I. ... 49.Nh6 Kg7 50.Nxg8 a5 51.Ne7 a4 52.Nf5+ Kf7 53.Nd6+ Kd7 54.Nc4 Rxc4 55.Rxb2 (55.Rb7+ Kd6 56.Rxb2... does not help White; and if 55.dxc4 a3 ..., then Black queens) 55...Rc7 56.Rb5 Ra7 57.Rxe5 Kd6 58.Rb5 a3 59.Rb1 a2 60.Ra1 Ke5 61.Kf3 Kd4 ... and Black wins. ...>

Nope. Ouch! Totally scratch that.

Here's the position in the line we're discussing.

click for larger view

I of course saw 52. Rxg8+ is forced and thought it looked winning, but of course 52. Rxg8+ Kxg8 53. Nxa4 Rb4 54. Nc3! Rc4 55. Nd5, and the e-pawn is held because of the fork after 55...Rxe4? 56. Nf6+.

So White gets a pawn for the exchange, which is why I said <very good winning chances> instead of <winning outright>.

After 52. Rxg8+ Kxg8 53. Nxa4 Kf7 54. Nc5 Rb5 55. Nd3 Kf6, White doesn't want to touch his pawns. He can try neutral moves like 56. Kf3 Rb3 57. Ke2 Rb1 58. Kf3 Rd1, etc. and I think Black can gradually infiltrate and tie White up. But of course it's a long process and very hard to give an exact "winning line" here.

Btw, I would have gone with the <47...a5>. But since <beatgiant> was already on the case and since I put lots of trust in his endgame skill, I decided to be useful exploring something else 'winning'.

(And now I am licking my wounds. And I have a headache.)

---

I at first thought Black had to win after 40...Bd2, but is it too much to imagine Chigorin foresaw the heroic knight tour starting some 9 moves later! with 49 Ne3-c4xb2xa4-c5-d3-g4) winning both pawns, and returning to set up a fortress?

As <Tamar> point out, it appears White's best chance of holding the draw in the 47...a5 line is heroic knight tour starting with 49.Ne3.

Tarrasch, however, was not through with this ending. In the 47...a5 line, he came up with an improvement for Black which is at first sight surprising, but then appears to be quite convincing. (..to be continued).

It starts with the line provided by <beatgiant> 40...Bd2! 41.Rf6 Bh6 42.Rh6 b3 43.Bd1 b2 44.Bc2 Rc8 45.Rd6 Rc2 46.Rd8+ Bg8 47. Rb8

47...a5 48 Kg2

but now

48...Rd2 (instead of the immediate 48...a4)

The idea is to hold back the a pawn until the knight has passed it by on its tour!

The other peculiarity is that Black can activate his King via g7, as the threat to the g8 bishop may not be real given the connected passed pawns that survive if White goes after that piece.

48 Kg2 Rd2

49 Ne3 Kg7 (To free the bishop to capture if 50 Nc4)

50 Nf1 Re2

51 Ng3 Rc2

52 Nf5+ Kf6

53 Rb6+ Be6

54 Ne3 Rd2

55 Nc4 Rxd3

56 Nxb2 Rb3!

click for larger view

Very well done. Your latest analysis shows that Black wins this ending.

I liked how the White rook was played, first 48...Rd2, and then it returned to clinch the game with 54...Rd2.

There are other variations in this ending that are favorable for Black, including some by Tarrasch, but none as convincing as yours. So I think the debate is over regarding this ending.

Tarrasch then indicated that Chigorin had improved on this line with the move 47.Rb8.

However, Tarrasch believed Black could still win by 47...a5 48.Kg2 a4 49.Ne3 Rc1 50.Rxb2 a3 51.Rc2 Ra1.

Black has given up his passed Pawn on b2, and placed his Rook in front of his passed Pawn on a3! However, the passed Pawn on a3 will now cost White a rook.

Here is a diagram of this unusual position after Black's move 51...Ra1:

click for larger view

Tarrasch gave the following variations after 51...Ra1:

52.Nd5 a2 53.Kf3 Bf7 54.Rc8+ Kg7 55.Ra8 Rg1 56.Rxa2 Bh5+ 57.Ke3 Re1+ 58.Kd2 Re2+ 59.Kc3 Rxa2 and Black wins.

52.Nc4 a2 53.Kf3 Kg7 54.Rb2? Bf7?(54...Kf6! 55.Rb6+ Be6 56.Ra6 Rd1!) 55.Nxe5 Rb1 and Black wins.

White can improve on this last variation with 52.Nc4 a2 53.Kf3 Kg7 54.Rxa2 Rxa2 55.Nxe5 Bb3 56.Ke3 Bd1, but Black is still winning.

Other variations not given by Tarrasch that are winning for Black:

52.Ng4 a2 53.Rxa2 Bxa2 54.Nxe5 Rd1 55.Kf3 Kg7 56.Ke3 Rh1.

52.Rc8 a2 53.Ra8 Rc1 54.Nc4 a1(Q) 55.Rxa1 Rxa1 56.Nxe5 Bb3.

However, at move 54, old Fritz discovered an improvement for White. After 54.Rxa2 Bxc4 55.Ra5 Be6 56.f3 Kg7 57.Rxe5, I could not find a way for Black to win. Do you have any analysis after 54.Rxa2?

In the meantime, <tamar> has discovered a nice win for Black. See his message of 04/11/07.

Chigorin now recommended 52.Nf5. Both players agreed on the following moves: 52...a2 53.Rc8 Rg1+ 54.Kxg1 a1(Q)+ 55.Kg2 Qa7 56.Re8 Qd7 57.Rxe5 Qxd3.

We now have the following position:

click for larger view

Chigorin thought this ending should be drawn. Tarrasch disagreed and stated: <Chigorin thinks that this ending of Queen and Bishop versus Rook and Knight will be drawn. I do not share this opinion. This large material advantage should be winning. Black only needs to bring his pieces, possibily including his King, into the game, and then advance his h-pawn.>

I do not know if Chigorin or Tarrasch ever provided any additional analysis of this position. Tarrasch may be correct, but it is a difficult position as White will try to set up a fortress position.

I do not recall giving <54.Rxa2> much more thought than just automatically rejecting it on the account of the <54...Bxc4>. It clearly is one of those mental 'shortcut tricks' of us humans that works most of the times, but still betray us often enough. As it was, ever since <tamar>'s engine highstepped its own path around my line, I was wordering, 'what in the world am I missing here'?! But I did not see it till I read your post.

If White could achieve this position,

click for larger view

it would be a draw after Kg1 or Kh1, with Kg3 losing.

However one rank higher

click for larger view

White loses no matter where he moves.

But I can find no way to keep the pawn on h3 without losing it.

This is Shredder's best play, ending in
the losing ending.

58 Re8 h5 59 h4 Kh7 60 Ng3 Bb3 61 Re7+
Kh6 62 Kh2 Bd1 63 Re5 Bg4 64 Rd5 Qf3 65 Kg1 Bh3 66 Rh5+ Kg7 67 Rg5+ Kf8 68 Nh5 Qxe4 69 Rg8+ Kf7 70 Rg7+ Ke6 71 Rg3 Qe1+ 72 Kh2 Qxf2+ 73 Kxh3 Qf5+ 74 Rg4 Qxh5

click for larger view

I have not discovered any detailed analysis of this position by Chigorin or Tarrasch, however, I did provide their respective opinions of the position in my posting of 04/14/07. Your analysis strongly supports Tarrasch's opinion.

Interestingly, I have found one additional move from Chigorin. That move is 58.Ng3. I have not been able to work out a definite conclusion for the position after 58.Ng3. You have indicated that 58.Re8 is White's best choice and Fritz also prefers 58.Re8.

After 40...Bd2 41.Rxf6 Bxh6 42.Rxh6 b3 43.Bd1 b2 44.Bc2 Rc8 45.Rxd6 Rxc2 46.Rd8+ Bg8 47.Rb8 a5 48.Kg2 a4 49.Ne3 Rc1 50.Rxb2 a3 51.Rc2 Ra1 52.Nf5 a2 53.Rc8 Rg1+ 54.Kxg1 a1(Q)+ 55.Kg2 Qa7 56.Re8 Qd7 57.Rxe5 Qxd3

Chigorin proposed 58 Ng3 and we get this position

click for larger view

Shredder follows Tarrasch's formula up to a point <Black only needs to bring his pieces, possibily including his King, into the game, and then advance his h-pawn> but does not advance the h pawn.

Here is the analysis.

58...Kg7 59 Nh5+ Kf7 60 Rf5+ Ke7 61 Re5+ Kd6 62 Re8 Qb3 63 Ng3 Be6 64 Ra8 Qb4 65 Rh8 Qb7 66 Rf8 Bf7 67 Rd8+ Ke5 68 Rd3 Qa7 69 Re3 Bg6 70 Kf1 Qd4 71 Kg2 Qa7 72 h4 Qd7 73 h5 -3.78/20

Reading between the lines, White tries to avoid advancing the h pawn, but finally must when the King reaches e5 and Black begins to coordinate his Queen and Bishop.