Hermann
Only in Summer 1945, the first derailment took place in Pittsburgh. Then, just one month after delivery of the first serial T1s, 5502 derailed on December 1, 1945, and another serial T1 the very next day. Did somehow, tragically, the two prototypes have the edge over the 5500s in curves?
If I remember correctly, the problem was caused at only one switch location; I dimly and perhaps imperfectly remember it as a double slip switch. It is possible that the 'problem' was created by track maintenance of some kind, but I think the situation is more likely what Hermann notes.
Thw two original engines had radically different equalization arrangements from the 'production' engines -- for, I think, very good objective reasons. It's been known since around the turn of the century that equalization of eight-drivered locomotives is better if the equalization is interrupted or 'tied' between the second and third driver pair, with the lead truck equalized with the forward sets and the trailing truck with the rear ones. This, with some progressive improvement in snubbing and auxiliary springing, is what was done with the production engines right up to 1948. The original design had a prominent walking beam carrying the equalization from front to rear engine, and this was thought to be causing some of the riding and slip issues -- I have not read the surviving correspondence and source material on this, but it is very clear that substantial changes were made to the arrangement during the war-years testing.
Now, something that is interesting when you look at the T1 spring-rigging arrangements is that a great amount of tinkering went on with the permitted lateral of the driver pairs, sometimes on the order of sixteenths of an inch -- there is a long list of emended numbers, perhaps notable in that by late 1947 there was free lateral on all four driver pairs. I find it highly suspicious that this would be done on a high-speed locomotive with lateral-motion devices unless it were intended to permit a certain amount of float, at low speed, over the kind of obstacle resulting in derailment in terminal trackage. (Note that this is very different from the reported issues with T1s losing critical adhesion over frogs and low joints when pulling long consists out of station traffic.)
Now, it would be one thing if Glaze-style stiff lateral compensation on lead and trailing trucks were applied to T1s to keep their required overbalance low (or, in fact, zero, which is a condition likely necessary for T1s to reach the speed their steam generation and valve gear would allow). That never, apparently, became the case; the assumption seems to have been that the long effective rigid wheelbase and duplex kinetics would keep nosing/hunting minimized. Interestingly the very large tenders might likely have minimized in-phase surge, too -- so a certain amount of overbalance might have been tolerated that 'shouldn't have been'. (The effect of increasing augment on propensity to high-speed slip should be no mystery to anyone following along with T1 history!)
There are stories from the Crestline history that describe how much of a penchant 6100 had for derailing (usually somewhere in the driver wheelbase, if I remember correctly) while being run around that servicing point. To my knowledge that situation was never satisfactorily worked out, even by restricting the engine to particular tracks and stalls. It would be interesting to examine experiments, if any, with lateral motion on the driver pairs to see what did and didn't help with that.