How to handle results from an unpublished paper? I am writing a paper right now, and part of the paper makes use of a (trivial) generalization of a number of really nice theorems and constructions from a paper that was never made public.  The author has left pure mathematics and has no intention of publishing the paper, but I received a copy directly from him several years ago. 
The results and constructions are crucial to my paper, and since I am working with a minor generalization, I think I do need to include the proofs, especially since they aren't available.  I don't want it to look like I'm taking credit for the results or plagiarizing, but some of the proofs are basically copies.  I have a bunch of disclaimers at the top of the section and continually remind the reader that all of these results are in the original paper. 
Do I need to not only give credit at the top of the section, but also give credit for every observation, statement, lemma, and diagram? 
PS It seems like the community wiki checkbox is gone.  I'd appreciate if a moderator could do that for me.
Edit: I contacted the author, and he agreed to put it on the arXiv himself. Going to leave the question up as a community wiki for anyone else who runs into this problem.
 A: First of all, if you received a paper from the author privately, and it is unpublished, you need his/her permission to use the result, and permission to mention his/her work.
When you ask the permission, you may also ask whether the author is willing to post the paper on the arXiv, and you may propose him/her to do this yourself.
If permission is granted but the rest does not work, you may include the paper in your reference list, marking it as "private communication". But this is the measure of the last resort (if the author cannot be contacted and/or gives you no permission to publish his paper).
Remark. I have some results which I consider important, and which I do not publish. The situation is exactly the same: I build on an unpublished result.
The author gave me his manuscript but
refuses to publish it, and does not give me permission to publish his result.
Myself, I stick to the following rule: everything I use in a published paper must be available (in principle) to every reader.
A: Why not you approach the author and ask to join forces to write as a co-author if the other person is willing to do so? That could be a fine resolution and fair to both parties.
