The trick is that whenever you launch lynx, it actually starts a web-browser, waiting for user input.. wich is generally a bad idea 😉 Lynx has a very nice -dump option, e.g.:
lynx -dump http://www.phpbuilder.com
This dumps the rendered output of www.phpbuilder.com to stdout. You should use this option, and if your script outputs some sort of result, error messages etc, you should cat the rendered page on to a log file. Otherwise, if the data outputted is void, send it to 'the black hole'! E.g.:
lynx -dump http://your.website.com >> /some/dir/cronlog
or;
lynx -dump http://your.website.com > /dev/null