#18 Reshevsky-Geller (click here to view on chessgames.com)
Before there was Fischer, there was Reshevsky, the first American to threaten the Soviet machine. The 1953 Candidates tournament was as close as he came to a world championship match, and indeed how close he came. This game was as important as Keres-Smyslov in determining the final standings. An endgame approached, and Reshevsky had every reason for optimism...
31. b6 Rb8?
Bronstein gives the amazing 31. ... Rxa3! 32. b7 Rb4 33. Rd8+ Kh7 34. b8=Q Rxb8 35. Rxb8 Rd3 36. Rf1 Rc3! with a likely drawn 4v3 rook endgame. Tragically easy for us humans to miss...
Analysis diagram after 36. .... Rc3!
Instead the game heads down more prosaic channels...for now.
32. Rd6 Ra4
33. Rxc2 Rxa3
34. h3 Rb3
35. Rcc6 Rb2
36. e4 h5
37. e5 h4
Bronstein: And here it might seem that nothing can save Black. Nevertheless, I would not have traded my b-pawn for the insignificant h-pawn. Couldn't White have relocated his rook along the seventh rank?
Bronstein gives 38. e6 f6 39. Rc7! Rxb6 40. R6d7 as winning, as indeed it is.
38. .... R2xb6
39. Rxb6 Rxb6
40. Rxh4 Rb1+
41. Kh2 Re1
Averbakh: Reshevsky did not conceal his surprise that Geller decided to play on. With an ironic smile he sat down at the board, ordered a cup of coffee and began slowly stirring it with his spoon. There was indeed no reason to hurry: the two extra pawns were a sure guarantee of victory.
Bronstein: In order to understand what follows, keep in mind that there are some rook endings in which two extra pawns are not enough to win...sometimes it is impossible to win the ending with rook and f- and h-pawns against rook, or rook and two connected passed pawns against rook, if the pawns can be blockaded. Geller is hoping to transpose into one of these endgames.
What follows is a series of suboptimal moves, each of which is individually insignificant but which lead collectively to a truly poor move.
42. f4 Re3
43. Rg4 Kh7
44. Rg3 Re2
45. h4 Re4
46. Rf3 f6
47. exf6 gxf6
48. Kg3?! Kg6 (48. g4!)
49. Ra3 f5
50. Ra6+ Kh5
51. Rf6?? Re3+ (51. Ra8 is still sufficient to win)
52. Kf2 Ra3 (Now the problem with Rf6 - stalemate is in the air!)
53. g3 Rf3+!! (See below for analysis of 53. Rxg5+)
Averbakh: It can be imagined with what pleasure Geller made this move, and how triumphantly he looked at his opponent. And Reshevsky? In the seconds remaining to the time control he began thinking intensively. At that moment Geller summoned a waiter, deliberately loudly ordered a glass of tea, and the unhurriedly began stirring in some sugar.
54. Ke2 Rxg3
55. Rxf5+ Kxh4
56. Kf2 Ra3
57. Rg5 Rb3
58. Rg1 Kh5
59. Ke2 Ra3
60. f5 Ra5
A tragedy for Reshevsky....but what about the simple 53. Rxg5+ Kxh4, going into a 2v0 rook endgame?
But this too is drawn! An unusual case, but with White's king cut off below the third rank, his pawns cannot advance unchallenged. If the f-pawn advances the Black king chases it, and the combination of king and rook will blockade the g-pawn. An unusual draw, requiring key conditions - connected pawns with the black king in front, an active Black rook and a mostly inactive White rook (in its worst position in front).
Try this against Fritz - it's oddly frustrating.