I've been reading K.E. Tsiolkovsky's actual 1903 paper (not later revisions).
First (cue violins) let me say what a pain it has been to find and read that work. I found the Russian version, but writing at the turn of the century, he used a bunch of old-slavonic letters that nobody uses today, as well as some old-fashion spellings involving a lot more hard signs and soft signs than modern Russian. This is bad when you are using OCR and computer language translation, but I massaged his antique Russion into mostly modern form.
So, Konstantin Eduardovich considered the heat of combustion and mass of various fuels and concluded that liquid oxygen and liquid hyrogen was the way to make a rocket. He derived the "rocket equation" and then calculated the idea exhaust velocity for LOX/LH2, by converting the heat of combustion into kinetic energy, he came up with 5700 meters/sec for the specific impulse.
I looked up current values for heat of combustion of hydrogen (141.9 megajoules per kilogram). Since it takes 8 kilograms of oxygen to burn a kilogram of hydrogen, I divided by (1+8) and get an equivalent kinetic energy of 5615 m/s, so he seems to have done the math right (I assume he just used an older and slightly different heat-of-combustion value).
The SSME is a closed cycle engine of great efficiency, and it's exhaust velocity is 4500 m/s. I guess we should be impressed that it comes that close to 100 percent efficiency of converting chemical energy into exhaust kinetic energy! I get 64% efficiency.
Oddly enough, Goddard claimed to have measured 64 percent efficiency in 1915, working with black powder engines and de Laval nozzles. I wonder if that was really possible?
Oh Tsiolkovsky mentions that he ignored the extra energy needed to account for the coldness of the liquified gases. I ignored that too.