I think this is a duty of the author to refer only on those results which are available. (Perhaps not published, but posted somewhere on the Internet). I think it is a duty of the editor and referee to make sure that the authors follow this rule. A result which uses a reference which is not available cannot be considered as proved.

In the case when there is an unpublished reference, which I need, I search on the internet, Google scholar, author's web page, etc. If not found I write to the author and ask her to send me a copy. In pre-computer era, I would write a letter to the author in the case I am really interested in the result.