The ten-player supergrandmaster tournament in Montreal was immodestly de scribed on the cover of the French-language tournament book as the most impor tant in the history of chess. With an average Elo rating of 2622 and a prize fund of $ 1 10,000, it sounded tremendous; yet for such an important tournament there were serious organisational faults. lt was played on the island of St. Helena at the mouth of the St. Lawrence River, the site of Expo 1967. The tournament hall was a fairly large theatre in a somewhat decaying building. The drafts were so strong that a few players sometimes put on their coats during the game. Often the boards and pieces were not set up until five minutes before the beginning of the round. Apparently, the tasks had not been very efficiently divided among the local organ isers. Or did they find it beneath their dignity to perform such chores? The arbiter, Svetozar Gligoric, usually had to do most of the work, but once I saw Spassky carrying the boards into the hall while Hort was setting out the name cards.
But let's forget all that. The attractive prize fund in any case ensured that the games would be extremely hard fought. Witness the following game .
1 d4 ltlf6 2 c4 g6 3 lL!c3 dS 4 cxdS lLlxdS s e4 ltlxc3 6 bxc3 ..tg7 7 i.c4
The Exchange Variation is still one of the most dangerous weapons against the Griinfeld Defence, but nowadays 7 ltlf3 is also played regularly; e.g., Karpov-Ljubojevic, Montreal 1 979: 7 ltlf3 c5 8 ..te2 0-0 9 0-0 ..tg4 10 ..te3 'WaS 1 1 'Wb3 cxd4 12 cxd4 lLlc6 1 3 .l:tadl 'Wb4 1 4 h3 ..txf3 15 i.xf3 .l:tfc8 1 6 'Wxb4 lLlxb4 17 e5 .l:tc7 1 8 .l:tc 1 .J:r.ac8 19 .l:txc7 .l:txc7 20 .l:tb 1 ltlxa2 2 1 .l:ta1 ! . 7 cS 8 lL!e2 0-0 9 0-0 lL!c6 10 ..te3 ..tg4 1 1 f3 lOas
Delaying the exchange on d4 has a dual purpose: first, White cannot pro tect his bishop on c4 with .l:ta1-c 1 ; sec ond, the variation with .ic4-d5 is not so attractive for White with the c-file closed. Thus White is more or less forced to play the classical exchange sacrifice after 12 .id3 cxd4 1 3 cxd4 .ie6 14 d5 . But Spassky, who has
been playing the Exchange Variation against the Griinfeld ever since his youth, has never shown any interest in the exchange sacrifice. He chooses an other continuation, which is associ ated with great risk.
12 i..dS i..d7 (D)
13 i..g5
The usual move is 1 3 l:tb1 , as in Hort-Timman, Niksic 1 978. After 1 3 l':.b1 'ii'c8 14 dxc5, Black could have had a good game with 14 . . . l:td8, as Hort suggested after the game. Need less to say, taking on c5 meant that White's strategy was a failure. There were no good alternatives, however, because Black threatened to get the better position with 14 . . . e6 15 i..b3 tLlxb3 .
Spassky therefore chooses a con tinuation that lets him keep his bishop on d5 for a while. It is a remarkable idea because the move i..g5 in con nection with the pawn exchange on d4 (and therefore played on move 14) is never mentioned as an alternative to
the usual 1 4 l:[b 1 (after the pawn ex change). Even now, after 1 3 .. . cxd4 14 cxd4 White's position remains very good and he can sacrifice his d-pawn with little to worry about. You can see that after 14 . . . i..b5 15 l:tb1 i..xe2 1 6 'ihe2 i..xd4+ 17 �h 1 Black has seri ous problems.
13 • • • i..bS
14 l:tb1 i..a6
Not so good is l4 ... 1i'd7, because after 15 l:txb5 'ii'xb5 1 6 i..xe7 White clearly has the better chances. But now Black is ready for l5 .. . 'ii'd7 fol lowed by 1 6 .. . e6, after which White's position would collapse. Therefore, the following supersharp attacking at tempt is born of necessity.
15 f4 'ii'd7
Black can insert l5 . . . h6 here, at a moment when White is forced to move his bishop to h4. But my intui tion warned me that it was too risky, and a closer look shows that this was the correct evaluation; after 15 ... h6 1 6 i..h4 'ii'd7 17 f5 gxf5 White launches an offensive full of bold sacrifices: 1 8 tLlg3 ! e6 19 ttlh5 ! ! (D) and now:
1) 19 ... i..xfl 20 i..f6! exd5 2 1 i..xg7 and White's attack quickly becomes decisive. The imperturbable manner in which White's attack is conducted in this variation is rather typical of this position; Black's pieces are active and well co-ordinated, but they are not well placed for defence.
2) 1 9 .. . f6 20 l:txf5 ! exd5 2 1 i..xf6 i..xf6 22 tLlxf6+ l:txf6 23 l:xf6 'ii'g7 24 e5. Although the attack has been
8
temporarily halted and the material situation has not turned out too badly for Black, White has the better chances because Black's minor pieces cannot be brought onto the battlefield very easily.
16 rs gxrs 17 l:.f3 (D)
It took Spassky quite a while to find this unsophisticated continuation of the attack. Now 17 lilg3 e6 1 8 lilh5 is not sufficient because of 1 8 .. . f6. Al though 19 l:.xf5 is still possible (as in variation 2 above), under these circum stances it is just too fantastic: Black reacts cold-bloodedly with 1 9 ... exd5
20 i.xf6 .ixf6 2 1 lilxf6+ l:.xf6 22 l:.xf6, and with the pawn on h7 instead of h6, Black's king position is ade quately defended and he has time for 22 ... dxe4 with a winning advantage.
17 ... fxe4
It took me almost an hour to find a fully satisfactory defensive scheme. I initially considered 1 7 .. . e6, but after I found the complicated refutation 1 8 l:.g3 �h8 ( 1 8 .. . f6 i s insufficient due
to 19 i.h6 l:.f7 20 lilf4 ! exd5 2 1 lilh5 with a winning attack) 19 lilf4 exd5 20 'ii'h5 fxe4 2 1 l:.h3 1i'f5 22 g4 ..ie2 23 lilxe2 1i'g6 24 1i'h4 e3 25 l:.fl , with the unavoidable threat of 26 lilf4, I re alised that 1 9 exf5 'iWxd5 20 f6 pro duces a much more plausible win. Remember this long variation, how ever, for it will come up again.
Choosing1>etween the text and the insertion of 1 7 .. . h6 gave me the big gest headache. After the subtle reply 1 8 .te l ! Black's position is critical. Taking on e4 is now forced, and after
1 8 ... fxe4 19 l:.g3 (D) I investigated:
1) 19 ... 'ii'xd5 20 .txh6 'fih5 2 1 .txg7 'ii'xe2 2 2 'ii'c 1 ! with the devas tating threat of 23 'fih6.
2) 1 9 .. . �h7 20 lLlf4 .td3 21 'ii'h5 .txbl 22 l:txg7+ �xg7 23 ltle6+ and White has a mating attack.
18 llg3 (D)
B
18 • • • .txe2!
A necessary exchange, because the white knight threatened to enter the thick of the fight by going to f4. Nev ertheless, after 18 ... �h8 19 lLlf4 Black is not lost because he still has the fi nesse 1 9 .. . .td3 . White's attack is not yet strong enough for him to sacrifice a rook: 20 1i'h5 .txbl and if now 2 1 .txf7 llxf7 22 'ii'xf7 llf8 or 2 1 .th6 .txh6 22 'ii'xh6 llg8 23 .txf7 e3 or 2 1 .te6 .txd4+ ! 2 2 cxd4 'ii'xd4+ 2 3 �h 1 .txa2, White's attack is repulsed and Black's material plus will be decisive. But the rook sacrifice is unneces sary. White's pieces occupy ideal at tacking positions and he can take the time to play 20 llc l . The most impor tant strength of Black's defensive
set-up is the fact that his e- and f pawns are not yet advanced, but aside from that it is surprising how few de fensive possibilities Black has. The main variation is 20 llc l 'ii'f5 2 1 'ifh5 (threatening 22 .th6) 2 1 ...e6 22 llh3 exd5 23 g4 .te2 24 ltlxe2 'fi' g6 25 'ii'h4 and we get almost the same posi tion as in the note after Black's 17th move. After 25 . . . h6 26 lLlf4 'ii'h7 27 .txh6 .txh6 28 1i'f6+ �g8 29 llxh6 White's attack has reached storm pro portions.
19 'ifxe2 �h8 20 .txe4
One of the ideas of Black's defence becomes clear after 20 'ii'xe4 f5 2 1 'ii'h4 'ii'xd5 22 llh3 'ii'g8 and Black wins.
20 . • .
21 .trJ
rs
Quite rightly refraining from fur ther attacking tries and placing the bishop on the long diagonal where it can exert the most pressure.
21 • . . cxd4 (D)
There is no time to protect the e pawn; after 2 1 . . .e6 22 dxc5 White's bishop-pair comes to life and after 22 ... .txc3 he would have more than sufficient compensation for the pawn .
22 .txe7!
The only way to stay in the game. The endgame after 22 'ii'xe7 'ilxe7 23 .txe7 llfe8 24 .tb4 ltlc6 offers White little chance of survival.
22 .•. d3
The beginning of a forced simpli fication leading to a roughly equal
w
endgame. The point of White's play becomes clear after 22 ... l:tf7 23 l:txg7 ! l:txg7 24 .tf6 dxc3 25 .txc3. White does not hurry to regain the exchange, while Black's task of defending him self becomes extremely difficult and unpleasant.
A good alternative is to sacrifice the exchange by 22 ... dxc3 (D):
w
If White accepts, then after 23 i.xf8 l:txf8 Black has excellent pros pects: his king is safe, and White's rook on g3 is therefore out of play. The advanced passed pawn on c3 is
extremely unpleasant for White. The attempt to improve White's play with 23 l:td 1 fails: 23 ... .td4+ 24 �fl l:r.ae8 ! 25 i.xf8 l:txe2 26 i.xe2 c2 27 l:tc l fie? and Black is winning. White's best bet is to sacrifice the exchange with 23 l:txg7. After 23 ... 'ii'd4+ 24 �fl "flxg7 25 .i.b4 ! c!Oc6 (25 . . . c2 26 l:tc l "flb2? 27 "fld2 is senseless, and White wins) 26 .i.xc6 bxc6 27 l:tb3 ! Black has nothing better than 27 . . . l:tae8, which gives White the opportunity to simplify to a drawish endgame with 28 i.xc3 l:txe2 29 �xe2 (29 ... l:tf6? fails to 30 l:tb8+ ), and with nothing but isolated pawns, Black has no real istic winning chances.
23 .txrs dxe2
Sacrificing the exchange now is out of the question as White immediately sacrifices back: 23 .. . l:txf8 24 l:txg7 (24 ... �xg7? 25 "fle5+ or 24 ... "flxg7 25 "flxd3 with a positional advantage for White). w 24 i.xg7+ 25 l:txg7 'it'xg7 �xg7 (D)
Attack and defence have balanced each other, and if White had now chosen 26 .i.xe2 a cease-fire would probably have been signed in short or der. One possibility is 26 .. . l:r.c8 27 l:tb5 b6 28 l:r.xf5 l:r.xc3 29 .i.f3; if Black now risks winning the a-pawn with 29 ... l:r.c l + 30 'ii.>f2 l:r.c2+, White becomes too active with 3 1 'ii.>e3 l:r.xa2 32 .i.d5 . The safest plan is to bring the knight into play immediately with 29 ... llX:4.
Instead, Spassky begins to play carelessly.
26 c,Pf2 l:r.c8
Based on a neat trap: 27 .i.xb7? fails to 27 . . . .1:.b8 28 .i.e4 e 11i'+ ! 29 'ii.>xel l:r.e8 winning a piece.
27 'ii.>xe2 b6 28 l:r.dl
White gives up a pawn to activate his rook. He is in no danger of losing because of this move, but neither is there any clear reason to play it. After 28 l:r.c l 'ii.>f6 Black has a somewhat better position due to his well-co-ordi nated pieces, but it would not be enough for White to worry about.
28 :.XcJ
29 :d7+ 'ii.>g6 30 l:r.xa7 l:r.c2+ 3 1 'ii.>d3 l:r.xa2 32 l:r.d7
White's rook is again in the most active position. The direct attack 32 l:r.a6 is bad due to 32 .. . l:r.a3+ 33 'ii.>c2 /Oc4 with excellent winning chances for Black.
32 • • • b5
33 l:r.d6+ <ii>g5 (D)
34 l:r.b6?
This is not the best way to exploit the active position of White's pieces. With 34 l:r.d5 b4 35 .i.e4 he can win the f-pawn and simultaneously bring his bishop to the bl -h7 diagonal, neu tralising Black' s passed b-pawn.
34 • • • llb2
35 g3
White's best chance under the cir cumsta.nces. The mating threat 36 h4 forces Black to exchange a pair of pawns on the kingside. Black, it is true, gets rid of one of his isolated pawns, but it is more important that the total number of pawns is reduced and that, except for the passed b pawn, Black is left only with the insig nificant h-pawn. 35 36 'ii.>e2 37 l:r.d6 l:b3+ f4
Forcing the capture on g3 . Too am bitious is 37 h4+ 'ii.>f5 38 g4+ 'ii.>e5 39 l:r.h6 as after 39 ... 10c4 40 l:r.xh7 l:r.e3+
4 1 �f2 �d2 White will be pushed even farther back, and he cannot ex change rooks (42 l:.e7+ �d4 43 l:.xe3 fxe3+ 44 �e2 �xf3 45 �xf3 �d3 and Black wins).
37 38 hxg3 39 l:.dS+ 40 �! (D) fxg3 �4 �f6
Now that he is in danger, Spassky defends himself very well. This move protects the g-pawn and frees his bishop to go either to e2 or to e4, de pending on Black's moves.
8
40 • • • l:.b2+?
It happens quite often that a player makes a mistake on his 40th move. The text-move helps only White, and Black's winning chances melt like snow in the sun.
40 .. . �e5 is no more effective, be cause of 4 1 .ie2. Black should have realised that f5, not e5, is the ideal place for the knight and that 40 ... �3 ! is the correct move. After 4 1 l:.d7 h6 Black maintains winning chances; for
example, 42 l:.h7 �g6! 43 .ie4+ �g5 44 l:.g7+ �f6 45 l:lg6+ �e5 and a draw is nowhere in sight for White: Black consolidates a solid extra pawn. The exchange sacrifice 4 1 l:.xb5 l:.xb5 42 �xe3 l:.b2 43 .ie2 �e5 is also not a watertight method of draw ing. White's best chance is 4 1 l:.c5 !, as Spassky suggested later, in order to keep as many checking options as pos sible. Black still has practical chances after, say, 4 1 ...b4 42 l:.c7 h6 43 l:.c6+ �e5 44 l:lxh6 �f5 followed by 45 . . . l:.b2+, but it is doubtful whether they are very real.
This ending bears a striking resem blance to an ending from the first game of the Spassky-Petrosian World Championship Match in 1 969. After Black's 5 1 st move the following posi tion arose:
w
After putting up a heroic resistance, Spassky went wrong with 52 l:.h6+ and had to resign after 52 .. . �e5 53 l:.b6 �4 54 l:.e6+ 'ii?d4 55 l:.e4+ �c5 56 l:.xa4 l:.al . Later analysis showed
that White could have drawn with 52 <ili'e3 ; e.g., 52 ... l0a4 53 .l:r.h4 ! l0c3 54 l:tb4 l0d5+ 55 ..txd5, etc.
Although I knew during the game that my practical chances were not as good as Petrosian's were, I still had to play the game and hope that it would turn out just as well for me.
41 ..te2 b4 42 <ili'f3
The sealed move. Home analysis revealed that Black has no winning chances.
42 • • . l0e5+
42 .. . l0a3 43 .id3 achieves even less.
43 �e3
White's pieces are again working in harmony, and Black has no objective winning chances at all. But I had one more trap up my sleeve.
43 .l:tc2
44 .l:tb5 o&:4+ 45 �d3
Exactly what I was hoping for! Af ter 45 �f3 Black has no possibility of making progress.
45 • • • .l:td2+
46 �xc4 J:txe2 (D) The outlines of the trap are becom ing visible. No matter how White cap tures Black's b-pawn, the resulting pawn endgame is a win for Black, as can easily be seen. Unfortunately for Black, the rook ending is still a draw.
47 .l:th5
The most convincing method. It would have been more difficult after 47 �d3 .l:tb2.
47 �g6
Unfortunately, 47 ... .l:te4+ 48 <ili'd3 �g6 is not playable because in the pawn end
g
ame after 49 .l:txh7, etc., White's king is within reach of the b pawn's queening square.48 lth4 (D)
Spassky needed a good half hour to convince himself that this move guar anteed the draw. After the game he said he feared that 48 .l:th3 might give Black winning chances after 48 .. ..l:te4+ 49 <i!tb3, but in the post-mortem we couldn't find any. The main variation we examined was 49 ... h5 50 !ita4 �g5 S l .l:th l lteS (threatening 52 ... �g4) 52 l:.h4 b3 53 'iPa3 ! .:te3 54 .l:th3 and Black has made no progress. It is re markable how the seemingly weak po sition of the white rook on h3 still manages to spoil whatever winning chances Black might have.
But a closer analysis shows that Spassky's intuition did not fail him. Black must not play the routine 49 ... h5 ? but must reserve that square for the rook; so 49 ... J:r.g4 SO �a4 .l:tgS !
5 1 Wxb4 llh5. In this position Black is again headed for a winning pawn end game.
48 llg2
48 . . . .:.e3 is stronger, but even then the draw is inevitable if White plays accurately. The next few moves are forced: 49 .:.g4+ �f5 50 .:.g7 (not 50 .:.h4 l:.e4+) 50 ... .:.e4+ 5 1 Wd3 h6
(5 1 . ..h5 52 .:.h7 Wg6 53 .:.xh5 draws immediately) 52 .:.h7 .:.e6 53 .:.b7 Wg4 54 l:lxb4+ Wxg3 55 .:.b5 ! Wg4 (D) and now it all depends on this po sition:
During the post-mortem analysis Spassky suggested 56 Wd2 h5 57 .:.bt ,
w
but now after 57 . . . Wf3 Black cannot exchange rooks with 58 .:.et . The cor rect method is 56 .:.b4+, and now after 56 .. .'itf3 White can let his king be cut
off because 57 l:.h4 l:.d6+ 58 Wc3 Wg3 59 l:.h l leads to a theoretical draw.
49 l:lg4+ eMS 50 .:.h4
With the black rook on e3 this would not have been possible (50 .. . .:.e4+ );
with the rook on g2 it leads directly to a draw. Only after 50 .:.g7 h5 followed by 5 l ... h4 could Black have won.
50 b3
51 .:.xh7 b2 52 .:.b7 1fz.1/z