Fitting a new draw cord and finding the right mainspring tension

Part 01

How a cord dies

The original line on most of what I see is a braided cotton or linen cord somewhere between eight tenths and one and a quarter millimetres thick, running from an anchor on the carriage, round one or two pulleys, to the mainspring drum. It spends its whole life being wound onto and off a drum under tension, bending round a pulley at every pass, and being abraded at exactly the same two points every time. Sixty or seventy years of that is a good innings.

It fails by fraying to a core and then parting, and it nearly always parts at the pulley rather than at a knot. Before it parts it gets thin, and a thinner cord winds a slightly different diameter on the drum, which changes the carriage speed across the line by a small amount. So a machine that has started returning inconsistently, or that runs briskly at the left margin and reluctantly at the right, is often a cord about to go rather than an escapement fault.

I replace the cord on any machine that comes in for a full service and has its original line, whatever condition it appears to be in, provided I can get to it without dismantling half the frame. On some designs that is ten minutes. On others the drum sits behind the whole keyboard assembly and it is an hour and a half, and on those I look carefully and leave a sound cord alone.

Part 02

What I use instead of the original

I have settled on braided fishing line of the sort sold for sea angling, in a one millimetre diameter, with a stated breaking load far beyond anything the mainspring can produce. It is dimensionally stable, it does not swell in damp weather, and it does not have a cotton cord's habit of stretching a little over the first fortnight and then needing the tension reset. It is also slippery, which is the one real objection to it and the reason the knots matter.

The alternative is braided linen cord of the traditional type, which I use when the anchor is a simple loop over a hook and the knot will be under constant load. It handles beautifully, it grips itself, and it will need re-tensioning once after about two weeks of use because it beds in. Neither choice is wrong. I choose by the anchor: a hook or a pin takes cotton or linen happily, a clamp screw or a hole with a knot pulled into it is better with the braided line.

Diameter matters more than material. Going up from one millimetre to one and a half changes how much cord the drum holds and how the effective radius grows as it winds, and on a wide carriage that shows up as a carriage that is noticeably livelier at one end of the line than the other. I keep a small paper gauge with holes drilled at tenth-millimetre steps so I can check what came off before I decide what goes on.

Part 03

Knots, the drum, and counting turns

The drum end usually takes a figure-of-eight stopper knot pulled into a slot or hole, and with slippery line I add a second stopper right against the first so the tail cannot work back through. The carriage end depends on the design. A simple loop wants a bowline if there is room and a doubled overhand if there is not. I singe the cut end of synthetic line with a match held close but not touching, then roll it between finger and thumb before it cools, which leaves a small bead that will not pass back through a hole.

Winding the drum is the part that makes people nervous and it does not need to. With the carriage free and the cord attached at both ends, I hold the drum with a bar through its winding hole, take up the slack, and then add turns one at a time, counting out loud and writing the count on the sheet as I go. On most portables the working figure is between four and six full turns beyond slack, and on wide standards it can be eight or nine. What the machine needs is the smallest tension that will return the carriage cleanly to the left margin with the return lever from the far right of the platen, plus about one turn for the rest of its life.

Too much tension is the common mistake and it is not harmless. It loads the escapement dogs, so the carriage becomes noisy and the letter spacing begins to look uneven under a loupe; it makes the machine tiring to type on because every character is fighting the spring; and it shortens the life of both the cord and the spring itself. I test with the carriage at the extreme right and one finger resting on it. If it wants to pull my finger along, there is too much in the drum.

Part 04

While the carriage is apart: spools and the reverse trip

A carriage job is the one moment when the ribbon mechanism is properly accessible, so I go through it while I am there. The spools themselves are a small standards problem: the common flanged metal spool of about fifty millimetres is nearly universal in the machines I see, but the drive is not, and the differences are in the core. Some have a slot cut in the hub that a pin engages, some have a square drive, some rely on friction alone with a spring washer under the spool, and a spool from the wrong family will sit there turning freely while the ribbon goes nowhere.

The reverse trip is what turns the ribbon round when a spool empties, and on most designs it is triggered by the eyelet at the end of the ribbon pulling on a lever. If an owner has wound a ribbon on without the metal eyelet, or has tied the ends with tape, the trip has nothing to catch and the machine will happily jam the ribbon solid against a full spool and keep hammering it. That is the cause of a good number of the ribbon faults described to me as the machine tearing the ribbon.

The trip mechanism lives where it collects dust, dried ink and fluff off the ribbon, and it is spring loaded lightly enough that a little resistance stops it working. It gets the same treatment as the segment: solvent on a brush, worked by hand, dried, and no oil afterwards. Then I fit the ribbon, wind it by hand through a full pass in each direction, and watch the trip actually operate twice before I put the panels back.

Part 05

The test, and what goes on the sheet

The finished test for a carriage job is a full page: sixty lines at sixty-five characters, returned with the lever every time, with a ruler check on the left margin at the top, middle and bottom of the page. A cord that is going to slip in a knot will slip within that page, and a tension that is too low will show as a margin creeping right by a millimetre or two over the last twenty returns.

The sheet records the line material and diameter, the measured length, the number of turns in the drum, the date, and the machine's serial. Six months later somebody writes to say the carriage is behaving oddly and I can look at the sheet and know whether I put five turns in or eight. Before I kept the sheets, that conversation was me trying to remember a machine among the dozens that came through, and the answer was always a polite guess.