• The document to be transformed is XML
• The XSLT document is XML
• The internal form of both is defined by the parser
• Essentially each Unicode character is a single internal character

R&D
R&D
R&D
<![CDATA[R&D]]>

will all have the same internal format for &, a single character and only the parser knows what it is

The serialiser will output the character in the form it chooses

<xsl:text>&#960; &#163; &#215; &lt;</xsl:text>

Is output as &#960; &pound; &times; < by Saxon and π £ × < by MSXSL