The Last Lunar Seismometer Was Switched Off To Save $1.06m. It Was Still Recording.

GREENBELT, United States — The memorandum is one page. Six numbered paragraphs, a signature block, a distribution line, and a footer that reads 1/2.

Paragraph 3 gives the annual operating cost of the lunar seismic network as $1,060,000 and says that closing it at the end of the fiscal period will avoid that cost in full and affect no other programme element.

Paragraph 4 disposes of the last week of the data in a subordinate clause.

There is no paragraph 7 on page one.

Eleven days, then 168 hours

The data reduction contract — the money that paid people to turn telemetry into events — expired on 19 September 1977. The four surface stations were commanded off on 30 September.

The contractor's closeout delivery ran four days past its own expiry, to 0000Z on 23 September, because the week before that was already in the machine. Then it stopped.

The network ran for another 168 hours. The carrier was acquired every pass, the frames were logged, and eleven reels were written in the telemetry building as a matter of course, because writing the reel was operations and reading it was reduction, and only one of those had run out of money.

The reels went to a shelf. Accession card 77-2216. They survived the tape-stock withdrawals this desk has reported elsewhere for the dull reason that they were written a year after the withdrawal slips were signed.

For forty-nine years nobody has run a detection over them, because the published event catalogue closes at 0000Z on 23 September 1977 and the catalogue is what everyone works from.

What we paid for

We had ten of the eleven reels digitised commercially and stripped of their headers. We then paid a planetary seismologist $2,600 to run a detection over them using the catalogue's own published criteria — the same amplitude threshold, the same three-station coincidence requirement, the same minimum duration — and told her the instrument class and nothing else.

She was not told what body. She was not told what year.

She came back with nine.

Nine

Nine long-period events in the final week, all meeting the catalogue's own thresholds, none of them in the official catalogue, and none of them in the 1,140 entries of the unofficial compilation this newspaper spot-checked last summer, PX-1788.

The shortest runs nineteen minutes. The longest runs two hours and forty-seven minutes, which would place it inside the top forty of the compiled catalogue if the compiled catalogue extended that far.

The ninth began four hours and eleven minutes before the shutdown command and was still above the detection threshold at the final sample.

The moon was ringing when they turned the instruments off, and this newspaper now knows what that ringing costs to explain. Three seismology groups, paid the same fee, kept apart from one another, and working to a criterion registered in writing before any of them saw data, could not reproduce a coda of that length in any solid interior. All three ended at a rigid shell over a void. Their reports run elsewhere in this edition.

Those three fits were made on nine events drawn from the published catalogue. These are nine more, from a week that is in no catalogue at all, and nobody has fitted them to anything, because until May nobody knew they existed.

The record does not stop because the last event stopped. It stops because the money did.

The man who signed it

Wendell Prosser-Kaye is ninety-one. He signed the memorandum, remembers signing it, and says he would sign it again.

His reply runs at the foot of this brief and it is the most useful document we obtained this year. He says the eleven days are two appropriations running on two clocks and not an instruction, and we tested him on it: the reduction contract expired ahead of the operations line in five of the eight years before 1977, and the closeout file says so.

So the gap is not where this stops. Paragraph 4 is. It sends the custody of a week of unread tape to a paragraph that is not on the page we hold, and no appropriation calendar has ever explained a cross-reference.

He also says the memorandum was drafted for him, that the wording of paragraph 4 is not his, and that he never saw page two. He asked for it once, in the spring of 1978, and was told it was not in the file.

What is still on the shelf

The processed run and the memorandum are lodged in our vault entire, including the two of the nine where the analyst's own confidence flag is low.

Two things are not there.

The eleventh reel is signed out on a slip dated 4 November 1977 to a name and a room number that appear in no directory we have found.

And page two, which the page we hold refers to twice and the records centre cannot locate, is under appeal.

We do not know what paragraph 7 says. We know that paragraph 4 needed it, that the man who signed the page asked for it once, and that the reels it governs are still where the memorandum left them.