> I don't think there is a problem, as far as I can see the 
> input and output have the same bytes for character data, 
> except that the input &#C; are output in text mode as literal 
> (xml 1.1) control-L characters.

Well, the output method is method="text", so we would not expect the #xC
characters to be escaped in the output.

So (for the originator): what exactly is the problem? 

Michael Kay
http://www.saxonica.com/