In the new issue of MRH Joe Fugate talks about operating long trains with helpers. It so happens that I just finished setting up a train with a steam puller and pusher to demonstrate on my HO layout at the Portland NMRA convention. I have a fairly long (30 ft) 2.2 % grade that I just completed that leads into Cajon Pass Summit. It has curves of 48, 60, and 70 inches. I wanted a train of at least 30 cars and to use BLI 2-10-2's. Following advice from Andy Sperandeo (former Model Railroader editor and Cajon Pass modeler) I found out how many cars the lead engine could pull up the grade and how many the trailing engine could push, by adding cars until I got wheel slip. The number was about 22 for each engine. I knew I wanted to run less cars than that and I decided to let each engine handle five (about 23%) fewer cars, so my train was 35 cars. I was nervous the first time I ran the train (my track in that area is about 5 1/2 ft. off the concrete floor!) but I was amazed at how well it worked, so much so that I started analyzing the dynamics, since this has been my career for forty years. I quickly realized that the system is dynamically stable due to inherent negative feedback. Simply put, if one locomotive either speeds up or slows down, instead of the train string lining or jacknifing,  the other locomotive will automatically do the same and the train continues up the hill in a smooth motion, either a little faster or slower than before. For example, if you open the throttle on the lead engine on a it will speed up slightly, but since the trailing engine is now working on fewer cars it also will speed up, and the possibility of string lining, if on a curve, is greatly reduced. The same thing happens if you open or close the throttle on either engine. Note, however, this only happens on an uphill grade. Downhill or level operation is another story. Of course the same laws of physics apply to the prototype.