I played with the start of a Northeast scenario to find the best Grasshopper engine loadout versus revenues. Setup details: Financier level, no opponents, starting at Harrisburg, connecting to York. The only resources then are Passengers (P) or Mail (M), of course.
After playing with a bunch of possibilities (lone P or M car, just enough cars to get all available stuff, or even longer trains), the surprising result was:
The best money to be had is via an 8-car train (e.g. 4 P and 4 M cars). Also, build up the York terminal a.s.a.p. (You can make a Station immediately, then make it a terminal with first delivery there.)
When doing this, it is important to clip the railroad track on the far side of your terminals (the far side relative to the other terminal), just past the terminal. (Make a tiny turnoff there, then remove both the turnoff and any excess straight track.) This causes your 8-car train to stop right at the terminal, without taking the time to run way past it so that the middle of the train is at the terminal.
But even more importantly, when the 8-car train turns around on track clipped at the terminal, it's already halfway to the other terminal, in the hinky game world of Railroads.
Clipping terminal tracks might seem cheesy to some folks. Don't do it if you don't want, shrug.
Clipping them makes a much bigger difference when your start cities are close together, like Harrisburg and York. But other info posted here shows that speed and not distance is much more important early in the game... so woe on you if your starting city is pretty far from any other Passenger and Mail city!
In case you didn't know, empty cars don't weigh anything at all. (Not ever.) The total weakling Grasshopper with eight empty Goods cars makes the York/Harrisburg roundtrip in exactly the same time as it does without a single car on it (and reaches the same max speeds, including 29 mph up the slight incline to Harrisburg). This is with your terminal track clipped, of course - otherwise it would have taken longer, simply because the middle of the 8-car train would need to reach the terminal, etc.
Likewise, partial cars (only half full with passengers or whatever) only count for half their weight. Therefore, the (partially) empty cars of an 8-car train don't slow it down one bit - all they do is make it that much closer to the other terminal. Also, of course, all the extra cars guarantee you'll pick up all paying resources.
Some prices comparisons... they all use the Total Revenue from a Grasshopper making ping-pongs from Harrisburg to York; Total Revenue counted as of Sept. 1832 (after several ping-pongs):
1 P car only: $251k
2 P & 1 M: $303k
4 P & 4 M: $520k
Some observations for the larger game might be made from this, combined with other info that I and others have posted:
1) For best results, make your P & M trains be 8-car trains doing short ping pongs between clipped terminals. Use tracks custom made for that train.
2) Only P & M seem to have speed & distance bonuses. Thus, terminal clipping is less important for all other resources. But still it can get other consists out of the way quicker, for both the reasons stated above (they don't run halfway past the terminal before unloading, and are already most of the way out of the terminal when they turn around). Signals alone right past the terminal may or may not help, if you can't actually clip the track... I haven't tested this yet.
There are two places that long trains can get you in trouble:
1) If terminal tracks are not clipped somehow, trains will use more "terminal time", as stated above.
2) If your railroad network has long trains crossing each other, they can block and ultimately snarl it. But many folks recommend trying to keep trains on their own personal track, without any possibility of crossing to other tracks and blocking other trains. (One important exception is where two different trains might come at a given terminal on the same track, but from different directions. Then they're not actually crossing each other, they're just sometimes waiting for the other train to use the terminal.) Sometimes it takes hard work to "force" a train to the track you want, such as temporarily making connections then removing them and/or starting the train off on some spur that it never uses again.
Has anyone seen situations where any of the above thoughts might be a bad idea?