Originally posted by Shallow Blue
All twenty-three of them.
It seems from Greenpawn's blog that some people have taken this remark to be based on actual research. It wasn't - it was just a throwaway quip about how few games Morphy is know to have actually played. I should've used 42 instead of 23 - that would have made it more obvious. (And closer to the truth - he actually played 59 real games.)
However. Due to this misunderstanding, I have now made the effort to find the real numbers. This wasn't as much of a job as it sounds like, because Tim Krabbé has providentially put all of them, as far as known, in PGN files. There's one containing all known Morphy games including blind simul demos, odds games, and whatnot; and one containing only the serious ones. And with a PGN, it's as simple as typing Ctrl-F/o-o-o/Again.
So here are the numbers.
In all his known games, Morphy himself castled queenside twenty-seven (close!) times. Twelve were in odds games, one in a blind game. Twice he castled with check (once in the blind game!), and one time it was the last, winning move of the game. He won sixteen of these, lost five and drew one. Six were in official games, of which he lost one and won the rest.
Contrariwise, only twenty times did someone commit the long castle against Morphy. Five of these were in an odds game and also five in blind games (including one against Paulsen, also playing blind, which ended in a draw). Of these he won twelve, lost five, and drew three. Two were in official games, which he both won.
And then there's the interesting one. In a Paris blind simul, against a certain Bornemann,
both players castled queenside. It didn't save Bornemann; Morphy won.
I have not bothered to break all this down between black and white. Someone even more geeky that I can do that - the PGNs are available from http://timkr.home.xs4all.nl/ChessTutor/morphy.htm .
Richard