After the shocks of the first two games, the next double-header was hugely important. Veselin Topalov may have lost the first two games, but the character of play certainly gave him enough reasons to believe the match is still wide open - he dominated proceedings for long periods in both games, and arguably could have been at least 1,5-0,5 ahead.

Therefore, it was imperative for him to continue in the same positive vein, and hope the final touch will not elude him if and when he gets another chance. From Kramnik's point of view, on the other hand, it was important to restore at least some sanity to the proceedings, and show his opponent not only who is in the lead, but also who is in charge.

Catalan Opening E02
V.KRAMNIK - V.TOPALOV
Game 3

1.d4 Nf6 2.Nf3 e6 3.c4 d5 4.g3 dxc4 5.Bg2 Nc6. Topalov is the first to deviate from the opening of the 1st game.

6.Qa4 Bd7. A relatively rare move - the main line starts with 6...Bb4+ 7.Bd2 Nd5. The text move is much more modest.

7.Qxc4 Na5. Black is hoping that the impending liquidation in the center will ensure that the awkward position of that knight will not be felt too much.

8.Qd3 c5 9.0-0 Bc6 10.Nc3 cxd4 11.Nxd4 Bc5.

A rare line, in which Black is not aiming for too much. In fact most of the lines end with a typical Catalan endgames - mostly, if not entirely, harmless. It would seem that Veselin decided that his priority for this game is to get off the mark...

12.Rd1 Bxg2?! ... or is it? My feeling, as I sat watching the game live, was that Veselin came to the board with the idea to trade as many pieces as possible as soon as possible and make a draw, but couldn't bring himself to do it - the whole concept is, as we've seen on numerous occasions, alien to him. It was much more prudent to play 12...Qxd4 13.Qxd4 Bxd4 14.Rxd4 Bxg2 15.Kxg2 Nc6 16.Rd1 0-0 17.Bg5 Rfd8, and it is hard to believe Black's problems are not temporary. 13.Qb5+! With this intermediate check White further disrupts Black's coordination, and his advantage becomes tangible.

13...Nd7 14.Kxg2 a6 15.Qd3. We are still following the stem game Tkachiev - Solozhenkin, France 2000, which White won convincingly.

15...Rc8. This is the first new move, and objectively speaking it is not an improvement. Solozhenkin played 15...Be7, and after 16.Bf4 Rc8 17.Nf3 White had a clear plus.

16.Bg5! With the Black king stuck in the center it is not surprising that White has a chance to start a direct attack.

16...Be7. Played immediately - but there was no real choice: 16...Qxg5 17.Nxe6 Qe7 18.Nxg7+ Kd8 19.Nd5 is a massacre, while 16...Ne5 17.Bxd8 Nxd3 18.Bxa5 Nxb2 19.Rdb1 Nc4 (19...Bxd4 20.Rxb2 b6 21.Bxb6! Bxc3 22.Rc2+-) 20.Nb3 Nxa5 21.Nxa5 Bd4 22.Ne4 0-0 23.Nxb7+/- is also ill-advised.

17.Bxe7?! After a long think, Vladimir decided against the attractive 17.Ne4!, which would have made Black's life very difficult: his only move seems to be 17...Nc4.

17...Bxg5 18.Nd6+ Ke7 19.Nxc8+ Qxc8 20.Nf3 loses on the spot - there is no defence against 21.Qa3 Qc5 22.Rd7!

But as Kramnik said after the game, he could not find anything decisive after 17...Ne5. However, White has a beautiful shot up his sleeve: 18.Bxe7 (18.Qb1 Qc7 19.Bxe7 Kxe7 is not as good for White as it looks) 18...Qxe7 19.Nf5!!, and Black is in a lot of trouble: 19...Nxd3 (19...exf5 20.Nd6+ Kd8 21.Qa3!+-) 20.Ned6+ Kf8 21.Nxe7 Rd8 (21...Rc7 22.Ndc8!+-) 22.Ndc8! Nc6 23.Nxc6 Rxc8 24.Na7 Ra8 25.Rxd3 g6 26.Rc1, and White is winning.

In case of 17...Nc4 18.b3 Bxg5! (18...Nce5 once again allows 19.Bxe7 Qxe7 20.Nf5) 19.bxc4 Be7 is quite acceptable for Black.

But after 18.Rac1! White is very much on top:

a) 18...Nxb2 19.Qb3! Nxd1 20.Nxe6!, and the attack is suddenly unstoppable;

b) 18...Nce5 19.Qb3!? (19.Rxc8 Qxc8 20.Qb3 is also a pleasant edge for White) 19...Rxc1 20.Rxc1! Bxg5 (20...Nf6 21.Qxb7 Qxd4 22.Rc8+ Bd8 23.Nxf6+ gxf6 24.Bxf6 0-0 25.Qa8!+-) 21.Nxe6! Bxc1 (21...fxe6 22.Qxe6+ Kf8 23.Nxg5 h6 24.f4!, and White is breaking through) 22.Nxd8 Kxd8 23.Qxb7, and the Queen should be much stronger than the disjointed pieces;

c) 18...Nde5

c1) 19.Qb1!? Bxg5 20.Nxe6 fxe6 (20...Qe7 is no longer an option: 21.Nxg7+! Kf8 22.Nf5+-) 21.Rxd8+ Bxd8 22.b3 Na3 23.Qb2 Rxc1 24.Qxc1 Nb5 25.Qc8 Nd7 26.Qxb7 Nc7 27.Qc6 Ke7 28.b4, and once again Black is paralyzed;

C2) 19.Qb3, which is also quite unpleasant: 19...Bxg5 (19...Qd5 20.f3 Bxg5 21.Nxg5 Ng4 22.Ndxe6 Nge3+ 23.Kh3 Nxd1 24.Rxd1 Qf5+ 25.g4 Qb5 26.Nxg7+ Kf8 27.Qxb5 axb5 28.Nf5±) 20.Nxe6 Qe7 21.N6xg5, and White is a pawn up for very little;

17...Qxe7. Now Black can relax - a little.

18.Rac1 Nc4. It was possible to play 18...0-0, but after 19.Nf5 (19.Nxe6 Ne5! 20.Qd5 fxe6 21.Qxe5 Nc4 is unclear) 19...exf5 20.Qxd7 Qxd7 21.Rxd7 b5 (21...Rfd8 22.Nd5!+/=) 22.Rcd1 Nc4 23.Nd5 Nxb2 24.Ne7+ Kh8 25.R1d6 Ra8 26.Nxf5 White retains some pressure.

19.Na4. Embarking on the forcing simplification which gives White a stable and safe plus. 19...b5 20.b3 0-0. 20...Nde5 was very dangerous: after 21.Qe4 f5 (21...bxa4 22.bxc4 Qc5 23.f4 Nd7 24.Nf3 is not very good either) 22.Qb1! bxa4 23.bxc4 Nxc4 (23...0-0 24.Qb6+/-) 24.Qd3 Qd7 25.f3! Nb2 (25...Ne5? loses to 26.Qe3) 26.Qxa6 0-0 27.Rxc8 Rxc8 28.Rb1, and Black is facing an uphill struggle.

21.bxc4 bxa4.

22.Nc6! With the knight on c5 Black would have nothing at all to worry about, so White trades it off immediately.

22...Rxc6 23.Qxd7 Qc5. The dust has settled. White is slightly better, and Veselin is facing perhaps the most unpleasant scenario in his match situation - a passive position with no real counterplay, and a realistic chance to lose if you're not careful.

24.Rc3. Hinting at a possible Rc3-f3 if the opportunity arises, and freeing the other rook for the b-file.

24...g6 25.Rb1.

25…h5. A logical move - Black wants to have the option of disturbing White on the K-side - but as Sergey Shipov and then Alexei Korotylev pointed out on this website, Black could try and nip White's advance in the bud by playing 25...Qf5!? 26.Qxc6 (26.Qd3 Qc5 27.Rb7 Rd6 is not very promising for White) 26...Qxb1. However, White still retains some edge after both 27.Ra3 Rd8 28.c5! (28.Rxa4 Rd1) 28...Qb5! (28...Qc2 29.Rxa4 Qxe2 30.Qxa6+/-; 28...Rd1 29.Qe8+ Kg7 30.Rf3+/-) 29.Qc7 Rd7 30.Qc8+ Kg7 31.Rf3!, and Black is not yet out of the woods, and 27.a3!? Qe1 28.Re3.

26.Rb7 e5. The machine, which dislikes passive defence almost as much as humans do, suggests 26...Rd6, hoping for 27.Qxa4 Rd2 28.Qxa6 (28.Re3 Rxa2! 29.Qxa2 Qc6+ is a draw) 28...Rxe2 29.Rf3 Rc2 30.Rf4 Rxa2!, and draw is not far away. However, White would probably ignore the a-pawns for a while and play 27.Qe7!, keeping Black tied up.

27.e4. 26 moves later than usual - and still this move is rushed. As many commentators pointed out, it was better not to commit in the center just yet. White has one more improving move left: 27.a3!?, securing the b4 pawn for the rook, and after 27...Rf6 (27...e4 28.Re3! Qxc4 29.Rb4+/-) 28.Re3 White has a tangible edge.

27...Rf6 28.Rc2 Qa3?! Black could not resist the temptation of forcing the Queen back. However it was better to play 28...Rd6! 29.Qxa4 (29.Qc7 Rc6) 29...Qd4, and White will regret the hasty e2-e4.

29.Qd1 Rd6 30.Rd2 Rfd8. If Black thought this wins him the d-file, he was mistaken.

31.Rd5! 31.Rb8 Rxd2 (31...Rxb8 32.Rxd6 Qxa2? 33.Qd5 is wrong) 32.Rxd8+ Rxd8 33.Qxd8+ Kg7 34.Qd2 also makes sense, but the text move is stronger.

31...Rxd5. What else?

32.cxd5?! But here, with around 15 minutes left, White backs off and opts for a safer option. It's an understandable decision given the overall situation - but 32.exd5! would have given him a very real chance to go 3-0 up, as the following lines show:

a) as GM Korotylev correctly points out, after 32...e4 33.Qe2 Re8 34.Rc7! (34.Qe3 Qxa2 35.Qf4 Rf8 36.Qxe4 Qd2 is good for White, but he does not really need to complicate matters so much) 34...Qd3 35.Qe3 White is almost winning;

b) 32...Qxa2 33.Qd3! (33.Qf3 Rf8 is wrong - White should keep the attack on f7 in reserve) 33...Qa1 (33...a3 34.d6 e4 35.Qd5 Rf8 36.d7 e3 37.Qxf7+!!) 34.d6 (34.c5 Qc1 35.c6 a3) 34...Qd4 35.Qxd4 exd4 36.c5 Kg7 (36...a3 37.c6 a2 38.c7 Rf8 39.Rb8+-) 37.Rb4, and White should win.

32...Qxa2. Now Black will escape.

33.Qf3. 33.Qd3 Qa1 34.Qf3 (34.Ra7 Rd6) 34...Rf8 35.Ra7 Qd4 36.Rxa6 Rb8 also does not promise White much.

33...Rf8 34.Qd3 a3! 34...Qa1 35.Ra7 is not as good now - Black does not have 35...Rd6 anymore. But the text move forces the draw.

35.Rb3. 35.Qxa6 Qc2 is also a draw.

35...f5! But the rook on f8 has its uses.

36.Qxa6. There was nothing better.

36...Qxb3 37.Qxg6+ Kh8 38.Qh6+ Kg8. In the end, Black had more reasons to be cheerful after this game - he managed to hold a very difficult position, and finally got on to the scoreboard. On the other hand, the distance became one game shorter.

Meran Variation D47
V.TOPALOV - V.KRAMNIK
Game 4

1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.e3. Once again Topalov is the first to deviate. Game 2 saw 4.Nf3 dxc4.

4...e6 5.Nf3 Nbd7 6.Bd3 dxc4 7.Bxc4 b5 8.Bd3 Bb7. The classical Meran - an opening Kramnik used to play a lot in the 1990s.

9.a3 b4 10.Ne4 Nxe4 11.Bxe4 bxa3.

12.0-0. A modern way of treating this position - White is hoping to stop Black from freeing himself with c6-c5. He is also aiming to keep the a-file open for his rook, so 12.bxa3 is certainly not part of the plan. Funnily enough, this idea was first played by Vassily Ivanchuk against Kramnik himself - way back in 1994.

12...Bd6. The stem game went 12...axb2 13.Bxb2 Nf6 (Black ran into trouble after 13...Be7 14.Qc2 h6 15.Bxc6 Rc8 16.d5! exd5 17.Bxd7+ Qxd7 18.Qb3 in Najer - Potkin, Ubeda 2001) 14.Bd3 Be7 15.Qb1 Rb8 16.Rxa7 Qb6 17.Qa2 0-0 18.Rb1 Ra8 Ivanchuk - Kramnik, New York 1994, and Black equalized.

13.b3. A renowned Meran specialist, Boris Avrukh, tried a rather drastic 13.b4!? here, but after 13...Nf6 (13...Bxb4 14.Qb3 a5 15.Bxa3 is what White had been hoping for) 14.Bd3 Nd5! 15.Bxa3 Nxb4 Black was fine in Avrukh - Al Modiahki, Biel 2003. The text is milder, and also makes good positional sense - if White can stop c6-c5, that is. If the dark-squared bishops are traded off, and the pawn is still on c6, White will have every reason to be happy.

13...Nf6 14.Nd2! And here is the novelty. Previously White tried the obvious 14.Bd3, but after 14...c5 15.Bb5+ Ke7 Black was doing reasonably well in Pogorelov - Vera, Andorra 2006. The text move introduces the new concept - White is not planning to leave the long diagonal at all. It is also a pawn sacrifice...

14...Qc7 ...although to accept it at that moment would have been reckless at best: 14...Nxe4 15.Nxe4 Bxh2+ 16.Kxh2 Qh4+ 17.Kg1 Qxe4 18.f3, and while Black still has defensive resources, it is clear this will not be a picnic.

15.Bf3! 15.h3 would have been too meek: a tempo is a lot in positions like that. After the possible 15...0-0 16.Bxa3 Bxa3 17.Rxa3 c5! 18.Bxb7 Qxb7 19.dxc5 Qb4 20.Qc1 Rfc8 Black will be fine.

15...Bxh2+. But now the sacrifice has to be accepted.

16.Kh1 Bd6 17.Nc4 Be7 18.Bxa3. White has achieved his objectives - both c6 and a7 are weak, and the bishop on b7 has very few prospects, at least for a while. However, the price he had to pay for that is quite steep - if the position ever opens his king will feel draughty.

18...0-0 19.Bxe7 Qxe7 20.Ra5. Keeping the possible break-out under control.

20...Rfd8 21.Kg1. This move would have to be made sooner or later, so White gets it out of the way as early as possible. The pitfalls of leaving this issue hanging for too long are evident after 21.Qa1 c5! 22.Rxa7 Rxa7 23.Qxa7 Ng4! - Black's problems (and the game itself) are over: 24.Kg1 Qh4 25.Bxg4 Bxg2 26.Kxg2 Qxg4+ with a perpetual.

21...c5!? A very sensible decision. Black is not willing to wait until White's pressure on the Q-side weaknesses becomes unbearable, and gives back his extra pawn to free up his bishop.

22.Rxc5 Ne4. Another option was 22...Bxf3 23.Qxf3 Rac8, but after 24.b4 (or 24.Rxc8 Rxc8 25.Ra1) 24...Nd5 25.Rb1 White retains some pressure.

23.Bxe4 Bxe4 24.Qg4 Bd3. An important finesse. Black is aiming for the heavy-piece endgame, preferring it to the positions like this one: 24...Bd5 25.Ne5 Rac8 26.b4, and White is somewhat better.

25.Ra1. White could force the trade there and then by playing 25.Rc1, but after 25...Bxc4 (25...Rac8 26.Ne5 Rxc5 27.Rxc5 Bf5 28.Qf3 is bad for Black) 26.bxc4 (26.R1xc4 is nice-looking, but a bit vague - Black has no weaknesses, and possession of an open file will not get White too much on its own) 26...a5 the position becomes double-edged. With the rook on a1 Black will not be able to push the a-pawn as fast as in the previous line - but he is no longer forced to take on c4.

25...Rac8 26.Raa5. White could try 26.Rg5, but after the precise 26...g6 (26...f6 27.Rga5 Rc7 28.Ra6 is riskier) White can't exploit the weakening of the dark squares.

26...Rb8!? With the 1st rank temporarily weakened, Black finally occupies the b-file - after 27.Ne5 Rxb3 28.Nc6 Qd6 the threat of 29...Rb1# means Black escapes unscathed.

27.Qd1. Since this move does not force Black to take on c4, perhaps it was better to play 27.Nd2, keeping the Queen active on the K-side.

27...Be4 28.Qa1 Rb7 29.Nd2 Bg6 30.Qc3 h6 31.Ra6 Kh7 32.Nc4. What can only get that far without involving the knight - but this allows the bishop back into play.

32...Be4 33.f3 Bd5 34.Nd2. 34.Na5 Rbd7 did not promise White anything concrete either. Now he is threatening e3-e4.

34...Rdb8. An attempt to break free immediately by playing 34...e5 would run into 35.Qd3+! (35.dxe5 Bxb3 is unclear) 35...Kg8 36.e4, and White wins material. However, 34...Rbd7 was a viable alternative.

35.Qd3+. After 35.e4 Bxb3 36.Ra3 Bd1 37.Qe3 Rb4 the bishop will escape.

35...f5. 35...Kg8? 36.e4 Bxb3 37.Rc3 Qh4 38.Ra1 would have been a fatal mistake.

36.Rc3 Qh4. Black continues to defend against e3-e4 by tactical means.

37.Ra1 Qg3 38.Qc2?! It is difficult to decide on 38.e4!? fxe4 39.fxe4 Qxd3 40.Rxd3 Bxb3 41.Rc3 three moves before the time-control, and a careful look confirms that Black is fine even here. After Kg1-f2-e2 the threat of Ra3 would become quite serious, but if Black plays 41...a5 immediately he should be fine after 42.Rxa5 Bd1.

38...Rf7 39.Rf1. Once again, 39.e4 does not quite work: 39...fxe4 40.fxe4 Qf2+ 41.Kh1 Bxb3 42.Qd3 Ba4! 43.Rxa4 Rb2 44.Rc2 Rxc2 45.Qxc2 Qe1+ 46.Kh2 Qh4+=.

39...Qg6 40.Qd3 Qg3. The pseudo-active 40th move. Black did not really have to stop e3-e4 anymore - it is only a threat when the bishop has nowhere to go.

41.Rfc1 Rfb7 42.Qc2 Qg5 43.Ra1 Qf6. Black is perfectly placed to meet anything White can throw at him, but Veselin, understandably, continues fighting.

44.Qd3 Rd7 45.Ra4 Rbd8 46.Rc5 Kg8?! Not the best square for the king, as will be obvious soon. The manoeuvering is not really getting White anywhere, so Veselin decides to bite the bullet and allow more simplifications. And here, for a very short period of time, things get interesting again.

47.Nc4 Bxc4?! 47...Rb7 was still fine for Black - the capture on c4 can wait.

48.Raxc4?! After this move, played rather quickly, draw becomes unavoidable. 48.Qxc4 was much stronger - after 48...f4 49.Re5! fxe3 (49...Rd5 50.exf4 Qxf4 51.Rxe6 Qxd4+ 52.Qxd4 Rxd4 53.Rxa7+/=) 50.Rxe6 Qf7 51.Re4! Qxc4 52.Rxc4 Black is suffering. A strange miss - perhaps Veselin himself no longer believed he could win this game?

48...f4. Now the board will become empty rather quickly.

49.Rc6 fxe3 50.Qxe3 Rxd4 51.Rxe6 Qh4 52.Rxd4 Qxd4 53.Re8+ Kh7 54.Qxd4. Draw agreed. Compared to the first two, these were a much more sedate couple of games - but the match is now two games shorter.