THE MAINTENANCE OF ARCHIVES

Rational search by humankind through a hyperspace of the size and complexity contemplated here requires systematic compaction and storage of the results as they come in. Uncommon care should be exercised in deciding what data to discard. In the near future, a plan should be prepared for a relatively long-lasting storage procedure, one providing easy access to the stored data, not only to colleagues in the search, but also to others who have an interest for any variety of reasons. Its construction will require careful thought, some experience, and recognition that this data bank has enormous growth potential. Probably, such a plan would itself grow through a number of stages in response to comments following dissemination of a preliminary plan.

It is attractive to suppose some appropriate journal would publish the plan and its development, and perhaps continue with periodic reports of the results to date in summary form, with clear information on how to gain the full details. The initial publication should be co-authored by representatives of several of the groups in the field. The Byurakan-founded international CETI committee could easily serve to invite such attention.

The magnitude of the permanent data storage problem presented by an extensive, state-ofthe-art SETI installation is worth illustration. Consider just the preposed receiver - 300 MHz instantaneous bandwidth; 1 Hz frequency resolution (bin width); 6 polarization channels; and 16 bits per output bin per second; 100 sec per target direction (for 100 unit observations). Thus there will be approximately 3X1012 bits to be disposed of every 100 sec. A large fraction of this torrent of data is redundant, since it is just plain noise. Sorting and compacting routines are an economic necessity for all recognizable spectral structures 2 or 3 sigma above the mean noise level. Then, the remainder of the data must be discarded.

What should be saved for future reference? Here is a partial list:

1. Housekeeping data when, by whom, with what installation; circumstances of the observation; essential system parameters, etc.

2. Identified (interfering) coherent signals - general characteristics (frequency, drift rates, bandwidth, etc.).

3. Unidentified coherent signals - general characteristics and estimated nature.

4. Astrophysical data continuum radiation as a function of frequency; a wealth of (primarily) interstellar line profiles; variable or intermittent radiations (pulsars, etc.); etc.

5. Pattern recognition and compaction algorithms in operation.

6. Comments by observers.

247-831 - 78 - 16

And all this should be available to interested parties for an indefinitely long time, easily, and without requiring highly specialized equipment.

As a final remark, we stress the need to favor storage over discard, but with good judgment. Any clear hint of structure above the calm sea of background noise may be the clue.