There's a long thread in m.t.r.a about the catenary network used on Amtrak's NE Corridor and elsewhere. [See the ref's if you want to start at the beginning..]
For historic [i.e. even older than Jim!] reasons, it's 11KV@25 Hz. over much of the distance. Amtrak is awaiting money to make it all 25KV@60Hz, and expects to do so about the time cold fusion power plants come on line in Fedwick. Until then....
But someone points out it is really 50KV center-tapped. One leg is the cat itself; the other a feeder in parallel. Periodically, a center-tapped autotransformer mounted track-side goes feeder-rails-cat. Railroad folks call this 2x25KV, I gather.
See a similar spec:
The claimed advantages are:
a) less droop under load, no small matter when an Acela draws about 10 MW.
b) EMI cancellation since the feeder and parallel cat are out of phase.
c) You can tap up on the auto-formers further from a feed.
On a) I'm trying to model in my head why it is better than just paralleling the feeder with the cat every few 100 meters. I can't see an advantage.
b) I'm dubious on but will let go.
c) bothers me; if T{mile 0} is tapped at 25 0 25 [feeder rail cat], and T{mile2} is tapped so as to provide more voltage there [25 0 26]; what happens at no load? I think the two transformers will fight it out, heating the 2 miles of cat resistance and driving up the bill.
Insight welcome...
ps: Which Jim is the question...