W
e are going t o look a t an interesting game, played by Sasha Chernosvitov. He annotated it in great detail: he gave numerous variations, and described what he was thinking about during the game, what he was afraid of and what he over looked . His analysis is not error-free. Of course, in such cases mistakes are inevita ble - after all, when you are examining complicated variations, it is easy at some point to become entangled . But behind the mistakes made, definite deficiencies in thinking, in the approach to the game, can sometimes be seen. The reason I have chosen this particular game for analysis was because some of the omissions, both in the moves and in the comments, seemed to me to be instructive.Denisov - Chernosvitov Moscow Junior Championship 1 99 1
Queen 's Gambit Accepted
1 d4 dS 2 c4 dxc4 3 e3 lDf6 4 �xc4 e6 S lDf3 cS 6 0-0 a6 7 a4 lDc6 8 'ife2 'ifc7 9 lDc3 �d6 1 0 dxcS
1 0 b3, 1 0 d5 and 1 0 l:r.d1 have also been played . In his commentary on the game Sasha described in detail his views on the theory of the given variation . But since his opening conceptions are to a certain extent a personal matter, there is no need to discuss them here. We are more interested in the problems which arose later.
1 0 . . . �xcS 1 1 h3?!
1 1 e4 is more critical, althoug h then 1 1 .. . lDg4 (threatening 1 2 .. . lDd4) or the immediate 1 1 .. . lDd4 has to be reckoned with . White wants to advance his e-pawn in comfort, but in the opening every tempo counts, and with such slow play he can no longer expect an advantage. 11 . . . 1 2 e4 1 3 lDxd4 0-0 lDd41 �xd4 The opening stage has concluded in Black's favour. He controls the central squares, the bishop on c4 is running up against the e6- pawn, the white e-pawn has been halted and in some variations it can even come under attack. It only remains for Black to develop his light-square bishop, and his position will be preferable.
14 �d2?!
Chernosvitov recommends 14 �d3, in order to prepare the development of the bishop at e3.
Analysis of a Game
tZJ
1 83 1 4 . . . 1 5 llac1 1 6 �d3 b6 �b7The threat of 1 7 ttJd5 (or 1 7 ttJb5) is very obvious. But what will White do after the queen moves off the c-file? Perhaps ex change the dark-square bishops by 1 7 �e3, but most probably prepare f2-f4 by moving the king to h 1 . Is it not possible to make it difficult for the opponent to carry out his plans?
1 6 . . . 'ii'g3! An excellent manoeuvre! 16 .. . 'ii'e5 1 7 �h1 ttJh5 1 8 'ii'g4 was much weaker.
1 7 �h1 'ii'h4
The queen is excellently placed at h4. From here it presses on the e4-pawn and simulta neously eyes the king. In some cases the knight will jump to g4, while if 1 8 f4, then 1 8 .. . ttJh5 1 9 �e1 ttJg3+ 20 �xg3 'ii'xg3 with the better chances for Black.
1 8 l:tce1 ?!
Now White's position becomes difficult. He should have decided on a second succes sive king move: 1 8 �h2 ! . The tactical basis of it is the elegant variation 1 8 .. . �xc3 1 9 �xc3 (weaker is 1 9 g 3 �e5 20 f4 ttJg4+! 2 1 �g2 'ii'e7) 1 9 .. . ttJxe4 (after 1 9 .. . �xe4 20 �xa6 White is only slightly worse) 20 �xg7!
�xg7 21 g3 with equality.
1 8 . . . l:.fd8
The white pieces on the d-file are beginning to hang . If 1 9 �c2 (with the idea of �h2 and g2-g3, or 'ii'f3-g3) Chernosvitov suggests replying 1 9 .. . g5! 20 'ii'f3 �e5 2 1 ':'e2 l:.d4 . If 1 9 ttJd 1 (preparing 20 �c3), then 1 9 .. . b5 20 axb5 axb5 21 �c3 e5 22 'ii'f3 �xc3 23 bxc3 (23 ttJxc3 b4) 23 . . . ':'a2, and Black has an appreciable advantage.
Sasha considers the comparatively best defence to be 1 9 'ii'f3!? �xf2 20 ':'xf2 ':xd3 21 'ii'xd3 'ii'xf2 22 'ii'e3!? 'ii'xe3 23 �xe3 ttJxe4 (23 .. . ttJd7 24 :d1 ) 24 ttJxe4 �xe4 25 �xb6 with chances of equalising (for exam ple, 25 .. .f5 26 ':'c1 �f7 27 �d4). However, if the concluding position of this variation does not satisfy Black, he can retain the advantage with the quiet move 1 9 .. . �e5!?, seeing as 20 ttJd 1 will be met by a little combination: 20 .. . J:txd3! 2 1 'ii'xd3 ttJxe4 with the threats 22 .. Jld8 and 22 . . . ttJg3+.
1 9 � h2?
It is a well-known tru ism that everything should be done at the right time. 'A move made one move later is often a mistake. ' Now the capture with the bishop o n f2, which we saw in the last variation, gains greatly in strength .
20 1lxf2 21 'iWxd3 22 �g1 23 �h1 l:txd3 lDg4+! 'ifxf2+ Chernosvitov played the first half of the game very strongly, completely outplayed his opponent, and gained an advantage quite sufficient for a win . But from this moment it was as though a substitute had taken his place.
What should Black play now? 23 .. . 'iWh4 suggests itself, but then the counterattack 24 'iWd7 is unpleasant. Therefore in the first instance the safer continuation 23 . . . lDe5! should be considered . The reply 24 'iWe3 is forced . Let us continue the variation : 2 4 .. . 'iWxe3 2 5 i.xe3 lDc4 2 6 i. c 1 (26 i.f2 lDxb2 27 i.xb6 ':'c8 is also cheerless, for example: 28 i.d4 I[c4 29 lIb1 ':xd4 30 J:txb2 l:td7) 26 .. . lId8 27 b3 lDe5 28 i.e3 (28 i.f4 lDd3) 28 .. . lId3 29 i.xb6 J:txc3 30 i.d4 l1xb3 31 i.xe5 f6 followed by . . . e6-e5, and Black converts his extra pawn without particular difficulty.
Why d idn't Sasha play this? Through inertia he wanted to continue the attack, fearing that in the endgame White would save himself thanks to the opposite-colour bish ops. But more importantly, he underesti mated the opponent's threat, assuming that the queen move to d7 was not possible due to the loss of the e4-pawn . But in fact after 23 .. :tIi'h4 24 'iWd7 lDf2+ 25 �h2 i.xe4 after exchanging on e4 White can then simultane ously attack two pieces with 'iWc6 or 'iWb7. One can launch into such adventures only on the basis of very accurate calculation. Here I would almost certainly have called a halt and rejected the entire variation (if I had not done this a move earlier, on seeing the move 24 'iWd7).
I n the conversion of an advantage you should try to strictly control all the oppo nent's active possibilities, not allowing any
unnecessary sharpening of the play. It can happen that your advantage hangs by a thread , and to maintain it you have to exert yourself and deeply calculate complicated variations. But here the situation is clearly different: Black is a sound pawn to the good and he has a safe way of retaining all the advantages of his position .
23 . . . 'iWh4? 24 'iid7!
What should Black do now?
24 .. Jlb8? 25 :f1 is completely bad for Black, and therefore he must choose be tween 24 .. Jld8 and 24 .. . lDf2+ .
On the previous move Sash a committed a typical psychological mistake: realising that he had gained a decisive advantage, he relaxed and played carelessly. Now he sensed that things were by no means as simple as they appeared earlier. And here came a second psychological mistake (re member, incidentally, Tarrasch's famous maxim: 'Mistakes never occur singly' . ) As often happens, the sudden difficulties spoiled his mood and prevented him from calmly analysing the variations, which , however, are now very complicated . Black also failed to solve this problem later, in analysis at home.
Analysis of a Game
ttJ
1 85':xd2 because of 26 1:.e2. But here it is too early to stop the calculation: Black can play 26 ... h51 (or 26 .. . h6!), opening an escape square for the king . Black's rook is immune and his threats look rather dangerous. I ncidentally, in similar situations the pawn is usually advanced not one square, but two, since it may come in useful in the attack. I n the given instance the two moves are roughly equivalent.
The opponent's reply is obvious: 27 'ii'xb6 (27 'ii'xa6? l:1c2! is bad). The next problem is this: how can Black exploit the advantages of his position?
The simplest solution - 27 .. . .l:.xe2 28 lbxe2 'iVe 1 + 29 lbg 1 lbf2+ 30 'itth2 lbxe4 - does not seem convincing to me. After 3 1 lbf3 (31 'ifb8+! 'itth7 32 lbf3 is even more accurate) the coordination of the black pieces is disrupted somewhat, and White is threaten ing the rapid advance of his pawn on the queenside. For example, 31 lbf3 'ii'g3+ 32 'ittg 1 'iVd6 33 'iVxd6 lbxd6 34 b4 followed by lbd4 .
Let us try 27 ... lbf2+! . If 28 'ittg 1 ? , then 28 .. . lbxh3+! 29 gxh3 :xe2 30 lbxe2 'iVe1 + 3 1 'ittg2 'ii'xe2+, and the queen endgame is easily won . I n this variation the pawn is better placed at h5 than at h6. However, even with the pawn on h6 the evaluation of
the position would not be changed (if 32 'ittg3, then 32 .. . 'iVe 1 + followed by 33 .. . a5 is possible, if there is nothing better).
And in the event of 28 'itth2 there is an excellent knight sacrifice: 28 ... lbxh31 29 gxh3 l:[d31 with an irresistible attack. For example, 30 'iVb8+ 'itth7 3 1 'ittg 1 ':xh3 32 .l:th2 'iVg5+! , or 30 'ittg2 ':xh3 31 'ii'd6 'itth7!, intending 32 .. .f6 and 33 ... e5 (31 ... 'ii'g4+ 32 'ittf2 'iVg5! with the same threat of . . . e6-e5 is equally good).
Attempts to avoid mate lead to a lost endgame for White: 29 ':xd2 lbf4+ 30 'ittg 1 'iVe 1 + 3 1 'itth2 'ii'xd2 , o r 2 9 'ii'b8+ 'itth7 30 'iVg3 'iVxg3+ 31 'ittxg3 ':d3+! 32 'itth2 lbg5. Instead of 26 ':e2 we must consider 26 .l:tf1 ! .
After 26 .. . f6 there i s the strong reply 27 'iVe?! . And if 26 ... h5!?, then White contin ues 27 'ii'xf7+ 'itth7 28 'ii'xe6 lbe3 29 lbd5! lbxf1 30 'ii'f5+ 'itth6 31 'ii'e6+ g6 32 'iVg81 , and the game inevitably ends in perpetual check.
How can Black's play be improved? Grand master Dolmatov found a rather unexpected idea : after 24 .. .l:ld8 25 'ii'xb7 he suggested avoiding the capture of the bishop in favour of 25 ... h6!.
26 l:le2 l:lxd2 transposes into the variation 25 .. J:txd2 26 l:le2 h6, in which , according to our analysis, Black's attack achieves its aim (true, we put the pawn on h5, but this is of no significance). And in the event of 26 l:lf1 Black is no longer obliged to capture on d2 - 26 . . . liJf2+! 27 l:lxf2 'ii'xf2 is much stronger, when the white bishop cannot escape. The best defence is 26 liJd51 exd5 27 :f1 (if 27 exd5, then 27 .. . liJf2+ and 28 .. . liJxh3!), but here too Black retains a big advantage, by continuing 27 ... liJf2+ 28 �h2 (28 �g 1 liJxh3+ 29 gxh3 'ii'g3+ 30 � h 1 'ii'xh3+ 3 1 �g 1 'ii'g4+! 3 2 � h 1 'ii'h5+ and 3 3 .. . dxe4 i s hopeless for White) 2 8 .. . dxe4 (weaker i s 2 8 .. . liJxe4 2 9 'ii'xf7+ � h 7 3 0 iLf4) 2 9 iLe1 'ii'f4+ 30 �g1 e3 31 'ii'xa6 :d1 32 'ii'e2 'iixa4 (32 . . . 'ifd4 33 iLxf2 exf2+ 34 'ifxf2 l:lxf1 + 35 �xf1 'ii'd 1 + 36 'ii'e 1 'ili'xa4 is also good , with an extra pawn in a queen endgame) 33 'ii'xe3 liJd3 .
There are good grounds for once again remembering Tarrasch , who maintained that you should resort to a combination in order to repair a mistake committed earlier. I would remark, however, that at the board it is not so d ifficult to decide to go in for tactical complications, if the method of elimination is employed - after first satisfying yourself that all other tries are unpromising. Strangely enough, although Sash a saw that he would
achieve nothing after 24 ... liJf2+ 25 �h2 iLxe4, nevertheless he quickly rejected 24 .. . .l:.d8. Apparently what showed itself here was the psychological effect of the mistake made earlier, about which I have already spoken.
24 . . . liJf2+ 25 �h2 iLxe4? As grandmaster Evgeny Bareev pointed out, it was not yet too late to switch to attack by 25 .. . :d8! 26 'ili'xb7 lIxd2. White loses after 27 l:r.f1 ? 'ii'f4+ 28 �g 1 liJxh3+ or 27 'ifb8+? l:r.d8 28 _g3 liJg4+! ' And in the event of 27 :e2 Black gains an advantage by 27 .. . h6 (or 27 .. . h5) 28 'iVxa6 liJg4+ 29 �g 1 (29 � h 1 ? :tc2 ! ) 29 .. . liJe3 ! , for exam ple: 30 llxd2 'ili'e 1 + 3 1 �h2 liJf1 + etc. The outcome of the complications arising after 27 .. . g5!? is less clear: 28 'ili'xb6! (28 'iVxa6?! l:txe2 29 'ii'xe2 g4! ) 28 . . . liJxh3!? 29 gxh3 l:td3 30 �g 1 ! lIxh3 31 'iVb8+ �g7 32 �g2. I ncidentally, with this move order, as in the variation by Dolmatov analysed earlier, Black would have deprived his opponent of the best defence l:tf1 ! .
26 liJxe4?
White misses his chance. He should have played 26 :txe4! liJxe4 27 'ii'c6 'ifg3+ (after 27 .. . l:td8 28 liJxe4 the bishop on d2 is defended - this is why he should take with the rook, not the knight) 28 �g1 .l:.d8 29 liJxe4. After 29 .. :"e5 there follows 30 'ii'xb6 'ili'd4+?! 31 iLe3, while if 29 .. . 'ii'b8 (as Chernosvitov wanted to play), then 30 iLe3! b5 31 a5 h6 32 'ili'b6 (32 'ii'xa6? 'ii'e5) 32 .. J:ld 1 + 33 �f2 'if e5 34 liJd2 with excellent drawing chances.
In endings a rook is often stronger than a bishop and knight, if it is able to break through at the right time into the opponent's position to attack the enemy queenside pawns and create there an outside passed pawn. But here the endgame is still a long way off, and besides, it is not only White's
Analysis of a Game
ltJ
1 87queenside pawns that are under fire, but also Black's. 26 . . . 27 'it'c6 28 �g1 29 �h2 tDxe4 'ii'g3+ 'ii'f2+ 'ii'g3+ It is useful, by repeating moves, to save time on the clock. 30 �g1 31 �h2 32 l:txe4 33 'it'xb6 'it'f2+ l:.d8 'ii'xd2
Black is a sound pawn to the good . Which is more correct: to exchange or retain the queens?
Of course, it is better to keep the queens on. After all, the white king is exposed and it can easily come under attack. Chernosvitov illustrates this assessment with the follow ing sample variations: 33 .. . h6 34 b4 (34 'iVxa6 'it'xb2) 34 . . . 1:.d6 35 'it'b8+ �h7 36 b5 axb5 37 axb5 f5! 38 ltc4 e5 (if 38 .. . 'ii'd5 the only defence is 39 1:.c3! 'ii'e5+ 40 ':'g3) 39 l:.c2 'iVf4+ 40 g3 1:.d2+ 41 1:.xd2 'ii'xd2+ 42 �g 1 'iVe3+ 43 �g2 e4 44 'ii'f4? 'ii'e2+ 45 �g 1 'ii'xb5.
It is unclear whether Black can win after 44 b6, but this is not so important: he can achieve his goal with 40 .. . 'ii'e4! (instead of
40 .. . 1:.d2+? ! ) 41 1:.g2 1:.d 1 or 4 1 'ii'xd6 'it'xc2+ 42 � g 1 'ii'b 1 + and 43 .. . 'ii'xb5.
In the game Black decided to go into an endgame. An incorrect assessment of the position! Even if the endgame is won , it is clear that with the queens on it would be much simpler to convert the advantage.
33 . . . 'ii'd3? 34 1:.e3
35 'it'xd6
'ii'd6+ ':'xd6 An interesting rook ending has arisen. In the analysis of it, numerous questions occurred to me; the answers to them were either lacking in the commentary, or did not seem convincing.
36 .l:[b3
White prepares a4-a5 and 1:.b6. A tempting idea , but the more standard plan beginning with 36 b4 should also have been consid ered (the rook will stand behind the passed pawn). After 36 .. . �f8 37 b5 axb5 38 axb5 �e7 39 1:.b3 it is bad to play 39 .. . �d7? 40 b6 �c8 in view of 41 l:lc3+ �b8 42 1:.c7. 39 .. . l:tb6 is correct, when Black most prob ably wins, although it is not so simple. If his king goes to c5, there follows 1:.c3+; this means that he will have to waste a tempo on . . . 1:. b6-b 7. During this time White will create counterplay on the kingside. How? Well, for
example, by 40 �g3 l:tb 7 41 �f4 �d6 42 g4 f6 43 g5 �c5 44 l:te3, or 42 . . . �c5 43 ':c3+ �xb5 44 �e5, intending l:tc8-g8.
36 . . . g6?
A very strange move, on which Sash a makes no comment. It is clear that Black will have to advance his kingside pawns, so why not advance the pawn two squares, why waste a tempo? It is quite possible that a race will develop, in which every tempo will count. Even if the delay does not affect the assessment here, the next time it will . O f course, 3 6 .. . g 5 ! ? was stronger than the move in the game. 36 .. .f5!? also looks tempting, intending to bring the king out to f6 followed by . . . h7-h5-h4, and . . . e6-e5-e4. For example: 37 a5 (37 l::tb7 a5!?) 37 .. . �f7 38 l:tb6 �e7! (gaining another tempo) 39 :b7+ �f6 40 l:tb6 l::td2 41 b4 l:1d4.
37 :b8+
Here Chernosvitov makes the following comment: 'It would appear that White could have immediately gained a draw by 37 a5 :d5 38 :b8+ �g7 39 b4 �f6 40 l:tb6 l:td4 4 1 ':'xa6 l:txb4 42 l:ta 7' .
Roughly such a position can arise in many variations and its assessment is important for the correct understanding of the entire endgame. But is it really drawn? The black
rook is excellently placed to the rear of the passed a-pawn . I don't see what can be done to oppose the advance of the pawn armada on the kingside. For example: 42 . . . h5 43 a6 l:ta4 44 �g3 h4+ 45 �f3 e5 46 ':'a8 (46 �e3 l:ta2) 46 .. . �f5 47 a7 l:ta3+ 48 �f2 g5 and 49 .. . �f4. Note the good position of the f7 -pawn on its initial square - after 49 ':'f8 ':'xa7 it will be defended by the rook. 48 .. .f6? would be a typical mistake - now after . . . �f5-f4 the reply l:ta8-f8 gains in strength .
37 . . . 38 a5
�g7 l:td2
Black's other plan is to bring his king to the centre of the board . But in this case he will have to give up one or two pawns on the kingside. Here is Chernosvitov's analysis: 38 . . . �f6 39 l:lb6 �e5 40 l:tb7 f5 4 1 l:txh7 'ud5 (4 1 . . .�d4 42 l:tb7 e5 43 l:tb6 �c5 44 l:txd6 �xd6 45 g4! fxg4 46 hxg4 �c5 47 �g2 , and the pawn endgame is drawn ) 42 b4 l:.b5 43 ':'a7 ':'xb4 44 l:txa6 l:ta4 45 l:ta8 �f4 46 a6 e5 47 a7 e4 (47 . . . g5 48 g3+ �e4 49 11g8 or 48 .. . �f3 49 l:tf8) 48 l:tg8 l:txa7 49 l:txg6 e3 50 ':'e6 with a draw.
This variation is interesting, but by no means forced . At the very end instead of 49 .. . e3? there is the far stronger 49 .. . �e3!' I am not sure that White can save himself here - the e-pawn really is too strong . On the other hand , it is not altogether clear why he initially wasted time advancing his a-pawn, and only then went for the g6-pawn . In reply to 45 .. . �f4 either 46 :a6 e5 47 l:txg6 or 46 l:te8 e5 47 g3+ suggests itself.
39 �g3 �f6
40 b4 41 l:tb6
':'b2 �e5? It is more natural to advance the kingside pawns: 4 1 . . .h5 or first 4 1 . . .g5. I ncidentally, after . . . h7-h5 the king move to e5 gains in strength - since when the white rook steps onto the 7th rank, Black simply replies
Analysis of a Game
ltJ
1 89. . . f7 -f5 and the h-pawn will no longer be under attack.
42 1:[b7
Chernosvitov comments: 'Neither White nor Black saw the "two-mover" 42 ':'xa6 1:[xb4 43 1:Ia7 with a draw'. Well, we have already talked about such a position - in fact after 43 .. . �f6 Black should most probably win . 43 .. .f5 4 4 ':'xh7 ':'a4 4 5 ':'a7 g5 followed by . . .1::t a3+ is also tempting.
When analysing endings it is very important to make a correct assessment of key positions, which are reached from different variations. Sash a made a mistake in his assessment of one such position , and as a result his perception of the entire endgame was distorted .
I should mention that White does not have time to take his king across to the queenside. Here are some sample variations: 42 �f3 f5 43 �e3 g5 44 �d3 f4 45 �c3 l:txg2 46 ':'xa6 f3 47 1:[a8 (47 �d3 .:tb2 or 47 .. . h5) 47 .. . h5 48 a6 :a2 49 �b3 :'xa6! 50 :txa6 f2 51 :a 1 g4 52 hxg4 hxg4 53 b5 �d5! 54 :1c1 g3 55 b6 g2 56 b 7 f1 'it' 57 b8'ii' 'it'xc1 , or 48 b5 :1g 1 49 :f8 �e4 50 b6 (nothing is changed by 50 �c2 l:ta1 or 50 �b2 l:.d 1 5 1 b6 %:td5) 50 .. J:tb1 51 �c2 l:.b5 52 �d2 (52 a6 ':'xb6 53 a 7 .1:.a6 54 a8'i1i' + l::txa8 55 %:txa8 f2 56 ':f8 �e3) 52 .. . g4 53 hxg4 hxg4 54 a6 %lb2+! 55 �d1 (55 �e1 �e3) 55 .. . g3 56 a7 1:[a2.
42 . . . f5 43 l::txh7
The delay in playing . . . h7-h5 has told. The outcome of the game is now in question .
43 . . . �e4
44 l:th4+?
A decisive loss of time. And meanwhile it is possible that there was no longer a win after any of the 'normal' continuations: 44 1:Ia7 f4+ 45 �h4 l::txg2 46 l::txa6, 44 .l:r.g7 f4+ (or 44 . . . l:txb4 45 ltxg6 �e5 46 h4) 45 �h2 f3 46 :txg6 e5 47 %:tf3 (47 �g 1 ), or 44 h4 f4+ 45 �h3 e5 46 ':a7 �e3 47 l::txa6 e4 48 %lxg6.
44 . . . �e3 45 1:1c4
46 l:tc3+
e5
White also loses after 46 l::tc6 f4+ 47 �h2 e4 (threatening 48 .. .f3) 48 �g 1 l::tb 1 + 49 �h2 l::txb4 followed by . . . �f2 and . . . e4--e3.
46 . . . �e2 47 ':'c6
48 :1xg6
1:[xb4 48 %1e6 f4+ 49 �g4 f3+ 50 �g3 f2 51 l::txe5+ �f1 , and the f2-pawn inevitably promotes to a queen. 48 . . . f4+ 49 �h2 e4 50 :1xa6 e3 51 l:tb6 11a4 52 a6 �f2 White resigned .
When analysing complicated endings, we sometimes stumble upon positions which , possibly, do not arise by force, but which are interesting in their own right. One such position is examined by Chernosvitov.
I n his opinion this is a position of mutual zugzwang. Well, it is not hard to see that if it is White to move he loses (1 g3 f3, 1 h4 g4, or 1 l::tgS ':xa7 2 ':xg5 'iti>f2 3 l::tf5 f3 4 gxf3 e3). But I cannot agree with the conclusion that with Black to move it is a draw. The white king is really very awkwardly placed . Black plays 1 . . . :ta2 2 'iti>g 1 l:Xa6! 3 'iti> h 1 (3 'iti>h2 ::ta1 ) 3 . . . g4! 4 hxg4 'iti>f2 5 g5 'iti>g3 with unavoidable mate.
There is also other, more spectacular solution : 1 . . . 'iti>f2 2 .:teS g4! 3 h4 (3 hxg4 lba7) 3 . . . l:ta3! (another way is 3 .. .f3! 4 as. l:th 1 + ! ! ) 4 'iti> h 1 (if 4 as. or 4 l:txe4 there
follows 4 . . . l::th3+! 5 gxh3 g3+) 4 . . . l::txa7 (4 .. . g3! 5 as" f3! mates more quickly) 5 l::txe4 l::ta 1 + 6 'iti>h2 g3+ 7 'iti>h3 f3 S :f4 l::th 1 + 9 'iti>g4 'iti>xg2 1 0 l::txf3 l::txh4+ .
It is time to sum up. Chernosvitov played the second half of the game uncertainly, and also his commentary, although detailed, was not too successful. Here two serious deficiencies of his are clearly seen:
1 ) Weak conversion of an advantage. Remember: in a winning position Black incorrectly allowed his opponent to sharpen the play. Then he did not even try to figure out the resulting complications. Finally, after incorrectly resolving an exchanging prob lem, he went into an endgame instead of playing for mate. Later any methodical player would surely have advanced his g pawn two squares, instead of one, and he would also have chosen an appropriate moment to advance his h-pawn , whereas Sasha left it at h 7, where it was lost. As a result, the opponent gained real chances of saving the game.
2) Uncertain orientation in rook endings. Chernosvitov overlooked some typical ideas and plans, and his general assessments and specific recommendations often proved incorrect.
Sasha can be advised to make a serious study of rook endings, and even better - of the theory and technique of the endgame as