[xml] language-lawyer question: namespace precedence rules for attributes?

[Craeg K Strong <cstrong arielpartners com>]

Hello:

Using libslt 1.0.27 and the following transformer:

<?xml version="1.0"?>
<xsl:stylesheet   xmlns:xsl="http://www.w3.org/1999/XSL/Transform";  
version="1.0">
 <xsl:output method = "xml" indent="no" omit-xml-declaration="yes"/>
 <xsl:template match="/">
   <result>
     <xsl:attribute name="foo" 
namespace="http://some-ns/";>bar</xsl:attribute>
     <xsl:attribute name="x:foo" 
xmlns:x="http://some-ns/";>baz</xsl:attribute>
     <xsl:attribute name="y:bar" 
xmlns:y="http://some-ns/";>baz</xsl:attribute>
     <xsl:attribute name="z:yow" 
xmlns:z="http://some-ns/";>baz</xsl:attribute>
     <xsl:attribute name="a:mum" 
xmlns:a="http://some-ns/";>baz</xsl:attribute>
     <xsl:attribute name="b:zip" 
xmlns:b="http://some-ns/";>baz</xsl:attribute>
     <xsl:attribute name="c:zap" 
xmlns:c="http://some-ns/";>baz</xsl:attribute>
  </result>
 </xsl:template>
</xsl:stylesheet>

on a simple xml file   <dummy/>

I get the following result:

<result xmlns:ns1="http://some-ns/"; ns1:foo="baz" ns1:bar="baz" 
ns1:yow="baz" ns1:mum="baz" ns1:zip="baz" ns1:zap="baz"/>

In other words
1- the *first* prefix for a duplicated namespace wins
2- the *last* value for a duplicated attribute wins

This seems fairly reasonable, but I can't find any basis in the W3C spec 
for heuristic #1.    
Can anyone offer me any pointers as to the letter of the W3C law on this 
point?

Thanks in advance,

--Craeg