A little perpetual motion device called the Feynman-Smoluchowski Rachet, also known as the Brownian Ratchet, was an early attempt at defying the laws of thermodynamics. It failed, but it was an entertaining failure.
Marian Smoluchowski was an experimental physicist in the late 1800s, which meant, unfortunately for him, he was leaving physics right at the dawn of its modern golden age. He still managed to get a lot of stuff done. He was one of the people who explained the color of the sky as the dispersion of light by the gases in the atmosphere. He discovered something called critical opalescence, which describes how some substances get beautifully cloudy as they approach a phase transition, and he attempted a perpetual motion machine.
This machine, called by all the different combinations of the names Smoluchowski, Feynman, and Brown, is named for its intellectual contributors. Smoluchowski invented the machine, which was meant to be powered by the motion discovered by Robert Brown. Brown noticed that, when he looked at grains of pollen in a solution, they jiggled around as if getting hit by invisible particles - that we now know to be atoms. Smoluchowski, after studying Brownian motion, came up with a device that might take advantage of it.
His device was a sort of horizontal windmill. A long stick would be suspended in a case. One end of the stick would have large fins, or paddles, attached to it, and would look like a weather vane. The other end would have a ratchet wheel with many small teeth in it. Holding that wheel in place would be a pawl - a little hook that would let the wheel turn one direction, but would slam back into place after the wheel turned to prevent it turning backwards.
Without the pawl, the random motion of the molecules of the solution should result in no work being done. The wheel would turn backwards as well as forwards. The addition of the pawl, Smoluchowski thought, might change this. It would prevent the wheel from turning back, and so useful work would come from random motion.
Even Smoluchowski knew he wasn't laying the groundwork for a thermodynamic revolution. The random motion of molecules in a lukewarm solution does not make for a great deal of power. When people talk about an object being lifted by the work of the wheel, they generally make the object a flea. Still, it was work from random motion. Until, half a century later, Richard Feynman tore it all down. He showed that, not only would the energy of the pawl clicking back into place push the wheel slightly backwards again, occasionally the wheel turned backwards while the pawl was still up. This would happen more and more often as the overall energy of the system increased.
The only way, Feynman discovered, to keep the wheel from turning backwards, would be to make the solution around the wheel significantly colder than the solution around the paddles. So the stick had to be suspended between two tanks full of solution - one hot tank and one cold tank. The end with the paddles had to be in the hot tank, getting a lot of energy to turn. Meanwhile, the pawl and wheel had to be in the cold tank, not getting jostled. In other words, T2 in the picture had to be much hotter than T1. This meant work - heating - had to be put into the device to get work out of it. So much for free energy. It was a beautiful dream.