Sheet 003 / Cleaning

Cleaning a sluggish segment without dunking the machine

More than half of what arrives here is described as sticking keys, and more than half of those need nothing but the segment cleaning out. It is the least skilled job on the bench and the one I have had to develop the most discipline about, because the fast version of it ruins machines.

Part 01

What a sticking key usually is

The segment is the fan-shaped slotted casting that the typebars pivot in. Each bar sits in its own slot with a clearance of a few hundredths of a millimetre, and that clearance is the whole reason the machine can type at speed. Nothing in the slot is supposed to be lubricated. Oil there attracts dust, dust and oil make a paste, the paste oxidises into something like soft varnish, and within a decade or two the bar is moving through a substance rather than through air.

So the standard complaint of keys that rise slowly, stay up, or come back only when the next one knocks them, is almost always the same thing: somebody helpfully oiled the machine at some point, or the machine was stored where the air carried enough grease for it to happen by itself. Kitchens do it, and so do workshops. I had one machine in 2024 from a flat above a takeaway on the other side of the canal that had a genuine film of cooking fat on every exposed surface, and every one of its forty-four bars was slow.

The other causes exist and are much rarer. A bent typebar rubbing its neighbour, a broken or unhooked return spring, or a bar that has been sprung sideways by somebody pulling apart a jam without releasing the tension first. I check for those by lifting each bar with a finger and letting it fall, before any liquid comes near the machine. A bar that falls freely and only misbehaves under power is dirty. A bar that is stiff when lifted slowly by hand is bent or fouled, and washing it will not help.

Ticked off at the bench

Lift and drop every bar by hand before deciding anything
Slow under power but free by hand means dirt, not damage
Stiff by hand means a bent bar, a sprung link or a broken spring
The segment slots are a dry-running fit and should never be oiled

Part 02

The flush, and keeping it off everything else

I do not put machines in baths. Full immersion carries the dissolved crud into the mainspring drum, the escapement and the key linkages at the back, and it takes felts, paper and decals with it. What I do instead is a controlled flush of the segment alone, with the machine standing on its back on a tray, the carriage locked, and the body panels off wherever they come off without a fight.

The solvent is odourless mineral spirits. It is slow, which people treat as a disadvantage and I treat as the point: it gives me time to see where it is going, and it does not attack the paint or the rubber feet in the fifteen seconds it takes to reach the tray. I apply it with a syringe and a short length of tube, directly down into the slots, a few millilitres at a time, and work the bars by hand between applications so the fluid carries the loosened material out rather than just relocating it. A run takes maybe twenty minutes of syringe, work, syringe, work, and the tray underneath tells you when to stop: the drips come out grey and then they come out clear.

Everything that is not the segment gets covered. Two strips of thin plastic sheet under the segment and over the keyboard linkages, one folded card behind the type basket to catch spray, and the ribbon and spools out of the machine entirely and on the shelf. The ribbon in particular will wick solvent and then hand it back to the platen and the type for days afterwards.

Part 03

Drying time, and the winter problem

Mineral spirits leave the slots slowly. In this room in July, with the window open and the machine tipped onto its back, a flushed segment is genuinely dry in about four hours. In January, with the room at seventeen degrees and the air off the canal at ninety percent, it takes a day and a half, and a machine I hand back at six hours will type well that evening and be tacky again by the weekend as the last of the solvent creeps out and takes the remaining dissolved residue with it.

I found that out with a 1953 portable that came back three weeks later with the same complaint and the owner reasonably wanting to know what he had paid for. The second clean took no longer than the first. The difference was that I left it standing on its back for two days in front of the radiator before I touched it again, and it has been fine since. So the shop rule is now a minimum of twenty-four hours, and in the damp months I tell people at the outset that the machine will be here over a weekend.

The temptation is to speed it up with a heat gun or compressed air. Air is fine at low pressure and from a distance, and I use a hand bulb blower on the slots. A heat gun near sixty-year-old paint, decals and a rubber platen is a bad trade for two hours saved.

Part 04

The panel I stripped, and what I use now

In the spring of 2023 I ran out of mineral spirits mid-job and finished with a stronger solvent from the shelf, on the reasoning that it would be faster and I would only use a little. It ran further than I expected along the underside of the segment and reached the painted front panel of a crinkle-finish machine. Crinkle paint from that era has a texture that was baked in, and once the solvent has softened it there is nothing to be done that looks like the original. I told the owner what had happened, I did not charge for the job, and the machine still has a patch on the lower right of the front panel that I can see from across the room.

What I changed afterwards is that the bench now carries only one solvent for this work, in a labelled bottle, and the stronger materials live in a cupboard on the far wall with the lid of the box taped down. I do not trust my judgement at the end of a long evening, so I have arranged the room so the wrong bottle is not within reach.

The other change is that I test on the underside of a panel first on any machine I have not worked on before. Thirty seconds, a cotton bud, and a look for colour lifting onto the white fibres. Most finishes from these decades are tougher than I feared. The two or three that are not will tell you immediately, and finding out on a square centimetre under the machine is very different from finding out on the front.