Some years ago, the publishers of mathematical journals began to request authors to transfer the copyright on their papers to them, the publishers. In this way, they used to argue, would they be better able to protect our rights as authors and ensure widest possible dissemination. Since that time, this request has in most cases seen a silent transformation into a requirement: without a transfer of copyright most journal publishers now refuse to print a paper already accepted for publication by the editors - or at least this is what they say officially.
With journal prices rising fast and library resources at best stagnating, no-one knows what role any given journal will play in, say, 10 years' time. Will an article published now still be available in the then current electronic format (on a server widely accessible to the mathematical community) even if the publisher has discontinued the journal for commercial reasons? It does not seem unreasonable to doubt this.
Yet I believe we have both the right and even a degree of duty to ensure that our work will continue to be accessible to the general public that funds most of us now. For all these reasons I stopped signing away my copyright on journal papers in the late 1990s.
Interestingly, almost all publishers reacted either positively or not at all when I did not return the copyright form signed as requested: in all cases did they print the paper in question, usually without additional delay, and sometimes with unexpected understanding and support. (Yes, there have been one or two cases where things were a little more difficult at first, but these too were resolved amicably in the end.)
In most cases where the publishers insisted on some sort of legal declaration I simply changed the words "I hereby transfer the copyright..." (or whatever) to "I hereby grant the right to publish..." or similar. While so far I have been doing this by hand and with ad-hoc wording, maybe if more authors would like to handle this matter similarly we ought to make any amended copyright forms that have been agreed with a publisher available for future use by others. In this way, one might hope, the precedent of one case where a particular wording has been used may make it easier for the publisher to accept the same wording again without spending too much time and worry over it.
Below is the beginning of a collection of amended copyright forms - please contribute to the list!
UPDATE on 27.9.04:. Some publishers, including Elsevier, now have a consent-to-publish form that can replace the copyright transfer if authors insist. See below.
UPDATE on 5.7.06: Wiley have informed me that, following
a high-level policy decision, they are no longer prepared to publish
papers without a copyright transfer. This marks a change from
their previous policy, and has been substantiated by at least
one case known to me where they did indeed refuse to print a paper
that had already been accepted for publication, in the Journal
of Graph Theory. The editors protested without
success. Thus, if you consider sending your paper to the Journal
of Graph Theory, or any other Wiley journal, be
UPDATE on 14.3.2010: Elsevier have modified their consent-to-publish
form, and appear to be happy now to use it without fuss. You can fill
in the details by
hand (including your papers identification number at the top, which
includes a code for the journal), or edit the .doc version provided
below. There are still some questionable clauses in the text, which you
may wish to discuss with them. (For example, in 1.4 you grant them the
exclusive(!) right to use all or parts of your paper in other works. If
this right is exclusive, you're signing away your right to use parts of
your paper again elsewhere. There's clause 2.8 to give you some of
those rights back, but then why bother in the first place?)
Wiley (but see note of 5.7.06 above)
Elsevier (pdf; doc)
MAA (Beware of the two indemnification clauses. After some discussion, the MAA agreed to their deletion.)