SIGNALS ANALYSIS Group Seven

The transmission opens with eight seconds of tone at 462 Hz, then the voice. Female, middle-aged, accent unplaceable. It is not live: the same breath falls in the same places every night, under the same tape hiss. We are listening to a recording, replayed by a machine.

Thirteen groups of five figures, read twice, closing with the single word “ends.” The groups do not change from night to night. A message that never changes is not traffic. It is a recitation.

Against current systems: nothing. So I went down into the older registers. The read format — the pause lengths, the doubled reading, the terminal word — matches VESPER, a one-time-pad system retired and destroyed in June 1974. I have never seen VESPER; it is in no training syllabus. I found it because the cadence survives in exactly one archived recording, shelf item AR/662, marked for disposal and never disposed of, and the archive card beside it is stamped DESTROYED.

Someone is broadcasting in the accent of a dead system.

One anomaly, and I want it on the record with the date attached. Last night, 25 March, group seven was not read. In its place: eight seconds of the tuning tone, precisely the length the group would have taken. The other twelve were read as always, and the voice went on as if nothing were missing.

If the message has been identical for fifteen nights, and the message is a list — then last night, something was crossed off it.