“does a longer train of empties require more HP than a short train of loads of the same trailing weight”
Same total weight on more axles, with more air resistance – RRs have always known a given tonnage of loaded cars roll easier than the same tonnage of empties, on curvy rail or straight. That’s why many RRs used “adjusted tonnage” ratings. Look at the back page of this IC timetable
A 4-8-2 was rated at 11325 adjusted tons northward to Chicago, meaning 9825 actual tons in 100 cars, or 8325 actual tons in 200 cars, or 6825 actual tons in 300 cars. But apparently Gilman to Clinton has a steeper grade, so a lower rating, and also a lower adjustment for empty cars, since the steeper the grade the more important actual weight is.
Note that IC claimed some rebuilt 2100-class 2-8-2s could take a 120-car 10475-ton coal train north to Chicago, which includes a 0.3% climb from Kankakee. Goes to show you can’t always believe tonnage tables.