Soltis in "Why Lasker Matters", stated that some sources show the last move as 55.Ra8+.

Part I

Soltis calls this <one of the most remarkable endgames of all time> and annotates it at length. Below are paraphrases and quotes from his notes, plus a few notes from Tarrasch's tournament book. Soltis quotes are in brackets.

The position after 5. Nf3 was reached in the Kasparov-Short rapid match: Kasparov vs Short, 1987. Schiffers went badly wrong with 6....b6 (6....Nc6 7. Qxb7 Nb4 at least draws; compare Tarrasch vs A Fritz, 1889). After 8....Na6? 9. cxd5 exd5 10. e4 Black should lose in short order. 10....dxe4 11. Bxa6 (Tarrasch thinks 11. Nh4 Bc8 12. Bxa6 Bxa6 13. Qe6+ Be7 14. Qxc6+ Kf7 15. Qxe4 is much stronger, but Soltis writes that White <deserves more> than he gets in Tarrasch's line) 11....exf3 12. 0-0 Bd6 13. Bb7?

<The adage says mistakes come in threes. This is No. 1: with 13. Re1+ Ne7 14. Ne4! White's initiative keeps growing (14....Bc7 15. Bb7! since 15....Bxh2+ 16. Kxh2 Qc7+ 17. g3 Qxb7 18. Nd6+).> Tarrasch, incidentally, recommends 13. d5.

After 15. g3?

<White still has chances for advantage after 15. Bf4! Qxf4+ 16. g3 Qg4 17. Kg1 or 15....Qxb7 16. Rfe1+ Kf8 (16....Ne7 17. Bd6) 17. d5>

After 17....0-0-0?

<Black returns the favor and allows 18. Nb5! (18....cxb5 19. Rac1+).> (Tarrasch overlooked this too.)

18. a5? Ng6!

<Now 19. Be3 h5! leaves Black with the attack and extra material, e.g. 19. axb6 20. Qxb6 axb6 21. d5 h4!.>

click for larger view

<From move 6 to move 22 each side made enough errors to lose three games. It's hard to imagine what follows is one of the most remarkable endgames of all times.

White is two pawns down and faced with losing his only positional assets, the fine bishop (....Nxf4) and the seventh rank (...Re7). He can get back one of the pawns with 24. Rxg7, but after 24....Nxf4 25. gxf4 Rd8 he is losing.

Among the factors is that his rook can't leave the g-file (26. Rf7 Rg8! 27. Rxf6 Rg2+ 28. Kh1 Bd3! and the f-pawn wins.) For example, 29. Rxc6+ Kd7! 30. Rxb6 Rxf2 and because Black's king stops Re6, there is no defense to ...Re2 and ...f2.>

24. d5!

<What saves Lasker repeatedly from here on are two qualities that every player envies in the truly great. The first is the ability to create winning chances in positions that seem barren. Like Marshall, Alekhine, Shirov and few others he seemed to be able to find tactics wherever he looked. Here, for example, the threat of 25. Ra8+ Kd7 26. dxc6+ allows White's knight to play.>

24....cxd5

<Black begins to avoid drawish endings, such as 24....Re7 25. Rxe7 Nxe7 26. Bd6 Nxd5 27. Nxd5 exd5 28. Bf8. He almost certainly believed that the game can only have two results -- Black win or draw -- and felt he must reject the candidates that led to the second.>

25. Nxd5

<The second quality that saved Lasker is his talent for scaring opponents. Black becomes concerned now by the threats of 25. Nxb6+ and 25. Ra8+ or even 25. Be3.> On the other hand, if Schiffers was so scared of Lasker, why avoid drawish lines?

25....Nxf4?

<This seems perfectly logical: Black makes sure that White has no counterplay and he trades an inactive knight for the good bishop.

But Tarrasch pointed out that after 25....Re2! Black could always meet Be3 with the crushing ...Ne5-g4+. He said 26. Nxb6+ Kd8 wins for Black (27. Be3 Ne5 or 27. Bd6 Rxf2+ 28. Kg1 Rg2+ 29. Kh1 Ke8).

Actually the win is harder because White can retreat the knight to e3. On 26. Kg1, which Tarrasch thought lost to 26....Re1+ 27. Kh2 Rf1, White wins material with 28. Ne3! Rxf2+ 29. Kg1. (Black should play 26....Rxb2 and win the hard way.)>

26. gxf4! Re2

<Now 27. Nxb6+ fails to 27....Kb8 28. Rxg7 Rxf2+ 29. Kh1 Rf1+ 30. Kh2 f2! and the pawn promotes (31. Kg2 Bh3+! 32. Kxh3 Rh1+)....Only after the game did Black realize that 25....Nxf4 damages his chances so much that the counter-intuitive 26....Re6! was best (27. Rxg7 Be4).>

click for larger view

27. Kg3!

<This the first of several positions in which White could force a draw. He has a perpetual check after 27. Ne7+ Kb8 28. Nc6+ Kc8 29. Ne7+. Black must accept because of 29....Kd8? 30. Nxf5 Rxf2+ 31. Kg3 Rxb2 32. Nxg7 f2 33. Kg2.

But why is White playing 27. Kg3 if he can force a draw? Objectively he has no right to play for more. True, but Lasker can see that his opponent is the one making the bigger mistakes again. And if both sides have chances, the better endgame player may win regardless of who has the theoretical advantage. >

27....Rxb2

<...[I]f Black wanted a draw, the right way to get it was to do what Lasker had done -- threaten to win. That means 27....Kd8 and then 28. Rxg7 Rxb2 29. Nxf6 h6 followed by ....b5-b4.>

28. Rxg7!

Soltis: <The most dangerous pawn on the board is now the one at f4.>

Tarrasch: "White prefers, despite his material disadvantage, to play for a win -- a bold stroke which finds a reward for those deserving of it."

28....Kb8

<Black had to stop Ne7 xf5 now that ...Kb8 no longer attacks a rook. He could reach the same position as in the note to 27....Rxb2 with 28....Kd8! 29. Nxf6 h6.>

29. Nxf6 b5 30. Rg5

Soltis: <This is strong because Black's ...Kb8 took his king in the wrong direction and made it harder for him to stop the f-pawn.>

Tarrasch: "It is interesting to see how White, here and on the next moves, works steadily with win of tempo."

30....Bd3 (Tarrasch thinks the bishop would be less exposed on c2) 31. Nd7+! 32. Ne5 Be4 33. f5!

click for larger view

<Black's pieces are well placed to push the b-pawn but not to stop the f-pawn, e.g. 33....Kd6 34. f6 Ke6 35. f7 Ke7 36. Rg8 and queens. It looks like 34....Bd5 is better (35. f7 Bxf7!). But White has the killing 35. Nd3! followed by Nxb2 or Rxd5+!.>

33....Ra2! 34. f6 Ra8 35. f7 Kd6 36. Rg8!

White will be lucky to draw if he has to settle for 36. Nxf3. A more attractive idea than that is 36. Kf4. But 36....Bd5! stops 37. Rg8 and threatens to win with ...Rb8 and ...b4.

36....Ke7 37. Kf4 Bd5 38. Rg7

<This is the same position as in 36. Kf4 Bd5 37. Rg7 except that Black's king is on e7, not d6. Why did White prefer this one? After all, Black can eliminate the f-pawn and draw with 38...Kf6 39. Rxh7 Ra4+! 40. Kg3 Bxf7! (41. Nxf7 Kg6).

But to play that Black has to assume the role of the defender seeking a draw and Schiffers wasn't ready to admit that.

The optimal move -- whether playing the board or the man -- was 38....b4!. Then the placement of Black's king would help White out after 39. Rxh7! because of 39....b3 40. Ng6+! Kd6 41. f8/Q+ Rxf8 42. Nxf8 (although 42....a2 still draws).

This is best because White also has to worry (after 38....b4 39. Rxh7) about the winning try 39....Ra6!?. (White saves himself with 40. Rh8 and Rb8.)>

38....Rh8!?

<Schiffers finds a tricky way to play for a win. His move prepares to push the h-pawn and enables him to meet the winning try of Kg5 with ...h6+.>

39. Kg5!

<White rejects 39. Nd7 Bxf7 40. Ne5 Rf8 41. Rxh7, threatening Ng6+. After 41....Ke6 his best is 42. Rxf7, leading K+P vs. K+P -- followed by king-vs.-king.>

39....h6+

<By now Black must have realized that White has a second way to win, a queen sacrifice. This is evident after 39....b4 because White has no good move other than 40. f8/Q+!.

Then 40....Kxf8 41. Kf6 threatens mate by 42. Re7 and Nd7+ as well as various ways of winning the rook and bishop (42. Rc7 and Rc8+). White wins after 41....Ke8 42. Re7+ Kd8 44. Rd7+ Kc8 45. Rxd5 because he picks up the remaining pawns while keeping his own. >

40. Kf5! Be6+! (preventing a knight check on g6 by forcing White's king there) 41. Kg6

<Now Black realizes the rook is badly placed because 41....b4, the move he had been counting on, allows 41. f8/Q+ Kxf8 42. Kf6 and only White can win.>

41....Rc8

click for larger view

<Black is finally ready for ...b4. There is nothing to 42. f8/Q+ Kxf8 43. Kf6 Bd5! (44. Rh7 Kg8!). >

42. Rh7!

<What could White's idea be? Is it Rxh6 and Rh1 to stop the b-pawn? Or is it to clear a square for his king so that he can play Ng6+? Either way, this innocuous-looking move gives Black reason to be optimistic again.>

42....b4?

<He must have looked at two or three scenarios and that turns out to be one too short. Naturally he examined 43. Rxh6 b3 44. Rh1? and saw 44....Bxf7+ 45. Nxf7 Rg8+ is a likely Black win.

He also would have checked out 43. Kg7. Then 43....Bf5 seems to win (44. Rxh6 b3 or 44. Rh8 Rxh8 45. Kxh8 Kf8). But in fact it loses (44. Nc6+! Kd7 45. Rh8).

After 43. Kg7 Black would play 43....Bxf7 44. Nxf7 Rc2 and ...Rxf2 guarantees him the moral victory of of knowing he had the better of the draw that follows 45. Kxh6.>

43. f8/Q+!!

<Black had to stop this mechanically, with 42...Rf8.>

43.... Kxf8 44. Kf6

<The main threat is not 45. Kxe6 but 45. Re7! followed by 46. Ng6+ and 47. Rg7 mate. Black also loses after 44....Bd5 45. Rh8+ Bg8 46. Ng6+ Ke8 47. Rxg8+ Kd7 48. Ne5+.>

44....Bg8

<The knight performs wonders after this. But 44....Kg8 45. Rg7+ Kf8 46. Ng6+ Ke8 was no better because of 47. Kxe6 Rc6+ 48. Kf5 Rc2 49. Rg8+ Kf7 50. Rf8+ Kg7 51. Rb8, so that 51....Rxf2 52. Rb7+ and Kf6 will mate.

Or 49....Kd7 50. Ne5+ and 51. Nd3, preserving the winning pawn.>

45. Re7! Bh7 46. Rxh7 Kg8 47. Rg7+ Kf8 48. Rb7!

White goes for zugzwang. Black's rook cannot leave the first rank because of Rb8+, while 48....Rd8 or 48....Re8 allow 49. Ng6+ and mates.

48....Ra8 49. Rf7+ Ke8

Tarrasch: "White announced mate in six as follows: 50. Re7+ Kd8! 51. Nf7+ Kc8 52. Nd6+ Kd8! 53. Ke6 Ra7 54. Rxa7 any 55. Rd7 mate. The whole endgame was conducted by Lasker with wonderful exactness and sharpness."

Pal Benko Winning with Chess Psychology pg 15.

<keypusher>

<Lasker announced mate in six-50. Re7+ Kd8! 51. Nf7+ Kc8 52. Nd6+ Kd8! 53. Ke6 Ra7 54. Rxa7 any 55. Rd7 mate-and Black resigned.>

Thanks for that excerpt.

Just as with Capablanca at New York three decades later, Lasker's chief advantage over his rivals at Nuremberg was not so much his ability to lose less but his ability to win more. The second-place finisher, Maroczy, lost only once, but had eight wins compared to twelve for Lasker. Tarrasch lost three games, the same as Lasker, but won only nine times. Schlechter only lost twice but managed just five wins, finishing in a tie for 7th.

To harvest his wins, Lasker played almost every game out to the bitter end. Remember, Nuremberg was pretty much a replay of Hastings the year before. All the top players in the world were present. This game was played in Round 2, but Lasker had every reason to think that a very high score would be necessary to win the tournament--at Hastings, he had finished +10, and that was only good enough for third place! So, against Schiffers, with a draw in hand, he decided to keep playing.

Another one of his valuable attributes is that he simply didn't get discouraged. In this game he had a big advantage in the opening, but blundered it away and reached a lost position. But he just kept coming, and finally won. His game against Chigorin followed the same course.

Lasker vs Chigorin, 1896

<But his knowledge of human nature-or of Schiffers specifically-led him to suspect that his opponent would get careless or complacent with his "winning" advantage.>

This strikes me as nonsense. Does it make sense to you to suppose that Schiffers was complacent or careless? He was playing the World Champion (who, incidentally, had beaten him at Hastings). Suppose a GM playing Kasparov in 1986 or Karpov in 1976 reaches a complicated ending two pawns up. Do you think that GM would be complacent or careless? I don't.

The fact that, after trading queens, Schiffers quickly allowed Lasker to reach a position where he could force a draw suggests to me that Schiffers wasn't overconfident at all.

I think Lasker's "psychology" in this game was simply this: "I am a better chessplayer than Schiffers, and I need this win. I've practically got a draw in hand. So I'm going to keep playing and see what happens."

<Lasker announced mate in six-50. Re7+ Kd8! 51. Nf7+ Kc8 52. Nd6+ Kd8! 53. Ke6 Ra7 54. Rxa7 any 55. Rd7 mate-and Black resigned.>

After 42...h5? Schiffers "will not accept less than a full point, but since a full point is not justified by the position, he loses."

In his Manual of Chess, Lasker writes that loss generally occurs "when the player overrates his advantage or for other reasons seeks to derive from a minute advantage a great return such as a forced win."

He also writes, "He who has a slight disadvantage plays more attentively, inventively, and more boldly than his antagonist who either takes it easy or aspires after too much. Thus, a slight disadvantage is very frequently seen to convert into a good, solid advantage."

After 42...h5? Schiffers "will not accept less than a full point, but since a full point is not justified by the position, he loses.">