and Peterson [8] in the general case. Their n r m a i n result gives the following splitting of certain of the D (R ,S ). It is provenby q using (ii) of T h e o r e m 2.2, the Thorn construction, and various structure m a p s f r o m [36} to write d o w n explicit splitting m a p s and then using F. Cohen's calcula-

tions in [14,111 and IV] to check that they do indeed produce a splitting. W e adopt the convention that Theorem D0(R n ,X) =

O" S
Then

Z.3. Let t >__1 (except that t > 1 if n = 2, 4, or 8).

Dq (Rn' stY(n- l)t - n) is h o m o t o p y equivalent to [q/Z] n-i z~(n- l)t-n) S2 ~(n- i) ÷it-n- I) V Dq_zi(R ,S A Di(R n, . i=0

As u s u a l , [m] d e n o t e s the g r e a t e s t i n t e g e r

<_m.

B r o w n a n d P e t e r s o n [8] cohomologi-

a l s o o b s e r v e t h a t the N i s h i d a r e l a t i o n s i m p l y the f o l l o w i n g 2 - p r i m a r y
cal periodicity. Lemma 2.4.

L e t ~ ( n - l ) be d e f i n e d by 2 J # ( n - l ) - I < n_< 2 ~ ( n - 1 )
~,* n z¢(n- i) + r) H Dq(R ,S y q?~(n-l) ~,Dq(Rn,sr).

Then

45

Since one can add a multiple number 2~(n-1)u-n, the previous

Zkb(n-l)s lemma

to any

z~(n-1)t-

n

s o a s to r e a c h

a

and theorem

have the following conse-

quence. Corollary Z.5. As a module over the m o d Z Steenrod algebra, ~* n Z ~(n- l)t n H Dq(R ,S - ) is isomorphic to

[~]i=0 H DN~= q-Zi( Rn-I,S Z¢(n-l)t-n) ~

H~*Di(R ,n sZ~(n-1)t-n -I )

Brown

and Peterson [8] note one further cohomological As a module over the m o d ) ~ ~ i=0

splitting.

Proposition 2.6.

Z Steenrod algebra,

H Dq(R ,S X~ K 0 = Z2and

~ (q-zi)zk~(nml)t K i ~I DZI(R n,S Z•(n- 1 )t)

where

~ K., i>0, is dual to the sub A - m o d u l e
1

of

spanned by a n m o n o m i a l s mental class of S Z~(n-l)t

in the

Ql(L)

not divisible by ~ (where L is the funda-

and the suspensions are realized by multiplication with

q-zi).
These last two cohomological metrically. splittings m a y or m a y not be realizable geo-

B r o w n and Peterson conjecture that they exhaust the Z-primary as

possibilities in the sense that ~eDq(Rn, S r) has no non-trivial direct s u m m a n d s an A - m o d u l e unless r - 0 or r ~ -n rood 2 ~(n-l), They point out explicitly at

the end of [8] that, at least w h e n

n = 3, there can be finer splittings than those are satisfied. The analysis is not yet

displayed w h e n the specified congruences complete and there remains m u c h

w o r k to be done.

In particular, virtually structure of the indecomposable

nothing is k n o w n about the explicit global A - m o d u l e surnmands w h e n n >_ 3.

The splittings of Cohen, together with g e m m a

Taylor, and myself in [17] (see T h e o r e m

3.7 below)

Z.1. and an easy homological inspection (compare [17, 3.3]) Z.3.

imply the following analog of T h e o r e m

46

Theorem

Z 7.

Dq(R n,S3n'qt) is stably equivalent to

~.jn, q(so qt
-

[o72]B(Rn,
i=2

V

2i)/B(R n ,2i-1))

.

Specializing n o w to the case imply that (i) Here Dq(RZ, S 2r+l)

n = Z, note that L e m m a

2.i and T h e o r e m

2.2(i)

--~ EZqrDq(R2,SI)

and

D (R2,S 2r) ~ Y,Z q r D (R2, S 0) . q q

Dq(R 2, S O) : B(R 2, c~ + .

(The disjoint basepoint was omitted in [15, p. 226].) Theorem 2.3 implies that

Since ~(i) = i and Dq("R i, S t,) --~S tq

D q (R z, s zt) --~ [o/21 V
i=0

EZt(q-2i)Di(R2, S4t+l) ,

t >_ 1 .

Setting t = i and combining with (i), w e find the splitting (Z) Z 2 q D q (Rz's°) -~ sZq ~ ( [o/Z] z Z q D i (R2, s:)) . V i=1 Z.7 and [17, 3.3]. Its original proof

This splitting is also immediate from T h e o r e m is in B r o w n and Peterson [7].

Clearly, then, analysis of the stable homotopy type of Zsr+2 V Dq(R 2,S r) reduces to analysis of the stable homotopy q> 1 R2 ' type of the spaces Dq( S l'). W e therefore abbreviate Xq= in what follows. Dq(a z,s:) The s

W e fix a prime p and localize all spaces and spectra at p.

results to follow are due to M a h o w a l d

[33], B r o w n and Gitler [6], and B r o w n and p >Z.

Peterson [7] at p : 2 and to R. C o h e n [ZI] at

The starting point of the analysis of the X rood p cohomologies. Define M(q) : A/A{x(~Spi)I i If p = 2, we let pi = Sq
xpP greases by direct i>q

Let X

is the determination of their q be the conjugation in the m o d p Steenrod algebra A.

and

~: 0 or i}. Davis' result [23] that
+ p

and suppress the Bockstein.

_(i+0

: (_l)i+lpp

i

pp

i-:

...

pppi

I

p(i+i) : i +p + -..

i
t

s o m e of t h e c o m p u t a t i o n s . inductive calculation

The following result is not too hard to prove Cohen's results on H Xq [14,III].

from F.

47 Mahowald's original a r g u m e n t w h e n Abbreviate M([q/p]) Z.8. = M[q/p]. H (Xq; Zz) --~ ZqM[q/2] and there is a stable 2-primary p = z [33] is s o m e w h a t different.

Theorem

c ofibration se quenc e EX2q_ i which on m o d sequence 0 " where ~ is the A - m a p Here ~2X2q_l M(q-I)~ f ~X z q g _ EZq X q

2 cohomology realizes the Zq ~

suspension of the short exact ~ EqM[q/2] * 0 ,

M(c 0

specified by ~(Eql)

= x(Sq q) and

~ is the natural A - m a p . EZX2q_2 . The

m a y be replaced by its 2-1ocal equivalent

key to the cofibration is the const@uction of g which, as M i l g r a m pointed out, is an easy exercise in the use of the classical J a m e s maps. the geometric and homological properties of the spaces The rest follows by use of C2S r in [36] and [14,111];

see [15, T h m . 2] or [33, 5.5]. The analog at odd primes is m o r e complicated but r r r+ 1 proceeds along similar lines [21]. Let M denote the M o o r e space S u e P Theorem hold. 2.9. Localize all spaces at p > 2. Then the following conclusions

(i)
(ii)

X

q

is contractible unless

q -= 0 or

q--- I rood p.

E l -~ S 1 and X pq+l --~ E X pq M 2r(p-l) --~ X p 2 q + p r

if q > O. if q > 0 and 1 <_r<_p-i .

(iii) X p 2 q ^

(iv)

H (Xpq;Zp)

~ Egq(p-I)A/A{x(~ 'P:I I pi+g > 0}, in particular, if q ~ 0 rnod p.

H'~>~(Xpq;Zp) =~ EZq(p-l)M[q/p]

(v)

There is a stable m a p g:Xpi+Z+p--which on rood p eohornology realizes Z i 0z:~ p (P-I)M(p:-I) -~ M ( p i) is the A - m a p 2 i i ~(Z p (P'I)I) = × ( P P ) 2 i+ l(p- 1)X >n p pi+l + p ~E2(pi+l + l)(p-l)~ , w h e r e specified by

48
In order to obtain hornotopical information f r o m these cohornological calculations, one wants to determine k-invariants. Now B r o w n and Gitler [6] have dis-

w h e n p = 2, and q R. C o h e n [21] has generalized their constructions to odd primes. C o m b i n i n g results, one obtains the following theorem. Theorem ing properties. Z.i0. T h e r e exist finite p-local C W - s p e c t r a B(q) with the follow-

played certain spectra with the s a m e c o h o m o l o g y as the X

(i) (ii)

H (B(c~;Zp) If i : B ( q ) - ~

is isomorphic K(Z,0)

to t h e A - m o d u l e the generator,

M(q). then the induced map X if e i t h e r

represents

i , : B ( q ) r ( X ) ~ H r ( X ; Zp) p = Z (iii) If M v and

is an epirnorphism or p > 2

for all spaces and

r <__2 q + l

r < 2 p ( q + l ) - 1. bundle

is a compact

smooth n-manifold Tv, then

embedded in R n+j with normal

and Thorn space

i , : B ( q ) r ( T v ) ~ H r ( T v ; Zp)

is an epirnorphisrn

if either p = 2 (iv) and n+j-r < 2q+l or p > 2 and n + j - r <__2 p ( q + l ) 1. in

~kB(q) is a known Z -vector P d e g r e e k) if e i t h e r p = 2 and k< Zq

s p a c e (a c e r t a i n

q u o t i e n t of t h e A - a l g e b r a

or

p > 2 and duality,

k<

2p(q+l)-2 M

.

H e r e (3) f o l l o w s f r o m (Z) b y A l e x a n d e r Tv. Particularly Let for the case T

since

i s a n n+j S - d u a l of

p > Z, a l i t t l e g e n e r a l i z a t i o n

of t h i s s i t u a t i o n i s and let

appropriate. v ~ H S ( T ; Z ). P (a)
(b)

be an m S-dual (T,v)

of s o m e f i n i t e C W - c o r n p l e x M(q) if m-s ~ 2p(q+l)is A { X ( ~ P i ) ]

Say that m-s

i s a d a p t e d to or p > 2

p = 2 and

< 2q+l

and

1;

and

the kernel of v-:A -~" H * (T; Zp) , v-(a) = av, Brown

i > q} .

and Peterson [7] at p = 2 and R. C o h e n [21] at p > 2 proved the followB(q).

ing characterization of the spectra Theorem 2.11.

Let • be a s p e c t r u m which is trivial at all primes other

than p and satisfies M(q) t h e r e
y SM(q) to

H (E; Zp) -- 2~SM(q). "~:EO°T -~E

S u p p o s e t h a t f o r s o m e ( T , v ) a d a p t e d to v carries the generator of

exists a map
v.

such that ESB(q).

T h e n E i s e q u i v a l e n t to

49

T h e y used this characterization to prove the following basic t h e o r e m on the structure of the Xq. to spectra. Theorem EC°X Pq 2.12. At p = 2, E°°X is equivalent to EqB[q/2]. At p > 2, Recall that E c° denotes the stabilization functor f r o m spaces

is equivalent to E zq(p-l) B[q/p] if q ~ 0 rood p. q 9(iii), it suffices to consider q-= Z rood p w h e n p > Z.
q

B e c a u s e of T h e o r e m

In both cases, one inductively constructs degree q o r 2q(p-l), together with m a p s which realize uq.

(Tq, Vq) adapted to M[q/p], with u u

of

from ~°°T to ~,°°X or ~°°X q q q Pq M o d u l o certain constructions and calculations based on the geo-

metric and homological properties of the C 2 S r in [36] and [i4,11I], the construction of the (Tq, Uq) and Theorem ~ q reduces to the verification of the following result.

F o r each i >__0, there is a c o m p a c t s m o o t h closed manifold i+ 2 of dimension 2i if p = 2 or p + p-2 if p > Z with stable n o r m a l bundle u. and a
1

2.13.

bundle m a p fi:vi-~ ~2,2i suchthat w . if p = 2 or fi:vi~Z,pi+2+p if p > Z

2i - 1

(vi) / 0 if p = 2 or

(Tfi) pp(i+

*

l)ui)

/ 0 if p > 2, w h e r e as an A - m o d u l e .

p(i+i) = l + p + ... + pi and

u.1 generate

H*(T~2,pi+2+p)

W h e n p > 2, the cohomological condition looks like but isn't a statement about W u classes (since the relevant bundles ~n,qare not orientable).

Unfortunately, the proof of this last t h e o r e m is not very pleasant, being based on a detailed study of the A d a m s spectral sequance for the suspension This analysis is based on 2.8 and 2.9 provide the

s p e c t r u m of T(~Z,Q)_ ^ K(Zp, I) for the relevant q.

ideas of M a h o w a l d [33]; the m a p s g displayed in T h e o r e m s key geometric input.

Several people have considered the p r o b l e m of directly constructing explicit manifolds as claimed in the t h e o r e m above. H o w e v e r , as far as I know, not even

the requisite four-dimensional manifold has been concretely identified. T h e m a p s f. are called M a h o w a l d orientations, or braid orientations.
i

O n the classifying

space level, one is asking for a lift of v.:M. -~ B O

to the appropriate braid space

B(R z, q) = K(Bq, 11.
mutative d i a g r a m

i

I

F. C o h e n [12] has used results of [36 and 37] to prove that there is a c o m -

50

i

~(t3oo, l)

* K(~oo, 1) ,

d
f22S3 where ~ ~ BO ]3
OO

I
~ )2
(DO

i is induced by the natural h o m o m o r p h i s m

, j is induced by the and ~- is the second

regular representation c o

-~ O, ~ i s a h o m o l o g y i s o m o r p h i s m , S I-~ IBO.

loop m a p induced f r o m the non-trivial m a p original proof of, (i) of T h e o r e m trum M~-

(This implies, and w a s the

2.2.) M a h o w a l d

[33] proved that the Thorn spec-

of ~- is I((Z2,0), and it follows that the Thorn spectrum of ji is also (See Lewis [31] for a thorough study of the Thorn spectra of maps. ) In for any space X is represented

K(Z2, 0).

turn, this implies that any element of H.(X, Z2) by a braid orientable manifold.

This fact says that such manifolds m u s t abound, F.Cohen [12] shows that solv-

but they are still very hard to find directly.

manifolds and certain Bieberbach manifolds admit braid orientations. In view of all this, it is very natural to ask precisely what it m e a n s metrically for an n-manifold M to admit a braid orientation. geo-

Sanderson [44] a canonical i m in

a n s w e r s this question by associating to a braid orientation of M mersion M N X I -+ M,

d i m N -- n-l, factored through an e m b e d d i n g of N X I can be recovered f r o m the

X R 2 and showing h o w the n o r m a l bundle of M

immersion.

I shall say no m o r e about this here since Sanderson's paper appears

in these proceedings. Incidentally, if ~-n is the restriction of ~ to ~ F C 2 S I C C 2 Si -~ ~22S3, then

Mahowald [33] also p r o v e d that H (M~n; 22) ~ M(n).
recently proven that M~n is in fact equivalent to B(n).

R. Cohen [22] has just

Returning to our m a i n line of development, lead to infinite families of elements in the ring spheres.

w e sketch h o w the results above s w. of stable h o m o t o p y groups of X q to ~2sr+2 on the w,B(q) on the

T h e essential point is the connection of the spaces

one hand and the relative ease of analyzing the h o m o t o p y groups other. M a h o w a l d ' s w o r k actually preceded that s u m m a r i z e d

above.

In particular,

he conjectured T h e o r e m Adams

2.12 (its implicit assumption being an error caught by given the results above,

in a preliminary version of [33]). H o w e v e r , arguments b:S 8-~ B O go as follows. generate w8BO Let = Z

Mahowald's Let

and let ~ : ~ S 9 --~ B O "~ B O map

be the unique loop

m a p which restricts to b on S 8"

~'.~2S9

be the adjoint of Qb, ~I,~2S9 "-~~2S 9. It is easy to

n a m e l y the composite of ~ with the evaluation

51

check that b (wzi) / 0 for i Z 3 [33, p.250].
and [31,2.3.4]) that the Thorn spectrum 9:~°°f22S9 -~ ~°°S0 = S Mb

It is w e l l - k n o w n

(see [14,II. 5.15]

is the cofibre of the m a p

of spectra adjoint to the composite m a p of spaces J ,, SF = QI S0 *[-i] - Q o S O C QS0 Zi ~t / 0 for

[2S 9 [2b ~ f20BO --~S O

(where the subscripts indicate the relevant components). i _> 3, w h e r e b is the Thorn class of M b . Now

Of course, Sq

V zcoDq(R2, sv)
q>_l hence ~is a s u m of m a p s ~q:~co(~6qXq) -~ S.

V zco( 6%q),
q>l Let

~ Z i - Z + 2 i-1
Y'x = Xzi_3 and fi = #zi-3""Zoo Yi ~ S, Let Cf i> 3. denote the

Mahowald's m a i n result [33,Thin. Z] reads as follows. cofibre of a m a p f (of spaces or spectra), Theorem properties. 2i (i) Yi has dimension connected. i sqZ ~t / 0, w h e r e 2i-i, H -l(Yi;Zz) = Z 2 , and Yi 2.14. The spaces Y. and stable m a p s
i

f. have the following
1

is (21-2i-3-i)-

~ generates H0(Cfi;Z2 ). 2i (iii) There is a (stable) m a p gi:S -~ Y. w h o s e composite with the projection
i

(ii)

Zi- Z i Yi -~ Yi/Yi -~ S 21-

is essential.

Here (i) is immediate f r o m the filtration and h o m o l o g y of C2S i and (ii) is immediate f r o m the paragraph above. Part (iii) requires a little m o r e w o r k since 2.I0

~rziYi = ~vzi_3B(Zi-4), and this is the first group beyond the range of T h e o r e m (iv) and the first in which 4-torsion can occur.

(The preprint version of [7] gave Nevertheless, the

a range one higher, this being an error caught by Mahowald.)

explicit construction of the B(q) m a k e s detection of the required gi via the A d a m s spectral sequence quite easy. Standard arguments show that fio~ Ez co gi:~ co 2i S ~ S projects to hlh i in i

2, 2 +2

of therood 2 A d a m s

spectral sequence.

cycles and represent non-zero elements

Thus the hlh i are permanent s ~i in w, . This was the starting point

of Mahowald's Aarhus talk, in which he noted that one could start the argument above with b:S 4-~ B O instead of b:S 8 -~ B O and so obtain a different family of

52 Z ~]i's. He asserted that 2~]i+i = I]i for both families, this being zero if i = 4 and non-zero if i >__5 (for both families) but the reader is referred to Mahowald's contribution [35] to these proceedings for further information. R. C o h e n [Zl] gives the following analogous development at odd primes. b:S 2p-Z -- B S O generate = 2 p _ z B S O = Z. -~ B S F The composite and ~ Bjob:S 2p-Z-~" B S F Let de-

termines a loop m a p Here

~ : ~ S 2p-I

has adjoint ~ : ~ 2 Z S 2p-I -~BSF.

b (w i) / 0 for i>__ I, w h e r e w k is the k th rood p W u class. The Thorn P spectrum M ~ is the cofibre of the adjoint @:~°°f~ZsZP-i -- S of the composite f~Zs2p-I pi P , - ~ B S F ~ SF = Q1 s 0

*[-t] ~ QoS0C

QS °

Of course, P

~ / 0 for i >_ i, w h e r e

~ is the Thorn class of M ~ .

Now

N°°~2ZsZP-1 ~
hence ~ is a s u m of m a p s

V
q>_l

NO°Dq(RZ,SZp-3) ~

V
q>_l

N°°(%(zP-4)qxq) ,
Let "~ S, i > i.

~q: ~°°(~(ZP-4)qXq) --~S, and fi : @pi" C O Y i

Y'I : N(ZP-4)PlXpI' Theorem Z.15, The spaces

Y. and stable m a p s
1

f. have the following pro1

perties (where cohomology is taken rood p). (i) Yi has dimension Z(p-l)p I- I, • H Z(p-I)pi- l(Yi) and y and x i H 2 (p-l)p-2(yi)

are both Zp with respective generators Yi is (2(p-l)p I- 2p I- - l)-connected. (ill ppl =~ and Fi_i~ - ~

such that ~(x) = y, and

(up to non-zero constants), where

~ generates

H 0 (Cfi) ,
1

and

* ~ are the images in l-I (Cfi) of ~ x and ~ y in H * (~Yi) , and

F. is the s e c o n d a r y

c o h o m o l o g y o p e r a t i o n a s s o c i a t e d to the A d e m r e l a t i o n

for

P (p- t ) p i PP i ; the s e c o n d e q u a t i o n h o l d s rnodulo z e r o i n d e t e r m i n a c y . i
gi:S 2(p +l)(p-l)-3 -~ y. such that Plxfl 0 in
--

(iii) There is a (stable) m a p

H*(Cgi),

where

x

pulls back to x c

H

*

1

(Yi).

H e r e (i) is a g a i n i m m e d i a t e

f r o m the f i l t r a t i o n and h o m o l o g y of C2S t , the a b o v e , and the s e c o n d p a r t of (ii) i f a c t o r i z a t i o n of PP v i a s e c o n d a r y 2.9(iii) a n d 2 . t 2 g i v e t h a t

f i r s t p a r t of (ii) i s i m m e d i a t e

f r o m the p a r a g r a p h

i s a d i r e c t c o n s e q u e n c e in v i e w of L i u l e v i c i u s ' c o h o m o l o g y o p e r a t i o n s [32].

F o r (iii), T h e o r e m s

X P

i • MZ(P-I)

~

X

s y Z(p_i)(pi-i+ i)B(pi_2 ) • --~ p1+p

53
This time the desired element is just within the range of T h e o r e m gives a h o m o t o p y class 2.10(iv), which

~ p ( i - i ) ¢ Wq B ( p l - Z ) '
The required m a p E2(p-l)(p1-1+l)

q = ZP(Pl-Z+ l )

4.

gi is obtained by suspension of the composite of with the projection Xp1+p -~ E Z P - I x . pX induced by the

~ p ( i - t)

projection

M 2 ( p - t ) -- S 2 p - t .
co gi: EcosZ(p- f)(p1+1) - 3 -~ S proHere

Standard arguments n o w s h o w that floe jects to h0Xi_ I in E 3,2(p-l)(p1+l) Z

of the rood p A d a m s

spectral sequence.

E 2'2(p-I)pi+I corresponds to F. and is also denoted b i (or bl, io r bi) ; up I ki ~ 2 i to sign, it is the p-fold s y m m e t r i c M a s s e y product <hi,...,h i>. Thus the h0X i are permanent cycles which represent non-zero elements Of course, ~
1

~i of order p in w.

S

.

extends to Y.: r z(p-0(p~+I+ i)- Z M - i
1

-, S, where

M -I is the

rood p M o o r e spectrum with bottom cell in dimension -I.

If w: M - I -~ S is the

projection onto the top cell, then C o h e n shows further that "~i is represented by w * (h0 h i+i) in E Z of the rood p A d a m s spectral sequence converging to the stable

cohomotopy of M -l, but the m e t h o d fails to detect the h0hi+ I themselves.

§3.

Splitting theorems;

J a m e s m a p s and Segal m a p s

The material to be discussed here is simpler, but in an earlier stage of development, than that discussed in the previous section. It is potentially at least

as rich, and should lead to a later generation of concrete homotopical applications. The calculations of the previous section presupposed f r o m iterated loop space theory only the geometric properties of the approximation existence of the stable splitting C ( R homologies of c(Rn, x)
n

C(Rn, X) -~ ~ n E n x ,

the

,X)

and the pieces

V Dq(Rn, X), and understanding of the q>__l Dq(Rn, X). The analogous information for NiX ~ 2 ~ X , the splitting

s

first loop spaces would be the approximation EMX

--~ V E x [q], and the homologies of M X and the X [q]. The latter informaq>_l tion is utterly trivial, and the J a m e s approximation acquires m u c h of its force jq:MX • --~ WLX[q] -~ ~ E X [q] w h o s e

f r o m homological calculation of the Jamesr ]maps adjoint J a m e s - H o p f m a p s

h :• M X -~ E X LqJ yield the splitting. F o r example, it q was just such homological information which led to the homological understanding of the key m a p s g of T h e o r e m s 2.8 and Z.9.

54

T h e deepest part of the theory to follow (and the part in m o s t rudimentary form) will in principle lead to complete information on the homological behavior of the J a m e s
S

maps

jq:C(Rn, X) -~ Q D q ( R n , x )

w h o s e adjoint stable J a m e s - H o p f However,

maps

h q :~C°C(Rn, X)-~ ~,c°5(Rn, x ) yield the stable splitting.

while the geo-

m e t r y leading to such computations is m o r e or less understood, begun the actual calculations.

w e have not yet

Thus the present state of the theory is analogous to

the status of the original approximation t h e o r e m after the w o r k of [36] but before that of [14]. Before proceeding further, I should say that virtually everything discussed in this section is joint w o r k of Cohen, Taylor, and myself [16-19] and also C a r u s o

[i0], the only exception being the closely related w o r k of K o s e h o r k e and Sanderson [3O]. I shall first explain the various splitting t h e o r e m s of [16 and 17] and then discuss the multiplicative properties of the J a m e s maps, m a p s and certain analogous shall see that an

the definition of which is based on ideas of Segal [46]. W e

unstable version of the K a h n - P r i d d y

t h e o r e m follows directly f r o m these proper-

ties, and w e shall obtain a result on the Z - p r i m a r y exponent of the h o m o t o p y group groups of spheres as an obvious corollary. Another fairly i m m e d i a t e application K(Z,0) as

is a simple proof of M a h o w a l d ' s t h e o r e m [34, 6.Z.8] on h o w to represent a Thorn spectrum. belong to the future. Return to the general context established in section one. Nevertheless, I am

sure that the m o s t interesting applications

A collection of

~..-spaces ~. with degeneracy operators is denoted ~ and called a coefficient J J system. A collection X = {Xq) of based spaces with all the formal properties th that would be present if X w e r e the q p o w e r of a based space X is called a q If-space. Given ~ a n d X , there results a filtered based space C X . See [16,§§l,Z] for details of this generalization of the construction C X of section one. be P ^ X q for based spaces P a n d X , splitting theorems. and this example leads to useful However, the reader m a y prefer to think of X q might

"parametrized" X q as X q

T h e splitting t h e o r e m s of [I 6] all fit into a single general f r a m e w o r k w e n o w sketch. Let ~ and ~' be coefficient systems and let q be given.
=

which Let

Dq(e,x)

~qcx/~_tc_x

-- eq ^Z q X[q] ,
a

+

w h e r e X[q] is the quotient of X q b y

the generalized fat w e d g e present i n X q f o r

55 H-space X. These spaces for ~' will be irrelevant, and we abbreviate

nq(~, _X) :
certain

nq_X.
system diagrams ~ ~ ' is a collection [16,4.1]. of maps A James C r -~ 09 ( % r - q ) system induces such that a James map commute

A James simple

j q : C X-- -~ C ' D q-X for any H - s p a c e X [16,4.2]. for generalll-spaces). In practice, C ' X is an H - s p a c e for spaces X (but not

If w e are given J a m e s s y s t e m s ~ - ~ ' for i<_ q<_ r, then r w e c a n d e f i n e k r : C X _ -~ C'( V DqX)_ to be the s u m over q o f t h e c o m p o s i t e s of q=l r the _jqwith the evident inclusions C ' D X -* C'( V DqX). W e can also restrict k q--r q=l to the finite filtrations of C X_. A n y c h o s e n basepoint of ~ I' gives rise to a natural m a p diagram:
L iT

~] : X -* C ' X

for spaces X.

We

can thus write d o w n the following

Fr_ l CX
(*) k

- FCX r --

DX r--

r-1 C'~ '

r-1

r

1
DqE)

C'( V DqX_) q:l

C'( V
q=l

Chr

I
, C'Dr_X
In to NX comes from

H e r e ~ a n d ~ are u s e d generically for evident cofibrations a n d quotient m a p s . practice, the left s q u a r e a l w a y s c o m m u t e s limits o v e r r w h e n w e h a v e J a m e s commutes We up to h o m o t o p y . are interested in h o m o t o p i c a l splittings, but w e digress m o m e n t a r i l y T h e infinite s y m m e t r i c product on the nose

(which allows p a s s a g e to

s y s t e m s for all q) a n d the right s q u a r e at least

discuss h o m o l o g i c a l splittings.

the coefficient s y s t e m ~ with e a c h ~j a point. give a J a m e s s y s t e m for e a c h q. Here

T h e unique m a p s ApplyingN

~ r ~ ~(q,r-q) to (*) and using

(*) c o m m u t e s .

the natural t r a n s f o r m a t i o n N N

-* N, w e obtain the c o m m u t a t i v e NL Nw

diagram

NF

r-I

CX --

~- N F C X r --

~ ND X
r--

r-1

1
DqX)
--

r

J
N.

Ii
~ ND X
r--

N( V
q=l

N~

, N( V nqX)
q= i

56 U s i n g the fact that N converts cofibrations to quasifibrations and the relationship b e t w e e n N and integral h o m o l o g y [24], and playing g a m e s with p a r a m e t e r [l 6, 4.10]. spaces P,

w e derive the following general h o m o l o g i c a l Theorem 3.1.

splitting t h e o r e m

F o r all coefficient s y s t e m s

~, H - s p a c e s X, Abelian groups G,

and

r >__I ( i n c l u d i n g r = co),

~,(FrCX;G

) is i s o m o r p h i c

These isomorphisms
operations.

a r e n a t u r a l in ~ , X ,

to ~ ~,(Dq_X;G). q=l and G and c o m m u t e w i t h B o c k s t e i n

T h e case duced symmetric

C. : ~

recovers

Steenrod's h o m o l o g i c a l

splitting [49] of the re-

powers

a stable homotopical

F NX and this e x a m p l e s h o w s that w e could not hope for r splitting without s o m e restriction on ~. splittings, ~ will always be E-free in the sense that E.

F o r the h o m o t o p i c a l

acts freely on ~. for each j. Returning to the general context, a s s u m e given a J natural m a p ~t:C'X -*[2tEtx for spaces X and s o m e t > i. T h e k e y e x a m p l e is

C ' X = C (R t, X)

and

~3 = c~tg - 1 as in s e c t i o n o n e . t

Composing the diagram diagram
~ • >ntD

(*) w i t h

~3 and t a k i n g a d j o i n t s , t Et-ID
X
r--

we o b t a i n a h o m o t o p y c o m m u t a t i v e
6

EtFr- 1 CX
- -

~

- ~tF CX
r - -

X
r--

kr_ 1

kr

r-I V EtD x q=1 q-

~ r ~- ..... ~- V ~ t D x q=l q-

~

,. E t D

x r-

Assuming that

inductively that ~

r-1

isan equivalence,

a trivial d i a g r a m r

chase implies

6 --'~0 in the top cofibration sequence.

This implies that ~

is an equiva-

lence.

T h e s a m e s o r t of a r g u m e n t
With ~ : ~i = ~, w h e r e

works when

t = oo.

~j : 2.j, this recovers and generalizes Milnor's

s p l i t t i n g [4Z] of N M X
Theorem 3.Z.

[t6,3.7].
F o r all H - s p a c e s
r

X
--

and

r > i (including

r = oo), E F
r

MX
--

nV_ E x =l [q] " of restrictions of J a m e s - H o p f maps h

is naturally equivalent to

T h e equivalence is given by s u m s q :EMX -- E X [q] . 61:MX

over q

Note that no connectivity hypothesis is needed; but do not require it to be an equivalence. there is a natural w e a k b a s e d spaces X. equivalence Eg~EX

w e use a m a p

-~ f~EX that

It is an i m m e d i a t e ~ V q>l E x [q]

consequence

for all connected

57 In [16,§5], w e introduce "separated" coefficient systems. F o r such ~, if / and ~tq) is the coefficient s y s t e m given by the configuration spaces = ~ ~Eq [~ F(~qDj), there are tautological J a m e s systems ~ - ~ ~ for each q >_I. W h e n

~q

= ~(Y)

is itself the configuration space coefficient s y s t e m of a space Y, w e shall If 8 q e m b e d s in R t for

write d o w n the resulting J a m e s m a p s explicitly below.

q<__ r, then ~j(q) m a p s to ~ ( R t) and w e can apply the theory above w i t h ~ ' = ~(Rt). T h e product of any coefficient s y s t e m with a separated s y s t e m is separated, and by use of tricks involving both pairs of projections • ~ X ~ ( R c°) ,C~(I~ °°) and ~(~= ~(gJ X ~ ( R 0°) - ~ ( R °°)

w e prove the following general splitting t h e o r e m [16, 8.Z]. The second pair of projections plays a critical role in the naturality, leads to a uniqueness assertion for the stable James-l-lopf m a p s , ings. and allows one to avoid any choices of e m b e d d -

This precision is crucial to the deeper theory discussed at the end of the O n the other hand, as discussed in [16,§5], use of e m b e d d i n g s gives a

section.

good hold on the destabilization properties of the J a m e s - H o p f m a p s . T h e o r e m 3.3. F o r all Z-free coefficient systems ~, ll-spaces X, and r >__i r (including r = oo), ~ ° ° F C X is equivalent to \/ ~C°DqX_, naturally in ~ and X. r -q'l -T h e equivalence is given by restrictions of staSle $ a m e s - H o p f m a p s h s : Z c ° C X -~ N ° ° D X. q -q-Specializing either to ~ = ~ ( R n) or @ = @n' compatibly in view of

g: @ n - ~ ~(Rn)' w e obtain the following i m m e d i a t e consequence [16,8.4]. Corollary 3.4. F o r all connected based spaces X, there is a natural equivaV ~ ° ° D (Rn, X), n >_I or q>__l q n = oo, and these equivalences are compatible as n varles. Such equivalences w e r e first obtained by Snaith [48], but our proof has a n u m b e r of advantages (discussed in [16, § 7]). In particular, it is not clear that ~ t F C X -- ~ t D X can be extended over all of ~tc X; r n n,q n that is, they are not given by globally defined S a m e s - H o p f m a p s . W e shall c o m e back to these splittings shortly, but I want first to explain the further splittings obtained in [17], which partially r e m o v e the restriction to connected spaces in the corollary above. In [17,§i], w e introduce the notion of a "directed" coefficient system. details are rather delicate and the range of e x a m p l e s is peculiar; ~ directed but the ~ n are not; ~(Y) is directed if Y and ~ The are Snaith's splitting m a p s lence in the stable category between ~°°~n~nx and

is an open manifold but is

58 not directed if Y is a c o m p a c t A N R . When ~ is directed, there are inclusions -~ C'r+l X

~r: ~ r X ~ r X r
for

a H-space X, and w e define

CX

Xr+ 1 r+l to be the resulting eolimit.

If ~, - ~ '

is a

James

system, there result J a m e s m a p s
j

X q : C X -~ C ' D q-- , -Just as before,

where

Dq_X is a certain spaCer equivalent to the cofibre of ~q-l" :CX-*
r --

w e define k the diagram

C'(q L-IV ~ q X )
_ --

by s u m m i n g t h e

~- for -H

q <__r and write d o w n

L

~r-i X ~riiXr_l kr-i r-I [

C~ r

X.X,
r
r

w

X r

P

DrX

c,( V
q=l

nq_X)
the derivation

C'~

r

.c,( V ~x_)
q=l

C'~

. c,5~x
the same as in

From

here,

of splitting theorems

is precisely [17,§Z].

the discussion
Theorem groups G, and

above,

and we obtain the following theorems

3.5. F o r all directed coefficient systems ~, ll-spaces X , r k i, there are i s o m o r p h i s m s
r

Abelian

~*(OrXE r which

Xr;G) ~

E ~,.(DqX;G) q=l

and

~.,,(C-X;G)., ~

E q~l

~,(DqX;G)

are natural in ~, X,

and G and commute

with Bocksteins.
[49]

With ~ = ~, this recovers Steenrod's i s o m o r p h i s m s
r

~ H,(X
for the unreduced
Theorem lences

r/~r;G)~

symmetric
3.6. r

E ~:.(X q/~q. xq-l/~q_l; G) q=l powersof a space.
X and r > i, there are natural equiva-

F o r all H-spaces

ZXr~ V Z(Xq/ImXq_I) and Z~X--~ V
q=l
For

Z(Xq/ImXq_1)"

q>_l

spaces X, IVlX is the w e a k infinite product of countably m a n y copies being e m b e d d e d as points

o r e and the successive quotients are x q / x q-l, X q-I with last coordinate the basepoint.

89 Theorem 3.7. F o r all E-free directed coefficient systems 0. and all

ll-spaces X, there are natural equivalences r ~°°(~-~rX3q X r) --~ V 32°°Dq-X and r q=l

~ ° ° ~ X-- --~ V o o ~ q-X. q>__l for a topological

F o r example, with 0. = ~(RC°), the case of spaces B G m o n o i d G gives that B ( E o o [ G )_

is stably equivalent to the w e d g e of the cofibres of

the n a t u r a l m a p s

B ( E q _ I ; G) ~ B ( E q ] G ) .

F o r the p r o m i s e d a n a l o g of C o r o l l a r y 3.4, we u s e a n a p p r o x i m a t i o n of the
form ~-(Rn, x ) , Here X 7 ~ n ( X +) ~" n n , [20 n ( x + ) .

is a connected space and X +

is the union of X and a disjoint basepoint.

T h e space < ( X

+) is the telescope of a sequence of "right translations" x X ' r Er+l is a h o m o l o g y i s o m o r p h i s m constructed at the end of [14,1§5]. n,r ~ n,r+1 x X
r

-~ ~

r+l

and the m a p While [

~n

is defined there for g e n e r a l X , it is only a h o m o l o g y i s o m o r p h i s m for n connected X; in general, the two-variable B r o w d e r operations m i x c o m p o n e n t s non-trivially in H.f20

n n(x+ )

but not in H,

(X+).

The m a p

~ is an equiva-

lence analogous to g [17, 3.1]. F o r all connected based spaces X, there is a natural equivaco n n. + V oon fence in the stable category between E ~0 E ( X ) and q>--i E Dq(R ,X), n >__2 or n = oo, and these equivalences are compatible as n varies. T h e s e results by no m e a n s exhaust the possibilities of the basic line of argument, and there are various other such splittings k n o w n to Cohen, Taylor, and myself but not written down. F o r example, Joe Neisendorfer r e m i n d e d us of En(X ,A) for a based pair Corollary 3.8.

[36,6.6], in which I introduced a relative construction (X,A). When A -~X

is a cofibration and A is connected, there is a quasifibering

C nA

--

En(X,A)

-- C n _ I ( X / A ) ,

n>__l ,

w h e r e C 0 is the i d e n t i t y f u n c t o r [36, 7.3].
sions C n A C E (X,A) C C X
n n

There are filtration preserving inclus-

and it is perfectly straightforward to trace through and see that it restricts to give a stable
n

the proof of the stable splitting of C X splitting of E n ( X ,A).

60 Theorem 3.9. Let A -~ X be a cofibration. F o r all r > l (including r = oo),

and all n >_ l (including n = co), there is a natural equivalence r

~co~En(X,A)
These equivalences the stable splittings of C A

--~

V
q=1 as

~ c o ( ~ n (x, a)/rq_iEn(X, a)).
r and n v a r y and are also compatible with

are compatible n and

C X. n and of C nI(X/A) is

T h e relationship b e t w e e n the splittings of E n ( X , A ) unclear and deserves Again, theorems, study.

it is a simple matter to give equivariant versions of our splitting

putting actions of a finite group G on all spaces in sight (see [39,§5]),

and this in turn is surely a special case of a general categorical version of the argument. We return to the original splitting t h e o r e m ~(Y), the case and specialize to configuration

space coefficient s y s t e m s Actually,

Y = R n being of m o s t interest. in the rest of the paper, as a topic of inde-

w e are wholly uninterested in splitting t h e o r e m s

being c o n c e r n e d instead with the analysis of the J a m e s pendent interest. A s in section one, think of points of C ( Y , X ) finite subset of Y and A s m e n t i o n e d above, maps jq: C ( Y , X ) Explicitly, jq(h,k) = (M,g), w h e r e M X" L -~ X is a function.

maps

as pairs

(L,X), w h e r e

L is a

Recall that systems

B(Y, q) = F(Y, q)/~q.

there are canonical J a m e s

w h i c h give rise to / a m e s

~ C(B(Y, q), mq(Y,X)). is the set of all subsets of b with q e l e m e n t s -~ D q ( Y , X )

(such a set of q elements of Y being a typical point of B(Y, q))and ~: M

sends a point m c Mto
it is not i m m e d i a t e l y combinatorial

t h e i m a g e i n D q ( Y , X ) of ( m , k l m ) • F q C ( Y , X ) .
apparent that j is well-defined.

Of c o u r s e ,

q description in [16,§5] is perhaps m o r e that B(Y, c~ e m b e d s map

T o c h e c k this, the m o r e appropriate. To p r o c e e d
corn-

further, one can a s s u m e pose with

in R t , say via eq, and then

C(eq, l) to obtain a J a m e s

J q: c(~ x) ~ C(R t,Dq(Y,x)).
WhenY = R , we may take
n

t : 2qn

(or (Zq-1)n, by [16,5.7]).
maps is due to K o s c h o r k e and Taylor. and (To see

This functional description of these J a m e s Sanderson [30], w h o discovered t h e m

independently of C o h e n

the comparison, their C~m(X) is our C(B(Rm, k),X).) Their emphasis is not on
the m a p s and their homotopical implications but rather on their geometrical

61 interpretation. Let V be a smooth manifold without boundary with one-point c o m -

pactification V . Also, let ~ be a vector bundle over s o m e space B, with Thorn c m Bk c o m p l e x T~ , and let ~ m , k be the evident derived bundle over Bk= F(R ,k) × Z k Consider i m m e r s i o n data consisting of a smooth closed manifold M, an i m m e r s i o n g l : M ~ V with n o r m a l bundle v, and a bundle m a p (gl'g2):M-* V X B ( R m , k) is" an embedding, w h e r e ~: v-~" ~ m , k such that

g z : M ~ B ( R m , k) is the c o m Let 4 ( V , ~ )

posite of the base space m a p of g and the projection B k-~ B(R re,k). be the set of b o r d i s m classes of such immersions. first prove that Ckm(T~) classifies this set,

Koschorke and Sanderson

Jkm(V,~)

~ [V c,Ck(T~)], jqabOve (for X a Thom space) in terms

and then explain how to interpret the maps
of certain operations sion g l , M ~ V Im(V,~) -~km(V,$)

specified by associating to an i m m e r k with n o r m a l bundle ~ an i m m e r s i o n gl:M(k) -~ V with n o r m a l M(k) C B(M, k) is the manifold of

bundle m a p p i n g appropriately to ~m,k' where k-tuple self-intersection points of gl"

In this context, they obtain geometric maps

proofs and interpretations of s o m e of the multiplicative properties of J a m e s w e are about to discuss. In [i0], w e shall discuss multiplicative properties of J a m e s axiomatic generality. Given suitably related J a m e s systems

m a p s in full

~-~ ~(q) and suitable

structure on ~ and the ~(q), there is a ring space structure on the infinite product

q>_0

X

C(~DqX

and the m a p

(jq): CX -is an e x p o n e n t i a l H - m a p . Here

× C(OJDqX qk0 CX to 1 ~S 0 r C(0)S 0.
, the trivial

D0X = S 0 and J0 c a r r i e s

F o r any coefficient system ~ with appropriate s u m s James systems C, -~ ~ used to prove T h e o r e m

Up X ~q - ~ p + q

3.1 satisfy the relevant axioms. systems ~ --~_,(~q) satisfy

F o r any separatedC, with sums, the canonical J a m e s the axioms.

If ~ = ~(Y), w h e r e Y admits an injection YJJ.Y -- Y each c o m p o n e n t

of which is homotopic through injections to the identity map, then~, admits s u m s of the sort required. all H - m a p s In particular, this applies to Y = R n. Here the following are

with respect to the appropriate multiplication on the infinite products:

62

x
× C(B(R n, q),Dq(Rn,X)) q>_O
(Jq)

C(eq, 0
~, X
q>_0

q>O

C(R Zqn, Dq(R n , x ) ) g

. C(Rn, X ) . . . . . . . . . . . .q). .
-i

(J

-IP x
q>O

[2Zq n E Z qnDq( R n ,x )

W e continue to write jqfor the composite

azqng

C(eq, l)jq.

W e could also have

stabilized, replacing Z q n by co on the right. The product on the loop space level is induced in an evident w a y f r o m s m a s h products f21YX~2JZ-~f/I+J(Y^Z) and the pairings % ( R n , X ) ^ Dt(Rn,x) --~ D s + t ( R n , x ) n ,X).

induced by the additive H - s p a c e structure on C ( R

W e digress to mention an application to Thorn spectra in [18]. There w e give a simple proof, based solely on use of Steenrod operations, of the following theorem. Let $ 3 < 3 > 3.10. denote the 3-connective cover of $3. (i) The Thorn s p e c t r u m associated to any H - m a p

Theorem ~Zs3 -~ B F

with non-zero first Stiefel-Whitney class is K ( Z Z,0). ~Zs3<3> ~ BSF with non-zero

(ii) T h e Thorn s p e c t r u m associated to any H - m a p

second Stiefel-Whitney class and non-zero first W u class at each odd prime is K(Z, 0). Part (i) gives a n e w proof of M a h o w a l d ' s result that M { : K ( Z z , 0 ). .At p > Z ,

Di(RZ, s Z q - 1 ) ~ 0 for l < i < p c u s s i o n above that is a p - l o c a l H - m a p .
spectrum.

and D (R2,SZq-l) "~ M 2pq-2. It follows f r o m t h e d i s L"

.

Z~Zq+I

CzsZq-1

QMZpq-1

As explained in [18], with q = 1 this e a s i l y leads to an H - m a p
is a Thorn

as prescribed in part (ii) and so gives M a h o w a l d ' s result that K(Z,0)

Returning to the w o r k in [i0], w e n o w head towards the K a h n ~ P r i d d y theorem. W e follow the ideas" of Segal [46], but w e w o r k unstably with general

spaces X and thus introduce a great deal of n e w structure into iterated loop space theory. W e want first to extend the J a m e s m a p s over [2 IE X.
n n

T h e r e is no prob-

lem whenX

is connected, but it is the case X = S O in which w e are m o s t interested.

A s Segal points out [46], the following obstruction theoretical observation allows use of the exponential H - m a p property above to extend the jq simultaneously for

63

all q.

Henceforward, all H - s p a c e s are to be h o m o t o p y associative and c o m -

mutative. Lemma 3.11. Let g:X -~ Y be a group completion of H-spaces, w h e r e w0X T h e n for any grouplike H - s p a c e Z and w e a k f:Y -~ Z such that is weakly

has a countable cofinal sequence. H-map

f:X -~ Z, there is a unique w e a k H - m a p

homotopic to f. The "weak" aspect is that w e are ignoring liraI terms. that, on finite-dimensional C W - c o m p l e x e s A, The interpretation is is universal with

g:[A,X]-~ [A,Y]

respect to natural transformations of monoid-valued functors f r o m [A,X] to groupvalued represented functors [A,Z]. classes. Zqn Zqn , n (1, × ~ z~ mq(m , X ) ) is g r o u p l i k e ; that q>_l i s , its m o n o i d of c o m p o n e n t s is a g r o u p . This gives the following generaiization W e take [ , ] in the sense of based h o m o t o p y

By a p o w e r s e r i e s

argument,

of r e s u l t s of S e g a l [46]. natural inclusion homotopy groups.
Theorem 3.12.

We a s s u m e

t h a t ~0 X is c o u n t a b l e and w r i t e ~](r,s) f o r the 1](r,s) i n d u c e s ( s - r ) - f o l d s u s p e n s i o n on

arzrx

-- f 2 S z S x , s > r;

F o r n >__2 and aliX, there exist m a p s

jq.. anznx

-

aZnqzZnq~(R n, x )

such that J0 is constant at i ¢ S 0, Jl is -q(n,2n), and Jr(O~+ ~3)= for if, ~ ~ [A,g2nEnx]. H e r e the s u m s are loop addition and the products are those specified above. ~ Jp(~)Jq(~) p+q= r

S e g a l [46] a l s o i n t r o d u c e d v e r y s p e c i a l c a s e s of the g e n e r a l m a p s
s :C(Z,Dq(Y,X)) -~ C(Z x Y q , X [q]) q specified by Sq(M,M) ment = (N,v) w h e r e if ~x(m) is the i m a g e in Dq(Y,X) of an elesuch that L m C Y has qelements,
. . . . .

(Lm, Xm) ~ FqC(Y,X)

then

N =
and u(m,~(l) .....

U
m ( M , £ i ~ L m , c~e 2 q

(m,~(~)

~(q)) C Z xYn

~(q))

= Xm[~(t))

^ "'" ^ lm(~(q)

)'

It is easy to analyze the additive and multiplicative properties of the S q C O m -

64

binatorially,
Let

and w e arrive at the following c o m p l e m e n t

to the previous result.

X [0] = S O.

Theorem

3.13.

For

m >_ Z, n >_ i, and aliX, there exist w e a k H - m a p s s q :e m ~ m Dq(R n ,X)-~ ~m+nq m+nqx[q ] and

such that so is the identity m a p

of f2ms m ,

s I is D(n,m+n),

Sr(~'f) = for ~ ~ [A,f~tp~tPD (Rn, x)] m m ~ for ~ ~ [A,~ ~ X], q where A:X -~ Y[q] and

(p,q)Sp(~)Sq(7),

r = p+ q, t >_ 2. Moreover,

? ~ [A,f~tq~tqDq(Rn, x)],

(s o~mzm~)(~)

= q:(~q(m,m+nojo~mEmA)(~),
-- Dq(Rn, X) is induced f r o m A (via

is the diagonal and ~ : X

any chosen basepoint in F ( R n, oJ). T h e product ~? of maps. Here s is that above, while that on the right is just s m a s h product

structures,

is obtained by application of L e m m a 3.11 to the additive H - s p a c e q and the uniqueness clause of that l e m m a implies the last formula. The

passage f r o m the combinatorial level product formula to the loop space level is more subtle and requires use of the following result (the need for which w a s over-

looked in [46]). Lemma ~0 X and ~0 X' 3.14. Let g:X -" Y and g':X' -" Y' be group completions, where

have countable cofinal sequences. f : X ^ X ' -- Z

T h e n for any grouplike H - s p a c e there exists a unique weakly is weakly hornotopic to f.

Z and weakly h o m o t o p y bilinear m a p hornotopy bilinear m a p Setting m ~ : Y ^ y ' -- Z

such that ?(gAg') S q w i t h jq.

= 2nq, w e can c o m p o s e

To analyze this composite, [ ]

w e need the m a p

k specified by k (U,~) q = (m,~)

q

:C(Y,X) M

-- G(Yq, X Lqj) is the set of allordered q-tuples of ele-

where

mentsofLand

~(~I ..... fq) = ~ ( ~ ) .... ^ ~(~q).
and use of L e m m a

Again, easycombinatorics,

a p o w e r series argument, Theorem 3.15.

3.11 give the following result.

F o r n>__ Z and aliX, there exist m a p s k q :$]nznx -- ~nqznqx[q] and

such that k 0 is constant at 1 c S O ' kr(~+8)

k I is the identity m a p , :

~ (p, q)kp(~)k(8 ) p+q= r

65
for e , ~
~ [A,f2n~r~]. Moreover,

(~q(nq, 3 n q ) o k q ) ( a )

= (Sq*jq)(C~) .

While a l l t h i s g e n e r a l s t r u c t u r e is b o u n d to p r o v e u s e f u l , the c o m b i n a t o r i c s f o r the l a s t s t e p t o w a r d s the K a h n - P r i d d y i d e n t i t y m a p , that is to s a y X = S O. Let theorem require A : X -~X [q] to be the

c. be t h e n u m b e r of w a y s of d i v i d i n g a lq set of q e l e m e n t s into i u n o r d e r e d s u b s e t s . The " t h e r e f o r e " i n the f o l l o w i n g r e s u l t

comes from a purely algebraic argument. Theorem 3.16. For n>_2 e Therefore
n n

a n d e e [A,~2 S ], = q ~. i=l Ciq(n (ni, nq) o ki) (c~). r c

n

n

q

kq(C~) = ~ ( e - 1 ) . . .

( a - q + 1) if A = B +, w h e r e If, f u r t h e r ,

[a,ans n] is

r times

t h e m a p w h i c h s e n d s B to 1. ~0 S , t h e n

B is a s u s p e n s i o n and c~ m a p s B to

kq(e) = ( - 1 ) q - I ( q - 1 ) ' 1 3 (n, nq)(~) . The last assertion holds since e~ = 0 by the standard argument that cup products are trivial for a suspension. n + Note t h a t Dq(R n , S O) = B(R , q) a n d let B(R n, q) to 1. that Let Fq(n,t) n 0 0 6 : D (R ,S )-~ S m a p 0 to 0 a n d t t qt t n 0 t t be the f i b r e of f~0 N 6:~0 ~, Dq(R ,S ) f20S a n d note

F (n, oo) ~ Q B ( R n, q). A n y c h o i c e of b a s e p o i n t in B(R n, q) y i e i d s q S 0 ~ Dq(Rn, S0), a n d t h e r e r e s u l t s a c o m p o s i t e e q u i v a l e n c e Fq(n,t) X a t s t -- a t N t D (Rn,S °) X a t ~ t D q ( R n, S °) -* at~.tD (R n, S °) . "~ q Let jq:f2ns n -~ e n q N n q D q(R n, S 0) have components Theorem nsn 3.15 gives a homotopy commutative (J . .'.J. q ) q •,, jlq and Jq in Fq(n,t) and ets t.

diagram a Z n q z Z n q D (Rn, S 0) q q

i- F (n, Znq) X a Z n q s Z n q ~ q

q an~nq ~](nq, 3 n ~ •

n3n~3nq

O n ~r~O Sn , r > 0, T h e o r e m s (si')(e) q-q

3.13 and 3.16 yield the formula
e

= (-l)q-l(q-l)! ~ 3 n q - n

-

q'

~

n q . jq ,

(4)

D

66 This is our unstable version of the K a h n - P r i d d y theorem. prime p, w e conclude that, up to a constant, iterated suspension h o m o m o r p h i s m . Since B ( R c°,p) ~ B E Taking qto be a

Sqj'q_iS congruent rood p to the

All m a p s in sight are compatible as n varies.

, w e obtain Segal's version [46] of the usual K a h n - P r i d d y P t h e o r e m on passage to limits.
.I

Theorem

3.17.

The composite

Q 0 S0

QBE P

s P~ Q o S 0

is a p-local

h o m o t o p y equivalence. is an infinite loop map. A c c o r d i n g to A d a m s P this is a necessary and sufficient condition that s agree with the map used by P K a h n a n d P r i d d y [28]. By construction, ~2n n 0S we have the commutative diagram It i s n o t c l e a r to us that s [1 ],

Q B ( R n, Z) ----- Q ( R P n - I )

~IS0 Q0
Theorem

Jz'
S

, QB(RC°,z) ~, Q ( R p °°) Q(Rpn-I).

1

sz

~ Q 0 S0 .

T h u s stabilization factors through

This has the following consequence.

3.18. If ~ ~ ~ is a Z-torsion element in the i m a g e under stahilir Zn+ 1 2n+~ zation of WZn+l+r S , then = 0, w h e r e g= 0 if n -- 0 or 3 rood 4 and E= 1 if n -= 1 or 2 m o d 4. has this order.

Indeed, T o d a [50] proves that the identity of E 2 n R p 2 n All of this is quite easy. We close with s o m e

r e m a r k s o n t h e deeper theory, maps is not just

to appear in [19], which explains what structure the J a m e s jq:C(Rn, X) "~ QDq(Rn, x) an H - s p a c e but a % - s p a c e . really carry. Since H,C(R

As mentioned before, C(Rn, X)
n

,X)

is functorially determined by H . X

via h o m o l o g y operations derived f r o m this structure, one wants to k n o w h o w this structure behaves with respect to the J a m e s m a p s . Consider the infinite product n X Q D (R ,X). W e have said that this is a ring space. In fact, it is an E ring q>0 q n space ( m o r e precisely, it has an equivalent subspace so structured). This m e a n s that there is an operad pair (~,~) in the sense of [37,VI.1.6] such that ~ is an (that is, ~, is equivalent to On) and there is an (For the

]72oo operad and .~ is an E n operad

action in the sense of [37,VI.I.I0] of (~.,.~) on qX 0>-- QDq(Rn,X). aficionados, ~ is the little convex bodies operad ~ao

and ~ = ~n X ~[, w h e r e ~ i s

the linear isometrics operad.) action.

The additive action, by ~, is the evident product

T h e multiplicative action, by ~,, is a parametrization of the multiplicative

67

H - s p a c e structure described earlier.

~ also acts on c(Rn, x)

(viathe pro-

jection ~ ~ % ) , and the crucial fact is that (jq):G(Rn,x)-~ i s a m a p of J - s p a c e s . coordinate Upon restriction X q>0 n QDq(IR , X ) th

of i t s t a r g e t to t h e u n i t s p a c e ( z e r o that the extension

1), t h e r e c o g n i t i o n

p r i n c i p l e of [36] i m p l i e s ~

n (jq):a Z n X

(i, X Q D q(R n, X )) q>_l

is actually an n-fold loop m a p for a suitable n-fold delooping of the target (not, of course, the obvious additive one). To c o m p u t e all the jq on homology, it suffices to determine the multiplicative h o m o l o g y operations on the target. In principle, these are completely determined rela-

by the k n o w n additive operations and general m i x e d Cartan and m i x e d A d e m tions for E n ring spaces like those developed for E

ring spaces in [14, II]. oo I have no doubt that such calculations will eventually b e c o m e a powerful tool for the working h o m o t o p y theorist, just as the earlier calculations of [14], which once s e e m e d impossibly complicated, are n o w being assimilated and exploited by m a n y w o r k e r s in the field. Bibliography i. Z. 3. 4. 5. 6. 7. 8. 9. J.F. A d a m s . 45-55. The K a h n - P r i d d y theorem. Proc. G a m b . Phil. Soc. 73(1973), lmroc. S y r u p . P u r e of

M.G. Barratt. A free group functor for stable homotopy, M a t h V o l . Z2, pp. 3 t - 3 5 . Amer. Math. Soe. t97l. M.G. Barratt and P.J.t~ccles. I~+ s t r u c t u r e s flc°G°°A. Topology 13(1974), 199-207. J. Beck. On H-spaces and infinite loop spaces. M a t h e m a t i c s V o l . 99, p p . 1 3 9 - 1 5 3 . 1969. ItI.

The stable structure Lecture

Springer

Notes in

J.M. Boardman and R.M. Vogt. Homotopy invariant algebraic structures on topological spaces. Springer Lecture Notes in M a t h e m a t i c s Vol 347. 1973. E . H . B r o w n , Jr. and S. Gitler. A s p e c t r u m w h o s e cohornology is a certain cyclic m o d u l e over the Steenrod algebra. Topology 12(1973), 283-Z95. E . H . B r o w n , Jr. and F . P . Peterson. O n the stable decomposition of f2ZSr+Z. Trans. A m e r . Math. Soc. T o appear. G. H. B r o w n , Jr. a n d F . P . Peterson. Quarterly J. Math. T o appear. J. Caruso. Thesis. Univ. of Chicago. The stable h o m o t o p y type of g2nsn+r. In preparation.

68 I0. 11. 1Z. 13. J. Caruso, F.R. Cohen, J.P. May, and L. Ro Taylor. maps, and the Kahn-Priddy theorem. In preparation. J. Caruso a n d S. W a n e r . An approximation to [2n~rLx. J a m e s maps, Segal To appear.

F.R. Cohen. Braid orientations and bundles with flat connections. Inventiones Math. 46(1978), 99-110. F. R. Cohen, Little cubes and the classifying space for n-sphere fibrations. Proc. Syrup. Pure Math. Voi. 32 Part 2, pp. 245-248. Arner. Math. Soc. 1978.

14. F. R. Cohen, T.J. Lada, and J.P. May. The homology of iterated loop spaces. Springer Lecture Notes in Mathematics Vol 533. 1976. 15. F. R. Cohen, M . E . Mahowald, and R.J. Milgram. The stable decomposition for the double loop space of a sphere. Proc. Syrup. Pure Math. V o 1 3 2 Part 2, pp.225-228. A m e r . Math. Soc. 1978. 16. F. R. Cohen, J.P. May, and L.R. Taylor. Math. Proc. C a m b . Phil. Soc. To appear. Splittingof certain spaces CX.

17. F. R. Cohen, J.P. May, and L.R. Taylor. Splitting of s o m e m o r e spaces. Math. Proc. C a m b . Phil. Soc. To appear. 18. 19. Z0. F . R. C o h e n , J . P . M a y , spectra. To a p p e a r . F . R. C o h e n , J . P . In preparation. May, and L.R. and L.R. Taylor. Taylor. K(Z,0) James and K(Zz,0 ) as Thom

maps and E n ring spaces.

F. R. Cohen and L. R. Taylor. Computations of Gelfand-Fuks cohomology, the cohomology of function spaces, and the cohomology of configuration spaces. Springer Lecture Notes in Mathematics Vol. 657, pp. 106-143. 1978. R. L. Cohen. 1978. R. L. Cohen. O n odd primary stable homotopy theory. Thesis. Brandeis. In preparation.

ZI. 22. Z3. 24. 25. Z6. Z7. Z8. 29. 30.

The geometry of ~2S3 and braid orientations.

D. Davis. The antiautomorphism of the Steenrod algebra. Math. Soc. 44(1974), 235-236.

Proc. A m e r .

A. Dold and R. Thorn. Quasifaserungen und unendliche symmetrische Produkte. Annals of Math. 67(1958), Z39-Z81. E. D y e r and R.K. Lashof. Math. 84(1962), 35-88. I.M. James. H o m o l o g y of iterated loop spaces. A m e r . J.

Reduced product spaces.

Annals of Math. 62(1955), 170-197.

D.S. Kahn. O n the stable decomposition of f/C°s°°A. Springer Lecture Notes in Mathematics Vol 658, pp. Z06-Z14. 1978. D.S. K a h n and S.B. Priddy. The transfer and stable homotopy theory. Proc. C a m b . Phil. Soc. 83(1978), 103-111. P.O. Kirley. Northwestern. O n the indecomposability of iterated loop spaces. 1975. Math.

Thesis.

U. Koschorke and B. Sanderson. Self intersections and higher Hopf invariants. Topology 17(1978), 283-290.

69 31. 32. 33. 34. 35. 36. 37. G. Lewis. The stable category and generalized Thorn spectra. Chicago. 1978. Thesis.

A. Ziulevicius. The factorization of cyclic reduced powers by secondary cohomology operations. M e m o i r s A m e r . Math. Soc. 42. 1962. 8 M. Mahowald. A n e w infinite family in Zw.. Topology 16(1977), 249-256. M. M a h o w a l d and A. Unell. Bott periodicity at the prime Z and the unstable hornotopy of spheres. Preprint. M. Mahowald. S o m e homotopy classes generated by~j . These proceedings. Springer Lecture Notes in

J.P. May. The geometry of iterated loop spaces. Mathematics Vol Z71. 1972.

J.P. M a y (with contributions b y F . Quinn, N.]Ray, and J.Tornehave). Eoo ring spaces and Eoo ring spectra. Springer Lecture Notes in Mathematics Vol 577. 1977. J. P. May. 456-494. Infinite loop space theory. Bull. Arner. Math. Soc. 83(1977),

38. 39. 40. 41. 42.

J.P. May. Infinite loop space theory revisited. Proc. conf. alg. top. Waterloo, 1978. D. McDuff. Configuration spaces of positive and negative particles. Topology Iterated loop spaces. Annals of Math. 84(1966), 386-403.

14(1975), 91-107.
R.J. Milgram. J. W. Milnor. O n the construction FK. In J.F. Adams. Algebraic Topology, a student's guide. London Math. Soc. Lecture Note Series 4, pp. I19-136. 1972. G. Saitati. 423-428. Loop spaces and K-theory. J. London Math. Soc. 9(1975), These proceedings. Invent. Math.

43. 44. 45. 46. 47. 48. 49. 50. 51. 5Z.

B. Sanderson. G. B. Segal.

The geometry of M a h o w a l d orientations.

Configuration-spaces and iterated loop spaces.

z1(1973), Zl3-Zzz.
G. B. Segah Operations in stable homotopy theory. Lecture Note Series ii, pp. 105-110. 1974. V.P. Snaith. A stable decomposition of ~2nsnx. London Math. Soc.

J. London Math. Soc.

7(1974), 577-583.
V.P. Snaith. OnK~.(~ZX;Zz). QuarterlyJ. Math. 26(1975), 421-436.

N. E. Steenrod. Co'homology operations and obstructions to extending continuous functions. Advances in Math. 8(1972), 371-416. H. Toda. Order of the identity class of a suspension space. Annals of Math. 78(1963), 300-323. n+l R. J. Wellington. The A-algebra H ~ ~]0 }2n+Ix, the Dyer-Lashof algebra, and the A-algebra. Thesis. Chic ago, 1977. S. W. Yang. Thesis. Brandeis. 1978.