Perhaps someone will find a promising continuation for White after 38...Rxa2!, but at this point I believe 38...Rxa2! would have given Black a position he could have drawn.

CharlesSullivan: <Pawn and Two> Yes, I believe we have shown that 38...Rxa2 leads to a draw after, for instance, 39.Rh8 b3! etc. By the way, I have spent several days looking at, after 38...Rxa2 39.Rh8, Fritz's two "drawish" suggestions: 39...Rc2 and 39...Rd2. I will post some small part of that analysis in the next day or so, but I believe both lose.

By the way, I believe that Rubinstein missed easy wins at moves 36 and 37. As far as I can tell, nobody has pointed this out. For example, in this position

< White played 37.Rh5+, but stronger is 37.h5! >

click for larger view

White has 37.h5 (passed pawns must be pushed!) 37...b3 38.axb3 a3 39.d4+ Kb6 (39...Nxd4 40.Ra6 is easy for White) 40.Bd5 Rc2 41.b4

click for larger view

and Black can resign (41...a2 42.Bxa2 Rxa2 43.d5 etc.). White also could have advanced the h-pawn one move earlier: 36.h5! leads to a crushing position.

CharlesSullivan: <Pawn and Two> Here is my analysis of 38...Rxa2 39.Rh8 Rc2 (analysis of 39...Rd2 will follow shortly). The best move here is 40.g4

click for larger view

and I cannot find a way to save the game for Black. Here are some variations:

(A) 40...Kc7 41.h5 Kd6 42.Bxc6 Rxc6 43.h6 winning position

(B) 40...Ne5 41.g5 Nxd3 42.Be4 Rxf2+ 43.Kg3 Re2 44.Rb8+ Ke5 45.Kf3 Rh2 46.Rc8+ Kb6 47.Bxd3 winning position

(C) 40...b3 (the main line) 41.g5

(C1) 41...Nb4 42.Rb8+ Ka5 43.Bc4 Rxc4 44.dxc4 a3 45.g6 b2 46.g7 b1=Q 47.g8=Q Qe4+ 48.Kg3 Qe5+ 49.f4 Qe3+ 50.Kg4 Qe2+ 51.Kg5 Qg2+ 52.Kf6 Qc6+ 53.Ke5 and Black does not have a perpetual check; White wins with his extra material

(C2) 41...Re2 42.g6 Re7 43.Rc8 Ne5 44.h5 b2 45.d4 Rb7 46.Be4 Ng4 47.Bb1 Ka5 48.Kf3 Nh6 49.Rh8 winning position

(C3) 41...Rc5 42.Bf7 Ne7 43.g6! (43.Rb8+ only draws) 43...Nxg6 44.Bxg6 Ra5 45.Bf7 Kc7 46.h5 Rg5+ 47.Kh3 Rf5 48.Bc4 Rf4 49.h6 Rxc4 50.h7 Rc6 51.Rc8+ Kxc8 52.h8=Q+ Kb7 53.Qg7+ Kb6 54.Qd4+ Kb5 55.f4 Rh6+ 56.Kg3 Rg6+ 57.Kf3 Rb6 58.f5 winning position

CharlesSullivan: <Pawn and Two> Now we look at sample variations after 38...Rxa2! 39.Rh8 Rd2(?). Here 40.Ra8! wins:

(A) 40...Na5 41.Rb8+ Kc5 42.Bf7 Nc6 43.h5 Ne5 44.d4+! Rxd4 45.Be8 Ng4 46.Rc8+ Kd6 47.Bxa4 Ke7 48.Rg8 Nh6 49.Rg6 Nf7 50.Rb6

click for larger view

and White wins (+5.98) with 2 extra pawns.

(B) 40...a3 41.h5 Rxd3 42.Bxc6 Kxc6 43.h6 Rd7 44.g4 Kd5 45.g5 Rh7 46.Kg3 Ke6 47.Kg4 Kd6 48.f4

click for larger view

and White's kingside pawns win (+6.31).

(C) Long but straightforward is 40...Ne5 41.Rxa4 Nxd3 42.Kf3 Ne1+ 43.Ke4 Re2+ 44.Kf5 Rxf2+ 45.Ke5 Kb5 46.Ra8 Nd3+ 47.Kd4 Nc1 48.Rc8 Rd2+ 49.Ke5 Nd3+ 50.Kd6 Nf2 51.Rb8+ Ka5 52.g4 Ne4+ 53.Ke5 Nc5 54.Bc6 Nd3+ 55.Ke4 Nc5+ 56.Ke3 Rd3+ 57.Ke2 Rc3 58.Rb5+ Ka6 59.Rxb4 Ka5 60.Rb5+ Ka6 61.Be8 Ne6 62.Rb4 Rc2+ 63.Kf3

click for larger view

and White wins (+6.37).

It looks as if the only draw for Black after 38...Rxa2 39.Rh8 is 39...b3! (See previous postings.)

CharlesSullivan: <Pawn and Two> In the 38...Rxa2 39.Rh8 Rd2 40.Ra8 variation, I failed to address the main point of 39...Rd2, namely 40...Rxd3. The refutation is 41.Bxc6! Kxc6 42.Rxa4 and White has a forced win; for example: 42...Kc5 43.Ra8 43.Rd7 44.Rc8+ Kd5 45.Rb8 Kc4 46.h5 Rh7 47.g4 b3 48.Kg3 Kc3 44.f3 b2 50.Rxb2 Kxb2

click for larger view

and, according to the endgame databases, it is mate-in-26 beginning 51.Kh4!

If a win for White cannot be proved after 42...Rxd3, then perhaps 41.Bf7 or 41.h5 should be considered. Also, 42.h5 may be stronger than 42.Bf7. I have only briefly reviewed these possibilities.

CharlesSullivan: <Pawn and Two> After 38...Rxa2 39.Rh8 Rd2 40.Ra8 Na5 41.Rb8+ Kc5 42.Bf7 Rxd3

click for larger view

43.Be8 will win. For example, 43...a3 44.Rb5+ wins Black's knight and White has plenty of extra material to stop the queenside pawns. Or, 43...Nc6 44.Bxc6 wins--perhaps 44...a3 45.Rb5+ and (a) 45...Kxc6 46.Rxb4 wins with two extra pawns or (b) 45...Kc4 46.Ra5 Kb3 47.h5 etc.

I'm quite surprised that Fritz seems to be struggling with these positions -- the analysis you post always is impeccable. Perhaps you are using only a small subset (or none?) of the available endgame tablebases? Of course, I could be overlooking something significant: I'm pretty much dependent on Zappa and Rybka, and they're certainly not infallible.

I am pleased with our findings for this endgame. Perhaps someone will still discover a promising continuation for White after 38...Rxa2!, but we have provided strong evidence that this move would have allowed Black to draw.

CharlesSullivan: <Pawn and Two> Yes, the endgame tablebases are a "must." I, too, am satisfied that we have come pretty close to the truth of this position. I plan to send some notes to John Donaldson (of "The Life and Games of Akiva Rubinstein" fame).

One last variation: after 38...Rxa2 39.Rh8 Rd2 40.Ra8 Rxd3 41.Bxc6 Kxc6 42.Rxa4 Kb5 43.Ra8 Rd5 44.g4 Rd4 45.Rg8 b3 46.h5 Kc6 47.h6 b2 48.Rb8 Rxg4+ 49.Kf3 Rh4 50.Rxb2 Rxh6

click for larger view

Black is only one pawn down, but 51.Rd2 (the only winning move) cuts off the Black king and is mate-in-61.

1.d4 otes by Jacques Mieses and Dr. Savielly Tartakower. *** Tartakower: In a game overflowing with the finer points of positional play, the outstanding feature is perhaps the problem-like turn 17.c1, by which White very elegantly saves all his unguarded pieces and remains with an extra pawn. d5 2.Nf3 c5 3.c4 e6 4.cxd5 exd5 5.Nc3 Nc6 6.g3 Tartakower: First introduced by Schlechter (against Dus-Chotimirsky, rague 1908), systematized by ubinstein, this positional maneuver has all but refuted the Tarrasch Defense. Be6 7.Bg2 Be7 8.O-O Rc8 Tartakower: Too dogmatic. lack devotes too much attention to the queenside. 9.dxc5 Bxc5 10.Ng5 ! Nf6 11.Nxe6 fxe6 12.Bh3 Qe7 13.Bg5 Mieses: A good move but not the best. eyond any doubt, with 13.e4! White achieves an advantage; for example, 13...d4 14.d5 exd5 15.xc8 dxe4 16.b3! or 13...dxe4 14.g5 O-O 15.xe4, etc. O-O
Tartakower: Too late, and yet - as the storm now breaks - not late enough. etter would be 13...d8. 14.Bxf6 Qxf6 Mieses: A mistake. In his calculation Capablanca has overlooked the opponent's 17th move. The right continuation was
14...gxf6. *** In "Chess Fundamentals" Jose aul Capablanca states that he overlooked 16.g2! and not White's 17th move giving the following combination: 16.g2 (the move Capablanca expected) 16...e5! 17.f4 (if 17.c1 xc1!! 18.xc1 xf2+ and wins) 17...g4 18.h3 (18.h3 xf2+ and lick wins a quality) 18...xf2 19.xf2 xf2+ 20.xf2 g5 and lack wins. 15.Nxd5 ! Qh6 16.Kg2 Rcd8 17.Qc1 !! exd5 18.Qxc5 Qd2 19.Qb5 Tartakower: Having cleverly won a pawn, White shows that he also can hold what he has gained. Nd4 20.Qd3 Qxd3 Mieses: lack cannot avoid the exchange of the queens: 20...xb2 loses a piece after 21.bf1, and if 20...b4, then 21.fd1 followed by e6+. 21.exd3 Rfe8 22.Bg4 ! Rd6 23.Rfe1 Rxe1 24.Rxe1 Rb6 Mieses: I don't like this move. The best seems to be 24...f7. *** Tartakower: Very cleverly lack obtains some counterplay which will bring in a pawn on the -side.

25.Re5
Rxb2 26.Rxd5 Nc6 27.Be6+ Kf8 28.Rf5+ Ke8 29.Bf7+ Kd7 30.Bc4 a6 31.Rf7+ Kd6 32.Rxg7 b5 33.Bg8 a5 34.Rxh7 a4 35.h4 b4 36.Rh6+ Kc5 37.Rh5+ Kb6 38.Bd5 ? All annotators agree on 38.c4! being better. b3 ? All annotators agree on 38...xa2 being better. According to Vladimir Vukovic White still wins with 39.h8! b3 40.h5! a1 (if 40...b4 41.h6! or 40...c2 41.h6! b2 42.a2 etc.) 41.xc6 c7 (not 41...xc6? 42.b8 c7 43.b4 and wins) 42.e4 b2 43.d4 a3 44.h7+ d6 45.b7 a2 46.xb2 g1+ 47.xg1 a1=+ 48.b1 xd4 49.f3, and White's position is still winning. 39.axb3 ! Tartakower: The result of very precise calculations. 39.xb3 is another way to win. a3 40.Bxc6 Mieses: This paradoxical capture is now feasible; e.g., 40...a2 41.b5+ a6 [41...xc6 42.a5] 42.b8!, etc. Rxb3 41.Bd5 a2 42.Rh6+ ! 1-0

CharlesSullivan: I now believe that 38...Rxa2 will not save Black (despite my analysis of July, 2009). After upgrading to an Intel i7-3930K (6 cores @ 4.4GHz) and adding an SSD with 145GB of endgame tablebases, I chose this position to analyze for about a month. A sample White win is 39.Rh8 b3 40.h5 Rc2 41.h6 Rc5 42.Bf3 b2 43.h7 b1=Q 44.Rc8 Qa1 45.h8=Q Qxh8 46.Rxh8 Na7 47.d4 Ra5 48.Rh6+ Kc7 49.d5 Nb5 50.Rc6+ Kd8 51.Bg4 a3 52.d6 Ra8 53.d7 Nc7 54.Rh6 Ke7 55.Rh7+ Kd6 56.Rh1 Ke7 57.Bf3 Ra6

click for larger view

58.Bb7! Rb6 59.Bc8 Rd6 60.Ra1 Rd3 61.Kh3 Kd8 62.Kh4 Rd4+

( < my analysis of 7 July 2009 went 62...Ne6 63.g4 Rf3 64.Ra2 Rc3 65.g5 Rc4+ 66.Kh5 Rc5 67.Rxa3 Rxg5+ 68.Kh6 Rg2 69.Rf3 Rh2+ 70.Kg6 Rg2+ 71.Kh5 Rh2+ 72.Kg4 Rg2+ 73.Kh4 Rh2+ 74.Kg3 Rh8 75.Kg4 Rg8+ 76.Kh5 Kc7 77.Rf6 Rh8+(?) [other Black moves also lose]

click for larger view

and 78.Kg6 led to a draw; but 78.Rh6! Nf4+ 79.Kg5 Rf8 80.Rf6 Nh3+ 81.Kg6 Rg8+ 82.Kf7 Rh8 83.d8=Q+ Kxd8 84.Bxh3 Rxh3 85.f4 is a winning ending > )

63.f4

click for larger view

63...Nb5

( < or 63...Ra4 64.Kh5 Nd5 65.f5 a2 66.Kg5 Nb6 67.Kg6 Nc4 68.f6

click for larger view

and White's connected pawns soon win on the kingside > )

64.Ba6 Nd6 65.Rxa3 Kxd7 66.Kg4 Re4 67.Kf3 Rb4 68.g4 Ke7 69.Bd3 Nb7 70.g5 Nc5 71.f5

click for larger view

71...Rh4 72.f6+ Ke6 73.Kg3 Rb4 74.Bg6 Kd6 75.Bf5 Rb8 76.Rc3 Na4 77.Rc8 Rb7 78.g6 Rb3+ 79.Kg4 Rb4+ 80.Kg5 Nb6 81.Rd8+ Ke5 82.f7 Rb2 83.f8=Q

click for larger view

and White wins.