That gap. It doesn’t look so dangerous, but once I installed phosphor-bronze wipers to pick up current, the fine wire would find the tiny gap between rail heads and launch poor under-weight Percy into the sky. Sometimes with disastrous effect.
Finally, with Percy’s split chassis assembled, and a few sharp corners filed away so they wouldn’t pick the shorting strips on the backs of the drivers, I could pull those wipers off the rail heads and roll across the gap. For the first time in months, Percy ventured into the southern end of the test track.
One interesting lesson along the way involved the keeper plate. I had thoughtlessly believed that this played the simple role of stopping the drivers from falling out when Percy was lifted from the rails. It seems obvious now, but because Percy doesn’t have a gearbox, the keeper plate also keeps the gears in mesh. If a lapse in care results in amputating one of its arms, Percy balks when asked to reverse. Perplexingly, when that happens, Percy still runs forward because the gear naturally pulls the wheel up into the frame.
Pembroke:87 
Your note leads me to ask: when all of this thorough experimenting is concluded, do you think the insights learned from Percy will translate naturally to the two Canada Atlantic locomotives, (or even to other steam models)? Or is a good portion of the learning more about the approach and methods for discovering issues, and that approach is what will have general application?
Interesting question, Rob. I had been counting on applying the new technique directly to the two CA locomotives. However, the key thing I’ve learned is where to look for problems. I feel that this approach with Percy has grown my understanding of steam locomotive models from neophyte to journeyman in a very short time.