DT-2026.04.029 / FIELD NOTEBOOK / VOL. I
DT-2026.04.029 MMXXVI

Every byte, every record, every backup carries an invisible cap. The cap frays a little with each migration, each format conversion, each warm-aisle year. We have spent a long time learning to count those frayings, and to hold the data steady while we count.

opening leaf — keeper's hand ~480 bp remaining (est.)
DT-2024.11.0142024

A telomere, in cytogenetics, is a small repeated sequence — TTAGGG, again and again — at the end of a chromosome. It exists so the chromosome doesn't lose anything important when the cell copies itself; it is the offering at the cliff edge, the part of the rope that we don't mind fraying, because the actual climb is held safe beneath.

Data has telomeres too, though we rarely notice them. A file copied from a 1996 hard disk to a 2003 RAID to a 2011 Time Capsule to a 2024 cold-store object will, with some non-zero probability, have lost a byte along the way. Sometimes the byte was a checksum and we caught it. Sometimes it was a pixel of a child's face, and nobody knew until twenty years later when somebody opened the photograph at four in the afternoon and noticed.

cf. The Telomerase Hypothesis, Greider 1985 ~457 bp remaining
DT-2023.08.2212023
Run 02 / Loss: 41 bp/cycle / Plotted 14 Aug

amplitude tapers — last 18% feathers to zero — chromosome cap fraying, sub specie aeternitatis

heartbeat sigil — see margin ~420 bp remaining
DT-2022.02.0412022

// fragment from logbook
0641 — array reseeded
0703 — checksum on cold-store batch 14
0719 — three sectors flagged, two recovered
0744 — one not recovered. (a single page from somebody's wedding album, 2008.)
0801 — coffee. brought to the lab by R.

— logbook excerpt ~390 bp remaining
DT-2021.07.0992021

There is a particular feeling, at four in the afternoon in late autumn, of light becoming a substance — slow, honeyed, almost weighable. The records vault has windows on the south side, and at four the light comes in low across the shelves. It picks out the dust. It warms the cardboard a little. The drives don't know about the light, of course; the drives are just spinning, dreaming their slow dreams of platter and head. But the curators know.

— from An Afternoon Vault, p. 14 ~360 bp remaining
DT-2020.05.0052020

fig. i — chromosome with intact caps

drawn in fountain pen, 2020 ~330 bp remaining
DT-2017.10.1142017

Format obsolescence is the slow part of the loss, the part that doesn't make a sound. A WordStar 4.0 file from 1985 is technically still here — the bytes haven't gone anywhere — but the grammar that turns those bytes back into a sentence has, for most readers, evaporated. We keep the bytes. We sometimes forget to keep the grammar.

(I have a friend who has spent eleven years writing emulators for dead word processors. She calls them séance machines. The phrase has stuck with me. There is something to it — the writer is gone, the format is gone, but the sentences come back, briefly, into a small green-on-black window.)

séance machines — keep this term ~290 bp remaining
DT-2014.03.0772014
Run 07 / Loss: 112 bp/cycle / Plotted 03 March

grief-curve — the long shoulder, the slow shoulder, the sudden cliff

a hard run. nothing recovered. ~255 bp remaining
DT-2011.06.1982011

The Time Capsule died on a Tuesday. It made a sound first — a small, polite click, like a book being shut on the other side of a wall — and then nothing. Most of what was on it was already on three other things, in three other places. But there was one folder, called scans, that had only ever lived there. The folder held seventy-two photographs of a woman's mother, taken between 1962 and 1981. The mother had been dead nine years.

We got sixty-eight back. The other four are the kind of loss that doesn't fit on a length-bar.

— oral history, A.K., 2011 ~225 bp remaining
DT-2007.09.0032007

An archivist I worked with in the early 2000s — a woman named Margarethe, late sixties at the time, raised in a town that had lost its archive twice in her grandmother's lifetime — used to say that the job was not preservation but translation across time. "You are not keeping a thing," she would say, "you are agreeing, on behalf of people not yet born, to keep translating it into whatever they will be able to read."

She retired in 2009. She still sends a letter every spring. The letters arrive on cotton-rag paper, and I scan them the same week. (I will keep translating them.)

M. — letters in folder /spring/ ~190 bp remaining
DT-2003.04.0442003

// from the 2003 flood, basement-level shelving
// 14 boxes affected, 3 catastrophic
// recovered: 1981–1989 logbooks (water-stained, legible)
// recovered: 1990s VHS reel-to-reel transfers (mostly)
// LOST: a single shoebox of 5.25" floppies, label illegible

the shoebox was J.W.'s thesis backups ~160 bp remaining
DT-2001.12.0122001

The thing nobody tells you about cold-store is that it has weather. Humidity creeps in through cable conduits. A bad summer in 2001 raised the dewpoint in our second vault by four degrees over six days. The drives were fine. The labels — paper, hand-written in 1994 — bloomed faintly along their edges. We re-labelled, by hand, in graphite. Pencil outlasts ink, in the long run, the way concrete outlasts paint.

pencil > ink, long run ~135 bp remaining
DT-1997.02.0271997
A

A note from the curator, in pencil, dated February 1997: "Began a small experiment today. Took the oldest tape we have — quarter-inch, marked THESIS BACKUP / S.K. / 1973 — and tried to read it on the new drive. The drive made a sound like a small bird. Nothing recoverable. I held the tape in my hand for a while afterward. It was warm — from the drive, not from any other reason — and it occurred to me that this was, in some sense, the last time anyone would ever hold this particular thought."

milestone — keeper's first loss ~105 bp remaining
DT-1993.??.???1993

[transcription incomplete]

transcription incomplete — page water-damaged in the 2003 flood

leaf 14 — only marginalia survives ~70 bp remaining
DT-1991.07.3011991

A photograph, lost. A printout of an autoradiograph, faded. The hand of someone who once worked here, in pencil at the corner of a logbook page: "k.k. — keeper of the warm aisle, summers 1989–1991. moved to montreal." Nobody remembers her last name. The autoradiograph was of a chromosome 17 telomere measurement. The numbers are still legible. The pencil has not yet given up.

k.k. — montreal — find her, if possible ~50 bp remaining
DT-1961.??.???⟨ ⟩

What we keep keeps us.

keeper@datatelomere.com
Continue the cycle →