As so often, I misunderstood this game at a quick glance. I have now looked at Hoffer's annotations in the tournament book and also reviewed the game by myself and with Shredder.

This was Round 28 of 30. Lasker had first place sewed up and had defeated Janowski in the first half.

Lasker's 4....c5 was a good reaction to the early 4. Bf4. A move later ...a6 prevents Nb5 and also sets up ...dxc4 and ...b5 later. Janowski's reaction, 6. dxc5 Bxc5 7. cxd5 Nxd5! 8. Nxd5 exd5 gives Black a favorable IQP setup. 9. Bd3 Bb4+ 10. Ke2!? probably would have transposed into Steinitz vs Lasker, 1896, but either Ke2 or Kf1 would have left White with an unpromising setup.

Lasker's 11....Be7! (threatening ...Nb4) and 12....Bf6 gave him promising play in the center and queenside, especially after Janowski's weakening 13. b3?!. After the further 13....Be6 14. Qe2 Rc8 15. Rd1?! Qe7 16. h3 0-0 17. g4?! g6 (strong here is 17....Nb4! 18. Bb1?! Bd7!) 18. Bb1 Rfe8 19. g5? (simply increasing the scope of Black's QB, as Hoffer pointed out in the tournament book) 19....Bg7 20. h4, White was in a bad way.

click for larger view

Lasker now forced the win of the exchange for a pawn with 20....Bg4 21. Rxd5 Nd4 22. Rxd4 Bxd4.

Hoffer wrote <The bargain is not to his best, as Lasker found subsequently. The better move, which would have probably won the game, was to advance the QP, a move which would have embarrassed White very much.>

It is true that White loses after 21. Nxd4? Nxd4 22. exd4 Bh3+. But Shredder thinks White is better off than in the game after 21. h5 Bg4 22. hxg6 hxg6 23. Rh4.

In any case, after 20....Bg4 21. Rxd5 Nd4 22. Rxd4 Bxd5 23. Kg2 Bg7 24. h5 Qe6 25. h6 Bh8 (also promising was 25....Be5 26. Bxe5 Bxf3+ 27. Qxf3 Qxe5 28. Qf4 Rc5, with excellent chances for Black in the endgame) 26. Kg3 Rc5 27. Rh4 Bf5 28. Bxf5 Qxf5 29. Qd1 Rd5 30. Qe1, Black seemed well on the way to victory.

click for larger view

Shredder likes 30....Qd7 here, but Lasker's 30....Qc2 certainly doesn't spoil anything. It seems that Black's queen and active rooks should overwhelm White's undefended queenside. But now Janowski begins showing enormous resourcefulness, while Black drifts. 31. Qb4 b5 32. Qa3 Qc6 33. Rh1 (getting the rook back into play) 33....Rc8 34. Qb4 Bc3?! 35. Qe4! Re8 (it's almost sacrilegious to say so, but perhaps Lasker had overlooked that 35. Qe4 Rxg5+?? 36. Nxg5 saves White's queen?) 36. Qc2!.

click for larger view

36....b4!? 37. Rd1 Red8? 38. Rd4 (perhaps Lasker overlooked this too, but 38. Qe4 is even stronger) 38....Qe6 39. Rc4 Rd1? (a rare Lasker blunder) 40. Nd4 Rg1+ 41. Kh2 Bxd4.

Forced. 41....Qg4 loses to 42. f3. Now White has the advantage!

42.Kxg1 Bc3 43. Bg3 Qd5

click for larger view

In light of Black's threats of ...Qxg5+ or ...Qf3 and ...Rd1+, both Janowski at the board and Hoffer in the book concluded that White had nothing better than the game's third exchange sacrifice, 44. Rxc3 bxc3 45. Qxc3. This leaves White with two pawns for the exchange and dangerous threats on the dark squares. But Janowski could have done better, taking into account another positional factor: Black's weak back rank. White could have placidly protected the g-pawn with 44. Bh4, since 44....Qf3 is well answered by 45. Qd3.

In the game continuation, Lasker kept trying to win. He started by getting his king away from the deadly g7 square: 45....Kf8 46. Qb4+ (46. Qh8+ Ke7 47. Qxh7?? Qf3 and Black wins) 46....Ke8 47. Qa4+ Ke7 48. Qa3+ Kd7 49. Bf4 (simply 49. Qa4+ probably would have led to a quick draw) 49....Rc8 50. Qa4+ Rc6 51. e4 (forestalling a later king march to g4, and also clearing e3 for the bishop) 51....Qd1+ 52. Kh2 Qf3 53. Be3 Qd1 54. Bf4 Qd3 55. Qb4 (threatening Qb7+) 55....Ke6 56. Bg3 Qf3 57. Qa4 Qc3 58. Bf4 Kd7 59. Be3 f6! (a final winning attempt).

click for larger view

60. Kg2 fxg5 61. Bxg5 Qe5 62. Bh4 g5! 63. Bg3 Qc5 64. f3 Ke7 (unpinning the rook) 65. Bf2 Qc3 66. b4 Qc4 67. Qa3 (better 67. Kg3=) 67....Kf7 68. Qb2 Rxh6 69. Qe5! Rf6 70. Bd4! (White has created dangerous threats to the Black king) 70...Qe2+ 71. Kg1 Qd1+ 72. Kh2. After adjournment, Lasker proposed a draw, which Janowski accepted.

keypusher has explained the ideas Lasker's 5...a6, but I prefer 5...cxd4 which leads to near equality and in many variations avoids the problems of the isolated Queen's pawn altogether. After Lasker's 5...a6, Janowski could have gotten the better game with either 6. Nf3 or 7. Rc1.

Janowski's 7. cxd5 was surely a mistake and the source of much of his later troubles. 7. Rc1 was best.

I agree with keypusher that Janowski's 15. Rd1 was questionable. The simple 15. h3 seems best.

After Janowski's 15. Rd1, Lasker's 15...Qe7 seems pointless. Why not just 15...0-0 at once! Once Lasker had played 15...Qe7, Janowski could have fought for the initiative with 16. e4.

Unlike keypusher, I like Janowski's 17. g4, which was the start of the excitement on the king's side. I see no better move for White.

But then Janowski went overboard. His 18. Bb1 was misguided. He should have cleaned up his messy king's-side with 18. Kg2. Lasker, in turn, would have done better with 18...Bc3 rather than his prosaic 18...Rfe8.

All of the above, however, was just the prelude to the battle that broke out with Janowski's 19. g5. keypusher has explained the flaws in this move. The Tournament Book's claim that White had no alternative is simply wrong. 19. Qd2 was best.

Lasker's 19...Bg7 was a slight inaccuracy. He should have posted his Bishop more agressively with 19...Bc3.

In any case, Janowski's 20. h4 was bad, and keypusher's statement that Janowski was in a "bad way" after this move is quite right. Janowski should have played 20. Kg2.

Lasker's 20...Bg4 was brilliant, and should have brought him victory, especially after Janowski's 21. Rxd5? Come what may, Janowski had to play 21. Kg2.

After Janowski's 21. Rxd5?, Lasker had a neat intermediate move: 21...Rcd8. There was no need to rush to win the exchange with 21...Nd4 (Lasker's move). Nonetheless, Lasker's move won the exchange for a pawn and was good enough for victory.

But Lasker, having outplayed Janowski and obtained a winning advantage, seemingly lost his bearings for a while. His 25...Bh8 was inferior to keypusher's 25...Be5. And Lasker's 26...Rc5 was inferior to 26...Red8.

Janowski should have brought his Rook into play immediately with 28. Rh1 rather than his 28. BxB. But despite these slight inaccuracies, Lasker's winning edge was still intact. But then he seemed to lose the thread of the game for a while (incredible for Lasker!).

Lasker's 32...Qc6 was not as accurate as 32...Re6, and his 34...Bc3 was doubtful (as keypusher has noted). 34...Qe6 seems better.

Lasker's 37...Red8 seems to have been an oversight for the reasons explained by keypusher. Lasker should have played 37...Qb5 or 37...Qd7. Now, the win was doubtful at best.

Lasker's advantage was gone entirely after what keypusher correctly calls a blunder: 39...Rd1.

click for larger view

Janowski played 43. Bg3? here, and the win was gone. The Tournament Book recommends 43. Bc7, but 43...Rd5 (and not 43...Re8 as analyzed by the Tournament Book) seems to give Black good chances to save the game. But I see no such chances for Black after the simple 43. Qe2.

After Janowski's 43. Bg3?, Lasker turned the tables on Janowski with 43...Qd5! I agree with keypusher that 44. Bh4 for Janowski would have been better than his (second!) exchange sacrifice with 44. RxB, but I don't think that keypusher's move wins for White. After 44. Bh4, Lasker could probably have held the game with 44...Kf8 or 44...a5 rather than the 44...Qf3 analyzed by keypusher.

After Janowski missed 44. Bh4 and played 44. RxB, a draw seemed inevitable. But nobody told Lasker and Janowski, both of whom continued slugging it out and seeking a win.

In his zeal to press for a win, Janowski's 62. Bh4 was an error. (He could have just played 62. Be3 if all he wanted was a draw). After Janowski's 66. b4 (again Be3 was safest), Lasker would have had real winning prospects had he played 66...g4! (instead of his actual 66...Qc4.

Now surely the game would peter out to a draw. But not with these guys!

As keypusher has pointed out, Janowski's 67. Qa3--though tricky--was inferior to 67. Kg3. Janowski's 68. Qb2 (instead of 68. Qa5) was also an error. Now, while Janowski had created threats to the Black king as keypusher has said, Lasker had real chances. I do not understand why he did not play 70...Qxa2+ (instead of 70...Qe2+). Maybe this would not have won, but Janowski would have had his work cut out for him.

After 70...Qe2+, the game was drawn two moves later.