Pin Torque and Tuning Stability: You're Reading the Wrong Number

Ask why a piano won't hold its tuning and you'll hear about the pins. Tight pins hold, loose pins slip, tighter is better. The first two are right. The third is where it falls apart, because the number people quote to prove it usually isn't measuring what they think it's measuring.

None of this is aimed at the bench. Working technicians quote a floor and a comfortable band, which is a different and far better-supported claim than the linear one. The linear version — more torque, more stability — lives in consumer piano-care articles and in rebuilding sales copy. That's what this piece is arguing with.

The floor is about 25 inch-pounds

Start with the part of the conventional view that holds up, because it's arithmetic.

A tuned string doesn't pull on its pin evenly. It applies a continuous torque trying to unwind the coil, equal to the tension times the sum of the pin radius and the wire radius. That torque never rests. If the pin's resistance doesn't beat it, the pin turns on its own and the note goes flat at once, and stays flat.

Two sources bracket the number. Harold A. Conklin Jr., Baldwin's chief acoustician, works through a 170 lb string in his 1989 patent (US 4,920,847) and lands near 30 in-lb. An unrefereed 2013 technical note by Kees van den Doel, measuring three A4 strings on one Heintzman upright, reports 22.6 to 24.3 in-lb, the same neighbourhood by a different method.

So there's a hard lower bound, and beneath it no technique and no lever will save you. Loose pins are a leading structural reason a piano can't be made to hold, every technician who's met a tired spinet knows it.

The measurement earns its keep for a second reason. Your pinblock is hidden under the plate. Delamination, compression set, a block that's relaxed toward its floor: all invisible. Torque is the only non-destructive window onto a concealed component, and telling technicians to stop measuring would take away their only probe. The argument here is narrower. Torque is a diagnostic. It's very good at identifying pianos that can't be stabilised at any price, and poor at ranking the ones that can.

Your wrench is reading two things at once

Here's where the quoted figures come apart.

Measure breakaway torque on a strung piano and the reading depends on which way you turn. Going sharp, you're working against the block's friction plus the string's tension. Going flat, you've got the string helping you, so you're working against friction minus that same tension. Those aren't two noisy estimates of one value. They're two different quantities:

• Sharp: block friction plus string torque
• Flat: block friction minus string torque

The two directions on one pin differ by 45 to 60 in-lb on a normal instrument, which is twice the string's contribution. Conklin gives a worked case: a pin measuring 105 in-lb unstrung reads 135 one way and 75 the other. Half the difference between your two readings recovers the string tension. The average recovers the block's grip as if the string weren't there.

That identity isn't new and this article doesn't claim it. Versions of it have been stated on the Piano Technicians Guild mailing list, and some technicians deliberately measure unstrung for exactly this reason. What hasn't travelled with it is the diagnostic conclusion: the average is the least useful of the three numbers you can compute.

What the string pulls against, all day and all night, is the flat-direction resistance. That's the number that has to survive. The average blends it with a term that works against the pin turning back, so a pin can average into comfortable territory while its flat reading is nearly gone.

A practice-room grand at 85 and 15

A technician on the Guild's Pianotech list described a practice-room grand reading about 85 in-lb turning sharp and under 15 turning flat. It wouldn't hold over even a short stretch. Average those and you get 47.5 in-lb, which most published guidance calls marginal but survivable.

Separate them and the picture changes. Block friction is about 50 in-lb. The string is contributing about 35. The ratio of grip to string pull comes out near 1.4. On van den Doel's three healthy strings, that same ratio runs 3.9 to 5.0.

There's a good physical reason to expect a static wrench reading to overpromise. Breakaway torque is a fast measurement taken once. Holding a tuning is a slow business under constant load, and wood under sustained stress creeps, with long-term capacity well below short-term strength. A block that shrugs off a quick pull can still yield to a year of one. Treat that as a hypothesis. We couldn't find a test of it.

Trinkle said 50 to 100. Blees says above 125 is too tight.

Numbers for acceptable pin torque have been in print since 1956, when a Baldwin patent (Trinkle, US 2,736,224) called 50 to 100 in-lb ideal for the life of the instrument, called 35 and below loose, and warned against initial figures near 300 because pins start breaking around 350. Conklin's 1990 patent reports that tuners generally prefer 75 to 125, and that factory pins driven to 200 to 300 leave instruments hard to tune accurately for years.

Set the working technicians' figures side by side:

• Farrell: 60 to 150 workable, 80 to 120 ideal
• Foote: 100 to 125 comfortable, above 150 uselessly tight, below 80 worth flagging to the customer
• Blees: below 40 replace, 40 to 60 questionable, above 125 too tight
• Emmery: 95 to 140 normal

The comfortable middle agrees well, roughly 90 to 125 across independent sources. The limits scatter by a factor of two. Blees's ceiling sits inside Foote's comfort zone and below Emmery's normal band.

You could call the edges folklore. There's a better reading, and it supports the argument here: those figures may not conflict at all. If one technician quotes a flat reading and another quotes a sharp one on a strung piano, they're reporting quantities that differ by 45 to 60 in-lb, which covers the entire spread. Published torque figures are largely incommensurable, because almost nobody states the protocol alongside the number.

What Cerisano argued in 2015

If more torque bought more stability, a tight block would be an easy sell. The modelling argument runs the other way, and it concerns a different failure.

Mark Cerisano, RPT, who holds a mechanical engineering degree, put the case in the Piano Technicians Journal in July 2015. A pin held tightly needs more force at the lever before it moves, that force produces more bending and twisting in the pin, and everything that bends and twists during the tuning has to unbend and untwist afterward. van den Doel's note reaches the same conclusion analytically, with maximum stored twist angle scaling linearly with breakaway torque.

Two cautions on that pairing. They aren't two independent confirmations: van den Doel's acknowledgements thank Cerisano, and the two are best read as one line of thought written up twice. And neither is a measurement. Both are analytical arguments, so nothing here has been demonstrated experimentally.

Cerisano's own treatment is symmetric, and quoting half of it would misrepresent him. He names excessively soft pins in older uprights as a real instability mode in the other direction, and notes that bending and twisting partly cancel. The claim was never that looser is better. It's that the relationship has a comfortable middle with penalties at both ends, which is what a threshold-and-ceiling variable looks like.

Notice the failure at the top end is a different kind. An over-tight pin is hard to set accurately. That's a tunability problem, and collapsing it into a holding problem is the exact error.

The agraffe, the pin, and the weather

Above the floor, other mechanisms take over. They add to what the pin is doing. The pin still has to hold.

The string moving at its termination

Cerisano and van den Doel both put the pitch-changing event at the upper termination, the V-bar or agraffe, and not at the pin. Tension differs between the speaking length and the segment running back to the pin, and until the string slips across that termination to equalise, your speaking length hasn't really changed. If the string doesn't slip, the pitch doesn't change.

Elastic recovery in the pin

The bending and torsional wind-up above gets released when you lift the lever. Worth separating from the rest, because it's a property of technique as much as of the instrument. It's what setting the pin is for. Published cents figures for residual wind-up depend on a parameter that spans more than an order of magnitude between its measured and fitted values, so we're not quoting one.

Humidity, and Young's 1949 measurement

Humidity is the largest seasonal mover of pitch. The reference measurement is Robert Young's 1949 study in the Journal of the Acoustical Society of America, which found roughly 5 cents of pitch change per 10 percentage points of relative humidity across the three central octaves. Two qualifications get dropped almost every time that figure is repeated.

First, it's time-weighted humidity. The piano integrates moisture over weeks, with a time constant Young put near 15 days, so today's hygrometer tells you less than last month did. Second, because of that lag, the room's raw 20 to 70 percent swing acted on the instrument as 35 to 62 percent, producing about 13 cents of total excursion. Multiply 5 cents across the raw range and you'll overstate it badly. That study covered one forty-year-old six-foot grand.

The study we could not locate

We could not locate any published controlled study relating measured pin torque to measured pitch drift over time. That search covered the academic indexes and the Guild's own literature. We'd like to be wrong, so if you know of one, tell us.

Take care with that absence, because it doesn't show torque is irrelevant. Two reasons. Any survey of pianos technicians service is drawn from a population already filtered above the floor, since pianos below it aren't on tuning routes. Within that surviving sample, torque would show little relationship to stability simply because the range has been restricted to the flat part of the curve. And measurement error in a predictor pushes correlations toward zero, which given the protocol problem above is a large effect here. Both license the statement we can't currently tell. Neither licenses there is no relationship.

Work on the pinblock does exist and deserves naming. Nick Gravagne, RPT, wrote on pinblock forces in the Piano Technicians Journal in 1992 and 1993. Larry Lobel, RPT, published on tuning pins in early 2020 and convened a Tuning Pin Symposium in 2016. A 2016 Mendel University thesis by Petr Zatloukal measured pin torque against hole diameter across three beech-plywood block types, which is real torque data without a stability variable attached. The Guild's own 2016 Piano Tuning Pin Study seems to have reached the prospectus stage and stopped. Several of these sit behind member access.

At the bench

The case here is for reporting torque in a form that means something.

1. State the direction and the string state. A torque figure without them isn't comparable to anyone else's. This alone would resolve much of the disagreement in the published ranges.
2. Read the flat direction. That's what the string pulls against continuously, and it's the reading that has to survive. A pin can average into comfortable territory with its flat reading nearly gone.
3. Treat a large sharp-to-flat gap as a signal. Half the difference is the string's contribution, and a ratio of grip to string pull normalises for tension, so you can compare a bass pin against a treble pin.
4. Use torque to rule pianos out. It's strong evidence an instrument can't be stabilised, and weak evidence about which of two serviceable pianos will hold better.
5. Keep the floor in view. Below the string's own back-torque nothing works, and that's still the most useful thing your wrench tells you.

The interventional evidence runs one way and deserves saying plainly. Oversize pins, consolidating the block with thin adhesives, and block replacement do restore tunability, and technicians have reported as much in the Journal for decades. Those are case reports. We found no trial, and a restringing changes several variables at once. They're still the best causal evidence the trade generates, and they support the floor.

Read torque with a wrench that shows you the peak: a beam, or a dial with a follower needle, fitted with a number 2 star socket. A click type gets set to a value and trips at it, so it's the wrong instrument for reading a number you don't already know.

So measure both directions. Half the difference is your string, the average is your block, and the flat reading is the one that has to hold. Ten minutes with a wrench you already own.