<Although this pretty move led to a win, Capablanca commented objectively that the better move , overlooked by all the analysts, was 26 Qg7-f6, whereby the King is provided with a flight-square at g7. This would have spared him some trouble.

<It does open the bishop; however, the bishop is not used for anything in the next few moves. > If you wanna castle your gonna have to bring the bishop out.

Edward Winter on Evans' "The 10 Most Common Chess Mistakes … and how to avoid them!":

click for larger view

12. Rfe1 dxc4 allows Marshall to maneuver his Queen to the 4th rank. 13. BXc4 b6 14. Qe4.... 18...Rbc8

click for larger view

Given his central and Kingside superiority Marshall embarks on a plan that looks quite reasonable, and which I believe many chessplayers would also take, to attack the Kingside beginnign with a rook lift 19. Re3.

Capablanca reacts vigorously, first trying to exchange off white's dangerous rook on the 3rd rank, 19... cxd4 20. cxd4 Rc3 21. Bb1

click for larger view

Then Capa plays the cold-blooded 21... g5, inviting the subsequent knight sacrifice by Marshall 22. Nxg5. Not many chessplayers would play such a move, as it leads to hair-rising complications and hanging pieces. Capa before playing this move must have seen that the line 26. Rxd4 leads to a win.

click for larger view

The reason why he played 26. Rxd4 instead of 26...Qf6

<Although this pretty move led to a win, Capablanca commented objectively that the better move , overlooked by all the analysts, was 26 Qg7-f6, whereby the King is provided with a flight-square at g7. This would have spared him some trouble.

'I was highly praised,' says he, 'because of the excellence of my play in this position, while in reality I could have done better.'>

is probably because he had already seen that this <pretty> move should win as well, even several moves back. Otherwise he probably would not have gone into the line leading into this position at all, beginning with 21...g5. For some psychological reason, once a clear way to an advantage has been calculated, most players just tend to follow their original calculated line; why waste time calculating other winning lines? If so, Capa must have preconceived the remarkable position after 28. e5+ even several moves back.

click for larger view

Black's King is under discovered check, while at the same time his Bishop and Rook are hanging. Capa must have seen that the resource 28... Be4 not only blocks the discovered check, but also removes one of the hanging pieces and neutralizes white's dangerous bishop.

click for larger view

click for larger view

Capablanca must have seen that his advantage should be enough to win, and probably already had a plan of how to go about it even during the middle game. A classic case of Fischer's hypothesis.

<"Capablanca was among the greatest of chess players, but not because of his endgame. His trick was to keep the openings simple, and then play with such brilliance in the middle game that the game was decided - even though his opponent didn't always know it - before he arrived at the ending." - Fischer>

In the above position, many players no doubt would have gone after the White outside passed a-pawn, and thus leave their exposed King to a perpetual check, something that Marshal must have counted upon. Unfortunately for Marshall, Capablanca carries out the only plan that could even conceivably win the game. He mostly leaves the passed pawn alone and instead goes for a direct attack on the White King with his Queen and Bishop, while shuffling his own King between h5 and g6 to avoid successive checks and so gain tempo for the attack; and he places his Queen in squares where it can ward of checks by the White Queen but still be available to attack White's King.

During the final Q and B attack Capa had to foresee the following remarkable position.

click for larger view

White is about to queen his passed pawn, which Capa had left alone to march up the a-file. However, now comes 52... Bxf3

click for larger view

Suddenly mate is threatened, and poor Marshall remarkably has no checks. With the same two pieces that he is threatening mate with, Capa economically also defends all squares in which Marshall can check the Black King (a6, c6, e8).

<With the same two pieces that he is threatening mate with, Capa economically also defends all squares in which Marshall can check the Black King (a6, c6, e8).>

click for larger view

Marshall can deliver a check with his Queen anyway on a6, c6, or e8 in order to draw away the Black Queen from mating his King, and then think of promoting his a7 pawn to a Queen.

But it is not to be. Astonishingly, however Capa captures, the a8 square is defended!

53. Qa6+ Qxa6

53. Qe8+ Qxe8

53. Qc6+ Bxc6

Everything falls in place with computer-like precision. And there is no doubt that Capablanca had seen all these even several moves before.

Twice that bishop goes to e4 and causes White nothing but pain.

One nice point after the brilliant 28...Be4 is that Marshall couldn't regain his piece with 29.Be4+ Re4 30.f3 Re2 31.fg Qg4 when Qb7 or Qa8 both fail after another little tactic, 32...Rg2+.

click for larger view

Lakadawala (in the book cited in my previous post, at p. 83) refrains from the comment I probably could not have resisted: "Will he see it?"

@Boomie - After 21...g5 22. Qg3 Lakdawala gives 22...Rxe3 23. fxe3 Nxe4 and white has insufficient compensation for the pawn deficit. Also, 21...e5 is not a shot that Capa missed; it is a mistake he avoided (22. Qxe5 Qxe5 23. Nxe5 and white is up a safe pawn)