[Walt Fricke]

I would not think a DAS bar (not cage) would do anything at all as far as twisting the chassis (or stiffening it, for that matter).

So I'd look elsewhere.

And given the hard pull you experienced, I would not think the weights are just a result of friction in the joints, so to speak.

Using the equal percentages approach to corner balance (ignores diagonals), you are 149 lbs out of perfection. Ideal would be
LF 498 RF 493
LR 776 RR 768

Because your car is so close to a 50/50 right/left distribution, this doesn't vary much from the equal diagonals approach, so as a practical matter you end up in the same place within the inevitable tolerance of corner balancing.

Unless these weights are without you in the car - which would explain the nice side to side split.

I'd be inclined to take Dad911's approach and get it within 10 or 15 lbs all around and go drive.

And then see where your ride heights are measured from some key suspension points: like a lower bolt where the rear spring plate attaches to the chassis, and the right and left sides of the reinforcing crossmember up front. And, for fun, from bolts holding the front of the A arm on. Doesn't matter what the absolute numbers are, just the side to side differences. That should tell you if the frame is twisted also.

I can see why you are puzzled. I had the front cut off my SC at about the shift lever, and a replacement welded on. I had left the rear suspension in the car for this. When I put the front suspension back on I just screwed the adjusters up equal amounts until the front was at a suitable height. To my amazement the corner balance was just about what it had been before the wreck, and I didn't have to change anything! Gave me a lot of confidence in that body shop's work.

If I had to guess it would be that something went wrong with setting the torsions in the rear - angles which seemed the same weren't somehow.

But, again, when I finally decided to put even bigger rear bars in the SC, measuring the "before" angles and the unloaded angles, calculating the reduction in angle appropriate for a bigger bar, and resetting to those calculated values worked out slick - the Wil Ferch method. Piece of cake, really (other than all the work generally). So no real reason for a huge problem here.

My 1.5 cents