Annotated chess game

This move order is possibly easy to overlook, because Black is not only exchanging his bishop for a Knight (and forgoing the chance of continuing with B vs N on an open board), but it's his 'Benoni' bishop at that!

White has no real choice but to take this Bishop. 19.Bxb7 Qxb7 20.bxc3 would, of course, transpose, but Black can do better: 19.Bxb7?? Bxb2 20.Bxa8 Bxa1 21.Qxa1 (else Black will play ...Bd4) 21...Rxa8. In this major piece end game, I would fancy the chances offered by Black's mobile Q-side majority (and extra pawn) over White passed and advanced, but isolated and immobile, d-pawn.

This position is on a knifes edge for white. One more slow mistake and it is over.

This position is in Black's favour, as the d-pawn is about to fall, after which the game comes down to whether and how Black can make use of his Q-side majority. I was acutely conscious from here on that a pawn extra was no guarantee at all of obtaining a winning game.

This is going to turn into a very common rook ending with a small but enduring advantage for black. However, as long as you can keep the material equal you have strong chance to draw.

This is not it. Best is 21. Re5 Qxd6 22. Qxd6 Rxd6 23. Rxc5 is more than likely a draw

This is quite a tricky situation. Though I was not unhappy to see this move, I felt that Black had reasonable chances after the expected 21.Re5 Qxd6 22.Qxd6 Rxd6 23.Rxc5 Rfd8. At this point, White has to deal with the threat of mate. If, say 24.g3, then ... Rd1ch 25.Rxd1 Rxd1ch 26.Kg2 Ra1 27.c4 b4 is looking pretty good for Black. I don't much like 27...bxc4 in this line as after 28.Rxc4 Rxa2 White can get his rook behind the passed pawn if he so chooses: 29.Rc8ch Kg7 30.Ra8. Such an endgame is very difficult for Black to win. So I was at considerable pains to avoid anything similar unless absolutely forced into it.

With a won endgame! As I have said in the club forms in the Rook endgame thread. Material equality is important. I also said dont move pawns when you can move pieces instead.

In my view 22.Qc2 was the losing mistake. That White didn't like 22.Qxd6 Rxd6 23.cxb4 axb4 was undertandable. Even though White can subject them to immediate pressure, Black can defend his Q-side pawn couple with rooks placed behind them, after which they would be very hard to stop. All Black has to avoid is this kind of construct with White to play: w In this position, White could win a pawn by Rxc4. However, better might have been 22.Qb3 so that if 22...Qd3 White is not forced to exchange Queens on d3.

The exchange is forced, one way or another. If 23.Qe2, Black could force a transposition into the game continuation by 23...Rd4. But even better might have been simply 23.(Qe2) Qxe2 24.Rxe2 bxc4!? (...b4!?) 25.Rc1 Rd4 26.Rc2 Rb8!! This mysterious looking rook move does not threaten mate 'on the forehand', but exploits the back rank mate motif indirectly - on the backhand, as it were - to protect the lead c-pawn (27.Rxc4?? Rxc4 28.Rxc4 Rb1ch and mates next move). The plan is to bring the rook to b4. If White tries to prevent this by 27.a3, then comes 27...Rb3 28.g3 Rxa3 and Black should win. Otherwise, 27.g3 Rb4 and with White's Q-side pretty much locked down, the Black King steps out from behind his shield wall and into the fight.

Well, experience does count for something. I have played a lot of rook endings on Gameknot, and I think I've lost about two. That doesn't make it easy, and I was spending quite a bit of time studying up this position. Mainly I was looking for lines to avoid... I did expect 24.cxb4 here, but that does look fine for Black. After the move played, cxb4 becomes a threat.

24...b4 may have been even more clear cut

On another day, I might well have played ...b4. But the opportunity has presented itself for Black to force off a pair of rooks (double rook endings adding problematical extra dimensions to the ending), so I took it.

At Black's 27th move came the big think. Neither pawn move (...b4 nor ...bxc4) recommended itself. It had to be a rook move: ...Ra1; ...Rc1; ...Rd2 or ...Rd4. Unfortunately, the pawn ending after 27...Rd4 is at best drawn, with plenty of ways for Black to lose. That got rejected. Of the others, 27...Rc1 seemed to me the most forcing, but 27...Ra1 is probably about as good. I spent hardly a glance at 27...Rd2, concentrating on the other moves.

I was fairly confident that I could win this end game - with due care.

This was clever, as, naturally, I did not want a trade. After 29...bxa4 30.Rxa4, once again the win becomes highly problematical, especially as it is White who can establish a rook behind the passed pawn. It might be an idea to discuss this type of rook ending in the Club forum: b

Yes, Black has allowed White a passed pawn, but his twin passers are much more dynamic, and have at least a limited capacity for self-protection. The white a-pawn lacks that vital feature.

A tempo gainer. That the rook protects the a-pawn's queening square was not lost upon me neither.

Not... exactly. The King was destined to intervene on the Q-side.