In the tournament, Torre himself always preferred this capture over 4....Nc6 transposing to the Two Knights Defense.
Both Torre and Marshall recognized the superiority of this move over the traditional 6.Bg5?! which allows Black to play for the center with 6....Nc6 7.Qh4 d5! -- a line that Torre used to advantage at least twice in the tournament (though only one of those games survives). According to their notes in the American Chess Bulletin, it was the New York chess writer and analyst C. S. Howell who first recognized the importance of 6.Nc3.
Now if Black tries 7....d5 White retains a clear though slight advantage after 8.Nxd5 Nxd5 9.Qxd8+ Nxd8 10.Bxd5 Be6 11.Be4. In his game with Marshall, Torre took advantage of the Knight's early development to try 7....Bb4!? 8.O-O Bxc3 (Black will need to make this move eventually, since after White plays Bg5 he will threaten to pile up on the pinned Knight at f6 with Nd5!) 9.bxc3 O-O. Marshall should then have demonstrated the power of the extra Bishop by playing immediately 10.Bg5! with good attacking chances. By playing his Bishop to e7, Black tries to nullify the pin by Bg5.
This is one of the most important positions for the Urusov Gambit. White has developed his pieces quickly toward the center, even castling long to make the open central files immediately available for his two Rooks. His pieces are placed for maximal effectiveness in attack, with his Bishop at c4 directing pressure on the always vulnerable f7 square. Black, meanwhile, plays his Bishop to e6 with the idea of exchanging or driving off White's powerful white-squared Bishop. The Bishop at e6 also places an extra piece on the open e-file leading to help blunt White's attacks with Rhe1. The position is dynamically balanced, and many contests from this position eventually end in a draw.
There are two standard moves in this position: 10.Bd3 to discourage Black from castling on the Kingside and 10.Rhe1 allowing the exchange of Bishops in return for complete mobilization. Also quite playable is 10.Bxe3 fxe3 11.Rhe1, since Black's extra central pawns are subject to attack, but this is much more rarely played. In 1924 the theory of this line was still in its infancy so not much was known about the tricks that can follow 10.Bd3 Qd7 11.Bb5! (exploiting the pin) 11....O-O (taking advantage of the chance to castle out of the center) 12.Ne5!? when the 1924 British Correspondence game between Griffith and MacDonald ended quickly with 12....Qe8 (it has since been discovered that Black does best to return the pawn with 12....Qc8) 13.Nxc6 bxc6 14.Bd3 h6 15.f4 hxg5? (better 15....Nd5!) 16.fxg5 1-0. Perhaps the players were familiar, however, with the fascinating game Mieses-Rubinstein, Beslau 1912, which continued 10.Bd3 Qd7 11.Bb5 O-O 12.Nd4 a6 13.Bd3 and ended in a spectacularly hardfought draw in 28 moves. It is likely that Torre was a bit suspicious of such tricky Bishop maneuvers, though, and preferred a more straightforward development of his attack. But 10.Bd3 does have the advantage of avoiding simplification of the position. And in a dynamically balanced position, the players do have to resort to such maneuvering as they await signs of weakness that can be exploited.
The great tactician Rudolf Spielmann classified two types of sacrifices: what he termed "sham" sacrifices, where a material is temporarily surrendered in the course of a well calculated combination, and "real" sacrifices, where material is sacrificed to gain rather intangible advantages of time, position, or attack but with no calculable recuperation of the material. Torre's exchange sacrifice clearly falls under the category of "real sacrifices," for it does not gain anything immediately other than the disruption of Black's King's field and, therefore, a chance to attack. Torre may have thought the idea was worthwhile for nothing more than its psychological effect. After all, it is always more pleasant to attack than it is to defend. The sacrificial idea was also not without precedent. In a very similar position, arising after 9....Bf5 (instead of the move 9....Be6 in the current game) 10.Rhe1 O-O the game Tartakower-Shoosmith, Ostende 1907, continued 11.Rxe7!? Nxe7 12.Bxf6 gxf6 13.Re1 (13.g4! looks like a better try) 13....Ng6 14.Qh6 c6 15.g4 Bxg4 16.Rg1 d5 17.Bd3 Bxf3 18.Bxg6 fxg6 19.Rxg6+ Kf7? (Black tries too hard to win: 19....hxg6 20.Qxg6+ Kh8 21.Qh6+ Kg8 =) 20.Rg7+ Ke6 21.Qe3+ with a position that Spielmann would no doubt have enjoyed tremendously. Tartakower went on to win in attractive fashion.
A less speculative way of handling the position was shown in the Correspondence match between the chess clubs of Berlin and Budapest (1937-1938), where White played to put piece pressure on the Bishop at e7 with 12.Re3 Nd7 13.h4 Re8 14.Bxe7 Rxe7 15.Rxe7 Nxe7 16.Ng5 Ne5 17.Qe4 N2g6 18.f4 h6! and now White should have tried 19.fxe5 hxg5 20.exd6 cxd6 21.h5 with unclear play.
Torre's move does have the advantage of not wasting any time in developing an attack while he has maximal piece mobility. Whether or not it can be proven completely sound is difficult to say, which is as it should be for a worthy "real sacrifice." But there is no question that Santasiere could have played much more precisely in the defense.
Better than 12....Qxe7?! when White still destroys Black's Kingside pawn position after 13.Nd5 Qe6 (not 13....Qd8 when White can play 14.Nxf6+ gxf6 15.Bh6 immediately winning back the exchange due to the threat of Qg4+) 14.Bxf6 gxf6 15.Qf4 (threatening both Nxf6+ and Nxc7), after which White has it even better than he does in the actual game.
The first fruits of White's speculative attack appear: Black plays less than perfectly in defense. Though 14....Ng6 looks like the natural move, allowing the Queen to defend the pawn at f6 while using the Knight to cover up the gaping hole in Black's kingside position, it nonetheless is not the best. Black should play the less-than-intuitive 14....Kh8! 15.Nxf6 Ng8! 16.Nh5 (no better is 16.Nd5 c6 17.Qd4+ f6 18.Nf4 Qa5!) 16....Qd7 17.g4 Qe6 18.Qd4+ f6 19.Kb1 (to defend the pawn at a2). Though White still has the initiative, Black has managed to mobilize his forces for the defense. More importantly, he has returned the pawn at f6 in order to hold the f6 square, which is much more important for the defense. It is likely that White does not have sufficient compensation for the exchange if Black plays precisely.
Instead of immediately surrendering the f6 pawn to help break the attack, Santasiere tries (though rather too half-heartedly) to defend it in order to keep White's Knights from penetrating his castle. The f6 pawn helps to block the long diagonal from c3 to h8 and it can keep a Knight from gaining a foothold at g5 to support the attack on h7. But at the same time, the pawn at f6 gets in the way of Black's own forces and prevents their coordinated mobilization to the kingside to support the defense. That said, however, while it was better to surrender the pawn than try to keep it, Black should now try with all his might to hold onto that pawn since it has become the cornerstone of his defensive edifice. If Black first goes wrong by not choosing the best plan, he now goes wrong by not sticking to the plan he chose.
White's Queen attacks powerfully along the long diagonal and is well positioned to move over to the kingside along the third rank, so it is now too late for Black to return the pawn in order to mobilize his forces. If 15....f5? 16.Nf6+ Kh8 White can choose between winning back the exchange immediately with 17.Nd7+ f6 18.Nxf8 or continuing the attack with 17.Ng5!? Ne5 18.Ngxh7. But if Black is instead trying to further the plan of defending the pawn at f6, he has chosen a poor defender in his King. A much better try would have been 15....Ne5 when the Knight can retreat to d7 or advance to g5 to defend the f-pawn after White plays the logical 16.Nh4 followed by f4. Though White would have a strong attack, Black would at least keep the f-pawn for defense. Play might go 15....Ne5 16.Nh4 Re8 (not 16....Kh8 17.f4 Nd7 18.Nf5! and there is little Black can do about the threat of Ng5 followed by Nxh7 blowing open the kingside) 17.Nf5 Kh8!? 18.f4 (logical, but better might be 18.Nh6) 18....Ng4! and things are quite unclear.
After 15....Kg7? the King will inevitably have to give way to a Knight check at f5. Therefore the King move makes little sense in the long run if Black wishes to defend the pawn at f6 and it probably shows how much Black has been shaken up by White's daring sacrifice.
There is no way Black can defend both the f6 and f5 squares from the marauding Knights, especially since he is using only the Queen and King to do so. These important pieces are too easily driven off from the defense and need to assistance of the minor pieces to hold the line. But it is now too late for Black to admit his mistake and revert to the defensive plan of holding f6 at all costs with 16....Ne5 17.Nf5+ Kh8 18.f4 Nd7 because White gets a crushing attack with 19.Ng5! threatening Nxh7.
Perhaps Santasiere had already made up his mind to make a run for it with his King at move 15. In any case, the Rook move is clearly intended to vacate the f8 square for his escape. It is certainly too late now to defend the f6 pawn, which can only be done with 17....Qd8?! when follows 18.Nf5+ Kh8 19.Qc3! (not 19.Qh5 Rg8 and it is difficult to see how White makes progress since 20.Rd3? is met by 20....Nf4!) and suddenly it is clear why Black cannot defend by moving his Queen back and forth. In this line, he must surrender the f6 pawn, after which White is surely making progress.
Santasiere tries to clear the way for his King's escape via e8 and d7 while defending the pawn at f7 with the Rook. If instead 19....Qd7 Black gets mated or loses his Queen after 20.Nxh7+ Kg8 21.Nf5, while 19....Qd8 loses the Queen immediately to 20.Ne6+ Rxe6 21.Nxe6+ since the pawn at f7 is pinned.
If Black tries to keep the exchange by 20....Rd7 he gets mated by 21.Nxh7+ Ke8 22.Ng7# or 21....Kg8 22.Qg7#. The only chance is for the Black King to run over to the Queenside, even though he must surrender a lot of material along the way. Now Torre could regain the exchange with a winning endgame advantage after 21.Nxe7 Qxe7 22.Nxh7+ Kg8 23.Qxe7 Nxe7 24.Ng5. But why should he give up the initiative when he has Black's King on the run? Also, there is no hurry to take the Rook.... As Lasker says, when you see a good move look for a better one!
The Rook at e7 is immobilized, since without it the King could not make his escape. If 21....Rd7? 22.Nxh7+ Ke8 23.Ng7#. Now White wins a piece instead of the exchange, keeping two pieces for a Rook while continuing the King Hunt.
Or 22....Nf8 23.Nxf7! Rxf7 (23....Qd7 24.N7xd6+! cxd6 25.Nxd6+ Kd8 26.Qxf8+) 24.Re1+ Kd7 25.Qxf7+ with a winning game.
The Knight will now be lost on this square, but there was no better retreat:
1) 23....Nc6? blocks the King's escape square: 24.Ng7+ and now 24....Kf8 25.N7e6+ wins the Queen and 24....Kd7 25.Qf5+ Re6 26.Nxe6 fxe6 27.Qxe6#
2) 23....Nd7? also blocks the escape and loses the Queen after 24.Ng7+ Kf8 25.N7e6+ Rxe6 26.Nxe6+ Ke8 27.Qh8+ etc.
3) 23....Nc4 24.Ng7+ Kd7 25.Qf5+ Kc6 26.Qd5+ wins the Knight, after which White's Queen is better placed for attack than in the game.
White has emerged from the skirmishes with two Knights for a Rook, which is only a slight advantage in terms of material. But Black's King is still exposed and White will therefore still have a strong initiative once he can coordinate his pieces.
The King must continue his harrowing adventures in front of the pawn chain. If he attempts to retreat behind pawn cover by 28....Kd7? there follows 29.Nc5+! Kd8 30.Nxb7+ Kc8 31.Nxd6+! winning.
Bringing another piece into the attack and threatening mate by 32.Qb5#. Black has few ways of avoiding immediate disaster. If 31....c6 32.Qb4+ Ka6 33.Ra5#, and if 31....a6 32.Qd4+ c5 (32....Kc6 33.Rc5+!) 33.Rxc5! wins.
Black has the unhappy choice of walking into a discovered check or losing his Queen after 32....Ka7 33.Rxa5+ Kb8 34.Rxa8+ Kxa8 35.Nxc7+ followed by Qxg8.
White threatened 34.Rxa5+ Kxa5 35.Qb5#.
Torre could have made the same combination on the previous move. Perhaps he chose to gain a bit of time to search for something better. After all, it probably seemed strange that he could not force check mate with the Black King so exposed and his heavy pieces hovering so nearby. The discovered check forces an easily won piece-up ending.
White easily picks off Black's isolated Kingside pawns, creating a passed pawn on that side of the board to force through the win. Black meanwhile tries desperately to create a passed pawn of his own on the Queenside.
This move allows a pretty finish, but there was no hope. On 48....Rb8 49.Nd6 helps win the c-pawn or forces through the h-pawn with 50.Nf7, and on 48....Rh8 49.Rc7 a5 50.Nh4 a6 51.Kb1 Kb6 52.Rb7+ followed by Ng6 White forces Black to surrender his Rook, after which the remaining Black pawns can be easily rounded up. Black's mate threat with 48....Re8 requires immediate action.
49. Rxc4+! 1-0
After 49....Kxc4 50.Nd6+ K-any 51.Nxe8 there is no way Black can stop the h-pawn from Queening.
A great attacking game for the young Torre that illustrates the psychological power of gambits. While the attacker can play by instinct, seeking out the most aggressive lines, the defender must anticipate everything and play with nerve-wracking care. Most importantly, though, the game demonstrates the importance of choosing a defensive plan and following through with it. Santasiere's uncertain defense was probably his undoing.