How the rate is calculated
A U-shaped anti-roll bar does two things at once when one wheel rises relative to the other: the straight centre section twists, and the arms bend like little cantilevers. Both give, so both belong in the answer:
1 ÷ k = (A² × L) ÷ (G × J) + A³ ÷ (3 × E × I)
where A is arm length, L the torsion section, J = π(D⁴ − d⁴) ÷ 32 the polar second moment, I = J ÷ 2 the bending one, G the shear modulus of steel (79.3 GPa) and E its Young's modulus (200 GPa). The first term is the twisting, the second the arm bending — leave the second one out and you overstate a long-armed bar by 10% or more.
That gives the rate at the end of the arm. The wheel is somewhere else, so:
wheel rate = k × motion ratio²
The ratio is squared for the same reason it is with a spring: the link moves less than the wheel and reacts through a shorter lever. A bar on a 0.7 motion ratio delivers 49% of its rate to the wheel, not 70%.
Why 2 mm changes everything
Stiffness follows D⁴. Going from 22 mm to 24 mm is a 9% increase in diameter and a 42% increase in rate. Going from 22 to 26 nearly doubles it. This is the single most useful fact about bars, and the reason "I'll just go one size up" so often ends with a car that skates over mid-corner bumps.
It is also why adjustable bars adjust the arm, not the diameter. Moving the drop-link to a hole 30 mm further out on a 200 mm arm drops the rate by roughly a quarter — a usable, reversible step, where the next bar diameter up is a 40% jump you cannot undo at the track.
Hollow bars are not a compromise
Take a 26 mm bar with a 4 mm wall. Its stiffness is proportional to 26⁴ − 18⁴, which is 352,000 against a solid 22 mm bar's 234,000 — it is 50% stiffer. Its mass goes with the cross-section area instead, 26² − 18² against 22², so it is also 27% lighter.
The material near the centre of a solid bar is barely stressed in torsion; removing it costs stiffness slowly and saves weight quickly. Every modern OE bar is hollow for this reason, and it is unsprung-adjacent weight on most layouts, so the saving is worth having.
What a bar actually does to the car
An anti-roll bar adds roll stiffness at one end of the car without adding ride stiffness in parallel bump — hit a bump with both wheels and the bar does nothing at all. That is its whole value: control roll without ruining the ride.
What it changes is the share of lateral load transfer each axle takes. Stiffen the front and the front axle transfers more load, its outside tyre saturates earlier, and the car understeers more. Stiffen the rear and it rotates more. The balance is set by the ratio between the ends, not by the absolute numbers — which is why roll stiffness in N·m/degree is the figure to compare front against rear, and why a "stiffer" bar on both ends changes balance far less than people expect.
| Change | Effect on that axle | Effect on balance |
|---|---|---|
| Stiffer front bar | More load transfer across the front | More understeer |
| Stiffer rear bar | More load transfer across the rear | More oversteer / rotation |
| Softer bar, both ends | More roll, more mechanical grip | Little balance change, slower response |
| Link moved outboard | Longer lever, softer rate | Same direction as a smaller bar |
The costs nobody mentions
A bar ties the two wheels together, so a single-wheel bump is partly transmitted to the other side — a very stiff bar lifts the inside wheel over kerbs and crests, and a lifted wheel carries no load at all. On a bumpy road or a rough track the fastest setup is usually softer bars and more spring.
Bars also load diagonally on an uneven surface, which is why corner weighting is done with the drop-links disconnected — leave them connected and you are measuring the bar's preload, not the car's weight distribution. Reconnect them at ride height with the car settled, or you build a permanent cross-weight error into the setup.
Finally, the bar is only one part of the roll-stiffness picture. Springs and their ride frequencies do most of the work; the bar trims the balance on top. If the springs are wrong, no bar will fix it.
FAQ
How do I calculate anti-roll bar stiffness?
Combine the twist of the centre section with the bending of the arms: 1/k = (arm squared times torsion length) divided by (shear modulus times polar moment), plus arm cubed divided by (3 times Young modulus times second moment). Then multiply by the motion ratio squared to get the rate at the wheel.
How much stiffer is a bar 2 mm bigger?
About 42% for a solid bar going from 22 to 24 mm, because stiffness follows the fourth power of diameter. Every millimetre is worth far more than it looks, which is why adjustable bars change the arm length instead.
Are hollow anti-roll bars weaker than solid ones?
Not for the same outside diameter and sensible wall thickness. Stiffness depends on the outside diameter to the fourth power minus the inside diameter to the fourth, and the material near the centre contributes very little in torsion. A hollow bar can be stiffer than a smaller solid one at the same weight.
Does a stiffer front bar cause understeer?
Yes, on its own. Stiffening one axle makes it take a larger share of lateral load transfer, so that end loses grip first. A stiffer front bar adds understeer, a stiffer rear bar adds rotation, and what matters is the ratio between the two ends rather than either number alone.
Should the drop links be connected when corner weighting?
No. Disconnect them so the bar is not preloading the corners, set the weights, then reconnect at ride height with the car settled. Corner weighting with the bars connected measures the bars as much as the car.