However, best here is 35. ..Rb8! to get that pesky passed pawn under control. Fritz gives 35. ..Rb8 36. Qc3 Qe7 37. Ra1 d5 38. Ra5 Ra8 39. Rxc5 Rxa6 40. Rxc6 Ra8 41. Qf3 e4 42. Qe3 with a slight black advantage (-0.38 @ 16/44 depth & 840kN/s).

Now, unfortunately, black is pretty much busted after white's 35. Re1 -- no matter what move he makes in response. Fritz 8 gives the following possibilities:

35...d4 loses as in the game continuation after 36. a7.

35...Re4 36. a7 Re8 37. Qxe3 Rc8 38. Qe8+ Qf8 39. Qe6+ Kh8 40. Ra1 Ra8 41. c3 (+2.06 @ 16/46 depth & 774kN/s)

35...e2 36. a7 Qf1+ 37. Kh2 Qf8 38. Qc7 Re4 39. Qxc6 Re8 40. Qxd5+ Kg7 41. Kg1 Ra8 42. Qb7+ Kh8 43. Rxe2 Qd8 44. Qe7 (+2.37)

35...Qf8 36. Rxe3 d4 37. Re6 d3 38. Qxd3 Rd4 39. Qe2 Rd8 40. Rxc6 Ra8 41. Qe6+ Kh8 42. Qc4 (+2.59)

35...Qf7 36. Rxe3 Qf1+ 37. Kh2 Qf8
38. Qe1 Qb8+ 39. g3 Re4 40. Rxe4 dxe4
41. Qxe4 Qa8 42. Qc4+ Kg7 43. Kg2
(+2.81).

35...Rf4, which <Dukenknight> wished to examine, loses quickly to 36. a7 Qf8 [36...Rf1+ 37. Rxf1 Qxf1+ 38. Kh2 Qf8 39. Qb8 ] 37. Ra1 Rf1+ 38. Rxf1 Qxf1 39. Kh2 (+9.06 & 16/55 depth & 766kN/s)

So given that black's game is hopeless after 35. Re1, where was the last chance for black to save the game? My thought is that it was by playing 32...Rb8! (not 35...Rb8??) instead of 32. Qe7?

Fritz 8 gives best play as 32..Rb8 33. Qg3 Re8 34. Ra1 Qg7 35. c3 Ra8 36. Qe3 Qa7 37. Qf4 Qe7 38. Qe3 Qa7 (+0.91) and it would appear black can take a draw by repetition or go for a bigger advantage if white wishes to complicate.

"So far Pillsbury has played the game splendidly, but now he begins to experiment, and finally loses a won game by careless play. The simple move 29...Re8, in combination with the advancing centre, must have won with certainty, as the analyses of the two players have shown. It is very often dangerous to move a rook away from the first line." Tarrasch

Siegbert is oversimplifying (as usual!). The move certainly contributed to defeat, but its not really clear that Re8 wins. For instance if Black charges forward with his pawn:

29...Re8 30.a5 d5 31.Qa4 e4 32.a6 e3 33.a7 e2 34.Re1 Ra8 35.Rxe2 Rxa7 (Crouch). Black is only a doubled pawn up with lots of play against his King.

True, but Lasker could have played Qg3 the previous move. Therefore 32.a6? is also an error allowing 32...Rb8. Several mistakes are bunched together here:

With all due respect to Tarrasch, I agree with you that Siegbert's claim of a win with 29...Re8 is an overstatement -- though 29...Re8 does give black a small advantage, as does 29...Kg7 and 29...Ra8, per Fritz 8:

29...Kg7 (-0.72 & 17 depth & 741kN/s)

29...Ra8 (-0.69 & 17 depth & 741kN/s)

29...Re8 (-0.63 & 17 depth & 741kN/s)

However, Fritz's analysis does not validate Tarrasch's claim that 29...Rb8 is "the error" (though I initially thougt it was without the aid of the computer analysis). After 29...Rb8 30. Qe1 black actually holds an advantage with the simple 30...Qc7 (-0.69 @ 16 depth & 722kN/s).

"The error" or losing blunder was 32. ..Qe7?, as 32. ..Rb8! was black's last chance to salvage his game.

----32.Qg3! Qe8 33.a6 Rb8 34.Qd6 Ra8 35.Qxc5 Rxa6 36.c4=

Lasker, however continued with
32.a6? This may give the draw away:

----32...Rb8! 33.Qg3 Ra8 34.Ra1 Qe7 35.c3 e3 etc

<One more offensive instead of defensive move. The Rook ought to have retired at least to b6, when 34. a7 would not have won: 33...Rb6 34. a7? Qxa7 35. Qd6 Rb8 36. Kg7!. If instead 34. Ra1, Black can stop the dangerous passed pawn with 34....Qa7 35. Qd6 e3 (not 35....Rxa6 on account of 36. Rf1 Qa8 37. Qe6+ and 38. Rf7) 36. Qxc5? e2 and 37....Rxa6, winning the pawn. But with 36. Rf1! Rb8 37. Qe6+ Kg7 38. Qxe3 White would still have had an advantage, as the a-pawn cannot be taken on account of 39. Qe5+, and Black's king is without defense.>

I haven't checked any of that analysis, though.

How about 41. Rxe1 Qxb7 42. Re8+ Kf7 43. a8=Q.

click for larger view