HTML::TokeParser (and HTML::TokeParser::Simple, too, as its subclass) fetches all tags in the document one-by-one. So, as the <TITLE> tag is always before any <H*> tags, the creation of new parser object makes little sense.