Meanwhile i solved the inconsistency. Further investigation yields for the xpath expression //table[1]/tr/td[2]/a/div/h3[\@class='sect2'].
The reason of the error was that i relied to heavily on the output of 'XPath Helper' (a chrome extension).
Is there any linux based browser which could help to extract a xpath of an website element correctly?
Comment on Re: can't extract node with HTML::TreeBuilder::XPath
I find using CSS selectors or relative XPath expressions to be more wieldly. Anchoring the expression on a "nearby" element with an id attribute is usually sufficient to get a short XPath (or CSS) expression.