Now forgive me if this seems pretty esoteric: it probably is. In fact, I’d never really even thought about it until a year or so ago; I’d never actually experienced it until today.

Most of you would be familiar with the idea of ‘toe’. Toe-in is where the wheels point inwards – when viewed from above, they’re constantly steering towards the centreline of the car. Toe-out, as you’d soon guess, is where the wheels are constantly steering outwards from the centreline. Zero toe means the wheels are parallel to the centre line.

Most cars these days run zero toe or just a very small amount of toe-in. Toe, usually measured in millimetres (although degrees would make far more sense), is at most only 1 or 2mm: the amount the wheels steer inwards or outwards is very small indeed.

OK – so that’s static toe. But what about when the suspension moves up and down?

If, during suspension travel, the wheels stay steering exactly in the directions they were originally steering in, the suspension is said to have zero bump steer. If the wheels steer inwards on bump, they’re said to have toe-in on bump. Toe-out on bump is defined as you’d expect it to be. (Note that in all these quoted cases, the steering wheel is held still – it’s the suspension itself that’s doing [or not doing] the steering.)

Now it would seem kinda obvious that you shouldn’t have any toe change in bump (or rebound): who wants to have the wheels steering when the suspension is just moving up and down? However, it gets more complex than this. Especially on custom and race cars, the amount of bump steer can be varied by altering the height of the outer or inner steering tie-rod ends.

So you can have at will zero bump steer, toe-out on bump or toe-in on bump.

The first point to be aware of is that on real world bumps, having toe-in, toe-out or zero bump steer doesn’t make a lot of difference to straight-line stability. (Or not that I am aware of anyway.) Instead, it’s the bump caused by roll, where the outer, more heavily laden wheel compresses its suspension, that’s important.

Take a car that has toe-out on bump. The driver turns-in and the initially vehicle rolls in rough proportion to the lateral acceleration. Because it is toe-ing out, the outer wheel doesn’t turn as much as the driver has requested. The response requires that the driver therefore feed in more steering.

Now take a car that has toe-in on bump. The driver turns in and because the outer wheel is toe-ing in, it turns more than the driver has requested. The response requires that the driver therefore unwinds the steering.

At least one textbook suggests that the former is better: it’s a more natural action to add lock rather than subtract it mid-corner.

But surely it would be better to have neither toe-in or toe-out on bump?

Not necessarily! What if the car is a natural understeerer? In that case, a system that automatically adds more steering to the most heavily loaded wheel will reduce the apparent understeer: the car will track more truly.

And the opposite example: what if the car doesn’t have much castor (steering self-centreing caused by the steering axis being inclined)? In that case, the steering will be light and lock easily applied. In fact, a car that applies even more steering as the steering wheel is turned could well feel nervous and darty. Steering input is easy for the driver – and then the suspension adds even more!

But then again, the amount of any bump steer depends on the roll stiffness – because a car very stiff in roll won’t compress the outer suspension so much, bump steer won’t be occurring in the same magnitude.

I don’t have a race car in my garage and I have never experimented with changing bump steer on a production car. But what I do have is an ultra lightweight, home-constructed three wheeled human powered vehicle. It’s got airbag suspension, deliberately designed-in toe-in under bump, an anti-roll bar and very quick steering.

And this morning, on amongst the first rides, the steering was so twitchy that at speeds of over 40 km/h, I thought I was going to have a very big lose. Like, crash and roll. That might sound like a bit of a joke, but plunging downhill in a t-shirt and jeans, only additional protection a foam crash helmet, the machine twitching all over the road – it isn’t fun. It’s bloody frightening.

For reasons I won’t go into here, the steering on one of these vehicles needs to be very quick in ratio: steering stability needs to come from castor and other suspension geometry. The castor angles I was using (‘angles’, because on this design they change with suspension movement!) were well proven. Bump steer was a little toe-in: better I thought to reduce the inherent understeer of this type of vehicle.

But gawd was the steering twitchy!

Then a guru in the field (Ian Sims) suggested that bump steer toe-in could give the affect I was experiencing. What about, he said, setting up the steering for toe-out on bump? I relocated the inner tie rod joints downwards about 8mm, a move that gave distinctive toe-out on bump. And the results? An immediate and dramatic improvement in steering. Enough change, in fact, to go from a vehicle I didn’t think I could control at 60 km/h, to one where while a sensitive pair of hands is certainly needed, you don’t feel you’re about to barrel roll off the road at any moment.

This bump-steer stuff is certainly very interesting…