Following Jim Stasheff's announcement of a unified maths eprint archive on the "categories" list, one issue which provoked a strong reaction was the requirement that papers be submitted in TeX source form. Mike Barr and others pointed out that they use versions of macro packages which cannot reasonably be expected to work under automatic control or in the hands of anyone apart from themselves. I agree very strongly with their views. I claimed in a previous email that the main site at Los Alamos accepts submissions in numerous formats but that the implementation at Duke University, where several existing maths servers exist, restricts this to TeX. I have spent most of today reading the documentation at xxx.lanl.gov and it has given me a headache: electronic methods have a very long way to go to compete with paper when it comes to reading the whole of a lengthy technical document. So, my claim was wrong. Paul Ginsparg has very strong views in favour of TeX (source): http://xxx.lanl.gov/help/faq/whytex.html and against alternative formats such as PostScript. I agree with a lot of what he says, and would even stick my neck out to suggest that Mike Barr probably does too. However, Mike and I feel very strongly that being forced to submit TeX source is a straitjacket, and you can expect us to continue arguing this vigorously (and informedly). Curiously, Paul Ginsparg doesn't discuss DVI (the output of (La)TeX) as an archive format, this being the one for which I would argue. TeX and PostScript are programming languages, but DVI is a very simple and robust "byte code". For anyone worried about the "doomsday scenario" that TeX will no longer exist in 100 years time, the structure of DVI is simple enough that it could be decyphered from existing binary files and a viewer recreated. After all, this was done for hieroglyphs in the 19th century, without the aid of the very sophisicated hardware and artificial intelligence techniques which we can expect to exist in the future. Ginsparg's archives have been running since 1991 and (according to his statistics) take a considerable volume of traffic. From TeX source he generates several other formats, configurably by the reader, apparently on the fly. To promise to do this and still keep your head above water requires an extremely robust system, as I know from having run a major TeX implementation for many years. There is also a lot of documentation about configuring your web browser to accept files in formats for which Netscape was never designed. In other words, he seems to have a very professional way of delivering files to readers. I take my hat off to him, because this is a conspicuous weakness of Hypatia. Whatever the arguments and counterarguments about this particular issue, I suggest that a more liberal attitude to archive submission formats be taken by this and other archives. Paul Taylor PS There was also some discussion of plagiarism, on which subject http://www.cs.cmu.edu/afs/cs/user/nch/www/koala-info.html is interesting.