A Guided Tour of Chess - The Chess Cafe, June 2000

The most elementary of all tactical devices is probably the fork. But however elementary, it was hotly discussed recently in the electronic newsletter of an Amsterdam chess club.
    This started when one contributor, commenting his Qd6 which attacked two Rooks on b8 and f4, innocently called that a fork. In the next issue, somebody else protested that this wasn't a fork at all because the teeth of a real fork do not point away from each other. When the discussion really got going, it turned out there were vastly different views about what a fork really is. Some thought both teeth of a fork must point forward; that only Knight and Pawn can fork; that the King can do it too, but not the Rook, the Bishop or the Queen; that a fork must be symmetrical; that empty squares can be forked too; that the attacker must be of less value than the attacked, 'because a fork cannot prod anything smaller than its teeth.' There was also somebody who thought a fork is only a fork when the forking piece moves to the square where it forks.
    That's something I want to refute at once. (See Diagram)
Black to play; NN - Mannheimer, Frankfurt am Main 1921. With 1...Re4!, Black activated the fork that Ke5 already gave. White resigned; he loses a piece.
    Just like the readers of that newsletter, chess encyclopedias disagree about the fork. They all say it is a double attack in different directions by one piece or a pawn, but sometimes that is the whole definition (and in that case empty squares count too, making any move, except h2-h4 and Ba1-h8, a fork) and sometimes again, it is only the Pawn which can fork.
    In my own definition that I use here, not so much to be pedantic as to limit the subject, I'm adding a clause that I saw nowhere else: the double or multiple attack must not be aimed at the King; in that case it is a 'Family Check'.
    The basic fork, agreed, is a pawn prodding two pieces; one of the first tactics a beginner encounters. Often a gross blunder allows one, and one soon learns the part it plays in standard pseudo sacrifices like 1.e4 e5 2.Nf3 Nc6 3.Nc3 Bc5 4.Nxe5 Nxe5 5.d4
    An elementary but surprising serial fork happened in Grigoriev - Panikovsky, Kurgan 1972: 1.e4 c5 2.Nc3 Nc6 3.g3 g6 4.Bg2 Bg7 5.d3 d6 6.f4 Nh6 7.Nf3 f5 8.O-O O-O 9.h3 Nf7 10.Be3 Nd4 11.Qd2 Rb8 12.Bf2 fxe4 13.Nxd4 A Zwischenzug. (See Diagram) But Black had the Zwischen-Zwischenzug 13...e3! and after Q or Bxe3 there would follow something resembling a multiple jump in draughts; cxd4 and dxe3 or dxc3. After 14.Ne6 exf2+, White resigned.

Another sort of serial fork is seen in the next diagram. In Kochiev - Tal, Moscow 1981, Black played 25...d2, forking the Rooks, but also vacating d3 where a second fork of those same Rooks will follow: 26.Qxd2 Nd3 winning the exchange and a little later the game.

A funny case of a fork met by a counterfork: (See Diagram) White to play; Tretyakov - Zhuravlev, Riga 1978. White saw a fork which wins a pawn: 1.dxe5 dxe5 2.Qc3 But he didn't see the counterfork: 2...Bb4 3.Qxe5 Qxe5 4.Nxe5 Bc3 which wins a piece.

An unusual fork happened in Razuvajev - Mestrovic, Keszthely 1981 (See Diagram) With 38.Nh7!, White won the exchange (Rxc8 39.Nxf6+) and soon the game.

And the dumbest fork that I ever saw happened in a game of an (ex-) world champion. In Christiansen - Karpov, Wijk aan Zee, 1993 (See Diagram) Black played 11...Bd6, and resigned after 12.Qd1.

There are several standard middlegame traps which are based on forks - here is one which has made many victims over the years. It was first seen in Norman - Vidmar, Hastings 1925: 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.Nf3 O-O 5.e4 d6 6.Bd3 Bg4 7.h3 Bxf3 8.Qxf3 Nc6 9.Be3 Nd7 10.Ne2? (See Diagram) And now 10...Nce5 won a pawn, and destroyed White's position. After 11.dxe5 Nxe5 12.Qg3 Nxd3+ 13.Kf1 c5 14.h4 Qd7 15.h5 Qe6 16.Rh4 Qxc4 17.hxg6 fxg6 18.Qh3 Nxf2 19.Bxf2 Bd4, White resigned.
    The exact position of the diagram has occurred more often - I found 12 games with it, the last one from 1998. In a slightly different form, in another opening, it has also happened 13 times - at surprisingly high levels. The premiere of that fork was:
Ljubojevic - Timman, Tilburg 1978: 1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.f4 Bg7 5.Nf3 c5 6.dxc5 Qa5 7.Bd3 Qxc5 8.Qe2 O-O 9.Be3 Qa5 10.O-O Bg4 11.Qf2 Bxf3 12.Qxf3 Nc6 13.Ne2 Nd7 14.c3 (See Diagram) 14...Nde5 15.fxe5 Nxe5 16.Qh3 Nxd3 17.Qd7 Qa6 18.Qxe7 Rae8 19.Qh4 d5 20.Ng3 f5 21.e5 f4 22.Bxf4 Qb6+ 23.Kh1 Nf2+ 24.Rxf2 Qxf2 25.Rf1 Qxb2 26.Qg5 and White resigned.
    You'd think this would have been a good warning, but with a minor modification (pawn h2 being at h3 because of the intermediate 11.h3 Bxf3 12.Qxf3 etc.) this has also happened in, among others:
Zapata - Chernin, Cienfiegos 1981
Hübner - Korchnoi, Skelleftea 1989
Korneev - Van Wely, Krumbach 1991
    All together, I found 34 games where this Ne5-fork was possible. It was missed 6 times - never in master games, except for this one:
Gligoric - Pirc, Bad Pyrmont 1951: 1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.f4 Qc7 7.Bd3 Bg4 8.Nf3 Nc6 9.h3 Bxf3 10.Qxf3 e6 11.O-O Be7 12.Be3 O-O 13.Ne2 Nd7 14.c3 (See Diagram) 14...Rad8 That 14.c3 looks like a trap, but I don't see why Ne5 couldn't have been played: 15.fxe5 Nxe5 16.Qg3 Nxd3 17.Bh6 Qc5+ followed by Qe5, and Black seems to be better. In reality, White won after: 15.Rad1 d5 16.e5 Bc5 17.Nd4 Bxd4 18.cxd4 f5 19.g4 g6 20.gxf5 gxf5 21.Kh1 Kh8 22.Rg1 Rg8 23.Qh5 Rde8 24.Qh6 Nd8 25.Bf2 Nf8 26.Bh4 Rg6 27.Bf6+ Kg8 28.h4 and Black resigned.

Indredibly, even one world champion fell victim to this fork, be it in a rapid play-off game. In Anand - Kramnik, Mainz 2001 (see diagram), Black played 16...Rb8, with a threat that Anand missed. 17.O-O Nce5 Winning a pawn, because after 18.fxe5 Nxe5 19.Qg3 Nxd3 the c2-pawn was overloaded. After 20.cxd3 Rxb3, Anand somehow miraculously managed to draw.
    This is the only extended example of the Ne5-fork - it does not regain the piece immediately, but exploits a defensive fork, as White's overloaded pawn c2 might be termed.

For forks aimed purely at winning material there is nothing like this study: (See Diagram) White to play and draw; A. Herbstmann & L. Kubbel, 1st Prize, Troitzky-Tourney, 1937. White cannot stop the pawn by ordinary means; moves are dictated by family checks on f3; knight-stalemates; pinning stalemates by a new Queen on e1; and by the knowledge that the endgame of three Knights against one Knight is won - something first demonstrated by Troitzky. 1.Ng1 Ne3+ (Nf4+ 2.Kh1 e1N 3.Nf3+ Nxf3 stalemate) 2.Kh3 Nf4+ (2...e1N 3.Nf3+ Nxf3 with a beautiful three-Knight stalemate) 3.Kh2 Ng4+ (Not 3...e1N 4.Nf3+ Nxf3+ 5.Kg3 and White forks two Knights; or again 3...Nf1+ 4.Kh1 e1N 5.Nf3+ Nxf3 stalemate) 4.Kh1 Nf2+ (Again both promotions are, or lead to stalemate) 5.Kh2 e1N Finally. No stalemate seems in sight. 6.Nf3+ Nxf3+ 7.Kg3 (See Diagram) A fork of three Knights - Black needs them all to win. So he must play 7...Ke3, but that is stalemate once again. Witty and beautiful, this belongs to my 10 favorite endgame studies.

A unique sort of fork is seen in this amazing study. (See Diagram) White to play and draw; F. Simkhovich, 1st Prize Turkmenskaya Iskra, 1940. After 1.cxd4 Rxe4 Black wins, and after random Bishop moves, the Rd4 just moves away, and again Black wins. 1.Bf5! however, restricts Black. 1...Rde4 then fails to 2.Kh3, and 1...Rg5 to 2.g7 Kxf7 3.g8Q+ Kxg8 (the Rg5 must keep attacking the Bf5) 4.Be6+, and White has time for cxd4. Therefore, Black must choose between 1...Rc4 and Ra4. But wherever the black Rook goes, there follows a sort of perpetual fork, e.g. 1...Rc4 2.Be6 Kf8 (2...Ra4 3.Bd7 etc.) 3.Kh3 Rge4 4.Bd5 Kg7 5.Kh2 Ra4 6.Bc6 Rec4 7.Bb5 Rg4 8.Bd7 Rae4 9.Bf5 and so on, always keeping both Rooks under attack and never capturing one. For each pair of squares that the Rooks have, there is a corresponding square for the Bishop. And if Black makes his waiting move again; 9...Kf8, White makes his waiting move: 10.Kh3! and the double attack continues. e.g. 10...Kg7 11.Kh2! Rc4 12.Be6 etcetera.
    That is not all in this ingenious study. One might wonder why 1.Bf3 does not work. The same perpetual fork seems possible from the other side; 1...Rc4 2.Be2 Ra4 3.Bd1 Rge4 4.Bc2 Rec4 5.Bb3 etc. In that case however, Black exploits a hidden weakness in White's position: after 1.Bf3? he plays 1...Ra4 (he can also shuffle around for a bit first) 2.Bd1 Kf8! and now it turns out that White's waiting move has a surprise drawback: 3.Kh3 Ra1! 4.Bxg4 Rh1 mate.

In the middlegame, an appealing sort of fork can be aimed at a Queen and a Rook defending against a back rank mate together. Two well-known examples:

Bernstein - Capablanca
Moscow 1914
Mikenas - Bronstein
Tallinn 1965

Bernstein had just captured a pawn on c3 and had to resign after 29...Qb2! while Mikenas, who had played 24.Rh4-b4, had certainly seen that Qe1+ was defended by Qf1, but not that 24...Rxa3! was a deadly fork, which forced him to resign.

A comparable fork, also indirectly aimed at mate, is hidden (not so well, you'd think) in the following position (See Diagram): Black to play, Alden - Nilsson, Sweden 1972. After 1...Qb7? 2.Qf1 White won, but with 1...Qc6!, Black could have turned the tables.

This type of fork, where two line-pieces are forked on the same line, is often motivated by a back rank mate as well, but of course it can have other points. (See Diagram) White to play; Csulits - Bade, GDR 1972. White is a piece and three pawns up, but the threatening black King makes his position very dangerous. And in fact, he lost after 37.Qd6? h5! 38.Be5? (38.Rd2 still draws) 38...Kh3! 39.c6 Rxh2+ and White resigned; it is mate next move. In the diagram, 37.Rc1 is winning, but much more forceful would have been 37.Rd2! - Black would have had to resign immediately; his forked Queen and Rook are overloaded.

    And a clear, be it somewhat schematic example from an endgame study: (See Diagram) White to play and win; V. Kalandadze, 1965. Direct attempts at promotion only lead to winning a Rook and losing all the pawns, and therefore to a draw. With the obvious 1.Re1 White forks the Rooks which are each tied to one white pawn, and that seems to be it: Black takes pawn; White takes Rook and wins. But Black has a temporary escape: 1...Rb2+ 2.Kxa3 Rff2 Which only leads to the next fork: 3.Re2! And so on: 3...Rb3+ 4.Kxa4 Rff3 5.Re3! Rb4+ 6.Kxa5 Rff4 7.Re4 and only the fourth fork finally does it.

    This could be seen as an elaboration of a much older repetitive fork (See Diagram): White to play and win; F. Sackmann, Deutsche Schachzeitung 1909. After 1.c7, Black can stop the promotion of that pawn in two ways, but in each case forks force a promotion anyway. 1...Nd6 (Nb6 2.Nd5!) 2.Ne4 and now a serial fork: Nd(f)xe4+ 3.Kf4 Nd(f)6 4.Ke5 and one pawn promotes.

A fork may also be aimed directly at mate, as in this elegant problem by the great American wizard: (See Diagram) White mates in 3; Sam Loyd, Leipziger Illustrierte Zeitung 1869. 1.Qf1! and now the Queen can always attack h7 while providing a diagonal QxB mate against the defense g6, e.g. Bb2 2.Qb1 or Bf6 2.Qf5. And after 1...g3, there is another mate with 2.Ng6+

Another well known multiple fork by a great American composer: (See Diagram) White mates in 3, W.A. Shinkman, Detroit Free Press 1882. After the waiting move 1.Kh1! the black Rooks are moving targets which White can always hit both at once, e.g. 1...Ra4 2.Qe4 or 1...Rh8 2.Qc3, or 1...h4 2.Qh5. In eight variations, it is mate on the third move by the Queen taking one of the Rooks or, if she is taken herself, by Ra7 or Rb8.

It is no coincidence that we already encountered Leonid Kubbel here: the fork, in all its simplicity, was one of his favorite themes. He also used it in problems; in that discipline too, he was one of the greats. (See Diagram) White mates in 4; L. Kubbel, Shakhmatiy 1937. Not an elegant position, but these forks come at a price - which is worth it. The hero is the inconspicious Nb3, which forks its way to three mating squares. After the key move 1.c4 (most other tries fail to Qxb3), 2.Rd5+ and mate next move is threatened. Black can meet this threat in different ways, leading to different Knight's forks and a cyclical permutation of mates:
1...Rxc4 2.e4! Rxe4 3.Nd2 and 4.Nxe4 or Nxc4 mate
                  2...Bb7 3.Na5! and 4.Nxc4 or Nxb7 mate
1...Bb7 2.Qf7 Re4 3.Nxc5 and 4.Nxb7 or Nxe4 mate

Many years later, two joint Russian composers showed a vastly improved rendering of this beautiful idea. The position was lighter, more elegant, there was more unity (the threat was a mating fork now, too), and they upped the number of variations from three to five, doubling the number of mating squares to six. (See Diagram) White mates in 4. A. Virtmanis & V. Chekarkov, 2nd Prize, Shakhmatiy v SSSR 1979. It might be obvious (after a few hours) that White must first block the Pe3, to rob Black of a possible flight square. After 1.Re2! there is already the beautifully thematic threat: 2.Ng4 Rxe6 3.Nxe3 and Nc2 or Nxf5 mate.
Black now has four defences, which all lead to a new Knight's fork:
1...Rxe2 2.Nh5! Rxe6 3.Nf4 and Nxe2 or Nxe6 mate
1...Rxf3 2.Nh7! Rxe6 3.Ng5 and Nxf3 or Nxe6 mate
1...Bxb3 2.Nd7! Rxe6 3.Nc5 and Nxb3 or Nxe6 mate
1...Bxb5 2.Ne8! Rxe6 3.Nxc7 and Nxb5 or Nxe6 mate

A simultaneous double fork, both parts satisfying all conditions, is the core of a well known study, worthy to conclude this survey: (See Diagram) White to play and win. G. Kasparyan, Sjachmatniy 1935. 1.Ne8 Threatening Ng7+ and Bf5 mate 1...Kg6 2.h5+ Rxh5 3.f5+ Rxf5 4.g4 Re5(f4) 5.Bf5+ Rxf5 6.Ng7 and the Rooks on f5 and h5 are doubly doubly attacked; mate next move. This is the Mother of All Forks. The fantastic final position is seen one move earlier after 1...Rxf4 2.Ng7+ Kg6 3.h5+ Rxh5 4.Bf5+ Rxf5 5.g4 (See Diagram)

© Tim Krabbé, 2000

Top | Main Chess Page |