The first one was the aelopile, of course, but the plan stuck, fairly rigorously, to the Victorian era, when, of course, most of the fun stuff was done. Watt's big improvement, instead of having the cylinder heat and cool off -- bad for metal -- was to have a separate condenser to cool it, so the metal wasn't stress. Also, of course, it shrinks when cooling. One panelist had a hot plate and a metal can which he brought to a boil, turned off the heat, and poured cold water on. This created a vacuum that crunched it all up. This can be used as well as the force of rising steam. (Another panelist pointed out that if the demonstrator had left it on the hot plate, he would have run away, because it would then be a bomb.)
The Carnot cycle, which works out how engines work, independent of what the particular stuff is. One panelist, later, was talking about how it really is individual molecules striking but you can treat them as a fluid, and I pointed out we know that about the molecules because of Einstein's work, at the time they didn't know whether they were such things of finite size, and another panelist explained that the Carnot cycle didn't rely on stuff being made of atoms. Indeed, one scientist theorized one made of nothingness.
Superheated steam can get more energy; you have to boil all the water first, but you can get it hotter after that. It does, however, create problems with pressure.
Trains used up more water than coal. Then, few trains had condensers and stuff to recover water. Therw was one in South Africa that did. Something about the climate.