Dom parsing with a attribute name with a number sign

Hey

I am trying to parse some xml from a ArcIms service, and in the result
xml, one of the attribute names contain a number sign, like #. When I
trying to parse this xml I get the folowing exception:
[Fatal Error] :1:49: Element type "FIELDS" must be followed by either
attribute specifications, ">" or "/>".
Exception in thread "main" org.xml.sax.SAXParseException: Element type
"FIELDS" must be followed by either attribute specifications, ">" or
"/>".

I have made this small demo program to illustratet my problem:

<java program>
import java.io.ByteArrayInputStream;

import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import org.w3c.dom.Document;

import org.xml.sax.InputSource;

public class Demo{

    public static void main(String[] args) throws Exception {

        String xmlString = "<?xml version='1.0' encoding='UTF-8' ?>"+
                "<FIELDS #SHAPE#='[Geometry]' />";

        byte[] byteArr =xmlString.getBytes();
        ByteArrayInputStream bais = new ByteArrayInputStream(byteArr);

        DocumentBuilderFactory factory =
DocumentBuilderFactory.newInstance();
        DocumentBuilder builder = factory.newDocumentBuilder();

        InputSource inputSource = new InputSource(bais);
        Document document = builder.parse(inputSource);
    }
}
</java>

What can I do to solve this problem.