Strange, I would have thought the jar would be downloaded to the file system somewhere. Another alternative is to type it into the URL. It would be relative to the page you're on, so in the place of pagename.html, put in client.jar, leaving the rest of the URL the same. I think that would do it.
It might be nicer if you requested the jar from the site's author, or checked if he/she made it available already for download. Who knows? You might even get source code that way.