Halfway down the tightening torque calculator there is a section headed "Friction is the biggest variable here". It is, and that section proves it for one bolt: the same M10 × 1.5 at class 8.8, one torque printed per lubricant condition.
What one bolt cannot show is the comparison that decides how a joint is specified — friction against property class, measured on the same axis. With a single class on screen the friction span is visible and the class span is not. Put the two side by side and the ordering is not what a drawing implies.
A property class is deliberate: you read 8.8 off a drawing, look it up, and get a yield strength back. Lubrication usually is not deliberate at all — the bolt came oiled from the box, the bracket is zinc plated, someone wiped grease on the threads because it goes in easier. Both end up as a number in the same equation, and only one of them is written on the part.
Two levers, one ruler
Tightening torque goes as T = K · d · F. Only K changes between conditions, so the ratio between any two of them is the ratio of two K values — which cancels the size, the class and the thread. That is why the table below has no "for M10" anywhere in it.
| What you change | Torque multiplier |
|---|---|
| Class 4.8 → 5.8 | ×1.235 |
| Class 5.8 → 8.8 | ×1.524 |
| Class 8.8 → 10.9 | ×1.469 |
| Class 10.9 → 12.9 | ×1.170 |
| Class 8.8 → 12.9 (two steps) | ×1.719 |
| Class 4.8 → 12.9 (full range) | ×3.235 |
| Dry → MoS₂, same bolt | ×1.833 |
Every adjacent step on the ladder is no larger than ×1.524. The two-step jump from 8.8 to 12.9 — what most people mean by going stronger — is ×1.719. Dry to MoS₂, same bolt, same wrench, everything else unchanged: ×1.833.
The one move on that list larger than a change of lubricant is the whole ladder at once, 4.8 to 12.9, at ×3.235. Anything short of that is smaller than deciding what is on the threads.
Why the ratio travels
Because K is a factor, not a term. Double it and the torque doubles; nothing else in the equation knows what K is, and K knows nothing about the bolt. As, Rp0.2 and d multiply together and cancel out of any comparison.
That cuts both ways. It makes ×1.833 a portable number — the same for every size in the database — and it makes K unreadable from the fastener. There is no head marking for friction, no class that implies a lubricant, and no way to tell by looking whether the joint about to be tightened sits at 0.12 or 0.22.
One size, both levers at once
Take one size and compute it every way round — down the classes, across the conditions. Every figure is torque in N·m at 90% of yield, for the M10 × 1.5:
| Property class | Rp0.2 (MPa) | K = 0.12 | K = 0.16 | K = 0.20 | K = 0.22 |
|---|---|---|---|---|---|
| 4.8 | 340 | 21.31 | 28.41 | 35.51 | 39.06 |
| 5.8 | 420 | 26.32 | 35.09 | 43.86 | 48.25 |
| 8.8 | 640 | 40.10 | 53.47 | 66.84 | 73.52 |
| 10.9 | 940 | 58.90 | 78.54 | 98.17 | 107.99 |
| 12.9 | 1100 | 68.93 | 91.90 | 114.88 | 126.37 |
Read across a row and you get a span: the lowest figure belongs to the best-lubricated joint, the highest to the worst. Read down and you get the effect of grade. The two directions are not the same size, and it is not close.
The rows do not nest
The worst-lubricated 8.8 — zinc plated, no oil — needs 73.52 N·m. A 10.9 on MoS₂ needs 58.90 N·m. The weaker bolt, in the worse condition, is tightened 25% harder, and it is still the weaker bolt: 640 MPa against 940.
Two classes up the margin thins rather than disappears: a 12.9 on MoS₂ needs 68.93 N·m against 66.84 for an 8.8 dry — 3% apart, with a yield strength 1.719 times higher.
So the figure on the wrench does not identify a class. It identifies a class and a friction condition together, and one reading cannot separate them. That is not a rounding effect — it is what the arithmetic does when the same variable sits in both comparisons.
The same number, two outcomes
The other direction is worse than a confusing chart. Set the wrench to the figure a generic table gives you — the dry one — and use it on a joint that is not dry:
| Real condition | K | Preload actually reached | Share of Rp0.2 |
|---|---|---|---|
| MoS₂ paste | 0.12 | 166.7% | 150.0% |
| Lightly oiled | 0.16 | 125.0% | 112.5% |
| Dry / as received (the assumption) | 0.20 | 100.0% | 90.0% |
| Zinc plated | 0.22 | 90.9% | 81.8% |
At the MoS₂ end the bolt reaches 166.7% of the intended preload, which is 150.0% of Rp0.2. The stress is past yield before the wrench stops turning: the bolt will not come back to length, and the next person to measure it will find something longer than the drawing says.
At the zinc end the joint reaches 90.9% of the target. Nothing breaks, nothing looks wrong, and the joint is 9.1% short of its design clamp load — which is where loosening starts.
What K is not
K is not a property of the bolt. It folds two friction paths into one number — the thread flanks and the bearing face under the head — and both depend on things the fastener does not carry: the lubricant, the plating, the washer, the roughness of whatever it is clamping. Change any of those and you have changed K without touching the bolt.
That is why the calculator exposes K as a control rather than baking it in. A table that prints one torque per size is not wrong; it is a table for one K, and it usually does not say which.
What this page does not tell you
- Your actual K. The values above are the generic conditions this database models. None of them was measured here, and a critical joint needs the torque calibrated against the real fastener, lubricant and bearing surface rather than read off a table, this one included.
- Temperature. Lubricant films change with temperature, and not one of these K values carries a temperature at all — they are used here as if every joint were at the same one.
- Reuse. A joint taken apart and retightened generally does not have the same friction the second time.
- Washers and flange heads. Both change the bearing face, and the bearing face is inside K.
- Loosening. Transverse vibration is a separate mechanism; a correctly preloaded joint can still come loose.
- Which class to use. This page compares what classes and friction conditions do to the number. It does not know your joint, your load or what happens when it fails.
The conditions are on the torque calculator, the yield strengths come from the property class reference, and the size worked through above is the M10 × 1.5 thread page. Choosing the class in the first place is its own article: how to choose a bolt property class.