Seven Ways to Refute the Traxler

The counter attack 1 e4 e5 2 Nf3 Nc6 3 Bc4 Nf6 4 Ng5 Bc5!? goes back to the
nineteenth century (Reinisch – Traxler, Prague 1890), but for decades many
theoreticians ignored the idea, with Reuben Fine ([1], [2]) being a rare exception.
Frank Marshall, wrongly believing that he was the first to analyze 4…Bc5, later
named it “Wilkes-Barre Variation,” after a town in Pennsylvania. In the early
1950s, Paul Keres published an excellent analysis and credited the move with an
exclamation mark. From that moment the Traxler Gambit gained deserved attention
and also found more supporters: Rohlicek, Pachman, Brinckmann, Estrin and many
others helped to develop its theory.

In the last decades White used to play 5

Over the Horizons Bxf7+ Ke7, followed by either 6 Bd5 or 6
Bb3. White may have a slight advantage,
but Black keeps substantial compensation
for his sacrificed pawn. Lev Gutman
Stefan Bücker concentrated on 6 Bd5 [4], published in
1996, which he analyzed on dozens of
pages. At that time the theoreticians still
preferred 5 Bxf7+ over the alternatives.
Only four years later Christophe and Moll
claimed in [5] that the real refutation of
the Traxler was 5 Nxf7 Bxf2+ 6 Kxf2
Nxe4+ 7 Ke3!, a courageous winning
attempt that had been underestimated by
theory. In the same year, Dan Heisman’s CD [7] appeared with plenty of
interesting computer analyses on the Traxler. And then Maarten De Zeeuw
presented six new “refutations” of the Traxler in a series of provoking articles [10].
According to him, the strongest line begins with 5 Nxf7 Bxf2+ 6 Kf1.
 Karel Traxler (1866-1936)
Source: Ceskoslovensky Sach 1936

The new winning attempts brought a fresh breeze to this opening. Adherents of the
gambit have already found valuable improvements for Black [6], [12], but some
questions remain. Let us see whether one of the seven propagated lines is a real
refutation. The following lines all start from the diagram above.

Refutation 1

5 Nxf7 Bxf2+ 6 Kxf2 Nxe4+ 7 Ke3!?

The alternative 7 Kg1 is known to lead to

a draw.

7…Qh4

But not 7…Qe7? 8 Nxh8 d5 (8…Qg5+ is

also hopeless) 9 Qh5+ (! Cramer) g6 10
Nxg6 Qc5+ 11 Ke2 Nf6 12 Qg5 Ne4 13
Qe3 [7] +/–.

8 g3 Nxg3 9 hxg3 Qd4+ 10 Kf3 d5 11

Rh4 e4 12 Kg2 0-0

Here Christophe/Moll detected the novelty

13 Nc3! dxc4 14 Qh5 [5], but their
claimed refutation of the Traxler met the
strong reply 14…Ne7! = (Bennedik [6]);
for example, 15 b3 Rxf7 16 Qxh7+ Kf8
17 Nxe4 Ke8 18 c3 Qe5 =. However, the
position is interesting enough to look at
some other possibilities.

13 Bb3
 Against 13 Rf4 dxc4 14 Qf1 (Oleksenko –
Malktsirtis, 1984), Gregor Cramer nicely
saved Black by means of 14…Be6! 15
Ng5 Bd5! 16 Nc3 e3+ [3].

13…Rxf7 14 Qg1

Heisman [7] doesn’t mention 14 Rf4!?, when Black has to play precisely: 14…
Rxf4 15 gxf4 Be6 16 d3 Rf8 17 c3 (17 Qg1 Qf6 18 Nc3 exd3 19 Qe3?! dxc2 20
Nxd5 Qg6+, followed by Bf7 =+) 17…Qf6! 18 dxe4 Qg6+ 19 Kf2 Qh6 20 Bxd5
Qh2+ 21 Kf1 Qh1+ 22 Ke2 Qg2+ 23 Kd3 Bxd5 =.

14…Qe5 15 d3

(a) 15 Nc3 (Heyken/Fette) 15…g5! 16 Rh6 Qf5 = [5]. More promising seems 16 d4
Nxd4 17 Rh5, but this doesn’t change the assessment: 17…h6! 18 Bxd5 Bg4 19
Bxf7+ Kxf7 20 Qf2+ Kg6! (20…Kg7? 21 Bf4 +–) 21 Rh2 Kg7, threatening Rf8,
with equal chances.

(b) The books forget to mention 15 d4!? Nxd4 16 Bxf4 Rxf4! 17 gxf4 (or 17 Rxf4
g5 18 Rf1 e3) 17…Qf6 18 Bxd5+ Be6 19 Bxe6+ (19 Nc3 Qxh4) 19…Nxe6 20 Rh2
Nxf4+ 21 Kh1 e3 22 Qxe3 Qc6+ 23 Kg1 Qg6+ 24 Kf1 Rf8 25 Rf2 Qh6 26 Qf3 Re8

Hoping for 27 Nc3? Re3!! –+. After 27

Rd2 Rf8, the result is a draw.

15…g5! 16 Rh5 h6 17 Rxh6 Bg4

Thanks to his superior development,

Black has nothing to fear; e.g., 18 Rg6+
Rg7 19 Rxg7+ Kxg7 20 Qe3 Bf3+ 21.
Kf2 Kg6 22 dxe4 dxe4 23 Nd2 Rf8 =.

Refutations 2-5

5 Nxf7 Bxf2+ 6 Kf1 Qe7 7 Nxh8 d5

In NIC Yearbook 68, Maarten De Zeeuw

gives a summary of his series on the
Traxler Gambit. He lists six refutations;
four of these begin with the moves above. He attributes both 5 Nxf7 and 6 Kf1 with
exclamation marks, thus downgrading his other refutations (6 and 7, below).
8 exd5

Another asserted refutation by De Zeeuw is 8 d3, “giving White the exchange for a
pawn” [10]. But in [12] Pietro Pastore showed that both 8…Bg4 9 Qd2 Bh4 and
8…dxc4 9 Kxf2 Bg4 10 Qe1 cxd3 11 h3 dxc2 (11…Be2 Montcaubeig [12] is also
strong) 12 hxg4 cxb1Q 13 Rxb1 (so far De Zeeuw, [10]) 13…0-0-0! are good for
Black.

8…Nxd4

9 d6!

Seems best. Other so-called refutations by

De Zeeuw:

(a) 9 h3 Bh4 (or 9…Bg3 10 a4 [!! 10] Qc5

11 b3 Nxb3 12 Bb5+ c6 Moncaubeig [12]
13 d4 unclear; 10…Ne4!? Pastore [12],
about =) 10 d6 cxd6 (10…Qxd6 11 Nf7
Qc5 12 Na3 +– De Zeeuw [12],
overlooking the attractive response 12…
e4 =; e.g., 13 g3 Qf5+ 14 Kg2 e3! 15 dxe3
Bxg3 Moncaubeig, as pointed out in [12])
11 g3 +– De Zeeuw [10].

“Here Black has many choices,” believes

Pietro Pastore and analyzes one line in
detail: 11…b5 12 Bxb5+ Nxb5 13 gxh4
Ne4 14 Qe1 Qf6+ 15 Kg1 Bb7 16 Nc3
Ng5! 17 Qf2 Nf3+ 18 Kf1 0-0-0 19 d3
Nxc3 20 bxc3 Rf8 21 Be3 (so far Pastore
[12]) 21…Qe6 –/+.

(b) 9 Be2 Bh4 10 c3 Nxe2 11 Qxe2 Bg4

12 Qb5+ Nd7 13 Kg1 “and White ends up
an exchange to the good,” De Zeeuw [10];
13…0-0-0 (or 13…Qf6 14 Qf1 Qb6+ 15
d4 0-0-0 16 g3 Rf8) 14 h3 (so far De
Zeeuw [10]) 14…Bh5 15 d3 Rf8 16 Qb4
Qf6 17 Be3 Bg3 18 Nd2 (18 Na3 Be2! Pastore) 18…Bf2+, with advantage for
Black.

9…Qxd6

9…cxd6 10 Kxf2! (Bennedik [6]) 10…d5 (Bruno Montcaubeig tries to rescue 10…
Bg4 11 Qf1 0-0-0 12 Nf7 d5 13 Nc3 Rf8 14 Nxd5 Nxd5 15 Bxd5 [10] by means of
15…Rxf7! = [12], but 13 Kg1! Rf8 14 c3 Nc2 15 Na3 Nxa1 16 Bxd5 Nxd5 17 Qc4
+ +/– looks more critical). For example:

(a) 11 Be2 Ne4+ 12 Kg1 Qc5 13 Kf1 +–

De Zeeuw [10], but 13…Qf8+ 14 Bf3 Bg4
15 d3 Nxf3 16 gxf3 Bxf3 17 Qe1 Bxh1+
18 Kg1 (Pastore [12]) is better for Black.
 (b) 11 c3!? Ng4+ 12 Ke1 Qf6 13 Rf1 Qh4
+ 14 g3 Qxh2 15 Bb5+ Ke7 16 Rf7+ Kd6
17 cxd4 Qxg3+ 18 Ke2 Nf6 19 Bd7 Bxd7
20 Rxd7+ Nxd7 21 Nf7+ Ke7! = Pastore
[12].

(c) 11 h3 dxc4 12 d3 deserves attention; e.

g., 12…cxd3 13 cxd3 Be6 14 Be3 +=.

10 Nf7

Another variation leading to +– by De Zeeuw was 10 Be2, but Da Costa Junior

[12] suggests 10…Bh4 11 Nc3 Qf8 12 Kg1 Bf2+ 13 Kf1 Bh4 =, with a draw by
repetition.

10…Qc5 11 d3 Bh4

Gambit Play
by Zygmunt Nasiolkowski, Germany

Heisman’s main line, which De Zeeuw mentions only in passing. The line is
extremely complicated, and Heisman believes that Black gets sufficient chances. 6
Kf1 may be strong, but until now the solution has not been demonstrated. The
following line is just an example.

12 b4 Qe7 13 Nd2

A new idea, not mentioned by Heisman [7].

13…Bg4 14 Nf3 Bxf3 15 gxf3 b5 16 c3 bxc4 17 Ng5 Nb5 18 dxc4 Nxc3 19 Qd3

Or 19 Qb3 h6 20 Qxc3 hxg5 21 Kg2 0-0-0 22 Rf1 Qe6 23 Qc2 e4!.

19…h6 20 Nh3
 20…Qe6! 21 Nf2 (21 Kg2 Nfe4!) 21…
Bxf2 22 Kxf2 Qh3 23 Qg6+ (23 Qxc3?
Ne4+) Kf8 24 Bd2 e4 25 Qg2 (25 Qg3 e3
+) 25…e3+ 26 Bxe3 Qh4+ 27 Kg1 Qxc4
+=.

Refutation 6

5 Bxf7+ Ke7 6 Bb3

There is enough theory on this line to fill a

book. Here I’ll only look at a single
variation recommended by De Zeeuw.

6…Rf8 7 d3 d6 8 Be3 Qe8 9 Bxc5 dxc5 10 Nc3

Maarten De Zeeuw continues with 10…

Nd4 11 f4! +/–. Instead, I’d suggest 10…
Qg6; for example, 11 Nf3 Bg4 12 Ba4
Bh5 13 Rg1 (13 Bxc6? Qxg2 14 Rg1 Qxg1
+) 13…Bxf3 14 Qxf3 Nd4 15 Qg3 Qxg3
16 hxg3 c6 =.

Refutation 7

5 d4!

If there is a refutation of the Traxler, it

may well start with 5 d4. Reuben Fine
recommended the move, as some sources
state correctly, but in 1939 he criticized it!
In [1] “his idea” only appeared as a
sideline (to 5 Bxf7+): “If 5 d4? d5!! 6
exd5 Nxd4 7 d6 0-0 =+.” Nine years later
[2] he recognized that 6 Bxd5 was an
improvement, and so “5 d4!” became his
main line. Not many authors have
followed him, but De Zeeuw [10]
considers the move “at least as good as 5
Bxf7+.”

5…d5! 6 Bxd5 Nxd4 7 Bxf7+ Ke7

7…Kf8 is new, but interesting. In some lines Black profits from the position of his
king:

(a) 8 c3 (since Qd8 is now unprotected), but after 8…h6 9 cxd4 hxg5 10 dxc5 Qxd1
+ 11 Kxd1 Kxf7 12 Nc3 g4, Black has no problems. Or 8 Bb3 Qe8 9 Qd3 Bg4 10
Qc4 b6 =.
(b) 8 Bc4 b5

(b1) 9 Bd3 (9 Bxb5 Bg4 10 f3 Nxb5 11 fxg4 Qxd1+ 12 Kxd1 Ke7 =) 9…Bg4 10 f3
(10 Nf3 Bxf3 11 gxf3 Nh5 =) 10…Bc8 11 c3 h6 12 cxd4 Qxd4 13 Qc2 hxg5 14
Bxg5 c6

15 Nc3 (15 Nd2 Qf2+ 16 Kd1 Be6 17

Nb3 Rxh2! 18 Qxf2 Rxh1+ 19 Ke2 Bxf2
20 Rxh1 Bb6 =) 15…Rh5 16 h4 (perhaps
16 Bxf6 Qe3+ 17 Qe2 gxf6, but here
White’s extra pawn isn’t too important)
16…Nh7! (with the king on e7 this knight
would be pinned) 17 0-0-0 Nxg5 18 hxg5
Rxh1 19 Rxh1 Qe3+ 20 Kb1 Be6, and the
position is about equal.

(b2) 9 Be2! h6 10 Nf3 Nxe4 11 0-0 Bf5

(or 11…Bb7 12 Nbd2) 12 Nxd4 (12
Nbd2? Nxf2!; the text move is also
stronger than 12 Be3 Qf6 13 c3 Rd8!
unclear, or 13 Bd3 Rd8 14 Nbd2 Nxd2 15 Nxd2 Bb6, almost =) 12…Qxd4 13 Qe1
Rd8 14 Nc3 (14 c3 Qd6 15 Bxb5 Qg6) 14…Nxc3 15 bxc3 Qa4 16 Rb1! +/–,
intending 16…a6 17 Bxb5! axb5 18 Qxe5, with a double attack on the black
bishops.

8 Bc4 b5

9 Bd3

De Zeeuw favors 9 Be2 (!, “with a sound

extra pawn,” De Zeeuw). However, his
observation that “the f3 square …is in
greater need of overprotection than the e4
pawn” is quite strange. 9…h6 10 Nf3
Nxe4 11 Be3!. Heisman [7] now gives
11…Qd5 12 Bxd4, which is repeated by
De Zeeuw [10]. But Black should choose
11…Bb7!? 12 Nxe5 (12 0-0 Nxf3+ 13
Bxf3 Qxd1) 12…Qd6

13 Nf3 (or 13 Nd3 Bb6 14 Nd2 Rad8 15 Nxe4 Bxe4 16 0-0 Rhe8, Black has
compensation) 13…Nxe2+ (or 13…Nxf3+ 14 gxf3 Ng5 15 Qxd6+ Bxd6 16 Nd2
a6 17 h4 Ne6 18 0-0-0 Rhf8 19 Rhe1 Rad8 +=. White’s extra pawn is doubled) 14
Qxe2 Rhe8 15 0-0 Kf7 with compensation; for example, 16 a4 b4 17 Nbd2 Nxd2
18 Nxd2 Qd5 19 Nf3 Kg8 20 Rad1 Qf7.

9…h6

(a) 9…Qe8 10 h3 Qg6 11 0-0 Rd8 12 Kh2, Black is in trouble.

(b) 9…Bg4 10 f3 Bc8 11 c3 h6 12 cxd4 Qxd4 13 Qc2 hxg5 14 Bxg5 +/– pins the
knight, an important difference to the same variation with 7…Kf8.

(c) Unfortunately, 9…Rf8 10 c3 Ng4 11 Rf1 Nxf2 does not work, because of 12
Rxf2 Rxf2 13 Kxf2 Nc2+ 14 Kf1! Nxa1 15 Ne6! +/–.

10 Nf3! Ng4

The main alternative is 10…Bg4 11 Nbd2 Qd6 (Pálkövi [8]; 11…Qe8 12 h3? Qh5
13 0-0 Rad8!, but 12 Nb3! Nxb3 13 axb3 Qh5 14 Be3 etc.) 12 0-0 [10]

“White can liberate himself by 12 0-0

intending 13 Nxd4,” De Zeeuw [10] in
2002. In the meantime Jan Pinski [11] has
continued the discussion: 12…Rhf8 13
Nxd4 Bxd4 14 Be2 Bd7 15 Nf3 Bc5 16
Qxd6+ cxd6! “Black has good
compensation because of the c-file and a
lead in development,” Pinski [11]. This is
optimistic. White should play 14 Qe1! c6
15 h3 Bh5 16 Nb3 Bb6 17 Be3 +/–.

11 Be3! +=.

Heisman’s [7] main line is 11 Rf1, but De Zeeuw [10] is right to prefer the direct
development of the bishop. White has an advantage. – When De Zeeuw compares 5
Bxf7+ and 5 d4, he writes: “Both moves yield White an extra pawn and leave
Black’s king stranded in the centre (e7), but 5 d4 also results in an open d-file.”
The last observation may seem a bit far-fetched at first, as long as Black holds the
initiative. However, the situation may quickly change; for example, 11…Rf8 [7] 12
Bxd4! Bxd4 13 0-0 Bxb2 14 Nbd2, and with every exchange Black’s king loses
some shelter.
None of the seven lines proposed in [5] and [10] comes close to being a theoretical
refutation. But those who want to practice Traxler’s Counterattack with success
have to do some homework concerning 5 d4!.

Sources:
[1] R. Fine (editor): Modern Chess Openings, sixth ed. 1939
[2] R. Fine: Practical Chess Openings, New York 1948
[3] G. Cramer: Traxler-Gegenangriff, Fritz-Variante, Ulvestad-Variante. Hollfeld
1993
[4] L. Gutman: (analysis in:) Schach-Archiv, Hamburg 1996
[5] N. Christophe, A. Moll: “Finally Refuted!,” in New in Chess Yearbook 55
(2000)
[6] M. Bennedik: “Really a Refutation?,” reader’s letter in New in Chess Yearbook
56 (2000)
[7] D. Heisman: The Traxler Counterattack CD-Rom, Wylie 2000
[8] J. Pálkövi: Two Knights’ Defence and Traxler Counter-Gambit, Kecskemét
2001
[9] T. Harding: “Turmoil in the Traxler Two Knights,” in: Chess Mail 1 and
2/2001.
[10] M. De Zeeuw: “Another Look at the Traxler Gambit,” in: NiC Yearbook 63,
65, 66, 67, 68 (2002-2003)
[11] J. Pinski: The Two Knights Defence, London 2004
[12] P. Pastore, B. Montcaubeig, L. R. da Costa Junior: “It’s Hard to Kill the
Traxler Gambit,” in NiC Yearbook 70 (2004)

Corrections
Otti Keller from Switzerland sends valuable comments on Don’t Name the
Gambit after Me, the article on the Closed Sicilian published in May:

– In Bronstein – Keres, my remark on Black’s 18th move contains a mistake,

because 18…Qe5 loses a queen. The reader is right. The moves that Erich
Eliskases had analyzed on two an a half pages were not “18…e5 and 18…Qe5!”,
but 18…e5 and 18…Nc6. Only after 18…Nc6 19 c3 follows 19…Qe5!, according
to Eliskases. My apologies.

– Halfway through the article, in the section “Food for Thought I,” I gave the
following line: 1 e4 c5 2 Nc3 Nc6 3 g3 g6 4 Bg2 Bg7 5 d3 d6 6 f4 e6 7 Nf3 Nge7 8
0-0 0-0 9 Be3 b6 10 d4 Ba6 11 Rf2 Qc7 12 a4 Rad8 13 Nb5 Bxb5 14 axb5 Nxd4
15 Nxd4 cxd4 16 Bxd4 e5 17 Be3 exf4 18 Bxf4 Bxb2 19 Ra4! Qd7 20 Qb1 Be5 21
Bh6. Here Mr. Keller writes: “My computer finds 21 Bxe5 +– (instead of 21 Bh6).
What is the idea behind 21 Bh6?”

The sequence 21 Bh6 Rfe8 22 Qf1 seemed logical, to create a weakness (pawn f7).
White then has a slight advantage. But any stronger suggestion is certainly
welcome! The immediate 21 Bxe5 dxe5 is interesting; perhaps White can in fact
find a way to bring his bishop to d5, beginning with 22 Qb3 or 22 Qb2. From d5
the bishop would exert plenty of pressure on Black’s position. But I fail to see a
concrete win; for example, 22 Qb2 Qc7 23 Bf1 Nc8 24 Bc4 Rd4. Perhaps the
solution is beyond the horizon – of my computer.

Send your games or comments to redaktion@kaissiber.de

 [ChessCafe Home Page] [Book Review] [Columnists]
[Endgame Study] [The Skittles Room] [Archives]
[Links] [Online Bookstore] [About ChessCafe.com] [Contact Us]

© 2007 CyberCafes, LLC. All Rights Reserved.

"ChessCafe.com®" is a registered trademark of Russell Enterprises, Inc.