While writing the research papers one quite often needs to get back to the full texts of old (pre-Internet or at least pre-arXiv) references. Of course, having access to a good library and/or the interlibrary loan usually solves the problem but can be somewhat time- and cost-consuming.

It is not that well known, however, that there is a fair chance to find the old paper or preprint you need online for free. Of course, the first thing to try is Google or perhaps another search engine of your choosing. However, if this does not work, you still have a fighting chance, at least as far physics and mathematics are concerned. The places to try are:

the KISS preprint server (you can also try the umbrella interface at SPIRES) allows you to search in (and get to the full text of) a huge database of scanned preprints going back to the 1970s at least. The database covers mostly high-energy physics and related areas, including a fair share of mathematical physics and mathematics. For instance, you can find there a number of preprints by Richard Feynman, including the unpublished ones.

All items but KISS are purelymathematical databases (to be precise, MathNet.Ru includes several physics, mechanics and mathematical physics journals as well).

If you know of other similar databases (be it in physics, mathematics, life sciences,…), please feel free to drop a comment with the relevant link(s).

I can’t think of any other databases than what you pasted but there is one more thing that has helped me in the past. Which is to go to the homepage of the author(s) of the paper and see whether they have posted a copy of the paper there (often scanned if the paper is old). In my experience, the chances of success are higher if the author(s) happens to be a well-known/important figure in the field :)

Your suggestion about going for the homepages is certainly correct but Google (and Google Scholar) is quite good in finding the papers available from such pages, so this is often unnecessary.

I can’t think of any other databases than what you pasted but there is one more thing that has helped me in the past. Which is to go to the homepage of the author(s) of the paper and see whether they have posted a copy of the paper there (often scanned if the paper is old). In my experience, the chances of success are higher if the author(s) happens to be a well-known/important figure in the field :)

Your suggestion about going for the homepages is certainly correct but Google (and Google Scholar) is quite good in finding the papers available from such pages, so this is often unnecessary.

I have some other relevant links on my math journal resource page.

Sorry, I forgot the link: http://homotopical.wordpress.com/links/journals/