<ph f="cmr"> </ph><ph f="cmbx">The Merkuriev-Suslin theorem for any semi-local ring</ph>

Raymond T. Hoobler

City College of New York, New York, New York 10031

• Abstract. We introduce here a method which uses étale neighborhoods to extend results from smooth semi-local rings to arbitrary semi-local rings $A$  by passing to the henselization of a smooth presentation of $A$  . The technique is used to show that étale cohomology of $A$  agrees with Galois cohomology, the Merkuriev-Suslin theorem holds for $A$  , and to describe torsion in ${K}_{2}\left(A\right)$  .

We introduce here a method which uses étale neighborhoods to extend results from smooth semi-local rings to arbitrary semi-local rings. Three applications are given. In the first and last, $A$  is a connected, semi-local ring containing a field $k$  while the second application holds for any connected, semi-local ring $A$  .
1) Let $X=Spec\left(A\right)$  , and let $n$  an integer with $\left(n,char\left(k\right)\right)=1$  . If $F$  is a finite, locally constant sheaf of $\mathbb{Z}/n$  -modules for the étale site on $X$  , we show ${H}^{p}\left(G\left({A}_{s}/A\right),F\left({A}_{s}\right)\right)\to \approx {H}^{p}\left(X,F\right)$  where ${A}_{s}$  is the separable closure of $A,$  the left hand side is the Galois cohomology of $A$  with coefficients in the $G\left({A}_{s}/A\right)$  -module $F\left({A}_{s}\right)$  and the right hand side is the étale cohomology group of the semi-local scheme $X$  with coefficients in $F$  .
2) We extend the Merkuriev-Suslin theorem to a connected, semi-local ring $A$  ; that is, for $n$  relatively prime to the residue characteristics of $A$  , the Galois symbol map ${K}_{2}\left(A\right)/n\to \approx {H}^{2}\left(A,\mathbb{Z}/n\left(2\right)\right)$  is an isomorphism where, as usual, $\mathbb{Z}/n\left(i\right)={\mu }_{n}^{\otimes i}$  . Since this implies the cup product map is surjective, we conclude that any Azumaya algebra of order $n$  in $Br\left(A\right)$  is similar to a tensor product of symbol algebras if $A$  contains a primitive ${n}^{th}$  root of unity.
3) We extend Suslin's computation of the $\ell$  -primary component of the torsion in ${K}_{2}\left(k\right)$  , $k$  a field, to $A$  .
Fix notation as follows. ${H}^{p}\left(X,F\right)$  (or, if $X=Spec\left(A\right)$  , ${H}^{p}\left(A,F\right)$  ) denotes the étale cohomology group of $X$  with coefficients in the étale sheaf $F$  . For connected $A$  , Galois cohomology will be indicated by ${H}^{p}\left({\pi }_{1}\left(A,x\right),M\right)$  where ${\pi }_{1}\left(A,x\right)$  is the algebraic fundamental group of $A$  with base point $x:Spec\left({k}_{s}\right)\to Spec\left(A\right)$  or, equivalently, by ${H}^{p}\left(G\left({A}_{s}/A\right),M\right)$  ( ${\pi }_{1}\left(A,x\right)=$  $G\left({A}_{s}/A\right)$  , the Galois group of the separable closure of $A$  , where the base point is defined by embedding the residue field of $A$  into a separably closed field) when $M$  is an abelian group equipped with a continuous action. ${X}^{\left(d\right)}$  stands for the set of generic points $x$  of irreducible components of $X$  of codimension $d$  . If $M$  is a fixed abelian group, we let $M/n$  stand for $M/nM$  , $M\left\{\ell \right\}$  stand for the $\ell$  -primary component of $M$  , and $Div\left(M\right)$  stand for the maximal divisible subgroup of $M$  .
I would like to thank Srinivas, Bruno Kahn, and Wayne Raskind for several very helpful discussions on these and related matters and Chuck Weibel for the last application and unending patience. I also want to express my delight and appreciation to the Tata Institute of Fundamental Research for providing such a fine location and so much stimulation.

1 Local functors

Let $\left(A,I\right)$  be a pair consisting of a commutative ring with $1$  and an ideal $I\subset A.$  An étale neighborhood of $\left(A,I\right)$  is a pair $\left({A}^{\prime },{I}^{\prime }\right)$  and an etale map $u:A\to {A}^{\prime }$  such that $u\left(I\right){A}^{\prime }={I}^{\prime }$  and $u$  induces an isomorphism $\overline{u}:A/I\to {A}^{\prime }/{I}^{\prime }$  . Geometrically etale neighborhoods of a closed set $W\subset X=Spec\left(A\right)$  look like a cartesian diagram where $\pi$  is etale:
 $\begin{array}{ccc}{W}^{\prime }& \subset & {X}^{\prime }\end{array}$
 $\begin{array}{ccc}↓\mathbb{\approx }& & ↓\pi \end{array}$
 $\begin{array}{ccc}W& \subset & X\end{array}$
The set of all étale neighborhoods of $I$  in $A$  is a filtered category which we denote $Et\left(I\right)$  . If we assume that $I\subseteq rad\left(A\right),$  then $Et\left(I\right)$  is used to define the henselization of the pair $\left(A,I\right)$  [15by setting ${{A}_{/I}}^{h}:=\underset{\to }{lim}\left({A}^{\prime },{I}^{\prime }\right)\in Et\left(I\right){A}^{\prime }.$  Then the pair $\left({A}_{/I}^{h},{I}^{h}\right)$  is a hensel pair where ${I}^{h}=I{A}_{/I}^{h}.$  If $I=\mathfrak{m}$  is maximal, we get the hensel local ring ${A}_{\mathfrak{m}}^{h}.$  Let $\mathcal{C}$  be a category containing $Et\left(I\right)$  as a full subcategory and ${A}_{/I}^{h}$  .
We introduce the definition of a local functor.
Definition 1. A covariant functor $F:\mathcal{C}\to Ab$  is said to be local (for the étale topology) with respect to a closed set $V\left(I\right)$  in $Spec\left(A\right)$  if
• (1) $I$  is a radical ideal in $A$  , $Et\left(I\right)$  is a full subcategory of $\mathcal{C}$  , and ${A}_{/I}^{h}\in \mathcal{C}$
• (2) the natural map  $\begin{array}{c}\underset{\to }{lim}\left({A}^{\prime },{I}^{\prime }\right)\in Et\left(I\right)F\left({A}^{\prime }\right)\to F\left(A/I\right)\end{array}$ (1)
is an isomorphism.
We say that $F$  is local for the étale topology on $X$  if, for all affine open subsets $Spec\left(A\right)\subset X$  and all closed sets $V\left(I\right)\subset Spec\left(A\right),$  $F$  is local with respect to the closed set $V\left(I\right)\subset Spec\left(A\right).$  In many cases of importance the limit condition above can be replaced with a condition involving the henselization of $A$  along $I$  . Recall that a covariant functor $F:\mathcal{C}\to Ab$  is said to be locally of finite presentation if for any filtered inductive limit $A={lim}_{\stackrel{\to }{i\in I}}{A}_{i}$  , the natural map ${lim}_{\stackrel{\to }{i\in I}}F\left({A}_{i}\right)\to F\left(A\right)$  is an isomorphism.
For such functors we are only interested in a hensel pair condition.
Definition 2. A functor $F:\mathcal{C}\to Ab$  is said to satisfy the hensel pair condition for $I$  , resp. epic hensel pair condition for $I$  , if $F\left(A\right)\to F\left(A/I\right)$  is an isomorphism, resp. an epimorphism, for any hensel pair $\left(A,I\right).$  The functor satisfies the hensel pair condition, resp. epic hensel pair condition if it does so for any hensel pair $\left(A,I\right)\in \mathcal{C}.$
Then if $F$  is locally of finite presentation and $F$  satisfies the hensel pair condition, $F$  is local for the closed set $V\left(I\right)$  . Thus in applying this definition we will first verify that $F$  is locally of finite presentation and then that $F\left({A}^{h}\right)\to F\left({A}^{h}/{I}^{h}\right)$  is an isomorphism when $\left({A}^{h},{I}^{h}\right)$  is a hensel pair.
We are primarily interested in three examples of functors satisfying the hensel pair condition.
Example 1.
• (1) ${H}^{i}\left(-,F\right)$  where $F$  is a locally constant sheaf of $\mathbb{Z}/n$  modules on $Spec\left(A{\right)}_{et}$  ([18or [5)
• (2) ${K}_{i}\left(-;\mathbb{Z}/n\right)$  where $n$  is relatively prime to the residue characteristics of $A$  ([4)and
• (3) ${K}_{2}\left(-\right)/n$  where $n$  is relatively prime to the residue characteristics of $A.$
The last example is easily seen to satisfy the hensel pair condition.
Surjectivity of ${K}_{2}\left(A\right)/n\to {K}_{2}\left(A/I\right)/n$  for a hensel pair $\left(A,I\right)$  follows immediately since ${K}_{2}$  is generated by symbols, and injectivity follows by a straightforward calculation done in [11,Appendixor from Gabber's theorem [4. Alternatively we could use the universal coefficient theorem, the second example, and note that the third term in the universal coefficient theorem is the $n-$  torsion in ${K}_{1}$  which is the ${n}^{th}$  roots of unity. These examples will be discussed at greater length in the applications.
Our applications are a result of the following theorem.
Theorem 1. Let $A$  be a semi-local ring essentially of finite type over a base ring $k$  . Suppose $A\sim =B/J$  where $B$  is a smooth semi-local ring essentially of finite type over $k$  . Let $C$  be a category of semi-local ring extensions of $B$  containing $A$  , ${B}_{/J}^{h}$  , and the full subcategory $Et\left(J\right)$  . Suppose ${F}_{1},{F}_{2}:C\to Ab$  are two covariant functors which are either both local for $V\left(J\right)$  or ${F}_{2}$  is local for $V\left(J\right)$  and ${F}_{1}$  is locally of finite presentation and satisfies the epic hensel pair condition. Let $\phi :{F}_{1}\to {F}_{2}$  be a natural transformation such that $\phi \left({B}^{\prime }\right)$  is an isomorphism if ${B}^{\prime }$  is smooth and essentially of finite type over $k$  . Then $\phi \left(A\right)$  is an isomorphism.
• Proof. If $A$  is essentially of finite type over $k$  , then $A$  has a presentation as $B/J$  where $B$  is essentially of finite type and smooth over $k$  . Thus $\phi \left(B\right)$  is an isomorphism as is $\phi \left({B}^{\prime }\right)$  for any ${B}^{\prime }\in Et\left(J\right)$  and so, by ( 1 ), $\phi \left(A\right)$  must be an isomorphism in the first case. In the second case, ${F}_{1}\left({B}_{/J}^{h}\right)\to {F}_{1}\left(A\right)$  must be an isomorphism since ${F}_{2}$  is local for $V\left(J\right)$  and $\phi \left({B}_{/J}^{h}\right)$  is an isomorphism.
Corollary 1. In the above situation ${F}_{1}$  is local for $V\left(J\right)$  if it is locally of finite presentation and satisfies the epic hensel pair condition.

2 Applications

We have three applications of this perspective. They all rely on first establishing an isomorphism for semi-local rings smooth over a base scheme and then using the examples above to extend the result to arbitrary semi-local schemes over the base.

2.1 Galois cohomology = Étale cohomology

As a first application we consider the relationship between étale and Galois cohomology of semi-local rings. While it is a generally accepted fact for smooth local rings over a field, the details do not, as far as I know, appear in the literature although there is an argument, due to Bruno Kahn, when the generalized Kato conjecture holds [11. We begin by reviewing and reinterpreting Galois cohomology and then proving the isomorphism in this case.
Let $S$  be an arbitrary connected scheme. Define a site ${S}_{ret}$  by considering the category of schemes $T\to S$  which are finite and étale (= revêtement étale) over $S$  . A covering morphism is a surjection of schemes over $S$  , ${T}_{2}\to {T}_{1}$  , which will necessarily be finite and étale.
This is the same as considering the class $E=\left(ret\right)$  of all finite, étale morphisms [12,ChapterII,Section1. ${S}_{ret}$  is then the small site $\left(E/S{\right)}_{E}$  where $E$  consists of ”revêtements étales” as above. This site is discussed in [12,Chapter1where it is called $FEt$  and used to describe the fundamental group of $S$  . Cohomology in ${S}_{ret}$  can be identified with Galois cohomology according to the following recipe.
Suppose $F$  is a sheaf on ${S}_{ret}$  . Then $F$  takes finite, disjoint unions of schemes to direct products. Thus the discussion in [12,ChapterIII,Example2.6applies to the covering $T\to S$  in ${S}_{ret}$  where $T$  is Galois over $S$  with group $G$  . Hence
 $\begin{array}{c}{\stackrel{ˇ}{H}}^{p}\left(T/S,F\right)\sim ={H}^{p}\left(G,F\left(T\right)\right)\end{array}$ (2)
where $F\left(T\right)$  is a left $G$  -module via the action of $G$  on $T$  .
Let $\iota :\mathbf{S}\left({S}_{ret}\right)\to \mathbf{P}\left({S}_{ret}\right)$  be the forgetful functor that regards a sheaf as a contravariant functor defined on ${S}_{ret}$  . When we pass to the limit over all coverings $T/S$  in ${S}_{ret}$  , the left hand side of ( 2 ) becomes ${\stackrel{ˇ}{H}}^{p}\left({S}_{ret},\iota F\right)$  . If we fix a geometric point $x:Spec\left({k}_{s}\right)\to S$  where ${k}_{s}$  is a separably closed field, then the theory of the fundamental group expresses any finite, étale covering $T\to S$  as a quotient of a Galois covering. Consequently the right hand side of ( 2 ) is $\underset{\to }{lim}{H}^{p}\left({G}_{i},F\left({T}_{i}\right)\right)$  where ${G}_{i}$  ranges through the finite quotients of ${\pi }_{1}\left(S,x\right)$  and ${T}_{i}\to S$  is the corresponding étale covering with group ${G}_{i}$  . Let $F\left({S}_{s}\right)$  denote the abelian group $\underset{\to }{lim}F\left({T}_{i}\right)$  . If $F$  is a locally constant sheaf on $S$  , then the subgroup of ${\pi }_{1}\left(S,x\right)$  defining the covering $T\to S$  such that $F{|}_{T}$  is constant is of finite index in ${\pi }_{1}\left(S,x\right)$  and acts trivially on $F\left({S}_{s}\right)$  . Hence $F\left({S}_{s}\right)$  has a continuous ${\pi }_{1}\left(S,x\right)$  -action. In general this need not be the case, but we have enough to conclude that for $F$  a locally constant sheaf on ${S}_{ret}$  ,
 $\begin{array}{c}{H}^{p}\left({S}_{ret},F\right)\sim ={H}^{p}\left({\pi }_{1}\left(S,x\right),F\left({S}_{s}\right)\right)\end{array}$ (3)
(Note that a finite, locally constant sheaf for the étale topology is the same as a finite, locally constant sheaf on ${S}_{ret}$  .) In view of the isomorphism ( 2 ) on ${S}_{ret}$  , this is essentially the statement that sheaf cohomology coincides with Čech cohomology. Corollary 2.5 of Chapter III [12gives sufficient conditions for this; namely, for every surjection $F\to {F}^{\prime \prime }$  of sheaves, the map $\underset{\to }{lim}\prod F\left({U}_{{i}_{0}...{i}_{p}}\right)\to \underset{\to }{lim}\prod {F}^{\prime \prime }\left({U}_{{i}_{0}...{i}_{p}}\right)$  is surjective where the limit is over all coverings $\left\{\prod {U}_{i}\to S\right\}$  of $S$  and ${U}_{{i}_{0}...{i}_{p}}={U}_{{i}_{0}}{×}_{S}\cdots {×}_{S}{U}_{{i}_{p}}.$  But if $T\to S$  is a Galois covering of $S$  with group $G$  , then $F\left({T}_{{i}_{0}...{i}_{p}}\right)=F\left({\prod }_{{G}^{×p}}T\right)={\prod }_{{G}^{×p}}F\left(T\right)$  . Thus taking limits of $F\left({T}_{{i}_{0}...{i}_{p}}\right)$  over all coverings means taking limits over all coverings of products of copies of $F\left(T\right)$  and so surjectivity is immediate.
This discussion reduces our first application to the following theorem about the change of sites morphism $\tau :{S}_{et}\to {S}_{ret}$  .
Theorem 2. Let $A$  be a connected semi-local ring containing a field $k$  . If $F$  is a finite, locally constant sheaf of $\mathbb{Z}/n$  -modules for the étale site on $A$  where $\left(n,char\left(k\right)\right)=1$  , then
 $\begin{array}{c}{H}^{p}\left({A}_{ret},{\tau }_{*}F\right)\to {H}^{p}\left({A}_{et},F\right)\end{array}$ (4)
is an isomorphism for all p.
• Proof. We wish to apply our extension theorem. We begin by treating the case of a smooth, semi-local ring $R$  essentially of finite type over $k$  , an algebraically closed field first. We will show that ${R}^{q}{\tau }_{*}F=0$  for $q>0$  and any finite, locally constant sheaf $F$  of $\mathbb{Z}/n$  -modules. Suppose $x\in {H}^{q}\left(R,F\right)$  with $q>0.$  We may assume $F={C}_{X}$  is a constant sheaf of $\mathbb{Z}/n$  -modules with value $C$  since $F$  becomes constant after a finite, etale extension of $R$  . By assumption there is a smooth, connected variety $X$  over $k$  , closed points ${t}_{i}\in X,1\le i\le m$  , such that $R\sim ={\mathit{O}}_{X,{t}_{1}...{t}_{m}}$  and an element ${x}^{\prime }\in {H}^{q}\left(X,{C}_{X}\right)$  such that ${x}^{\prime }{|}_{Spec\left({\mathit{O}}_{X,{t}_{1}...{t}_{m}}\right)}=x$  .
We construct a finite, étale covering $\rho :{X}^{\prime }\to X$  such that ${\rho }^{*}\left({x}^{\prime }\right)=0$  using induction on $dim\left(X\right)$  and the existence of Artin neighborhoods.
If $dim\left(X\right)=0$  , the assertion is obvious. If $dim\left(X\right)=d$  , the existence of an Artin neighborhood relative to $k$  means there is a diagram ([12,p.117or [16)  $\begin{array}{ccccc}X& \to j& \overline{X}& ←i& Y\end{array}$
 $\begin{array}{ccccc}& f↘& ↓\overline{f}& ↙g& \end{array}$
 $\begin{array}{ccccc}& & {X}_{0}& & \end{array}$
in which
• (1) $j$  is an open immersion, dense in each fibre and $Y=\overline{X}-X;$
• (2) $\overline{f}$  is smooth and projective with geometrically irreducible fibres of dimension one;
• (3) $g$  is finite and étale and each fibre of $g$  is non-empty.
(Although the statement of the existence of Artin neighborhoods given in [12refers only to a local ring, the proof as given in [16,ExposeXI,Section3clearly extends to the semi-local case. In fact, the proof of Proposition 3.3, the key statement, is footnoted to that effect. Since base extension of a local ring from an arbitrary field $k$  to its separable closure ${k}_{s}$  may produce a semi-local ring we need this more general result.) Now $\left(Y,\overline{X}\right)$  is a smooth ${X}_{0}-$  pair. The Gysin sequence [12,ChapterVI,Corollary5.3andRemark5.4(a)becomes ${\overline{f}}_{*}{C}_{\overline{X}}\sim ={f}_{*}{C}_{X},$  $0\to {R}^{1}{\overline{f}}_{*}{C}_{\overline{X}}\to {R}^{1}{f}_{*}{C}_{X}\to {g}_{*}\left({C}_{Y}\otimes {T}_{Y/\overline{X}}\right)\to {R}^{2}{\overline{f}}_{*}{C}_{\overline{X}}\to {R}^{2}{f}_{*}{C}_{X}\to 0,$  and ${R}^{p}{\overline{f}}_{*}{C}_{\overline{X}}\sim ={R}^{p}{f}_{*}{C}_{X}$  for $p>2$  . The base change maps for the proper morphisms $\overline{f}$  and $g$  are isomorphisms. Consequently the base change map for $f$  is also an isomorphism. Since the fibres of $f$  are non-complete curves, we find ${R}^{p}{f}_{*}C=0$  for $p>1$  . Moreover $\overline{f}$  is smooth and so ${R}^{1}{\overline{f}}_{*}{C}_{\overline{X}}$  is finite and locally constant on ${X}_{0}$  as is ${g}_{*}\left({C}_{Y}\otimes {T}_{Y/X}\right)$  since $g$  is an etale covering space and ${T}_{Y/X}$  is locally constant on $Y$  . Therefore we conclude that ${R}^{p}{f}_{*}{C}_{X}$  is a finite, locally constant sheaf of $\mathbb{Z}/n-$  modules for $p=0$  or $1$  .
This information shows that the Leray spectral sequence for $f$  degenerates into the long exact sequence  $\begin{array}{c}\begin{array}{cccc}...\to {H}^{q}\left({X}_{0},{f}_{*}{C}_{X}\right)\to {H}^{q}\left(X,{C}_{X}\right)\to & {H}^{q-1}\left({X}_{0},{R}^{1}{f}_{*}{C}_{X}\right)& & \to {H}^{q+1}\left({X}_{0},{f}_{*}{C}_{X}\right)\to ...\end{array}\end{array}$ (5)
and we may apply the induction hypothesis. Then for any $y\in {H}^{q-1}\left({X}_{0},{R}^{1}{f}_{*}{C}_{X}\right),$  there is, after shrinking ${X}_{0}$  if necessary, a finite, étale covering space ${\rho }^{\prime }:{X}_{0}^{\prime }\to {X}_{0}$  such that ${\rho }^{\prime *}\left(y\right)=0$  . Now by using the pullbacks of appropriate coverings of ${X}_{0}$  to $X$  and a diagram chase, it is a straightforward matter to produce a finite, étale covering $\rho :{X}^{\prime }\to X$  such that ${\rho }^{*}\left({x}^{\prime }\right)=0$  .
This argument is, of course, the essence of Remark 3.16 [12,ChapterIIIand shows that ${R}^{q}{\tau }_{*}F=0$  for $q>0$  . Hence the Leray spectral sequence for the change of site morphism $\tau :{X}_{et}\to {X}_{ret}$  collapses, and we conclude ${H}^{p}\left({R}_{ret},{\tau }_{*}F\right)\sim ={H}^{p}\left({R}_{\stackrel{´}{e}t},F\right)$  Now suppose that $R$  is a semi-local ring which is smooth and essentially of finite type over $k$  , a not necessarily algebraically closed field. Then for any scheme $X$  there is a Hochschild-Serre spectral sequence [12,ChapterIII,Remark2.21(b)  $\begin{array}{c}{H}^{p}\left(G\left({k}_{s}/k\right),{H}^{q}\left({X}_{s},F\right)\right)⇒{H}^{n}\left(X,F\right)\end{array}$ (6)
where ${X}_{s}=X{×}_{k}Spec\left({k}_{s}\right)$  . Similarly the theory of algebraic fundamental groups constructs a profinite group extension $1\to G\left(\left(R{\otimes }_{k}{k}_{s}{\right)}_{s}/R{\otimes }_{k}{k}_{s}\right)\to G\left({R}_{s}/R\right)\to G\left({k}_{s}/k\right)\to 1$  and so there is a corresponding Hochschild-Serre spectral sequence  $\begin{array}{c}\begin{array}{ccc}{H}^{p}\left(G\left({k}_{s}/k\right),{H}^{q}\left(G\left(\left(R{\otimes }_{k}{k}_{s}{\right)}_{s}/R{\otimes }_{k}{k}_{s}\right),F\left(\left(R{\otimes }_{k}{k}_{s}{\right)}_{s}\right)\right)\right)& ⇒& {H}^{n}\left(G\left({R}_{s}/R\right),F\left({R}_{s}\right)\right)\end{array}\end{array}$ (7)
The change of site morphism defines a homomorphism $\Psi$  from ( 6 ) to ( 7 ).
Since $F$  is a sheaf of $\mathbb{Z}/n$  -modules where $n$  is relatively prime to char $\left(k\right)$  , we conclude, as usual, that ${H}^{p}\left(X{×}_{k}Spec\left({k}_{s}\right),F\right)\sim ={H}^{p}\left(X{×}_{k}Spec\left(\overline{k}\right),F\right)$  where $\overline{k}$  is the algebraic closure of $k$  . Consequently, by the first case, $\Psi$  is an isomorphism on the ${E}_{2}^{p,q}$  terms. This shows that ( 4 ) is an isomorphism when $A$  is essentially smooth over a not necessarily algebraically closed field $k.$  Some preparation is required in order to apply Theorem  1 to an arbitrary semi-local ring $A$  whether we work with the small or large étale site. Our hypothesis on $F$  shows that $F$  is representable in either the small or large site by a semi-local scheme $\mathbb{F}$  which is finite and étale over $A.$  Now $A$  is a filtered limit of connected semi-local rings ${A}_{i}$  essentially of finite type over $k.$  Consequently there is a finite, etale covering space ${\mathbb{F}}_{{i}_{0}}$  over ${A}_{{i}_{0}}$  for a sufficiently large ${A}_{{i}_{0}}$  such that ${\mathbb{F}}_{{i}_{0}}{×}_{{A}_{{i}_{0}}}A\mathbb{\approxeq }\mathbb{F}$  . We use ${\mathbb{F}}_{{i}_{0}}$  to define a locally constant sheaf ${F}_{{i}_{0}}$  over ${A}_{{i}_{0}}$  whose restriction to $A$  is $F.$  Next find a presentation of ${A}_{{i}_{0}}$  by a smooth connected semi-local ring ${B}_{{i}_{0}}$  essentially of finite type over $k$  so that ${A}_{{i}_{0}}\sim ={B}_{{i}_{0}}/I$  . Since any étale covering space of ${A}_{{i}_{0}}$  may be lifted to an étale covering space of ${B}_{{i}_{0}}$  (by the theorem of the primitive element all one has to do is lift a separable polynomial), we may assume there is a finite, locally constant sheaf $G$  on $Spec\left({B}_{{i}_{0}}\right)$  whose restriction to $Spec\left(A\right)$  is $F$  .
Let $C$  be the category of connected semi-local rings over $B$  , ${F}_{1}={H}^{p}\left({-}_{ret},{\tau }_{*}G\right)$  , and ${F}_{2}={H}^{p}\left(-,G\right)$  where now we consider $G$  on the large site over $Spec\left(B\right)$  . Note that $G$  is represented by a finite étale covering space ${B}^{\prime }$  of $B$  in ${B}_{et}$  and so $G$  is also represented by ${B}^{\prime }$  on the big étale site over $B$  . Since ${B}^{\prime }$  also defines a covering in ${B}_{ret}$  the same observation applies to the sheaf ${\tau }_{*}G$  in the big site ${B}_{ret}$  . After showing that the hypotheses of our extension theorem are met we can conclude that ( 4 ) is an isomorphism for any $B-$  algebra $A$  with the property that $G{|}_{A}=F$  and ${\tau }_{*}G{|}_{A}={\tau }_{*}F$  .
Both ${F}_{1}$  and ${F}_{2}$  are locally of finite presentation. For ${F}_{2}$  this is [12,ChapterIII,Lemma1.16suitably interpreted. For ${F}_{1}$  we need a different argument. Suppose $R=\underset{\to }{lim}{R}_{i}$  with ${R}_{i}$  a connected semi-local ring over $B$  where $i\in \mathit{I}$  , a filtered set. Any étale covering space of $\underset{\to }{lim}{R}_{i}$  comes from an étale covering space defined over one of the ${R}_{i}$  and homomorphisms between any two étale covering spaces of $\underset{\to }{lim}{R}_{i}$  descend to a homomorphism between covering spaces over one of the ${R}_{i}$  . Thus we have an isomorphism ${lim}_{{R}_{i}}{lim}_{{T}_{i}/{R}_{i}}{\stackrel{ˇ}{H}}^{p}\left({T}_{i}/{R}_{i},G\right)\to {lim}_{T/R}{\stackrel{ˇ}{H}}^{p}\left(T/R,G\right)$  since the Čech cohomology groups depend only on the Galois group of the covering by ( 2 ) and $G$  is locally constant. Thus, using the Galois cohomology interpretation of ${F}_{1}$  , we also have ${H}^{p}\left({R}_{ret},{\tau }_{*}G\right)\sim =\underset{\to }{lim}{H}^{p}\left({R}_{iret},{\tau }_{*}G\right)$  and so ${F}_{1}$  is locally of finite presentation.
Now we turn to the hensel pair condition. Suppose $\left(R,J\right)$  is a hensel couple and $G$  is a locally constant sheaf of $\mathbb{Z}/n-$  modules on $Spec\left(R\right)$  .
Then $G$  is represented by a finite, étale group scheme over $R$  , and, by Example  1 , we have an isomorphism for a hensel couple $\left(R,J\right)$  and such a $G$  , ${H}^{p}\left({R}_{\stackrel{´}{e}t},G\right)\sim ={H}^{p}\left(R/{J}_{\stackrel{´}{e}t},G{|}_{R/J}\right).$  Thus ${F}_{2}$  is a local functor for the etale topology satisfying the hensel pair condition. As for ${F}_{1}$  , there is a one-to-one correspondence between étale covering spaces of $R/J$  and étale covering spaces of $R$  , and so we have an isomorphism ${H}^{p}\left({R}_{ret},{\tau }_{*}G\right)\sim ={H}^{p}\left(R/{J}_{ret},{\tau }_{*}\left(G{|}_{R/J}\right)\right).$  Hence ${F}_{1}$  is a local functor for the etale topology and also satisfies the hensel pair condition. We can now apply Theorem  1 to conclude the proof.
Corollary 2. Let $A$  be a semi-local ring containing a field $k$  .
Suppose $F$  is a finite, locally constant sheaf of $\mathbb{Z}/n$  -modules where $\left(n,char\left(k\right)\right)=1$  . Then for any $x\in {H}^{p}\left(A,F\right)$  , $p>0$  , there is a Galois extension ${A}^{\prime }/A$  such that $x{|}_{{A}^{\prime }}=0\in {H}^{p}\left({A}^{\prime },F\right)$  .

2.2 Merkurjev-Suslin theorem

Our second application is an extension of the Mercuriev-Suslin theorem to an arbitrary semi-local ring.
For a semi-local ring $A$  and any $n$  which is relatively prime to the residue characteristics of $A$  , Kummer theory provides a natural isomorphism ${K}_{1}\left(A\right)/n\to {H}^{1}\left(A,\mathbb{Z}/n\left(1\right)\right).$  The Galois symbol map [20is obtained by extending this map multiplicatively to ${K}_{i}\left(A\right)/n\to {H}^{i}\left(A,\mathbb{Z}/n\left(i\right)\right).$
Theorem 3. Let $A$  be a semi-local ring. Then if $n$  is relatively prime to the residue characteristics of $A$  , the Galois symbol map
 $\begin{array}{c}{K}_{2}\left(A\right)/n\to {H}^{2}\left(A,\mathbb{Z}/n\left(2\right)\right)\end{array}$ (8)
is an isomorphism.
Corollary 3. Let $A$  be a semi-local ring, and suppose $n$  is an integer which is relatively prime to the residue characteristics of $A$  . Then the cup product map ${H}^{1}\left(A,\mathbb{Z}/n\left(1\right)\right)×{H}^{1}\left(A,\mathbb{Z}/n\left(1\right)\right)\to {H}^{2}\left(A,\mathbb{Z}/n\left(2\right)\right)$  is onto. In particular, if $A$  contains a primitive ${n}^{th}$  root of unity, the $n$  -torsion in the Brauer group of $A$  is generated by symbol algebras $\left(a,b{\right)}_{n}$  .
Corollary 4. Let $X$  be a scheme, and suppose $n$  is an integer which is relatively prime to the residue characteristics of $X$  .
The symbol map defines an isomorphism of Zariski sheaves $\underline{{K}_{2}/n}\to \underline{{H}^{2}\left(\mathbb{Z}/n\left(2\right)}\right).$
The proof requires some preliminary material on K-theory with coefficients, Chern classes, and a discussion of results of Gillet which will be used to prove the theorem for semi-local rings smooth and essentially of finite type over a semi-local Dedekind ring.
Fix a ring $A$  , a prime $\ell$  that is a unit in $A$  , and an integer $n$  which is at least $2$  if $\ell =2$  (See [21,Proposition2.4for a detailed discussion of the case $\ell =2$  .) Algebraic K-theory with coefficients, ${K}_{r}\left(A;\mathbb{Z}/{\ell }^{n}\right)$  , is a bifunctor in $A$  and the cyclic group $\mathbb{Z}/{\ell }^{n}$  , and, for a pair of integers $1\le i,0\le j\le 2i$  , there is a theory of Chern classes given by natural transformations ${c}_{i,j}^{{\ell }^{n}}:{K}_{2i-j}\left(A;\mathbb{Z}/{\ell }^{n}\right)\to {H}^{j}\left(A,\mathbb{Z}/{\ell }^{n}\left(i\right)\right)$  satisfying certain properties. Among the properties these objects satisfy are:
• (1) (universal coefficient theorem) There is a natural short exact sequence  $\begin{array}{c}0\to {K}_{r}\left(A\right)/{\ell }^{n}{K}_{r}\left(A\right)\to {K}_{r}\left(A;\mathbb{Z}/{\ell }^{n}\right){\to }_{{\ell }^{n}}{K}_{r-1}\left(A\right)\to 0\end{array}$ (9)
• (2) (functoriality) For any $n>m$  , the diagram  $\begin{array}{ccc}{K}_{2i-j}\left(A;\mathbb{Z}/{\ell }^{m}\right)& \to & {K}_{2i-j}\left(A;\mathbb{Z}/{\ell }^{n}\right)\end{array}$
 $\begin{array}{ccc}↓{c}_{ij}^{{\ell }^{m}}& & ↓{c}_{ij}^{{\ell }^{n}}\end{array}$
 $\begin{array}{ccc}{H}^{j}\left(A,\mathbb{Z}/{\ell }^{m}\left(i\right)\right)& \to & {H}^{j}\left(A,\mathbb{Z}/{\ell }^{n}\left(i\right)\right)\end{array}$
defined from the inclusion $\mathbb{Z}/{\ell }^{m}↪\mathbb{Z}/{\ell }^{n}$  commutes.
• (3) (naturality) For any valuation ring ${\mathcal{O}}_{v}$  and prime $\ell$  distinct from the residue characteristic of ${\mathcal{O}}_{v}$  and any $n$  , we have a commutative diagram  $\begin{array}{ccc}{K}_{2i-j}\left(F;\mathbb{Z}/{\ell }^{m}\right)& \to \overline{\partial }& {K}_{2i-j-1}\left(\kappa \left(v\right);\mathbb{Z}/{\ell }^{n}\right)\end{array}$
 $\begin{array}{ccc}↓{c}_{ij}^{{\ell }^{n}}& & ↓\left(1-i\right){c}_{i-1j-1}^{{\ell }^{n}}\end{array}$
 $\begin{array}{ccc}{H}^{j}\left(F,\mathbb{Z}/{\ell }^{n}\left(i\right)\right)& \to \partial & {H}^{j-1}\left(\kappa \left(v\right),\mathbb{Z}/{\ell }^{n}\left(i-1\right)\right)\end{array}$
where $\kappa \left(v\right)$  is the residue field of ${\mathcal{O}}_{v}$  and $F$  is its field of fractions. In particular the Chern classes fit together to define a map from the Quillen-Gersten complex for ${K}_{*}\left(-;\mathbb{Z}/{\ell }^{n}\right)$  to the Bloch-Ogus complex for ${H}^{*}\left(-,\mathbb{Z}/{\ell }^{n}\left(\cdot \right)\right)$  . (Quillen's argument in [14applies unchanged to ${K}_{i}\left(-;\mathbb{Z}/m\right)$  where $m={\ell }^{n}$  .)
Details of the above properties can be found in [17.
Now the other tool we need is a reformulation and extension to a semi-local ring of a result of Gillet-Levine[7and Gillet[6.
Theorem 4 (Gillet-Levine, Gillet). Let $B$  be a connected, semi-local ring with quotient field $K$  which is smooth and essentially of finite type over a semi-local Dedekind ring $D$  . Then, for any integer $n$  relatively prime to the residue characteristics of $B$  and any $q\ge 0,$  we have a Gersten-Quillen resolution of ${K}_{q}\left(B;\mathbb{Z}/n\right):$  $0\to {K}_{q}\left(B;\mathbb{Z}/n\right)\to {K}_{q}\left(K;\mathbb{Z}/n\right)\to {\prod }_{x\in {B}^{\left(1\right)}}{K}_{q-1}\left(\kappa \left(x\right);\mathbb{Z}/n\right)\to \cdots$
• Proof. The argument of Gillet and Levine and Gillet immediately extends to semi-local rings as is clear from going through their arguments.
We also need a weak version of this result for etale cohomology. This has been demonstrated by Gillet but remains unpublished[8. A proof for $n$  an odd integer relatively prime to the residue characteristics of $B$  is given in an appendix at the end of this paper.
Lemma 1. Let $B$  be a connected, semi-local ring which is smooth and essentially of finite type over a semi-local Dedekind ring, $\mathcal{D}$  . Let $K$  be the quotient field of $B.$  Then, for any integer $n$  which is relatively prime to the residue characteristic of $\mathcal{D},$  ${H}^{2}\left(B,\mathbb{Z}/n\left(2\right)\right)\to {H}^{2}\left(K,\mathbb{Z}/n\left(2\right)\right)$  is a monomorphism.
We will need the universal coefficient theorem to produce a Gersten-Quillen sequence for ${K}_{2}\left(-\right)/n$  using work of Gillet and Gillet-Levine and then the compatibility of the Chern class map with the Quillen-Gersten complex and the Bloch-Ogus complex will do the rest in the smooth case.
• Proof. We deal with a semi-local algebra $A$  over a mixed characteristic Dedekind ring $\mathcal{D}$  first. We may assume $A$  is connected. Both ${K}_{2}\left(-\right)/n$  and ${H}^{2}\left(-,\mathbb{Z}/n\left(2\right)\right)$  are locally of finite presentation over $Spec\left(\mathcal{D}\right),$  and so we may assume $A$  is essentially of finite type over $\mathcal{D}.$  Now realize $A\mathbb{\approxeq }B/J$  where $B$  is a smooth, semi-local ring which is essentially of finite type over ${\mathcal{O}}_{v}.$  Let $\mathcal{C}$  be the category of connected semi-local rings essentially of finite presentation over $\mathcal{D}.$  We wish to apply Theorem  1 to this situation with $k=\mathcal{D},$  ${F}_{1}={K}_{2}\left(-\right)/n,$  ${F}_{2}={H}^{2}\left(-,\mathbb{Z}/n\left(2\right)\right),$  and $\phi :{F}_{1}\to {F}_{2}$  being the Galois symbol map.
First observe that ${K}_{2}\left(-\right)/n$  satisfies the epic hensel pair condition since it is generated by symbols and ${H}^{2}\left(-,\mathbb{Z}/n\left(2\right)\right)$  satisfies the hensel pair condition by Gabber or Strano's result ([5or [18). It remains to show that $\phi \left(B\right)$  is an isomorphism if $B$  is smooth and essentially of finite type over $\mathcal{D}.$  The exactness of  $\begin{array}{c}{K}_{2}\left(B;\mathbb{Z}/n\right)↪{K}_{2}\left(K;\mathbb{Z}/n\right)\to {\prod }_{x\in \left(SpecB{\right)}^{\left(1\right)}}{K}_{1}\left(\kappa \left(x\right);\mathbb{Z}/n\right)\end{array}$ (10)
where $K$  is the quotient field of $B$  was shown by Gillet [6using work of Gillet-Levine [7(see Theorem 2.6). We see that ${K}_{1}\left(\kappa \left(x\right);\mathbb{Z}/n\right)\mathbb{\approxeq }{K}_{1}\left(\kappa \left(x\right)\right)/n,$  ${K}_{2}\left(B;\mathbb{Z}/n\right)$  is an extension of ${\mu }_{n}\left(B\right)\left({=}_{n}{K}_{1}\left(B\right)\right)$  by ${K}_{2}\left(B\right)/n,$  and similarly for ${K}_{2}\left(K;\mathbb{Z}/n\right).$  We can now reinterpret ( 10 ) as the exact sequence of the first line below  $\begin{array}{cccc}& {K}_{2}\left(B\right)/n↪& {K}_{2}\left(K\right)/n\to & {\prod }_{x\in \left(SpecB{\right)}^{\left(1\right)}}{K}_{1}\left(\kappa \left(x\right)\right)/n\end{array}$
 $\begin{array}{cccc}& ↓& ↓\mathbb{\approxeq }& ↓\mathbb{\approxeq }\end{array}$
 $\begin{array}{cccc}& {H}^{2}\left(B,\mathbb{Z}/n\left(2\right)\right)↪& {H}^{2}\left(K,\mathbb{Z}/n\left(2\right)\right)\to & {\prod }_{x\in \left(SpecB{\right)}^{\left(1\right)}}{H}^{1}\left(\kappa \left(x\right),\mathbb{Z}/n\left(1\right)\right)\end{array}$
. Here the bottom row is not necessarily exact but is a complex, and the first map is a monomorphism by Lemma  1 , the first vertical isomorphism is the Merkuriev-Suslin theorem for fields, and the second vertical isomorphism is the observation that both groups are isomorphic to ${\prod }_{x\in \left(SpecB{\right)}^{\left(1\right)}}\kappa \left(x{\right)}^{*}/\kappa \left(x{\right)}^{*n}.$  The diagram commutes by the naturality condition above. Since the bottom row is a complex we conclude that the first vertical map is an isomorphism as desired. Theorem 1.1 now finishes this case.
The case of a semi-local ring containing a field is similar but simpler since we can use Grayson's version of the Gersten-Quillen sequence [9and the Bloch-Ogus sequence [3in place of the argument involving algebraic K-theory with coefficients and Lemma  1 .
Suppose the Bloch-Kato conjecture holds for fields; that is, the Galois symbol map
 $\begin{array}{c}{K}_{i}^{M}\left(K\right)/n\to {H}^{i}\left(K,\mathbb{Z}/n\left(i\right)\right)\end{array}$ (11)
is an isomorphism where ${K}_{i}^{M}\left(K\right)$  is the ${i}^{th}$  Milnor K-group of the field $K$  and $n$  is relatively prime to $char\left(K\right)$  . Then the same argument applies since the Gersten-Quillen and Bloch-Ogus sequences hold for $i$  as well as 2, at least up to $\left(i-1\right)!$  torsion and Gillet's unpublished result.
Thus the Bloch-Kato conjecture for fields in degree $i$  and $i-1$  would show that ( 11 ) is an isomorphism for smooth local rings essentially of finite type over $k$  . The rest of the argument is then identical, and so we would conclude that ${K}_{i}^{M}\left(A\right)/n$  $\to {H}^{i}\left(A,\mathbb{Z}/n\left(i\right)\right)$  is an isomorphism if $A$  is a semi-local ring containing $k$  and $\left(n,\left(i-1\right)!char\left(k\right)\right)=1$  .

2.3 Torsion in K ${}_{2}$

The final application was suggested by C. Weibel. Recall that Suslin in [19used Chern classes to construct an isomorphism ${K}_{2}\left(F\right)\left\{\ell \right\}\to \approx {H}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Div\left({H}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\right)$  where $F$  is either a field of positive characteristic and $\ell$  is a prime distinct from $char\left(F\right),$  or $F$  is a field of finite type over $\mathbb{Q}$  . Note that Suslin is using continuous Galois cohomology since he uses the identification ${H}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Div\left({H}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\right)={H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  where continuous cochains must be used to correctly identify ${H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  as ${H}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\right)\left(2\right)\right)$  modulo its maximal divisible subgroup.
Theorem 5. Let $A$  be a semi-local ring containing a field $k$  and suppose that $A$  is either essentially of finite type over $\mathbb{Q}$  or $char\left(k\right)>0$  . Let $\ell$  be a prime distinct from $char\left(k\right)$  . Then the Chern class ${c}_{2,1}$  induces an isomorphism ${K}_{2}\left(A\right)\left\{\ell \right\}\to \approx {H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Div\left({H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\right)\to \approx {H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  where ${H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\right)\left(2\right)\right)$  is continuous étale cohomology [10.
Corollary 5. Suppose $A$  is a unibranch, e.g. normal, local ring containing a field $k$  with $char\left(k\right)=p>0,$  and let $\ell$  be as above. Then ${K}_{2}\left(A\right)\left\{\ell \right\}\to \approx {H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$
The conclusion of Theorem  5 is independent of the characteristic of $k$  , but the argument required in the two cases is different. We begin by recalling the structure of Suslin's argument in [19and the role of continuous cohomology. He first observes (in the proof of Proposition 3.8 $\right)$  that ${c}_{2,1}$  appears in a commutative, exact diagram
 $\begin{array}{c}\begin{array}{ccccc}{K}_{3}\left(F\right)\to & {K}_{3}\left(F;\mathbb{Z}/{\ell }^{n}\right)& \to {c}_{2,1}& {H}^{1}\left(F,\mathbb{Z}/{\ell }^{n}\left(2\right)\right)& \\ ↓& & & ↓& \\ {H}_{cont}^{1}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\to & {H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }\left(2\right)\right)& \to & {H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\to & {H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}\to 0\end{array}\end{array}$ (12)
where the right hand vertical arrow comes from the coefficient sequence defined by multiplication by ${\ell }^{n}$  on ${\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)$  . Continuous cohomology is used here to make the bottom sequence exact (the first arrow is defined on $\ell$  -adic sheaves by sending $x↦{\ell }^{-n}x$  ) and to identify the image of ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }\left(2\right)\right)$  with the maximal divisible subgroup of ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  and the image of ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  with the $\ell$  -primary torsion in ${H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  . Now by exactness and the universal coefficient theorem ${c}_{2,1}$  factors through ${K}_{2}\left(F\right)\left\{\ell \right\}$  as
 $\begin{array}{ccc}{\overline{c}}_{2,1}:& {K}_{2}\left(F\right)\left\{\ell \right\}\to {H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Image\left({H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }\left(2\right)\right)\right)& \end{array}$
 $\begin{array}{ccc}& \sim ={H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}.& \end{array}$
Suslin has already shown that ${H}_{cont}^{1}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is finite when $char\left(F\right)>0$  [19,Corollary2.8, and so ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }\left(2\right)\right)=0$  and ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\sim ={H}_{cont}^{2}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  in this case. He then shows that ${\overline{c}}_{2,1}:{K}_{2}\left(F\right)\left\{\ell \right\}\to {H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is an isomorphism [19,Theorem3.9by an induction argument. In characteristic zero ${H}_{cont}^{1}\left(F,{\mathbb{Q}}_{\ell }\left(2\right)\right)$  is only zero if ${F}_{0}$  , the algebraic closure of $\mathbb{Q}$  in $F$  , has only real embeddings in $\mathbb{C}$  , but he can reduce the theorem to ${F}_{0}$  [19,Proposition3.3and here the desired result was proven earlier by Tate. Thus extending this result to a semi-local ring containing a field which is essentially of finite type over $\mathbb{Q}$  in characteristic zero requires first defining the factorization of ${c}_{2,1}$  through ${\overline{c}}_{2,1}$  (which will require continuous étale cohomology [10) and then showing that ${\overline{c}}_{2,1}$  is an isomorphism.
${c}_{2,1}$  is defined for arbitrary semi-local rings $A$  (at least if $n>2$  when $\ell =2$  ) and Suslin's factorization argument applies which allows us to define
 $\begin{array}{c}{\overline{c}}_{2,1}\left(A\right):{K}_{2}\left(A\right)\left\{\ell \right\}\to {H}_{cont}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Di{v}^{1}\left(A\right)\end{array}$ (13)
where, for simplicity, $Di{v}^{1}\left(A\right)$  denotes the maximal $\ell -$  divisible subgroup of ${H}_{cont}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  . The exact sequence needed for this factorization requires using continuous étale cohomology which Jannsen [10has developed.
It agrees with continuous group cochain cohomology when $A$  is a field and has all the needed properties to construct the exact sequence in ( 12 ) and so to show that ${\overline{c}}_{2,1}$  is properly defined. Moreover, for a coefficient system of sheaves ${F}_{n}$  , when ${H}^{i-1}\left(A,{F}_{n}\right)$  satisfies the Mittag-Leffler condition, ${H}_{cont}^{i}\left(A,{lim}_{←}{F}_{n}\right)$  is the usual $\ell$  -adic cohomology. Thus ${H}_{cont}^{1}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)={H}^{1}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  and ${H}_{cont}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)={H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  .
In addition, [10,Theorem5.14is the identification ${H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}\sim =Im\left({H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\right).$  The proof of Theorem 3, while different in $char\left(A\right)>0$  and $char\left(A\right)=0$  , starts from the comparison of the Gersten-Quillen exact sequence with a modification of the Bloch-Ogus sequence which is still exact.
Assume $A$  is a semi-local ring which is smooth and essentially of finite type over a field $k$  ( $k=\mathbb{Q}$  in characteristic $0$  ). Consider, as in the proof earlier,
 $\begin{array}{c}\begin{array}{ccc}0\to {K}_{2}\left(A\right)\left\{\ell \right\}\to & {K}_{2}\left(F\right)\left\{\ell \right\}\to & {\prod }_{x\in Spec\left(A{\right)}^{1}}{K}_{1}\left(\kappa \left(x\right)\right)\left\{\ell \right\}\\ ↓{\overline{c}}_{2,1}\left(A\right)& ↓{\overline{c}}_{2,1}\left(F\right)& ↓=\\ 0\to {H}^{1}\left(A,\left(2\right)\right)/Di{v}^{1}\left(A\right)\to & {H}^{1}\left(F,\left(2\right)\right)/Di{v}^{1}\left(F\right)\to & {\prod }_{x\in Spec\left(A{\right)}^{1}}{H}^{0}\left(\kappa \left(x\right),\left(1\right)\right)\end{array}\end{array}$ (14)
where ${H}^{1}\left(A,\left(2\right)\right)$  denotes ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  , etc. for typographical reasons. If $char\left(A\right)$  is positive, then ${H}^{1}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is finite and so ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }\left(2\right)\right)$  which maps onto the maximal divisible subgroup of ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  vanishes. Thus the bottom sequence in ( 14 ) is just the Bloch-Ogus sequence and so ${\overline{c}}_{2,1}\left(A\right)$  is an isomorphism. If $char\left(A\right)=0$  and $A$  is essentially of finite type over $\mathbb{Q}$  , then ${H}^{0}\left(\kappa \left(x\right),{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(1\right)\right)$  has no divisible part for any $x\in Spec\left(A{\right)}^{\left(1\right)}$  . Thus $Di{v}^{1}\left(A\right)=Di{v}^{1}\left(F\right)$  , and so the bottom sequence is an exact quotient of the Bloch-Ogus sequence.
Thus in either case ${\overline{c}}_{2,1}\left(A\right)$  is an isomorphism.
Now suppose $A$  is an arbitrary semi-local ring of positive characteristic.
Since both ${K}_{2}\left(A\right)$  and ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  commute with direct limits of rings we may assume that $A$  is a semi-local ring essentially of finite type over a field. We can then use Theorem  1 since ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is local for the étale topology with respect to closed sets by 1) in Example ( 1 ) and ${K}_{2}\left(A\right)\left\{\ell \right\}$  is locally of finite presentation and satisfies the epic hensel pair condition. This latter condition follows since ${K}_{3}\left(A;\mathbb{Z}/{\ell }^{n}\right)$  is local for closed sets by 2) in Example ( 1 ) and so ${}_{{\ell }^{n}}{K}_{2}\left(A\right),$  being a quotient, satisfies the epic hensel pair condition. Note that the Corollary to Theorem  1 shows that ${K}_{2}\left(A\right)\left\{\ell \right\}$  is then local for the étale topology with respect to closed sets.
The characteristic $0$  case is essentially the same but is complicated by the fact that ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Di{v}^{1}\left(A\right)$  does not behave nicely with respect to limits of rings. However Jannsen has shown that ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)/Di{v}^{1}\left(A\right)\mathbb{\approxeq }{H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}.$  In addition he shows that the sequence, $0\to {{lim}_{←}}^{1}{H}^{2}\left(B,\mathbb{Z}/{\ell }^{n}\left(2\right)\right)\to {H}_{cont}^{2}\left(B,{\mathbb{Z}}_{\ell }\left(2\right)\right)\to {lim}_{←}{H}^{2}\left(B,\mathbb{Z}/{\ell }^{n}\left(2\right)\right)\to 0,$  is exact where ${{lim}_{←}}^{1}$  refers to the first left derived functor of the system $\left\{{H}^{2}\left(B,\mathbb{Z}/{\ell }^{n}\left(2\right)\right)\right\}.$  Consequently ${H}_{cont}^{2}\left({B}^{h},{\mathbb{Z}}_{\ell }\left(2\right)\right)\mathbb{\approxeq }{H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  if ${B}^{h}$  is the henselization with respect to $I$  of a presentation $A\mathbb{\approxeq }B/I$  as a quotient of a semi-local ring $B$  smooth over $k.$  This and the corresponding isomorphism ${H}^{1}\left({B}^{h},{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\mathbb{\approxeq }{H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)$  then allow us to conclude that ${H}_{cont}^{2}\left({B}^{h},{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}\mathbb{\approxeq }{H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}.$  Now we may apply Theorem to show ${\overline{c}}_{2,1}:{K}_{2}\left(A\right)\left\{\ell \right\}\to {H}_{cont}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  is an isomorphism.
Corollary 4 is obtained by noting that Suslin [19,Corollary2.8also proved that ${H}^{1}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is finite if $F$  is a field of positive characteristic.
But if $A$  is a unibranch local ring, e.g. a normal local ring, then ${H}^{1}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)↪{H}^{1}\left(F,{\mathbb{Z}}_{\ell }\left(2\right)\right)$  is also finite. Consequently ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }\left(2\right)\right)=0$  , and so we have an isomorphism ${H}^{1}\left(A,{\mathbb{Q}}_{\ell }/{\mathbb{Z}}_{\ell }\left(2\right)\right)\sim ={H}^{2}\left(A,{\mathbb{Z}}_{\ell }\left(2\right)\right)\left\{\ell \right\}$  .

A

In this appendix ${\zeta }_{n}$  will always stand for a primitive ${n}^{th}$  root of unity. We wish to prove the following theorem. (Note that this is part of [11,Theorem1and is proved without assuming the generalized Kato conjecture.)
Theorem 6. Let $A$  be a connected, regular semi-local ring with quotient field $K,$  and $n$  an odd positive integer relatively prime to the residue characteristics of $A.$  Then ${H}^{2}\left(A,\mathbb{Z}/n\left(2\right)\right)\to j{H}^{2}\left(K,\mathbb{Z}/n\left(2\right)\right)$  is injective.
We begin with a series of lemmas to calculate the cohomology of the extension obtained by adjoining an ${n}^{th}$  root of $1$  . Fix the following situation in order to describe $\mathbb{Z}/n\left(k\right)$  and the action of the cyclotomic character on this module. Fix an integer $n$  such that $n=dm$  and $m=de$  for some positive integers $d,e,$  and let $A$  be a connected ring with ${\zeta }_{m}\in A$  and $d$  a unit in $A.$  Assume $A\left[{\zeta }_{n}\right]$  is connected. Then ${X}^{d}-{\zeta }_{m}{=}^{d}{\prod }_{i=1}\left(X-{\zeta }_{n}{\left({\zeta }_{m}^{e}\right)}^{i}\right)$  in $A\left[{\zeta }_{n}\right],$  and $A\left[{\zeta }_{n}\right]$  is a Galois extension of $A$  with cyclic Galois group ${G}_{d}:=\mathbb{Z}/d=〈\sigma 〉\subset {\left(\mathbb{Z}/n\right)}^{×}.$  Construct a dictionary for the action of ${G}_{d}$  on $\mathbb{Z}/n\left(k\right)$  by $\sigma$  sends ${\zeta }_{n}↦{\zeta }_{n}\left({\zeta }_{m}^{e}\right).$  Since $n|{m}^{2},$  we have
 $\begin{array}{cc}{G}_{d}\text{on}\mathbb{Z}/n\left(k\right)& \mathbb{Z}/d\text{on}\mathbb{Z}/n\end{array}$
 $\begin{array}{cc}\left({\zeta }_{n},{\zeta }_{m}={\zeta }_{n}^{d},{\zeta }_{d}={\zeta }_{m}^{e}\right)\in \mathbb{Z}/n\left(1{\right)}^{×3}& \left(1,d,m=de\right)\in \mathbb{Z}/{n}^{×3}\end{array}$
 $\begin{array}{cc}\sigma \left({\zeta }_{n}\right)={\zeta }_{n}\left({\zeta }_{m}^{e}\right)={\zeta }_{n}^{1+m}\in \mathbb{Z}/n\left(1\right)& \sigma \left(1\right)=1+m\in \mathbb{Z}/n\end{array}$
 $\begin{array}{cc}{\sigma }^{k}\left({\zeta }_{n}\right)={\left({\zeta }_{n}^{1+m}\right)}^{k}={\zeta }_{n}^{1+km}\in \mathbb{Z}/n\left(k\right)& {\sigma }^{k}\left(1\right)={\left(1+m\right)}^{k}=1+km\in \mathbb{Z}/n\end{array}$
 $\begin{array}{}\end{array}$
This dictionary makes the computation of the cohomology of ${G}_{d}$  on $\mathbb{Z}/n\left(k\right)$  straightforward.
Lemma 2. ${H}^{0}\left({G}_{d},\mathbb{Z}/n\left(k\right)\right)=\mathbb{Z}/\left(m\left(d,k\right)\right),$  and ${H}^{2r}\left({G}_{d},\mathbb{Z}/n\left(k\right)\right)=\left\{\begin{array}{ccc}\mathbb{Z}/\left(d,k\right)& \text{if}& d\text{is odd or}k\text{is even or}4|m\\ \mathbb{Z}/2\left(d,k\right)& \text{if}& d\text{is even and}k\text{is odd and}4\mathbb{\nmid }m\end{array}$  and ${H}^{2r+1}\left({G}_{d},\mathbb{Z}/n\left(k\right)\right)=\left\{\begin{array}{ccc}\mathbb{Z}/\left(d,k\right)& \text{if}& d\text{is odd or}k\text{is even or}4|m\\ \mathbb{Z}/2\left(d,k\right)& \text{if}& d\text{is even and}k\text{is odd and}4\mathbb{\nmid }m\end{array}$
• Proof. The standard resolution used to calculate the cohomology of a cyclic group ${G}_{d}$  on $\mathbb{Z}/n\left(k\right)$  is used to carry out this calculation. Our dictionary allows us to calculate ${N}_{k}={\sum }_{i=0}^{d-1}{\sigma }^{ik}:\mathbb{Z}/n\to \mathbb{Z}/n$  and ${T}_{k}=1-{\sigma }^{k}:\mathbb{Z}/n\to \mathbb{Z}/n$  as  $\begin{array}{ccc}{T}_{k}\left(j\right)& =& j-j\left(1+km\right)=-kmj\end{array}$
 $\begin{array}{ccc}{N}_{k}\left(j\right)& =& j\left({\sum }_{i=0}^{d-1}\left(1+ikm\right)\right)\end{array}$
 $\begin{array}{ccc}& =& j\left(d+km\frac{d\left(d-1\right)}{2}\right)\end{array}$
 $\begin{array}{ccc}& =& \left\{\begin{array}{ccc}jd& \text{if}& d\text{is odd or}k\text{is even}\\ j\left(d-m\frac{d}{2}\right)& \text{if}& d\text{is even and}k\text{is odd}\end{array}\text{}\end{array}$
In the case when $d$  is even and $k$  is odd, we note that $\left(m,1-\frac{m}{2}\right)=\left\{\begin{array}{cc}2& \text{if}4\mathbb{\nmid }m\\ 1& \text{otherwise}\end{array}.$  Then the sequence that calculates the cohomology groups is $\cdots \mathbb{Z}/n\to {N}_{k}\mathbb{Z}/n\to {T}_{k}\mathbb{Z}/n\to {N}_{k}\mathbb{Z}/n\to {T}_{k}\mathbb{Z}/n\cdots ,$  and we find  $\begin{array}{ccc}\text{Ker}{T}_{k}& =& \frac{d}{\left(d,k\right)}\mathbb{Z}/n\sim =\mathbb{Z}/\left(m\left(d,k\right)\right)\end{array}$
 $\begin{array}{ccc}\text{Im}{T}_{k}& =& km\mathbb{Z}/n=\left(d,k\right)\cdot m\mathbb{Z}/n\end{array}$
 $\begin{array}{ccc}\text{Ker}{N}_{k}& =& \left\{\begin{array}{ccc}m\mathbb{Z}/n& \text{if}& d\text{is odd or}k\text{is even or}4|m\\ \frac{m}{2}\mathbb{Z}/n& \text{if}& d\text{is even,}k\text{is odd, and}4\mathbb{\nmid }m\end{array}\text{}\end{array}$
 $\begin{array}{ccc}\text{Im}{N}_{k}& =& \left\{\begin{array}{ccc}d\mathbb{Z}/n& \text{if}& d\text{is odd or}k\text{is even or}4|m\\ 2d\mathbb{Z}/n& \text{if}& d\text{is even,}k\text{is odd, and}4\mathbb{\nmid }m\end{array}\text{}\end{array}$
The rest is straightforward.
Corollary 6. Let $A$  be a regular, connected semi-local domain, and let $n$  be an odd positive integer relatively prime to the residue characteristics of $A$  . Suppose that $m$  is the largest divisor of $n$  such that ${\zeta }_{m}\in A$  . Let ${A}_{n}=A\left[{\zeta }_{n}\right],$  and let $G=$  Gal $\left({A}_{n}/A\right).$  Then ${H}^{p}\left(G,\mathbb{Z}/n\left(2\right)\right)=\left\{\begin{array}{ccc}\mathbb{Z}/m\left(2\right)& \text{if}& p=0\\ 0& \text{if}& p>0\end{array}$
• Proof. If $m=n,$  we are done. Suppose that $n=\ell {n}^{\prime }$  where $\ell$  is the smallest prime divisor of $\frac{n}{m}.$  Then there is a cyclic subgroup $H⊳G$  such that ${A}_{{n}^{\prime }}:={A}^{H}=A\left[{\zeta }_{{n}^{\prime }}\right],$  and ${A}_{{n}^{\prime }}$  is a Galois extension of $A$  with abelian Galois group $G/H.$  If ${\zeta }_{\ell }\in {A}_{{n}^{\prime }},$  then ${A}_{n}$  and ${A}_{{n}^{\prime }}$  satisfy the hypothesis of the lemma.
Hence ${H}^{p}\left(H,\mathbb{Z}/n\left(2\right)\right)=\left\{\begin{array}{ccc}\mathbb{Z}/{n}^{\prime }\left(2\right)& \text{if}& p=0\\ 0& \text{if}& p>0\end{array}.$  If ${\zeta }_{\ell }\ne {A}_{{n}^{\prime },}$  then $\left[{A}_{n}:{A}_{{n}^{\prime }}\right]=\ell -1$  is relatively prime to ${n}^{\prime }.$  Hence ${H}^{p}\left(H,\mathbb{Z}/n\left(2\right)=\left\{\begin{array}{ccc}\mathbb{Z}/{n}^{\prime }\left(2\right)& \text{if}& p=0\\ 0& \text{if}& p>0\end{array}.$  Now by using the Hochschild-Serre spectral sequence, ${E}_{2}^{p,q}={H}^{p}\left(G/H,{H}^{q}\left(H,\mathbb{Z}/n\left(2\right)\right)\right)⇒{H}^{p+q}\left(G,\mathbb{Z}/n\left(2\right)\right),$  and the above calculations, we conclude that ${H}^{p}\left(G/H,{H}^{0}\left(H,\mathbb{Z}/n\left(2\right)\right)\right)\to \mathbb{\approx }{H}^{p}\left(G,\mathbb{Z}/n\left(2\right)\right).$  Since ${H}^{0}\left(H,\mathbb{Z}/n\left(2\right)\right)=\mathbb{Z}/{n}^{\prime }\left(2\right),$  we may apply an induction hypothesis and thus prove the corollary.
Lemma 3. Let $A$  be a regular, connected, semi-local domain with quotient field $K,$  $n$  a positive integer relatively prime to the residue characteristics of $A$  . Let $B$  be a Galois extension of $A$  with quotient field $L$  , and let $G=Gal\left(B/A\right).$  Suppose ${\zeta }_{n}\in B.$  Then ${H}^{1}\left(G,{H}^{1}\left(B,\mathbb{Z}/n\left(2\right)\right)\right)\to {H}^{1}\left(G,{H}^{1}\left(L,\mathbb{Z}/n\left(2\right)\right)\right)$  is injective.
• Proof. Since $A$  is a UFD, we have an exact sequence $0\to {A}^{*}\to {K}^{*}\to \prod or{d}_{x}{\prod }_{x\in {A}^{\left(1\right)}}{i}_{x*}\mathbb{Z}\to 0$  where ${A}^{\left(1\right)}=\left\{x\in Spec\left(A\right)/ht\left(x\right)=1\right\}$  and $or{d}_{x}:{K}^{*}\to {i}_{x*}\mathbb{Z}$  calculates the order of $f\in {K}^{*}$  at the discrete valuation associated to $x,$  and ${i}_{x}:Spec\left(k\left(x\right)\right)\to Spec\left(A\right)$  is the inclusion of the residue field of ${A}_{x}.$  If we map this sequence to itself by multiplication by $n$  , the cokernels form the short exact sequence $0\to {A}^{*}/{A}^{*n}\to {K}^{*}/{K}^{*n}\to \prod or{d}_{x}{\prod }_{x\in {A}^{\left(1\right)}}{i}_{x*}\mathbb{Z}/n\to 0.$  There is a sequence of $G$  -modules obtained by twisting the sequence for $B$  with the cyclotomic character $0\to {B}^{*}/{B}^{*n}\left(1\right)\to {L}^{*}/{L}^{*n}\left(1\right)\to {\prod }_{y\in {B}^{\left(1\right)}}{i}_{y*}\mathbb{Z}/n\left(1\right)\to 0,$  relating $B{/}^{*}{B}^{*n}\left(1\right)={H}^{1}\left(B,\mathbb{Z}/n\left(2\right)\right)$  to ${L}^{*}/{L}^{*n}\left(1\right)={H}^{1}\left(L,\mathbb{Z}/n\left(2\right)\right)$  since ${\zeta }_{n}\in {L}^{*}.$  Thus we need only show that ${H}^{0}\left(G,{L}^{*}/{L}^{n*}\left(1\right)\right)\to {H}^{0}\left(G,{\prod }_{y\in {B}^{\left(1\right)}}{i}_{y*}\mathbb{Z}/n\left(1\right)\right)$  is onto to complete the proof of the lemma. But ${H}^{0}\left(G,{\prod }_{y\in {B}^{\left(1\right)}}{i}_{y*}\mathbb{Z}/n\left(1\right)\right)\mathbb{\approxeq }{\prod }_{x\in {A}^{\left(1\right)}}{H}^{0}\left({D}_{y|x},{i}_{y*}\mathbb{Z}/n\left(1\right)\right)\mathbb{\approxeq }{\prod }_{x\in {A}^{\left(1\right)}}{i}_{x*}\mathbb{Z}/{m}_{x}\left(1\right)$  where ${D}_{y|x}$  is the decomposition group of a choice of $y\in {B}^{\left(1\right)}$  lying over $x\in {A}^{\left(1\right)}$  and $\mathbb{Z}/{m}_{x}\left(1\right)$  is the group of roots of unity of order $n$  in $k\left(y\right)$  fixed by ${D}_{y|x}$  (expressed in the above notation).
Any element $\alpha \in {\prod }_{x\in {A}^{\left(1\right)}}{i}_{x*}\mathbb{Z}/{m}_{x}\left(1\right)$  can be written as $\alpha ={\sum }_{x\in {A}^{\left(1\right)}}{\delta }_{x}\cdot {\zeta }_{n}^{{r}_{x}}$  where the sum is finite and ${\delta }_{x}=0$  for ${x}^{\prime }\in {A}^{\left(1\right)}$  if ${x}^{\prime }\ne x$  and ${\delta }_{x}=1\in {i}_{x*}\mathbb{Z}/n$  at $x\in {A}^{\left(1\right)}.$  Then there is ${f}_{x}\in {K}^{*}/{K}^{*n}$  with $or{d}_{x}\left({f}_{x}\right)={\delta }_{x}$  and $or{d}_{{x}^{\prime }}\left({f}_{x}\right)=0$  if ${x}^{\prime }\ne x.$  Hence ${\sum }_{x\in {A}^{\left(1\right)}}\left({\sum }_{\sigma \in G/{D}_{y|x}}{f}_{x}\sigma \left({\zeta }_{n}^{{r}_{x}}\right)\right)$  has image $\alpha .$
Remark 1. The proof amounts to first reducing to the case of a dvr $A$  by using the UFD sequence and then solving the problem at the completion where ${D}_{y|x}$  becomes the Galois group and finally spreading that solution around using the transitivity of the action of $G$  on $\left\{y\in {B}^{\left(1\right)}/y\text{lies over a fixed}x\in {A}^{\left(1\right)}\right\}.$
• Proof. If ${\zeta }_{n}\in A,$  the theorem reduces to the well known result $Br\left(A\right)\subset Br\left(K\right)$  since ${H}^{2}\left(A,\mathbb{Z}/n\left(k\right)\right)\mathbb{\approxeq }{H}^{2}\left(A,\mathbb{Z}/n\left(1\right)\right){\mathbb{\approxeq }}_{n}Br\left(A\right)$  and similarly for $K.$  We reduce the theorem to this case by analyzing the cohomology of a cyclic Galois covering $A\left[{\zeta }_{n}\right]/A$  with Galois group $G.$  There is a spectral sequence for this covering ${E}_{2}^{p,q}={H}^{p}\left(G,{H}^{q}\left(A\left[{\zeta }_{n}\right],\mathbb{Z}/n\left(2\right)\right)\right)⇒{H}^{n}\left(A,\mathbb{Z}/n\left(2\right)\right).$  Suppose $x\in Ker\left[{H}^{2}\left(A,\mathbb{Z}/n\left(2\right)\right)\to {H}^{2}\left(K,\mathbb{Z}/n\left(2\right)\right)\right].$  Then $x=0$  in ${H}^{2}{\left(A\left[{\zeta }_{n}\right],\mathbb{Z}/n\left(2\right)\right)}^{G}$  by the above remarks. But Lemma  3 shows that $x=0$  also in ${E}_{\infty }^{1,1}\subseteq {H}^{1}\left(G,{H}^{1}\left(A\left[{\zeta }_{n}\right],\mathbb{Z}/n\left(2\right)\right)\right).$  Finally Corollary  6 shows that $x\in {E}_{\infty }^{2,0}\subseteq {H}^{2}\left(G,\mathbb{Z}/m\left(2\right)\right)$  where ${\zeta }_{m}\in A.$  But ${H}^{2}\left(G,\mathbb{Z}/m\left(2\right)\right)$  is the same for the Galois covering $A\left[{\zeta }_{n}\right]/A$  and $K\left[{\zeta }_{n}\right]/K.$  Hence $x=0$  and the theorem follows.
References

1. S. Bloch and A. Ogus “Gersten's conjecture and the homology of schemes”, Ann Sci Ecole Norm Sup., 4th sér. 7(1974). 181-202.
2. J-L Colliot-Thélène and M. Ojanguren “Espaces Principaux homogènes localement triviaux”, Publications Mathematiques de l'IHES, 75(1992), 97-122.
3. J.-L. Colliot-Thélène, R.T. Hoobler, and B. Kahn, “The Bloch—Ogus—Gabber Theorem”, Algebraic K-Theory, Fields Inst. Commun., Vol. 16, Amer. Math. Soc., Providence, RI, 1997, 31–94.
4. O. Gabber, “K-theory of henselian local rings and henselian pairs”, a letter from Ofer Gabber to M. Karoubi, 83, AMS Contemp Math, 1989.
5. O. Gabber, “Affine analog of the proper base change theorem”, Israel J. of Math., 87(1994),325-335.
6. H. Gillet, “Gersten's conjecture for the K-theory with torsion coefficients of a discrete valuation ring”, J Algebra, 103(1986), 377-380.
7. H. Gillet,.and M. Levine, “The relative form of Gersten's conjecture over a discrete valuation ring:The smooth case”, J Pure Appl Algebra, 46(1987), 59-71.
8. H. Gillet, personal communication, July 1995.
9. D. Grayson, “Universal exactness in algebraic K-theory”, J Pure Appl Algebra, 36(1985), 139-141.
10. U. Jannsen, “Continuous étale cohomology”, Math Ann, 280(1988), 207-245.
11. B. Kahn, “Deux théorèmes de comparision en cohomologie étale; applications”, Duke Math J, 69(1993), 137-165.
12. J. Milne, Étale Cohomology, Princeton University Press, 1980, Princeton, NJ.
13. C. Pedrini, and C. Weibel, “Invariants of real curves”, Rend Sem Mat Univers Politecn Torino, 49(1991), 139-173.
14. D. Quillen, “Higher Algebraic K-theory I”, Algebraic K-theory I: Higher K-theories, Lecture Notes in Math, 341, Springer-Verlag, Berlin, 1973, 85-147.
15. Michel Raynaud, Anneaux Locaux Hensélien, Lecture Notes in Math, 169, Springer-Verlag, Berlin, 1970.
16. SGA4:Theorie des Topos et Cohomologie Étale des Schemas, Lecture Notes in Math, 305, Springer-Verlag, Berlin, 1973.
17. C. Soulé, “K-théorie des anneaux d'entiers de corps de nombres et cohomologie étale”, Inventiones math, 55(1979), 251-295.
18. R. Strano,“On the étale cohomology of hensel rings”, Comm in Alg, 12, 2195-2211.
19. A. Suslin, “Torsion in ${K}_{2}$  of fields”, K-theory, 1(1987), 5-29.
20. J. Tate, “Relations between ${K}_{2}$  and Galois cohomology”, Inv Math 36(1976), 257-274.
21. C. Weibel, “Étale Chern classes at the prime 2”, Proceedings of the Lake Louise Conference on Algebraic K-Theory, NATO ASI Series C, 407, Kluwer Academic Publishers, Dordrecht 1993, 249-286.

City College of New York, New York, New York 10031