First off, I think it is instructive (as is often the case with endgames of the great players) how A. lets some pawns go (such as e4 and c4 on moves 27 and 28) in the interests of concentrating on what's really important in the position - white's 'h' pawn, assisted by the 'g' pawn.

I suppose the principle is similar to that in a pure Q ending - given how powerful the queen is, the speed of passed pawns is more important than the number of them. Here, the 'h' pawn ties down the black Q, so while one of the rooks supports the pawn, the other is free to invade the black position.

35.Re6! sets up 36.g5, pulling the BQ out of its blockading position.

All the way through, of course, Alekhine avoids the greatest danger in these types of positions, that of losing a rook to a Q-fork.

52.Rh4+ stops the BQ checking on c4 once the 'h' pawn queens.

Overall, the ending is an important illustration of one aspect of the 2R v Q debate - it might be easier to handle the queen than the rooks, but if the rooks have a passed pawn that can tie down the queen, they can overrun the enemy position. It helps if you are Alekhine, too, of course.

click for larger view

<51.Rc8> throws away the win after <51...Qxc8 52.Rh4+ Kf3! 53.h8Q Qe6+>. I don't know who discovered this first; it seems to have escaped contemporary notice. I do have a collection of Alekhine's games from around 2000 that mentions it.

Just to ally <marljivi>'s fears, the facts are as presented here. The score is correct, Alekhine was White and he won the game.