Pulleys adding three motions onto a single wire M2 crankS2 crank K1 cranksum → pen
Fig. 5 — Each crank moves its pulley up and down; the single wire collects every motion. The free end is the predicted height. Drawn by Daniel.

A mechanical tide predictor is not a static sculpture. It is an instrument of interlocking gear trains, counterweights, and silvered indicator dials built to dynamic tolerances. When an institution takes custody of a Kelvin or Doodson machine, the central curatorial dilemma emerges immediately: if you leave the mechanism stationary for decades, oil oxidizes and bearings freeze; if you run it under electric motor power for visitors, century-old bronze teeth wear down irrecoverably.

Over the past two years, I examined conservation reports and spoke with museum technicians in London, Liverpool, and Munich to document how these machines survive their second century.

The wire problem: tension without deformation

The most vulnerable component on any operational predictor is the summing wire. In operational service, instruments used fine phosphor bronze wire or flexible nickel-steel ribbon threaded over dozens of grooved pulleys (see my primer on how wires add waves).

When an instrument enters museum storage or exhibition, conservators face an immediate physical trade-off:

  1. Slack storage: Leaving the wire completely unweighted prevents creep and work-hardening. However, unspooled wire slips out of pulley grooves, snags on crank frames, and kinks irreparably when disturbed.
  2. Reduced ballast: The standard practice at institutions like the Science Museum in London is displaying the instrument under roughly 20% to 30% of its historic operating tension. This keeps the line seated in every sheave without applying continuous elastic strain across the pulley pivots.
  3. Synthetic alternatives for demonstrations: Where an instrument is occasionally turned for academic demonstration, historic bronze wires are frequently retired to archival enclosures and replaced with modern braided polyethylene or nylon lines of equivalent diameter.

Lubrication: the shift from sperm oil to micro-crystalline wax

Operating logbooks from the 1920s and 1930s (such as those from the Bidston Observatory machines) show that operators lubricated high-speed dial pinions with refined sperm whale oil or light mineral watchmaker oil.

In a display environment, liquid oils present two destructive failure modes:

  • Acidification: Organic lubricants degrade over decades, developing organic acids that corrode copper alloys, forming green verdigris crusts in pivot holes.
  • Dust attraction: Static museum air contains particulate dust. Liquid oil traps these abrasives, turning oil film into lapping paste that grinds gear teeth during demonstration turns.

Modern conservation replaces wet oils on non-running surfaces with Renaissance micro-crystalline wax. The wax seals bare brass against atmospheric sulfur and moisture without leaving a sticky layer. For the few internal steel arbors that must turn during scheduled maintenance, conservators rely on synthetic fluorinated oils (such as Krytox) that remain chemically inert over multi-decade intervals.

The monthly turn: preventing flat spots and bearing seizure

Metals under static pressure migrate. An arbor resting in a bronze sleeve bushing for thirty years without motion can seize through galvanic adhesion and micro-welding of oxide layers.

Curatorial teams at technical museums implement a strict “exercise schedule.” Once a month, a conservator unlocks the primary hand-crank clutch and rotates the gear train by hand through 720 degrees of main shaft rotation. This slow manual turn:

  • Redistributes synthetic boundary lubricant across the journal surfaces.
  • Ensures the escapement and harmonic constituent pinions engage smoothly across all pitch points.
  • Confirms that no pulley shaft has drifted on its bracket (a common fault covered in where error lived).

Archival documentation: recording the baseline state

Before touching an inherited predictor, conservators produce a photographic index of every constituent setting. When an office retired a machine in the 1960s (see why the machines died), the amplitude pins and phase dials were often left at the final operational run—frequently predicting high water for a specific naval harbour or commercial estuary.

Documenting these dial positions preserves the final working profile of the port before cleaning begins. A scratch on an amplitude slide is not museum damage; it is physical evidence of where a wartime operator locked a constituent pin in 1944.

Frequently asked conservation questions

Can visitors safely turn the crank on surviving tide predictors? No. All surviving tide predictors on public exhibition have their drive shafts mechanically disengaged or locked with acrylic shields. The bronze alloys used by Légé and Kelvin contain lead additives for machinability, making teeth susceptible to shock damage if turned abruptly.

Why do some museum machines lack their long paper rolls? Historic recording rolls were manufactured from wood-pulp paper that yellows and becomes brittle under gallery lighting. Conservators remove historic rolls and store them flat in acid-free archival boxes, occasionally fitting inert cotton-rag facsimile paper on the display drums.

Are original constituent calculation tables kept with the machines? Most surviving machines are separated from their port constant ledgers. The machines themselves resided in computational offices, while the harmonic constants were calculated in hydrographic archives. Reconstructing which machine served which ports requires cross-referencing national observatory ledgers.