- Elsevier, on its ScienceDirect Server, offers a single "sample issues" of each of its journals from the last year. That's of no real interest for Phys.Lett. B, but in the case of the Studies in History and Philosophy of Modern Physics, with just four issues per year, you may by chance come across some interesting stuff, for example about Scalar Gravity and Whitehead Gravity - once considered, but now disproved relativistic theories of gravitation (OK, in this case, both papers are also on the arxiv ...).
The free sample issue policy is also used by the Taylor and Francis Group (Contemporary Physics, Annals of Science) on its informaworld server, but a registration is required.
- Springer, on its SpringerLink server, offers some papers via an open access model which they call OpenChoice. In this case, the author, or her/his institution, has to pay a fee of $3000, and the paper will then be given away for free. According to the recent agreement by the German research organisation Max Planck Society and Springer, all Max Planck authors can now publish at Springer under this model. So far, open access articles are rare at SpringerLink (and there is no list of recent open access articles), but that may change in the future.
- A similar model to Springer's OpenChoice is offered by the American Physical Society. APS's Physical Review papers are usually subscriber-only, unless offered to open access under the FREE TO READ scheme. Here, the fee is less than at Springer, $975 for articles in PR A-E and $1300 for a PRL. However, I have the impression that there are only very few open access articles at the APS journals. That's of course no wonder, since I guess all the papers are on the arxiv anyway. The Physical Review Letters will celebrate its its 50th anniversary in July 2008, and for that reason the APS presents for free a series of selected highlight papers, the Milestone Letters (see the Update).
- As for the professional societies, the British Institute of Physics (IOP) on its Electronic Journal server offers for free all papers published within the last 30 days. There are a few journals whose papers usually don't make it to the arxiv, for example Metrologia, so that may be a way to effectively read these journals without formal subscription. The SISSA-owned Journal of High Energy Physics, founded as an alternative channel to over-prized commercial journals, is now published by IOP, but with no special conditions of access. And, of course, IOP runs the first general-scope peer-reviewed open access physics journal, the now ten years old New Journal of Physics (article charge: $1280).
- Other professional societies offer for free their backlist of old journal issues. This seems to be quite common for astronomy and astrophysics and in the Earth sciences:
There is, of course, the fabulous NASA ADS server with its bibliographic database and links to a huge amount of scanned astronomical papers. The European Geosciences Union allows free access to past articles of its journal Annales Geophysicae (ANGEO) and the proceedings series Advances in Geosciences, the Mineralogical Society of America grants free access to older issues of the American Mineralogist, and the Mineralogical Society of Great Britain and Ireland offers the back issues of their Mineralogical Magazine for free. Even the American Geophysical Union, whose journals seem to be so elitist that the Frankfurt University has not subscribed to a single one of them, grants unrestricted access to the 1963-2002 issues of Reviews of Geophysics, so that everyone can learn about Geological Constraints on the Precambrian History of Earth's Rotation and the Moon's Orbit, for example.
- Speaking about back issues, some science journals rely on the service of JSTOR, a "not-for-profit organization with a dual mission to create and maintain a trusted archive of important scholarly journals, and to provide access to these journals as widely as possible." Papers stored at JSTOR have a high chance to show up in google searches, and one can have a look at the first page, but what really annoys me about JSTOR is that it is not even possible to search or browse tables of contents without subscription.
- Such a displeasure is of course not a problem with peer-reviewed open access journals.
Founded at the same time as the NJP, the Living Reviews in Relativity, published by the Max Planck Institute for Gravitational Physics in Golm near Potsdam, Germany, offer free access to review articles on General Relativity. The Living Reviews derive their name from the feature that published articles can be edited to keep them up-to-date. SISSA runs Proceedings of Science (PoS), an open access collection of conference proceedings, mostly in physics.
Then, there is PLoS ONE (with a publication fee of $1250), the general science open access journal of PLoS, the Public Library of Science. But even after running for more than one year, there are just 20 articles classified as "physics" - and, well, most of them are about physics-related life science topics. And a quite recent peer-reviewed open access venture is PhysMath Central, with the journals PMC Physics A on high-energy/nuclear/astroparticle physics, cosmology and gravity, and PMC Physics B on condensed matter, atomic, molecular and optical physics. Currently, PMC Physics A levies a charge of $1,320 for articles accepted for publication, while "waiving charges for authors from low-income countries". I am not sure if PMC Physics A/B will have a bigger impact for physics than PLoS One, but it seems that they offer the option to comment on papers online, which might be a bonus.
Publishing comments and replies by the authors could become something quite exciting, and it is of course easier to organise for newly founded electronic journals. The most striking case in point I have come across is Climate of the Past, an "Interactive Open Access Journal of the European Geosciences Union". At this journal, it is even possible to access referee reports and the author's replies - that's transparency in publishing papers.
Open access to peer-reviewed scientific literature is for sure gaining momentum, but it seems to me that "traditional" publishing models might still persist, or coexist, for quite a while. I was quite surprised to find a very large diversity of access models used in different communities and disciplines. But that's a good thing, I guess: It could allow for the evolution of solutions that best fit the needs of authors and their readers.
And in the meanwhile, there is already plenty of interesting stuff freely available to you, in case you have some time to waste.
Update (Feb 22): Originally, I had complained that the Milestone Letters from the PRL's fifty years' history are hidden behind the subscription wall of the PROLA server. I have learned from Gene Sprouse of the APS that the PDF files are freely available when clicking the "Read Letter" link on the "Letters from the Past" page. There seems to be a temporary problem with the most recently added links to the Letters of 1964, which still ask for the PROLA authentification, and somhow my Safari browser does not like to show up the other PDF files, although I can download them using lynx, as I have checked in the meantime. That was the reason for my quibble. Sorry about the confusion!
Declaration of competing interests: I am currently on the payroll of a big commercial scientific publisher.