Publishers can facilitate Open Access in two main ways. The publisher may, of course, publish the work with free, online access in an Open Access journal or as an Open Access monograph. Alternatively, if the publisher's business model is to sell monographs or subscriptions to journals, then the publisher can still facilitate Open Access by  permitting the author to self-archive the work in an institutional or subject repository.

Developing a self-archiving permissions policy

Not many (but some) publishers permit their own published version of an article to be self-archived (often referred to as the 'published PDF' because publishers normally provide copies of published articles to authors in the PDF format rather than HTML), but many do allow the author to self-archive their own final version of the manuscript after the corrections and revisions required from the peer review proces have been incorporated. Currently, around 60% of publishers and 95% of journals registered in SHERPA permit self-archiving in some form.

There are two forms that articles can take - a preprint form and a postprint form.

A preprint is a version of the article before it has been peer-reviewed for publication. The term may refer either to articles at an early stage of preparation or to articles at the last stage before submission for peer review. In the latter case the article is, of course, well-developed and ready for critical review and, if only minor revisions are needed as a result of peer review, a late-stage preprint may be very little different from the final article.

A postprint is the final version of the article that the author sees before it is published. A postprint has been peer-reviewed and the changes and revisions required by the reviewers have been incorporated. it is identical to the published version except that the publisher will lay out the article in the house style, assign page numbers (if the journal is still published in print), add the publisher's logo and copyright details and so forth. In content, however, the author's final postprint and the published version are effectively the same.

Publishers choose which of these forms they will allow to be self-archived. Most journals allow the postprint to be self-archived but a minority only permit authors to make a preprint Open Access. There are pros and cons from a publisher's point of view regarding postprint self-archiving. The value of having the postprint available to those who would not otherwise see the article is that the readership is increased, citations can be won and the journal reaps the benefit of the extra impact. Some publishers consider, however, that the free availability of the postprint may undermine sales of the journal. In fact, there is no evidence at all to support this notion: indeed the only evidence available supports the opposite conclusion, which is that sales are not harmed at all by the free availability of articles in repositories.

The evidence on this comes from two learned societies in physics, a discipline where self-archiving has been the norm since 1991 through the use of the subject repository, arXiv. arXiv contains over half a million articles, principally in nuclear physics, condensed matter and astronomy, and with some content in mathematics, quantitative biology and computer science. Over half of these are postprints. The experiences of these two societies, the American Physical Society and the Institute of Physics Publishing (UK), of doing business alongside the arXiv, are reported here.

Publisher policies on self-archiving are collated and presented by a number of services. The two largest are the SHERPA RoMEO and EPrints RoMEO services, both covering international journals: there are also some services providing information on national or regional journals. See the publisher permissions services page for more information on this topic.

Accommodating funder Open Access requirements

As Open Access policies from large research funders have appeared, publishers have been moderating their policies on self-archiving to accommodate funder requirements. The two funder policies that have had most attention in this sense so far are those of the National Institutes of Health (NIH) in the US and the Wellcome Trust. Both are large funders of biomedical research. The Wellcome Trust provides about 600m USD of funding per year; the NIH is the world's largest research funder, currently spending around 27 billion USD a year.

The policies differ slightly, but in essence both require their grant holders to make their work Open Access by making them available in PubMed Central or a related biomedical Open Access repository (e.g. UK PubMed Central), within a short period of time after publication. These policies have implications for publishers of biomedical journals and many are developing procedures to ensure that authors comply with these requirements. See more on what publishers are doing to accommodate funder requirements here.