Scientific journal publishing will at first sight seem bizarre to those accustomed to other forms of publishing: scientists critique their peers' work for free, often resulting in dramatic improvements to their papers. Journals do not pay their authors, but instead often charge publication fees to defer production costs. Many charge for subscriptions to keep submission fees affordable. How is this apparently topsy-turvy world possible? The reason is, of course, that peer-reviewed papers represent the validated measure of a researcher's output, which directly affects funding and career progression. Referees graciously provide their services because they see the value in community-based assessment of research output and they know that their research also benefits from good peer review. Authors and readers generally appreciate that papers benefit from a thorough editorial process that improves accessibility of specialized data and concepts. Equally, they understand that this process incurs significant costs. However, it is debatable whether copyright protection should extend to publishers. Since 2002, authors retain copyright at the Nature journals but grant them an exclusive licence to publish their work. Nevertheless, a constructive grass roots movement has emerged that argues for open access to all primary scientific data. Much has been written about the merits of various publishing models and indeed most permutations are now being subjected to real-world testing. Suffice it to say that 'open access' should apply to both reader and author, as excessive submission or publication charges can become prohibitive to all but the best funded laboratories.

PubMed Central (PMC), an open access archive set up by the US National Library of Medicine and its European counterpart UKPMC, provides a valuable repository for published work and is endorsed by many key funding bodies. NPG has encouraged archiving of accepted papers at PMC (or other appropriate repositories or websites) since 2005. The challenge is to make scientific literature accessible while recognizing the need to finance the publishing process. For this reason, the accepted, rather than the copyedited manuscripts, are uploaded six months after publication. The rationale is that the journal remains the formal point of reference, whereas PMC provides essential information centrally and at no cost. Links from PMC provide seamless access to the enhanced journal article.

As this dual publication approach gains momentum, NPG has launched an automated free service at the Nature journals for authors whose funding bodies require PMC posting. All it takes is to opt in at the submission stage (see www.nature.com/authors/author_services/deposition.html). Before posting, PMC confirms the deposition with the authors, and manuscripts are uploaded after six months. In the future, this will be extended to other NPG journals and repositories.