Long terms storage of data sets

[VISBMS at UBVMS.BITNET]

One way to ensure long term storage of data sets is to publish them
in an archived electronic journal. Flora Online (ISSN 0892-9106) is a
peer-reviewed journal available at huh.harvard.edu via anonymous FTP,
at Taxacom Dialup via 716-896-7581 (8-n-1), and on diskette (write
Buffalo Museum of Science, 1020 Humboldt Parkway, Buffalo NY 14211 for
info). We publish data-intensive text and programs (MS-DOS only for now)
on botanical subjects. We recommend that data sets be in comma-delimited
ASCII (e.g. "field1","field2","field3", etc.cr/lf per record). Most data
managers can import this. Relational data sets can be combined into
one flat file (when possible) and re-normalized later.

Regarding the problem with fading magnetic storage: if the data sets are
published as formal archived issues of a journal, then the responsibility
for keeping the mass of data organized and transfered to fresh media
in a timely fashion falls to librarians, a responsibility that should be
theirs (of course there are problems with gigantic data sets that should
not be dumped into librarians' laps). For data sets of a few megabytes,
publication in a journal is a better way to ensure saving the information
than encouraging yet another giant bureaucracy. Also, peer-review
encourages those with data sets they would like to share to submit
them for archiving in a cleaned-up form.

R. Zander, editor, Flora Online
visbms at ubvms.cc.buffalo.edu     visbms.ubvms.bitnet