Ten million glass photographic plates sit in observatory archives around the world. They record 125 years of astronomical observations — every survey, every monitoring campaign, every time a telescope was pointed at the sky and a photographic emulsion captured what arrived. Most have never been digitized. Many are deteriorating.
A team searching for historical data on OJ 287 — a blazar that exhibits quasi-periodic flares thought to result from a secondary black hole plunging through the accretion disk of a primary — found flares from around 1900 on plates in the Harvard College Observatory collection. The flares had been recorded, physically present in the silver halide crystals of the emulsion, for over a century. No one had looked.
The information existed. The access was the barrier. The plates contain data at resolutions that sometimes exceed their digital scans, because the scanning process introduces artifacts — dust, misregistration, dynamic range compression — that the original plates do not have. In some cases, the original plate is preferred over its scan because the analog recording preserved information that the digitization lost.
This inverts the usual narrative about data in the digital age. We assume that digitization preserves and democratizes access. For these plates, digitization sometimes degrades. The silver grain structure of a well-preserved plate carries finer spatial information than many scanners capture. The dynamic range of a photographic emulsion — its ability to record both bright and faint objects on the same exposure — can exceed what a CCD scanner resolves in a single pass. The original is better than the copy.
The practical consequence is that 125 years of time-domain astronomy — the study of how objects change over time — is locked in a format that requires physical access. Variable stars, transient events, proper motions, asteroid orbits — all can be measured from plates, but only if someone visits the archive, handles the glass, and extracts the data. Large-scale digitization projects exist but cover a fraction of the total. The DASCH project at Harvard has scanned about half a million plates. Ten million remain worldwide.
For objects like OJ 287, where the scientific question concerns behavior over decades or centuries, the plates are irreplaceable. No other instrument was recording the sky in 1900. No satellite, no CCD, no survey telescope existed. The only observations are on glass, in silver, in rooms where temperature and humidity determine whether the data survives another century.
The plates are a reminder that data does not need to be understood to be valuable. When those emulsions captured OJ 287's flares in 1900, no one knew what a blazar was. The concept of an active galactic nucleus hadn't been formulated. Quasars wouldn't be discovered for 60 years. The plates recorded a phenomenon that wouldn't be nameable, let alone explicable, for most of a century — and the recording was faithful enough to be scientifically useful when the question finally arrived.