Study DipCWS content as a chain of decisions, not a list of parts. Each section below isolates one decision point — what to record before dismantling, which escapement family you are looking at, how tooth counts localise a rate fault, why wind state changes power delivery, how to trace a keyless chain, and how lubrication follows the joint's job. Two worked scenarios show a plausible wrong turn and the better decision, and a final section gives a practice exercise, a self-check rubric, and a preparation sequence you can adapt.
The servicing habit that starts before any screw is removed
A servicing decision is only as good as the observations behind it. Record how the movement runs — rate, beat, wind-state behaviour — before dismantling, because teardown destroys the as-found evidence you will need to justify your work.
Before touching a movement, capture what it tells you: how much it gains or loses over a fixed interval, whether the error is steady or drifting, whether the tick sounds even or uneven, and whether anything changes between fully wound and nearly run down. Sketch the layout and note anything unusual — a marked tooth, a bent part, a previous repair. Once the movement is apart, none of that can be re-created, so these notes become your working evidence.
Contrast this with dismantling straight away. A movement that improves after a clean may still have had a damaged tooth or a worn pivot hidden by dirt, and if you intervened first you can no longer tell what the clean fixed and what it merely masked. Build the habit deliberately: an as-found record, then the intervention, then an as-left record using the same measurements. The difference between the two records is what you can actually claim your work achieved.
Recoil, dead-beat and Swiss lever: three escapements, three different checks
Recoil, dead-beat and lever escapements differ in how their pallets meet the wheel teeth. Identify the family first, because the checks that suit a recoil anchor do not transfer to a Swiss lever.
In an anchor escapement, watch the pallets against the wheel with a loupe while turning it slowly by hand. A recoil design lets the wheel back up slightly as a pallet releases — the teeth rock backwards — whereas a dead-beat design uses impulse faces shaped so the wheel stops still between impulses. That visible difference matters practically: lock and drop are set on the pallet faces, and uneven drop on one side shows up as an uneven tick long before it shows up as a rate error.
A Swiss lever in a watch works on a different principle: an impulse jewel in the balance staff engages the fork of a lever, and pallet stones exchange impulse with the escape wheel. Checking only lock, as you might on a dead-beat anchor, misses the draw that holds the stones against the teeth between impulses. Trace the sequence instead: unlocking, impulse through the lever to the jewel, and lock on the far side. Each family has its own vocabulary, and using the wrong family's checks produces confident but meaningless conclusions.
- Recoil anchor: wheel visibly backs up as each pallet releases
- Dead-beat anchor: wheel rests still between impulses; check drop and lock on both pallets
- Swiss lever: impulse passes through fork and jewel; check unlocking, impulse face contact, and draw
Counting the train: how tooth arithmetic localises a rate fault
Tooth counts fix the ratio from the barrel to the escape wheel. When a clock gains or loses uniformly, arithmetic on the train tells you which wheel could be responsible before you start guessing at parts.
The hands are geared together at fixed ratios, so a correct movement keeps them in agreement. Learn one anchor relationship from the going train: in a standard clock layout the centre wheel drives the third wheel, the third wheel drives the fourth wheel, and the fourth wheel often carries the seconds hand — so the tooth counts along that path determine how fast that hand turns. In some indirect centre-seconds watch layouts the drive relationship runs the other way, so confirm the actual mesh path of the movement in front of you before applying ratio arithmetic. If a wheel in that path was replaced with the wrong tooth count during an earlier repair, the seconds hand runs at the wrong speed relative to the minute hand — and arithmetic, not trial and error, is what finds it.
Scenario one: a pendulum clock gains about ten minutes a day, evenly. The tempting move is to adjust the pendulum until it is close — and it will run on time, for a while. The better decision is to check the hands against each other first: if the seconds hand disagrees with the minute hand over an hour, no pendulum setting can make both correct, and the fault is a train ratio, not the regulator. This matters because a train fault keeps drifting and hands that disagree make the clock unreliable to use, while a correctly diagnosed ratio fault points to a specific wheel to verify.
The same logic separates a train fault from a friction fault. A minute hand that is loose on its cannon pinion can slip and drag the hour hand with it, mimicking a rate error. Compare the hands during the as-found record: uniform error with all hands agreeing points at the train or regulator; a hand lagging or jumping points at its friction fit.
| Decision point | Pendulum clock movement | Stem-wound watch movement |
|---|---|---|
| Power storage | Weight or going barrel, often fusee-equalised | Mainspring in a barrel inside the case |
| Escapement family | Usually recoil or dead-beat anchor | Usually Swiss lever |
| Rate fault traced by | Pendulum, suspension, and train ratios | Balance, hairspring, and escapement condition |
| Position sensitivity | Fixed installation; one rate | Multiple positions; compare dial-up and dial-down |
| Lubrication emphasis | Slow, high-load pivots and sliding surfaces | Small quantities at fast pivots and keyless joints |
Going barrel versus fusee: why the error changes with the wind state
A going barrel delivers more force when fully wound; a fusee evens that out. A rate error that varies between full and empty points at the power side, not the escapement or the pendulum.
In a going barrel, the spring pushes directly on the train, and its force is greatest when fully wound and weakest when run down. A movement with a tired spring or dirty pivots may therefore run noticeably differently near full wind than near empty, and the strike may weaken as the spring relaxes. A fusee counters this with a conical pulley and chain: the spring pulls on a short lever arm when strong and a long one when weak, so the torque reaching the train should be far steadier across the wind.
Use this as a diagnostic fork in your as-found record. If the error varies with wind state, the promising territory is the power train — spring condition and set, barrel teeth, chain and hook condition, and pivot freedom under load — rather than chasing an average with the pendulum. Setting the pendulum to split the difference leaves you with a clock that is accurate only in a slice of its winding cycle, and it hides the power-side fault that will otherwise worsen.
Tracing the keyless works: stem, castle wheel, and the winding chain
In a stem-wound watch, winding and hand-setting pass through a lever chain. When winding feels rough or the hands will not set, trace the chain in order — crown, stem, winding pinion, castle wheel, yoke — rather than replacing parts on suspicion.
The keyless works is a small mechanism with two jobs that share parts. Turning the crown drives the stem and winding pinion to wind the mainspring; pulling the crown out moves the yoke so a sliding pinion disengages from winding and engages setting. Engagement points are precise, so a worn stem, a damaged winding pinion, or a yoke that no longer seats cleanly changes where the mechanism engages — which is why the symptom often appears only in winding or only in setting, not both.
Scenario two: a stem-wound watch stops after someone wound it past firm resistance, feeling something give. The plausible mistake is to assume a broken mainspring and fit a new one. The better decision is to examine the keyless parts under a loupe first: forced winding commonly damages the winding pinion or castle wheel teeth. A new spring in a damaged keyless will jam again the same way, and the client returns with the same fault. Order of inspection matters because the cheapest assumption is not always the one the evidence supports.
Direction of the symptom offers leads rather than verdicts. Hands moving while winding can point at a setting engagement that is not releasing cleanly, but it can also indicate a broken or badly set mainspring allowing barrel recoil, among other causes — so treat it as narrowing the search, and confirm with observation before concluding. If winding turns freely but the spring never tightens, the winding engagement is a natural first place to inspect. Matching each symptom to candidate engagement points narrows the inspection to a few parts before anything is disassembled.
Lubrication follows the joint: load, speed, and quantity before brand
Oil choice follows the joint's job: fast, low-load pivots need a light oil; slow, high-load points need a heavier application. Name what the joint does before deciding what belongs on it.
Map the movement by joint type. A balance staff turns quickly under a tiny load, so a heavy oil there adds drag to the most speed-sensitive part of the watch. A barrel arbor or a clock's great-wheel pivot turns slowly under high torque, where a light oil can be squeezed out or creep away, so a firmer lubricant is appropriate. Sliding contact points, such as pallet stone faces, need a minimal fixed quantity. Reasoning from the joint's job lets you justify each application rather than memorising a list.
Quantity matters as much as type. Excess oil creeps along pivots and arbors, collects dust, and eventually congeals exactly where it should never be, changing rate and wearing non-bearing surfaces. A correct application sits as a small meniscus at the bearing face. After assembly, inspect under the loupe for creep onto dial sides, teeth, or the balance — an oil check is an observation exercise, not just an application exercise, and it closes the loop with your as-left record.
A documented practice cycle: exercise, rubric, and preparation sequence
Build the evidence-chain habit on practice movements. Take an as-found record, service, re-record, and score yourself against a rubric. Alternate between a pendulum movement and a stem-wound watch so both vocabularies develop.
Exercise: take a spare pendulum movement with no historical value. Before touching it, run it and record the rate over one hour, the beat sound (even tick-tock or uneven), and its behaviour when fully wound versus half run down. Service it, then repeat the same record. Expected observation: the beat should still be even — if it became uneven after servicing, you set uneven lock or disturbed something during reassembly, so re-check drop and lock on both pallets before adjusting rate — and the rate error should be smaller and steadier across the wind.
Self-check rubric (learning milestones, not pass predictions): you can state your as-found rate and beat without re-reading your notes; you can identify the escapement family within two minutes of observation; you can compute the train ratio and confirm the hands agree; your as-left record quantifies the change from as-found. Score yourself on each cycle and target the weakest milestone next.
A preparation sequence you can adapt: alternate weeks between clock-side and watch-side topics so the two vocabularies reinforce rather than blur; close each week with one documented servicing cycle on a practice movement; keep a terminology notebook for drop, lock, draw, endshake, and engagement points, writing one sentence for each from your own observation. Consider yourself ready for the qualification's demands when you can complete a full as-found and as-left record unaided, and when every adjustment you made can be traced to a recorded observation.
Readiness checks: your latest practice cycle has all four rubric milestones met; you can explain, out loud, why a wind-state-dependent error points to the power side and why a seconds-hand disagreement points to the train; and your oil applications show no creep under loupe inspection. For administrative details about the DipCWS itself — structure, entry, and arrangements — consult the BHI directly at bhi.co.uk.
- Milestone 1: as-found rate and beat stated from memory
- Milestone 2: escapement family identified in two minutes of observation
- Milestone 3: train ratio computed and hands verified to agree
- Milestone 4: as-left record quantifies the change from as-found
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.