Brief
The client — a regional university library, named privately on request — held a 22,400-piece collection of 1880s–1920s glass-plate negatives, gathered from three regional photographic studios that had operated commercially in Pittsburgh between 1882 and 1928. The collection had been in the library’s storage since 1971, originally at general-archival conditions of approximately 22 °C / 55% RH. A 2023 condition survey by an independent photographic conservator had identified silver mirroring on roughly 38% of the plates and emerging emulsion lift on a smaller subset. The conservator’s recommendation: cool storage at 5 °C / 30% RH, with a slow ramp.
The library’s mechanical group had a small budget for a walk-in cool-storage box; what they did not have was a plan for the ramp, an instrument-quality sensor mesh, or a written equilibration protocol. We were retained in March 2024 to design the box, run the ramp, and produce the binder.
Why slowly
Glass-plate negatives have two vulnerabilities that constrain cooling rate. First, the silver-gelatin emulsion contracts on cooling at a different rate than the glass support; a fast cooling ramp can cause the emulsion to crack laterally (this is the failure mode usually called “tile crack”). Second, the glass plate itself, particularly the older 1880s–1890s thin-plate stock, has a glass-transition zone in its thermal expansion curve that aligns uncomfortably with the lower part of a typical cool-storage ramp; cooling too fast through that zone induces fracture along stress lines that may already exist as latent flaws in the glass.
The conservator-recommended ramp was 0.5 °C and 1.5% RH per week. Petra reduced this to 0.4 °C and 1.0% RH per week after running the IPI dew-point envelope against the collection’s realistic equilibrium and against the building’s realistic stability. The slower ramp added eight weeks to the equilibration. It was, in retrospect, the single most consequential decision on the project.
Mechanical (Gate M3)
A 1,200 ft² walk-in cool-storage box, retrofit into a basement-level back-of-house space the library had previously used for general dry storage. Box construction: 6-inch insulated-panel walls, 6-inch ceiling, 4-inch insulated floor on a frost-isolating slab, vapor-tight throughout. Glycol-loop split system: outdoor air-cooled condenser sized at 18 tons, indoor evaporator with a low-temperature ECM blower, glycol secondary loop, electronic-expansion-valve metering. Munters MX desiccant on the make-up air line. A four-foot vapor-tight vestibule with mechanical-door closer and a small dehumidified air-curtain. A sensor mesh of 11 LoRaWAN nodes, of which two are inside the boxes (the conservator wanted in-box readings) and the remainder distributed across the vault and vestibule.
One detail worth recording: the existing basement was tracking 18 °C / 65% RH year-round, which placed the vestibule’s ambient at a dew point uncomfortably close to the vault’s wall surface. We added a 320-cfm DOAS unit serving only the vestibule, conditioning it to 13 °C / 35% RH. The DOAS pays for itself, in our calculation, in roughly 11 years against avoided wall-condensation events.
The 312-day ramp (Gate M5)
The collection was moved into the new vault, in stages over six weeks, beginning January 2025. The vault was held at 22.0 °C / 55% RH (the pre-existing setpoint) during the move. The ramp began on the day the last box was on the shelf.
| Date | T (°C) | RH (%) | Note |
|---|---|---|---|
| 2025‑02‑22 | 22.0 | 55.0 | Ramp begins (day 0) |
| 2025‑04‑21 | 18.4 | 46.7 | Pause 1, 28 d |
| 2025‑06‑23 | 14.8 | 39.4 | Cold-snap pause, 14 d (unscheduled) |
| 2025‑08‑18 | 11.2 | 35.6 | Glass-transition pause, 28 d |
| 2025‑10‑27 | 7.4 | 31.6 | Pre-final pause, 14 d |
| 2025‑12‑31 | 5.0 | 30.0 | Setpoint reached (day 312) |
| 2026‑05‑01 | 5.0 | 30.0 | Holding (day 433) |
The unscheduled cold-snap pause in late June 2025 was a response to a four-day −3 °C ambient event that drove the building’s basement temperature down out of band; we paused the ramp for two weeks while the building stabilized. The glass-transition pause at 11.2 °C was scheduled; it is the most stress-prone region of the ramp and the four-week pause was, in Petra’s judgement, cheap insurance.
In-box sensor readings (the conservator’s requirement) confirmed that the boxes themselves tracked the vault setpoint with about a 14-hour lag and a 0.6 °C damping. We adjusted the ramp accordingly: the vault was always running about three weeks ahead of the boxes. At the December 31 setpoint date, the in-box readings were at 5.4 °C / 30.7% RH. By May 1, 2026, in-box and vault are tracking within 0.2 °C and 0.8% RH.
Today
The vault is in its fifth month at setpoint. The library’s photographic conservator has spot-checked 220 plates as part of an early follow-up condition survey; she reports that emulsion and substrate are stable and that no new fractures have emerged from the move or the ramp. We are scheduled for the standard one-year condition survey at the year mark (December 2026).
This case taught the practice three things, written into our shop manual:
- 0.4 °C/wk is the right rate for early-period glass-plate negatives. The shop’s previous heuristic was 0.5; we have lowered it.
- Pause through the glass-transition zone. Four weeks at the 11 °C waypoint is, in our view now, a non-negotiable feature of an early-period glass-plate ramp.
- Vestibule conditioning earns its keep against wall condensation. A 320-cfm DOAS in a basement-level vestibule pays back in roughly a decade against avoided incidents. We will spec it standard going forward.
The collection has been digitized in part — the regional university library is part of DLF’s digital library federation — but the originals are now stable for, by IPI’s natural-aging projection, another century or more.