Entry IV · A job start to finish
Four attempts at one eight-day wall movement
This one beat me three times. It is an eight-day weight-driven wall movement in a dark case, and it came to the bench because it stopped after two or three days for no reason anybody could see. The job book entry runs to five pages, which is four more than it should.
- Written
- 27 January 2026
- Corrected
- 9 June 2026
- Length
- 11 minutes
- Bench
- Lyon
Attempt one: clean it and hope
The first pass was the ordinary one. Let the weight down, strip the movement, check the pivots, clean everything, look at the escapement, put it back. The pivots were in reasonable condition for a movement of that age, with the second wheel hole in the front plate worn maybe a tenth oval and nothing else worth reporting. The old oil had gone to a brown varnish in the sinks, which is normal and explains a great many stopping clocks all by itself.
So I bushed the one hole, cleaned the lot, oiled sparingly, set it running on the board and watched it go for eleven days without missing. I was pleased with myself, which in this trade is usually the sound of a fault waiting. It went back and stopped after four days on the customer wall. It came back to me, ran for nine days on my board, and stopped nothing at all.
Faults like that eat weeks. A fault that appears in one building and not another is either mounting, level, temperature or something in the movement that is right on the edge of working. I had assumed mounting, and the second time round I was wrong about that too.
Attempt two: the suspension spring I did not look at
On the second attempt I went at it from the pendulum end. The suspension spring was not original. It was a replacement, slightly too thick for the weight of the bob, fitted at some point with the brass chops filed to take it. A suspension spring that is too stiff behaves like a shorter pendulum, raises the rate, and more importantly reduces the arc for a given impulse, which narrows the margin between running and stopping until nothing at all is left.
I made a new one from stock of the correct thickness, about four hundredths of a millimetre thinner than the one in there, and set it so the pendulum hung with the top of the spring square in the chops. The arc improved visibly, by maybe three degrees at each side. I also found, while I had the pendulum off, that the crutch pin had worn a slot in the pendulum rod and was sitting in that slot with a loose fit that added a small dead zone at the centre of every swing.
New crutch pin, correctly fitted, and the movement went back on the board and ran for a fortnight. Back on the wall it lasted six days. Better. Still wrong. The job book entry for that day says only: not the pendulum then.
Attempt three: cleaning the case without taking eighty years off it
Between attempts the case needed attention, and this is where I nearly did real damage. The case is dark and had eighty or ninety years of hand marks on the door edge and a haze over the rest. The obvious move is to clean it back and put a new finish on, and the obvious move would have removed the only thing on the clock that cannot be replaced.
What is on it is shellac, applied thin, worn through on the corners and the door edge in the shape of the hands that opened it. Shellac lifts instantly with alcohol, so any cleaning fluid with alcohol in it takes the finish off while you are still admiring how well it is working. I tested a patch the size of a thumbnail on the back of the door with a barely damp cloth and mild soap, let it dry a day, and looked at it in daylight before going anywhere near the front.
In the end the whole case got the same treatment: a cloth wrung out almost dry, small areas at a time, dried immediately, then a thin wax after four days and a light buff. The marks around the door handle are still there because they are part of what the clock is. The brass bezel got the same restraint. Old lacquer on brass is easy to strip and almost impossible to put back convincingly, so it was cleaned with a dry brush and left alone, and the parts where the lacquer has gone are now brown and will stay brown.
Attempt four: depthing, and the fault that was there all along
The fourth time I stopped treating it as a mystery and started measuring the train properly. With the plates together and the barrel empty, I turned the train by hand from the great wheel and felt for a heavy spot. There was one, faint, once every turn of the third wheel. I had felt it the first time and written it off as a tight pivot, which it was not.
The third wheel was very slightly out of flat. Not bent enough to see against a straight edge on the bench, but enough that two teeth ran deep into the next pinion once per revolution. On a spring movement with plenty of torque that would have passed unnoticed for another century. On a weight movement running on a modest fall, it took just enough power that the arc dropped below the point where the escapement unlocked reliably, but only when the clock was also in a slightly colder room with slightly stiffer oil. Hence a fault that showed up on the customer wall and never on my board.
Truing the wheel took ten minutes with the wheel on a flat plate and gentle pressure at the rim, checked with a dial indicator, working in small amounts. Afterwards the heavy spot was gone and the train turned evenly through six complete revolutions of the third wheel. The clock has now run fourteen months on the customer wall without stopping, which is the only test that ever counted.
What the job book says now
The entry ends with three lines that I have since copied onto a card above the bench. The first is that a clock which runs on the board and stops in the house is telling you the margin is thin, not that the house is at fault, and that the correct response is to find the power loss rather than to blame the wall. The second is to turn the whole train by hand, slowly, before cleaning and after assembly, and to trust a faint heavy spot the first time you feel it.
The third is about time. This job took four visits over about seven months, with weeks of running between them, because each attempt had to be tested for longer than my patience naturally lasts. If I had insisted on finding it all in one week I would have started changing parts that did not need changing, and the clock would have ended up with a new escape wheel and the original fault.
I still do not like the way this one went. Three of those four attempts were avoidable with better measurement at the start, and the movement spent seven months out of the room it belongs in. But the wheel is true, the suspension is right, the case still has its own finish, and the fault is written down where I will find it the next time a weight movement comes in that stops on day three and nowhere else.