The Plate Book
The observatory kept one star.
Between 1951 and 1994 the Marra Ridge field station exposed 3,117 photographic plates of a ninth-magnitude star with no name outside our own catalogue — H 812, four degrees east of the Coalsack — and measured every one of them on the same engine, in the same room, mostly by the same man. Eric Voysey had done the measuring for forty-four years. He had the steady hands the job wanted, and by 1994 he had to rest them on the table edge to keep them that way.
I was the night assistant. I mixed the developer, ran the plates through the fixing bath, kept the weather log, and changed the drum on the barograph every Monday. On clear nights I sat at the guide scope and held a star on the crosswires for the forty-minute exposures while the astrograph burned the sky onto glass. The plate book went back to 1951: date, plate number, seeing, exposure, and the initials of whoever guided. EV, EV, EV. My initials arrived in 1993 and looked wrong on the page, the way a new key looks wrong in an old lock.
The survey’s money ended in June. A committee was coming to weigh the archive — crates of glass, they said, to be digitised, which is a word for making a photograph of a photograph and calling it ours. Eric was retiring before the committee arrived. His farewell dinner was booked for the staff room downstairs, on the last clear night of his last roster.
The forecast said the front would clear by nine.
He was due to guide the final plate himself. It was his star; the secretary had even printed its number on the menu. But the dinner began at seven, the speech was his, and his hands had been bad all week — three plates spoiled, star images trailed like comets. He stood in the doorway of the measuring room with his tie already on and asked me, not the sky, whether I would take the last one. The series would end well. That was the whole thought.
I guided Plate 3117 from 9:40 to 10:20 and initialled the book.
1994 May 14 — P3117 — 40 min — seeing 3 — guided A.C.
Four Arcminutes West
A plate is not looked at; it is compared. Three nights later I paired 3117 with 3116 under the blink comparator, which flickers two plates fast enough that anything moved between them leaps on the screen. You use it to catch variable stars. H 812 leapt.
Not far. Four arcminutes, due west.
I measured it twice on the engine, then dug the earliest plates out of the 1951 crates and measured those. Every plate since the survey began put H 812 in the same place, within the wobble of the Earth’s own motion, which we subtract because we know it. Plate 3117 put the star four arcminutes west: a round, sharp, honest star image, with the other stars on the plate exactly where they belonged. The guide star was round. A slipped exposure drags everything. This had moved one star and left its neighbours fixed.
Eric came in at ten with his farewell flowers still in his lapel and measured it a third time without a word. The next clear night we took 3118. The star sat where it had sat for forty-three years.
Tilt, emulsion slip, a grain of dust, a shadow on the plate — we spent a week on explanations, and the explanations kept dying on us. What stayed was this: the move exists on one plate. The plate is numbered. The book holds my initials, and the archive does not withdraw plates, because the archive is not a record of the sky. It is what the sky was. Whatever H 812 did on the night of May 14, it did once, for forty minutes, to glass, and it will not do it again for anyone.
The Exceptions Book
Packing Eric’s desk took an afternoon. Bottom drawer, under the graph paper: a school exercise book, forty years of entries in his hand, headed in fountain pen — EXCEPTIONS.
12 Aug 58 — Barograph stepped +0.4 mb at 21:07. Reset itself. No front that night.
3 Mar 71 — Shortt clock gained 2.0 s in one night. Never repeated. Sent for service; returned perfect.
19 Jun 79 — Rain gauge 1.3 mm on a dry night. Gauge empty at 21:00, full at dawn. No dew.
5 Feb 90 — Seismo pen swung once, 4 mm. Sydney registered nothing.
Forty years, thirty-one entries. Not one repeated. Instruments misbehave — that is ordinary; gauges stick, clocks drift, that is why you service them. Eric had kept only the ones that happened once and never again, each dated, timed, initialled.
I did what you do with a suspicion you cannot afford: I checked it against the paperwork. The duty rosters go back as far as the book does. Every entry falls on a night when one observer relieved another — Ray covering Eric’s tea break, Eric sitting Fred’s shift so Fred could get to the coast, a student sent up at midnight so the regular could sleep. Every exception, on a handover. The barograph stepped four minutes after Ray took the pen from Eric’s hand. The clock gained its two seconds the first night Eric sat down at Fred’s chair.
The star moved on the one plate I guided, on the one night Eric’s watch passed to me mid-series, forty minutes of it — the exact length of my hand on the crosswires.
Eric’s last page was dated May 14. He had written it the morning after, while the flowers were still fresh: P3117. Below it he left one blank line, and the book was addressed to nobody, and it came to be in my drawer, because my initials had been the newest thing in the plate book when he packed.
The Relief Roster
The committee digitised the weather logs, the clock comparisons, the plate measurements. Paper forms went into crates. The relief roster — who sat down at whose station, on whose break, at whose request — was on the list to close, because no database has a column for it.
I took it home.
Then I started taking the reliefs. Not the shifts — the handovers. Doug’s tea break on Tuesdays. The students’ storm watches, so the students could eat. An hour here, ninety minutes there, my initials in the roster next to someone else’s start time. The instruments did what Eric’s book said they do, once each, and never again: the new barograph stepped in July, one tick, at 3:12 in the afternoon, while I held Doug’s pen. The renovated rain gauge delivered 0.9 millimetres on a clear night in August and went back to zero forever. Nothing broke. Every technician we called wrote the same words on the same form: no fault found.
H 812 has not moved again. I measure it every quarter, against 3116, against 1951, against 3117, and the star holds both of its places with the patience of glass. The digitised series will carry the displacement forever, flagged — when someone finally flags it — as a single unexplained value. There is no column for the roster that dates it.
There is a new assistant now, a good kid who mixes the developer strong and wants the permanent post, the way I wanted it. Last week, when my hands were bad, he stood in the doorway of the measuring room and asked — not the sky — whether I would like him to cover my tea break.
At 21:07 the barograph stepped once, plus 0.3, and settled. I wrote the date in Eric’s book, on the line below P3117, in ink.
