Actually many grant agencies do exactly this and count the references in the page limit - it of course makes life difficult for reviewers, but is a fact of life for those of us applying to big agencies such as BBSRC and ERC. A method for compacting the references as suggestted earlier in this thread would be really useful.