The content is expected to be a well-formed XML document (a document node with a single child element) in its character string representation. Attempts to load a document that is not well-formed results in an error.
When loading the document as a VARCHAR or CLOB, the driver encodes the document in the session character set. The constructors of XML type values from VARCHAR or CLOB transcodes this to Unicode (UTF-8) and ignores the encoding specified in the XML declaration.