That cone was a pointer of some kind, I think ... I only hit 1 cone on my first run (none on any other runs ...) and it wasn't that one, it was just a little later in the turn in the picture (a bit too much speed with a tight turn into an off-camber finish) /ubbthreads/images/graemlins/smile.gif
As to weight transfer, Dave, I think you're assuming the car will lift the other wheel immediately upon addition of a rear bar. However, if the rear suspension allows too much roll, while the front allows significantly less, it will actually transfer weight to the rear (supposedly). This is what I suspect is happening. It is entirely possible that my rear shocks are too stiff, but I don't think that's the only answer. While softer shocks may reduce breakaway, they certainly arent going to put any meaningful weight on the front inside tire, even if the tire was actually on the ground. it would simply let the tire follow the ground a bit longer, with no load on it (meaning it doesn't actually provide any cornering force, despite the fact that it's touching) If the vehicle's center of mass remains closer to the centerline of the track, it will tend to transfer weight to the inside of the car. This indicates that instead of rolling over (picking up inside tires), it will tend to slide. Of course, this all depends on how much cornering the outside wheels provide, etc (and assumes the vehicle behaves as a block, not suspended by springs etc). However, this assumption is increasingly valid as we stiffen the roll-bars because of the reduced body roll. While the handling may indeed be strange, as Nial said, it will put the inside front tire down with a load on it, and will likely provide better grip right up until the car starts to lift both wheels at the same time.
Needless to say, I intend to test these theories extensively, hence why I would like to experiment with a mumford link (fully adjustable of course) so that I can change roll centers, rear bar stiffness, and even shock stiffness (with adjustable rear shocks) and compare times on the skidpad. That won't tell me how it behaves under more transient conditions, but it will tell me how many g's the car can pull in steady state, which would be a good start. I can then use that information to adjust for transient conditions to make the car more predictable -- i.e. less twitchy.