Well, Capablanca had played ...Bb7 earlier, so he couldn't play ...Bxe6 anyway.

<Calli> posted this comment from Capablanca higher up the thread:

<Capablanca after 15.Nd4
"It is a curious but true fact that I did not see this move when I played 13...Bb7, otherwise I would have played the right move 13...Bxf4." >

Here are some comments from Capablanca and one from Tarrasch on this famous game. My source is Dale Brandreth's/Caissa Edition's reprint of Tarrasch's 1914 tournament book, translated with Robert Maxham, and with additional annotations from the contemporary press and later annotators.

I have three of Brandreth's reprints, Nuremberg 1896 (Tarrasch), Carlsbad 1907 (Marco and Schlechter), and St. Petersburg 1914 (Tarrasch). Each one is a treasure.

After 4. Bxc6

Tarrasch: "Why have you only chosen the Exchange Variation?" I asked Lasker in the noon break. "You must play sharply for the win." "I had nothing else," retorted Lasker. "I have found nothing against your defense, which you employed against Bernstein and me." The prospects for attack and defense have changed just so radically in a few years in the formerly so greatly feared Spanish Game.

<Here are the games Lasker is talking about. Lasker vs Tarrasch, 1914 O Bernstein vs Tarrasch, 1914 . Tarrasch, incidentally, was far too prone to this sort of sweeping conclusion. In his book about the 1908 match with Lasker, he wrote that the Open Defense was unsound. Now he thinks his innovation in the Open Defense means the Ruy Lopez is no good!

Brandreth quotes Reti's claim in <Masters of the Chessboard> that, as a rule, Lasker only adopted the Exchange Variation when his opponent wanted to play only for a draw. Like so much in <Masters of the Chessboard>, this is as false as grandma's eyelashes. Lasker played it against Tarrasch in the next to last round at Nuremberg 1896, when Tarrasch needed a win to have a chance of catching Lasker. Lasker also played it in matches with Steinitz and Janowski, neither of whom ever played for a draw, and played it in the first game of his 1908 match with Tarrasch. The only thing that can be said with any confidence is that Lasker didn't play the Exchange Variation very often, but did employ it in some important games.>

After 10. f4

Capablanca: This move I considered weak at the time and I do still. It leaves the e-pawn weak, unless it advances to e5, and makes it possible for Black to pin the knight by ...Bc5.

After 11....f6 <given a question mark by Tarrasch>

Capablanca: preparatory to ...b6, followed by ...c5 and ...Bb7 in conjunction with ...Ng6, which would make it extraordinarily hard for White to maintain the center pawns.

After 13....Bb7

Capablanca: Played against my better judgment. The right move was of course 13....Bxf4. Dr. Lasker gives the following variation: 13....Bxf4 14. Rxf4 c5 15. Rd1 Bb7 16. Rf2 Rad8 17. Rxd8 Rxd8 18. Rd2 Rxd2 19. Nxd2 and White, he claims, has the best of it. But as Nimzowitsch pointed out immediately after the game, 16....Rad8 is not the best. If 16...Rac8!, then White will have great difficulty in drawing the game, since there is no good way to stop Black from playing ...Nc6, followed by ...Ne5, threatening ...Nc4. And should White attempt to meet this maneuver by withdrawing the knight at b3; then the Black knight can go to d4, and the White pawn at e4 will be the object of attack. Taking Dr. Lasker's variation, however, whatever advantage there might be disappears at once if Black plays 19....Nc6, threatening ...Nb4 and also ...Nd4, neither of which can be stopped. If White answers 20. Nd5, ...Nd4 for Black will at least draw. In fact, after 19....Nc6, Black threatens so many things that it is difficult to see how White can prevent the loss of one or more pawns.

<Shredder observes that Capablanca gets a bit carried away here. After 13...Bxf4 14. Rxf4 c5 15. Rd1 Bb7 16. Rf2 Rad8 17. Rxd8 (17. Rfd2 is adequately met with 17....Rxd2 18. Rxd2 Nc6 19. Rd7 Rc8 and ...Ne5) 17....Rxd8 18. Rd2 Rxd2 19. Nxd2 Nc6 it's dead equal after, say, 20. Nd5 Nd4 21. Ne3. White is in no danger of losing anything. But Shredder agrees that Black has a small but real advantage after Nimzowitsch's 16....Rac8. Very surprising!

After 15. Nd4

Capablanca (as quoted previously): It is a curious but true fact that I did not see this move when I played 13....Bb7, otherwise I would have played the right move 13.....Bxf4.

After 15....Rad8 <again given a question mark by Tarrasch>

Capablanca: The game is yet far from lost, as against the entry of the knight, Black can later on play ...c5, followed by ...d5.

After 17....Nc8 <which Heidenfeld called the first real mistake of the game>

Capablanca: I was now on the point of playing 17....c5 to be followed by ...d5, which I thought would give me a draw, but suddenly I became ambitious and thought I could play the text move, 17....Nc8, and later on sacrifice the exchange for the knight on e6, winning a pawn for it, and leaving White's e-pawn still weaker. I intended to carry out this plan either before or after playing ...g5 as circumstances demanded. Now let us analyze: 17....c5; if 18. Nd5 Bxd5 19. exd5 b5; and a careful analysis will show that Black has nothing to fear. Black's plan in this case would be to work his knight around to e5 via c8, b6 and c4 or d7. Again, 17....c5 18. Rf2 d5 19. exd5 Bxd5 20. Nxd5 (best, since if Rfd2, ...Bxe6 gives Black the advantage) 20....Rxd5 21. Rxd5 Nxd5, and there is no good reason why Black should lose.

<Shredder finds a subtle improvement in Capa's second line: 17....c5 18. Rf3, and if 18....d5 then 19. Rg3 forces concessions. Since ...d5 remains unplayable for a time, White keeps the advantage.>

After 21....Ba8

Capablanca: Once more changing my plan and this time without any good reason. Had I now played 21....Rxe6 22. fxe6+ Rxe6 as I had intended to do when I went back with the knight to c8, I doubt very much if White would have been able to win the game. At least it would have been extremely difficult.

After 26....Rae7

Capablanca: This, of course, has no object now. Black, with a bad game, flounders around for a move. It would have been better to play ...Ra3 to keep the open file, and at the same time threaten to come out with the knight at b6 and c4.

After 28....g6

Capablanca: Again bad. White's last moves were weak, since the white King does noting here. He should have played his rook to g3 on the 27th move. Black should now have played 28....g5+. After missing this chance White has it all his own way, and finishes the game most accurately, and Black becomes more helpless with each move. The game needs no further comment, excepting that my play throughout was of an altogether irresolute character. When a plan is made, it must be carried out if at all possible. Regarding the play of White, I consider his 10th and 12th moves were very weak; he played well after that up to his 27th move, which was bad, as well as his 28th move. The rest of his play was good, probably perfect.