There is no way to turn off parsing of numbers in the StreamTokenizer(). By
default, it is on, and the parseNumbers() method just turns it on again.
parseNumbers() should take a boolean flag argument which indicates
whether numbers should be parsed or not.

=======================================================
Another report:

the public method in java.io.StreamTokenizer:
public void parseNumbers();

should probably be: public void setParseNumbers(boolean)

Currently the parseNumber method which is defined
as public is called from the private
StreamTokenizer constructor, so it is actually
not possible to create a TokenStream that does
not parse numbers.
If this feature was desired then parseNumbers
should be a private method, currently its useless
and causes confusion.

brian.klock@eng 1997-11-05
======================================================

ronan.mandel@Eng 1997-12-09
Here is some source demonstrating the problem. If a leading space is inserted
into the StringReader, then it works as expected. Otherwise, 1 is parsed as a number and the remainder of the string is parsed a whole token.
public class StreamTokBug {
    static public void main(String[] argv) {
        try {
//StringBufferInputStream source = new StringBufferInputStream("1 2 3 4");
Reader r = new StringReader("1 2 3 4");
            StreamTokenizer ST = new StreamTokenizer(r);
            ST.ordinaryChar(' ');
            ST.wordChars(' ', ' ');
ST.parseNumbers();
            while (ST.nextToken() != ST.TT_EOF) {
if(ST.ttype == ST.TT_NUMBER){
System.out.println("It's a number!");
}
              System.out.println (ST);
            }
        }
        catch (IOException x) { x.printStackTrace(); }
    }
}