The Kibitzer

Tim Harding

Two Knights Defense:
See the Wood, Not the Trees!

A recent trend in chess publishing is the “electronic book” in ChessBase
format. These are rapidly replacing the little printed monographs one used to
see on minor variations written by minor masters or experts and catering for a
fairly narrow audience of amateur players.

Main-line strategic openings are generally best left to the GMs or to strong
masters with a good track record for writing and research, such as IM John
Watson and FM Graham Burgess. With a few exceptions (such as GM Soltis)
grandmasters are usually not so interested in writing about bizarre variations
or primarily tactical openings that are not much played in master events.

In sharp openings when the assessment of variations mostly depends on a
piece count (after verifying there is no mate or perpetual check) a fairly
strong player (aided by computer) can probably come to much the same
conclusions as a GM would, anyway.

The main danger I see in relying on these “electronic books” (or their printed
equivalent for that matter) is that they make it hard to understand what is
going on overall, because the reader is smothered in detail. Take, for
example, what is probably the best of its kind so far US master Dan
Heisman’s coverage of the Traxler Two Knights, 1 e4 e5 2 Nf3 Nc6 3 Bc4
Nf6 4 Ng5 Bc5!? (see www.chesscentral.com for more information). This
includes games, text and variation notes, and a “tree” database to show how
the material is linked, but and I think this metaphor is appropriate it is
extremely hard to see the wood for the trees.

Heisman includes in the database almost every (not quite all!) game or piece
of analysis ever published on the variation, and there are text surveys to help
you navigate through the material, but even so there are sometimes
contradictions between what is said about a sub-variation in the survey and in
the game notes, and there isn’t really any discussion of how the Traxler fits
into the general scheme of things.

This article, and the second part next month, will be an exercise in the
opposite approach, which is an approach more suited to the beginner and
young player, and the typical amateur who doesn’t have time for detailed
opening research (or the memory).

A correspondence player, able to refer to books and databases between
moves, may appreciate the level of detail in a project like Heisman’s. Even
so, it is a fallacy to think that you can win many games against competent

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (1 of 10) [3/12/2001 11:44:32 PM]

opposition in email or postal just by having ever-more-thorough
encyclopaedic reference works. From time to time, we do all win a game
against less informed opposition by repeating a known game or trap, but such
games are very much the minority and are outweighed by the occasions
where our attempt to do so is upset by a new move from the opponent that is
not in the encyclopaedia.

For people playing over-the-board against human opponents, or on real-time
Internet servers, tactical alertness and a good general understanding of the
opening is of much more practical value than encyclopaedic reference works
that you can only consult after the game, to see what you forgot or where the
opponent improved!

If chess opening theory as a whole is the ecology of a planet, then the 1 e4 e5
open games are the old continent of Europe, and the open games minus the
mighty Ruy Lopez (Spanish Opening) are thought to be a sub-continent if
dwindling importance. (Although that may be changing, as I will try to assess
next month.)

In that scheme of things, the 3 Bc4 complex including the Giuoco Piano and
Two Knights Defence is a medium-sized group of countries (the
German-speaking heart of Europe), within which the 4 Ng5 lines of the Two
Knights are just a region (Bavaria?). Within that region, the Traxler (or
Wilkes-Barre) variation represents a small mountain region to the south-east
of Munich (Berchtesgaden?).

In other words, it is an interesting and rather notorious part of the chess
world, but compared with the megalopolis of the Sicilian Defence, the wide
open prairies of the English Opening or the vast oceans of the Queen’s
Gambit and Indian Systems, tiny Traxler is seen as a tiny spot on the map
where the vast majority of chessplayers never need to (and never will) set
foot.

This Kibitzer (and its sequel) will, however, attempt to map the 4 Ng5 Two
Knights Defence (Bavaria) at the sort of level of detail which the majority of
players could find of some help. You probably won’t want to “live” or”
work” there, but you may decide it is an interesting region to visit sometimes
for a “holiday”, with White or Black or both. My map will concentrate on the
forests you will find there and say very little about individual trees or clumps
of trees; don’t expect to find secrets or innovations by the dozen here.

1 e4 e5 2 Nf3 Nc6 3 Bc4

Personally I think this move is just as strong as 3 Bb5 and 3 d4, and I play it
partly because it is not as popular as those moves. Of course I have been
known to play other open games as well, not to mention different first
moves...

3...Nf6

This reply is more fun but I think that 3...Bc5 makes it harder for White to get
an advantage. (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (2 of 10) [3/12/2001 11:44:32 PM]

4 Ng5

This is the celebrated “Duffer’s Move”
or “Bungler’s Move” as it was termed
by Tarrasch. (Can anyone tell me when
and where, and what the original
German phrase was? If so, I shall write
about that in the next article).

Actually, 4 Ng5 is White’s strongest
move here, and the only one that can
challenge the soundness of the Two
Knights Defence. I know that I have
written elsewhere in favour of 4 d4

exd4 5 e5 but that is because those lines are easier to learn for most people,
but I also think that the 4 d4 lines are closer to being “solved” after having
received quite a lot of exposure in the past 20 years. That line is still
playable for White but is less likely to surprise the opponent. The last time
I played 4 d4 against a master in postal chess we had a draw in about 20
moves without anything very original occurring.

The other approach for White is to play the slow 4 d3 and continue rather
as in the Spanish. In our 1977 book, The Italian Game, IM George
Botterill and I wrote that “4 d3 must be taken seriously as a positional
reply to the Two Knights Defence!” In the 1980s that is exactly what
happened frequently via the move order 1 e4 e5 2 Bc4 Nf6 3 d3 Nc6 4
Nf3 whereby White avoids the Petroff Defence. 4 d3 remains perfectly
playable but it has now been rather deeply studied.

However, we are going to “Bavaria” to look at 4 Ng5!? which of course
creates the direct threat to capture Black’s f-pawn: either with the bishop,
giving check, or with the knight, forking king and queen.

The fundamental idea of 4 Ng5 is that it virtually forces Black to give up
material, for which he may or may not have adequate compensation. Also,
because 4 Ng5 has been out of fashion, some Two Knights Defence
players are even surprised when they have to meet it and do not have their
line so well prepared as against 4 d4 or 4 d3.

The drawbacks for White if he wants to play 4 Ng5 are twofold:

a) he may not want to be in the position of defending a gambit;

b) there are so many different ways that Black may continue at moves 4-8
that White has to know quite a lot, whereas Black needs only know his
chosen variation.

4...d5

My two-part article will deal only with this move, which is the soundest
continuation for Black. There are two other moves which I shall briefly
mention.

a) 4...Bc5 is the Traxler, or Wilkes-Barre, mentioned above. For this, see

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (3 of 10) [3/12/2001 11:44:32 PM]

my Chess Mail magazine issues 1/2001 and 2/2001 with an update to
follow later in issue 5. Here I will just make a general comment.

You might think that White’s threat on f7 is so strong that Black cannot
ignore it, yet the Czech player Karel Traxler got the idea in 1896 that
4...Bc5 5 Nxf7 Bxf2+! will draw out the white king and bring the black
queen into play with gain of time. Alternatively the white bishop can take
on f7, winning a pawn and depriving Black of castling rights, but then the
white bishop must loses time retreating and maybe Black can
counter-attack down the f-file. About 30 years later, players of the
Wilkes-Barre Chess Club in the USA independently got the same idea.

If you are prepared to absorb tremendous amounts of detailed analysis and
trust your memory, and if you are willing to walk your king all over the
place and give up your queen for a rook and assorted other lumps of wood,
it may well be the case that 5 Nxf7 is objectively a winning reply.

The vast majority of sensible players will (except perhaps in
correspondence chess) opt for 5 Bxf7+ Ke7 6 Bb3 (or 6 Bd5) and hope
that Black knows no more about the Traxler than they do.

b) 4...Nxe4?! is another way of ignoring the threat to f7 in preference for
counter-attack. I don’t discuss it further here as it was already analysed in
Kibitzer 5 with a follow-up in Kibitzer 33. You can find these in

The

Chess Café Archives

. I will just summarise here by saying that White’s

best line is 5 Bxf7+ Ke7 6 d4! h6!? (if 6...d5 7 Nc3!) 7 Nxe4 Kxf7 8 dxe5!
(The final edition of the Handbuch des Schachspiels has about five pages
on 4...Nxe4 without finding this.) 8...Qe8 9 f4! as in Van Steenis-Vlagsma,
cited in Kibitzer 33.

5 exd5

Of course White has no choice since 5 Bxd5? Nxd5 6 exd5 Qxg5 7 dxc6
Qxg2 is painful. (See Diagram)

Now Black is a pawn down and his
c6-knight is attacked. The main line
from this position is the counter-attack
on White’s bishop by 5...Na5, which
will be part of the subject of next
month’s Kibitzer column. Another
important line which I will discuss in
Kibitzer 59 is the Fritz variation,
5...Nd4, which is a rival to 5...Na5 for
soundness. This month I look at all the
less credible alternatives, namely:

A: 5...Nb4

●

B: 5...Bg4;

●

C: 5...Nxd5 6 Nxf7 (or 6 d4);

●

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (4 of 10) [3/12/2001 11:44:32 PM]

D: 5...b5 (Ulvestad lines)

●

Line A: 5...Nb4 (See Diagram)

A forgotten move, which was probably
first suggested by German postal player
Manfred Kloss in 1954 in the magazine
Fernschach and has lately been seen,
mostly in Internet games. Black intends
to recapture on d5 with the queen’s
knight, but of course White renews the
threat to f7 by pushing 6 d6 (also 6 d4
and 6 Nc3 come into consideration).

After 5...Nb4 6 d6 cxd6 7 Bxf7+! (7
Nxf7 Qc7 8 Bb3 Rg8) 7...Ke7 8 Bb3
d5 9 Qe2 Qc7 Black counter-attacks

against c2, but instead of 10 c3 Bf5 (Estrin) either 10 Nc3 d4 11 Nb5 or 10
d4 seems good for White.

The reply seen most in Internet games is 6...Nbd5 but after 7 dxc7 Black
has nothing really for the pawn. If instead 6...Nfd5 then 7 Nxf7! Kxf7 8 a3
seems very strong e.g 8...Nxc2+ 9 Qxc2 c6 10 00 Bxd6 11 Nc3 Be6 12
Ne4 Be7 13 d3 h6 14 f4! exf4 15 Bxf4 Nf6 16 Bxe6+ Kxe6 17 Rae1! with
a big advantage to White in J.Silva-P.Antunes, Portugal 1995.

Line B: 5...Bg4

This was described as a “relatively new continuation” by Estrin in the
1980s but in fact it was suggested by Max Lange in the 19th century. The
attack on White’s queen is met by 6 f3 after which Black has two pieces en
prise. (See Diagram)

After 6 f3 Black must counter-attack a
white piece so there are two
possibilities:

a) 6...Nxd5 7 Nxf7! Kxf7 8 fxg4 Bc5 9
d4! with a strong initiative for White;
b) 6...Na5 7 Qe2 and again White has
an extra pawn with nothing to fear.

Instead of 6 f3 White can also consider
Estrin’s move 6 Nxf7!? but this is more
complicated after 6...Qe7! (not
6...Bxd1? 7 Nxd8).

Line C: 5...Nxd5

This is the oldest variation of the Two Knights Defence. Black does not
offer a pawn sacrifice but it reopens the diagonal for White’s bishop
pointing towards f7. White’s knight is threatened but he can sacrifice it or

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (5 of 10) [3/12/2001 11:44:32 PM]

advance his d-pawn with tempo so that the c1-bishop protects the knight.
In effect, White has a free move to launch an attack.

Usually 5...Nxd5 gets a question mark in the books but periodically there
are attempts to revive it. (See Diagram)

Apparently this variation will be the
subject of Dan Heisman’s next
electronic book so I won’t go into
much detail about it here.

There are two distinct lines now:

C1: 6 Nxf7 (“Fegatello)

●

C2: 6 d4 (Lolli)

●

c1) 6 Nxf7!? Kxf7 7 Qf3+ Ke6 8 Nc3

This is the famous “Fegatello” or Fried
Liver Attack. Is it correct? (See

Diagram)

The enormous amount of detail on this
position can be summarized briefly.
White has sacrificed a piece but the
d5-knight is pinned and trebly attacked.
If White regains the piece, he
obviously has a winning position so the
other knight must move to defend d5.

There are two variations and the
question is whether White has a clear
win in both. If he does not have a clear
win then is it not better to play the 6 d4
line (C2) which gives a strong attack

without the risk of sacrificing a piece?

a) 8...Nce7 9 d4 c6 10 Bg5 h6 11 Bxe7 Bxe7 12 0-0-0 Rf8 13 Qe4 and the
white forces are massed to slaughter the black king in the centre.) .

b) 8...Ncb4 is critical. The recent book (in German) by the Hungarian
master Palkövi claims White is winning by 9 a3 Nxc2+ 10 Kd1 Nd4 11
Nxd5 but he doesn’t mention 11...c6! as analysed by German theoretician
Stefan Bücker in his Kaissiber magazine. Moreover, even 10...Nxa1 may
be playable according to Bücker.

Therefore, until I see a clearer proof, I do not trust 6 Nxf7.

C2: 6 d4 (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (6 of 10) [3/12/2001 11:44:32 PM]

This move, suggested by Lolli, has the
main point that after 6...Be7 White can
play the Fegatello under improved
circumstances, 7 Nxf7!, because the
queen’s bishop is ready to come out
and e5 is under pressure.

If 6...exd4 7 0-0! and again White is
ready to launch an attack against the
black king caught in the centre, e.g.
7...Be6 (7...Be7?! 8 Nxf7! Kxf7 9
Qh5+!) 8 Re1 Qd7 9 Nxf7! (pointed
out by Major Jaenisch in the mid-19th

century) 9...Kxf7 10 Qf3+ Kg8 (10..Kg6? 11 Rxe6+! Qxe6 12 Bd3+) 11
Rxe6! (given by Steinitz in his Modern Chess Instructor) 11...Rd8 and
now most books give 12 Bg5, Palkövi recommends 12 Re4 while I played
12 Qe4!? and went on to win in Harding-Knol, Heidenfeld Memorial corr
2000.

Therefore the only real attempt for Black to defend from the last diagram
is: 6... Bb4+ 7 c3 Be7 Black “wastes” a tempo to deny the c3 square to the
white knight. Now White plays a delayed Fegatello by 8 Nxf7 Kxf7 9
Qf3+ Ke6 when new ideas for White requiring testing are 10 a4 (Bücker)
and 10 0-0 (Nunn).

The book move 10 Qe4 is no longer considered clear because of 10...b5!
11 Bxb5 Bb7 12 f4 g6! (Kalvach-Drtina, Czechoslovakia corr 1986). For
more detail on this, see my book Startling Correspondence Chess
Miniatures.

In view of these possibilities, it cannot be said for certain that 5...Nxd5 is
a bad move. However, in practice the positions arising from 5...Nxd5 are
harder for Black than White because White has the initiative and can
afford an inaccuracy more than the defender can.

D: 5...b5!?

Although many lines in the Two Knights Defence are very ancient, this
move was only suggested by American master Ulvestad as recently as
1941. Both sides make paradoxical looking moves in this variation so a
little explanation is necessary. (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (7 of 10) [3/12/2001 11:44:32 PM]

The main point of the Ulvestad move
5...b5 is to divert the white bishop from
its defence of the d-pawn. White can
decline the second pawn and play 6
dxc6 but the messy position after
6...bxc4 probably favours Black who
has the bishop pair. Then if 7 Qe2
Black can defend his pawns by 7...Qd5
but 7...h6! 8 Qxe5+ Be7 is reckoned to
be even stronger. He gets castled first
and his pieces come out rapidly to
active squares.

After the other “obvious” answer, 6 Bxb5, White has two extra pawns, but
6...Qxd5 Black forks both b5 and g2 so that White is left with an
unattractive choice:

a) the miserable retreat 7 Bf1 which nobody plays

●

b) 7 Qe2 Qxg2 8 Qxe5+ Be7 9 Rf1 0-0 and it is the white king that
will be attacked in the middle;

●

c) 7 Bxc6+ Qxc6 8 0-0 Bb7 9 Qf3 e4 10 Qb3 0-0-0! or 8 Qf3 e4 9
Qb3 Bc5! when in each case Black has the bishop pair and good
attacking chances.

●

d) 7 Nc3 Qxg2 8 Qf3 returns one pawn to get the queens off, but
Black has a lead in development after 8...Qxf3 9 Nxf3 Bd7.

●

In view of these possibilities, it is generally accepted that the best answer
to 5...b5 is another paradoxical move, 6 Bf1! which is not a “miserable
retreat” because soon the bishop will be coming out again with gain of
tempo. Nevertheless, it may be the case that White can find an
improvement or two in the foregoing lines, for example 6 Bxb5 Qd5 7 Be2
has not been well explored.

Now we return to 6 Bf1. (See Diagram)

The first point is that Black has two
recaptures on d5 but each has a major
drawback.

If 6...Qxd5 then 7 Nc3 threatens the
queen which cannot capture on g2
since the bishop guards the pawn. So
the queen must move defensively, e.g.
7...Qc5, after which White has 8 Bxb5
pinning the black knight and preparing
castling. After 8...Be7 9 d3 White
remains a pawn ahead and Black has
no genuine compensation

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (8 of 10) [3/12/2001 11:44:32 PM]

(W.Heidenfeld-F.Bohatirchuk, South Africa v Canada corr 1951).

Ulvestad’s original conception was 6...Nxd5 but now White makes his
third bishop move, 7 Bxb5, because the reply...Qd5 is no longer available.
Black must defend his pinned knight, and the usual line is 7...Bb7 8 d4!
exd4 9 0-0 followed by Nf3 with a good game for White.

Possibly 7...Bd7 is better (to meet 8 d4 by 8...Nxd4!) but it can be met by 8
d3 with a promising game for White according to Palkõvi who gives
8...Be7 9 Qh5! Bxg5 10 Bxg5 obtaining the bishop pair.

Curiously, in a recent Dublin league game I had to meet yet another twist
in this line. After 5...b5 6 Bf1 Nxd5 7 Bxb5 my opponent Ciaran Quinn
surprised me with 7...Qf6!? which also defends the knight and keeps
options open for the bishop. This may be just as good as the other moves?
The normal formula of d2-d4 followed by 0-0 may be good but I wasn’t
sure.

Perhaps White should immediately reply 8 Qf3 because after my continuation
8 0-0 Bd7! (threatening to castle queenside) 9 Qf3 Black found 9...Ndb4! and
I had to defend the fork on c2 by 10 Na3. Then 10...Qxg5 leads to great
complications after 11 c3 or 11 d4!? but my opponent chose 10...Rb8 which
does threaten...Nd4 but permits 11 Qxf6 when White keeps some advantage.
However, this certainly wasn’t as clear as I would have liked or expected
after his 6...Nxd5.

Following 6 Bf1, Black normally transposes to the Fritz Variation, by 6...Nd4
7 c3 reaching a position that can also arise via 5...Nd4 6 c3 b5 7 Bf1 but with
different side-possibilities on the way. This is the Fritz/Ulvestad main line
which I shall discuss next time.

Black does have one other interesting possibility after 5...b5 6 Bf1. The move
6...Bg4? is simply answered by 7 f3 but 6...h6!? (discovered in the 1960s) is
very interesting. It provokes the thematic 7 Nxf7!? Kxf7 8 dxc6 but then
comes 8...Bc5! with great complications.

Alternatively, White can retreat the knight by 6...h6 7 Nf3 but then 7...Qxd5 8
Nc3 Qe6 9 Bxb5 Bb7 is more attractive to Black than the 6...Qxd5 line
because the advanced knight has been driven back. In fact, 9 Nxb5 may be
superior. None of these lines with 6...h6!? have received sufficient analysis or
practice for me to come to definite conclusions about them yet.

To sum up at this stage, none of the lines that we have looked at in this
article, except maybe the very last one (5...b5 6 Bf1 h6) offer Black real
compensation for the sacrificed pawn. Either White holds his extra material
without losing the initiative, or the first player seizes control by a sacrifice of
his own.

Therefore the Two Knights player, unless he dares the Traxler, must try to
justify his third move with one or other of the variations that will be
considered in my next column.

I hope you have found this scenic tour of the “wood” interesting. I have

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (9 of 10) [3/12/2001 11:44:32 PM]

deliberately avoided including complete games that could have introduced
“trees” to confuse the issue, but I am providing on my website a file of
relevant games that you can download. Here is the URL:

http://www.chessmail.com/freegames.html

This file will be added to next month after the coverage of the Fritz main line
and 5...Na5 line is published in Kibitzer 59.

[

The Chess Cafe Home Page

]

[

[

] [

]

"The Chess Cafe®" is a registered trademark of Russell Enterprises, Inc.

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (10 of 10) [3/12/2001 11:44:32 PM]

The Kibitzer

Tim Harding

Two Knights Defence without Tears:
Part 2

Last month I began my survey of some lines in the Two Knights Defence (1
e4 e5 2 Nf3 Nc6 3 Bc4 Nf6). I emphasised in Part 1 that this is an opening
where a player or student can easily become overwhelmed by the mass of
unclear complications that can arise. Publications that pile example on
example, omitting no practical example tactical detail do have their place -
for advanced and correspondence players. However, they leave the majority
of readers not being able to see the wood for the trees.

My objective in this series of articles is to give an overview of the Two
Knights. If you want detailed examples to analyse, you can then download the
games from my website and study them. This file now contains material of
the variations discussed this month in addition to the games already in the file
last time. Here is the URL: http://www.chessmail.com/freegames.html.

1 e4 e5 2 Nf3 Nc6 3 Bc4 Nf6 4 Ng5

Before we get on with the main topic of the article, there are two digressions
relating to Kibitzer.

Firstly, I asked readers for the source of Dr Tarrasch’s famous “duffer’s
move” put-down on 4 Ng5 and several of you gave me the answer. The
colorful phrase used by Tarrasch, not easy to translate into English is “ein
richtiger Stümperzug”. A “Stümper” is not an Australian cricket
wicket-keeper but (according to my dictionary) is a botcher or bungler or
clumsy fellow.

The most detailed comment came from Stefan Bücker, editor of the German
theoretical/historical magazine Kaissiber. He informed me that Tarrasch's
remark was made in his Die moderne Schachpartie, e.g. 2nd. ed. (Leipzig
1916), p.294, commenting on 4 d3 in Salwe-Schlechter (Carlsbad 1907):
“Wohl nicht so energisch wie d2-d4 oder 0-0, aber sicherlich viel besser als
der so häufig an dieser Stelle gemachte Zug Sg5, den ich für einen richtigen
Stümperzug halte.”

Which I loosely translate as follows “Probably not as energetic as 4 d4 or 4
0-0, but surely much better than the move so frequently made in this position
4 Ng5, which I hold to be a real bungler’s move”.

Modern theory doesn’t often agree with Tarrasch. For example in his 1965
book on Open games in his 4-volume openings series, GM Ludek Pachman
wrote that:

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (1 of 13) [4/10/2001 9:12:14 AM]

“Opinions about the soundness of the move 4 Ng5 have often changed over
the course of time. While it found little approval with Dr Tarrasch... in
modern practice, it is considered a very sound continuation, even the best
reply to 3...Nf6”.

It was preferred by the 7th Correspondence World Champion Yakov Estrin,
who enjoyed playing the resulting positions with either colour both in postal
and regular play, while in recent GM practice, for example, GM Alexander
Morozevich has been employing this move against the Two Knights.

4...d5 5 exd5 (See Diagram)

My second digression, before I go on
to discuss 5...Na5 and 5...Nd4,
concerns the Lolli Variation 5...Nxd5 6
d4 which was briefly mentioned last
month.

Dan Heisman tells me he has a lot of
new analysis to show that Black’s best
line is not 6...Bb4+ or 6...exd4 but
rather to “fall into the trap” by
6...Nxd4. Then after 7 c3 he should
play not 7...Ne6? 8 Qxd5 but either
7...f6 or 7...b5. His new electronic book

from Pickard & Son, dealing with the Fegatello and Lolli variations, will
include his proof of this and I look forward to it. Mr Heisman also points
out that his Traxler electronic book is not just an encyclopaedia of old
games and analysis but contains more new (computer-assisted) analysis
than most opening books

Now for the heart of this article. This second part covers the lines where
Black, after 4 Nf5 d5 5 exd5, employs either the standard 5...Na5 move.

I had originally intended to include the 5...Nd4 Fritz Variation (possibly
reaching the Fritz via 5...b5) in this article also but by the time I had
completed 5...Na5 this Kibitzer was already longer than the last. Since
5...Nd4 in itself is quite a big subject, it will be reserved for a third and
final part of the series next month.

Both 5...Na5 and 5...Nd4 answer the threat from White’s d-pawn by
moving the Knight while 5...b5 counter-attacks the white Bishop. I am
going to look at 5...Na5 first because it is the move that obtains the most
coverage in books, and is played most often of Black’s options against 4
Ng5. Also, I suspect that I shall find it hard to maintain my “wood not the
trees” approach so easily when dealing with the Fritz, which is almost pure
tactics. However, I will try.

5...Na5 (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (2 of 13) [4/10/2001 9:12:14 AM]

With 5...Na5 Black avoids losing a
tempo and plays to get developed
before launching a full-blooded tactical
bout. This probably explains why it is
more popular than the alternatives.

6 Bb5+

This check is virtually forced. It is
possible for White to concede the
Bishop pair by 6 d3 but this move is
hardly in keeping with the previous
aggression. Black should answer 6 d3
by 6...h6 7 Nf3 e4 when 8 Nd4 can be

met by 8...c6 and the piece sacrifice 8 dxe4 Nxc4 is unsound. So White
plays 8 Qe2 Nxc4 9 dxc4 and now Morphy’s move 9...Bc5 (hindering
Nd4) is still considered best.

6...c6

There is a strange reluctance to try Polerio’s move 6...Bd7!? which has not
really been tested or refuted yet. The upside for Black is that he may very
easily regain his pawn without compromising his queenside pawn
structure. The downside is that the pawn on d5 cramps him, and leaves the
a5-Knight vulnerable with no move at present. For both sides, this
variation is relatively unknown territory so if Black prepares 6...Bd7 he
could score well with it, and White needs to be aware of the possibility.
(See Diagram)

If White exchanges Bishops then after
7 Bxd7+ Qxd7 Black immediately
regains his pawn with a good game
(Ciric-Nesis, corr 1982) so 7 Qe2 is
necessary. Then Black has a choice
between 7...Be7 and 7...Bd6. The
former move keeps open the possibility
of capturing early on d5, whereas the
latter secures the e5-pawn and intends
to “play around” the advance d-pawn.
It also sets a trap.

Morozevich-I.Sokolov, Sarajevo 1999,

went 7 Qe2 Bd6 8 Nc3 0-0 9 Bxd7 (Avoiding the trap 9 0-0? Nxd5! 10
Bxd7 Nf4 11 Qg4 h5 12 Qh4 Be7.) 9...Qxd7 10 a3!? (If 10 0-0 then 10...b6
or 10...c6.) 10...b6 (necessary to save the Knight) 11 d3 (If 11 0-0 instead,
then 11...Nb7 12 b4 a5! according to Jozsef Palkovi, not 11...Rae8 12
Qa6!) 11...c6?! 12 b4! Nb7 13 dxc6 Qxc6 14 Nce4 Nd7? (14...Be7 leaves
White somewhat better.) 15 Qf3 Be7?! 16 Nxh7! and White won. As
Palkovi shows in his 2000 book Zweispringerspiel bis Traxler
Gegengambit, all this happened before in a 1996 Spanish game
Arnold-Iruzubieta, which doesn’t say much for GM Ivan Sokolov’s
preparation! Maybe he was misled by the fact that the Spanish game was a

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (3 of 13) [4/10/2001 9:12:14 AM]

short draw but Morozevich improved on it. Instead of 11...c6, Palkovi says
11...Rae8 is the right move since Qa6 is not possible in this line.

English GM Glenn Flear prefers to play 7...Be7. Then after 8 Nc3 0-0
White often chooses the safe 9 Bxd7. However, when he doesn’t hold the
gambit pawn he doesn’t really obtain winning chances e.g. 9...Qxd7 10 0-0
(if 10 d3 Nxd5) 10...Rfe8. On the other hand, 9 0-0 leads to obscure
positions. Herbrechtsmeier-Nunn, Bundesliga 1985, went 9...Bg4!? 10
Qxe5 Bd6 but this may not be good enough. Instead 9...c6 10 dxc6 Nxc6
seems to offer compensation. It may seem strange to avoid ...c6 earlier and
then play it after all, but White’s move Qe2 is not especially constructive.

Returning to 6...c6, play invariably goes 7 dxc6 bxc6 reaching the next
diagram. (See Diagram)

White must now either retreat the
Bishop or pin the c-pawn along the
diagonal. We look at the latter
possibility first.

8 Qf3

White tries to induce either a passive
defence of the c-pawn or else the
incorrect exchange sacrifice 8...cxb5?!
9 Qxa8. Instead Black can reply with a
move that was analysed by an English
prisoner-of-war in a Japanese camp
during World War II:

8... Rb8!?

It would now be foolhardy in practice (whatever your computer might say)
to grab the second pawn at c6. After 9 Bxc6+ Nxc6 10 Qxc6+ Nd7 White
has no piece in play except his Queen and the g5-Knight which is under
attack. Black will bring out his King’s Bishop, castle and then enjoy a big
lead in development and open lines.

Therefore the prudent move is to retreat the Bishop:

9 Bd3

Not to e2, because you could have done that last move if you wanted. Now
we have a fairly unusual position that Black (in his enthusiasm for the
Bxc6+ lines) may have neglected to study. I have never had any
experience with this line but it looks to me that White’s pieces are placed
awkwardly and his influence on the centre is reduced because his d-pawn
is blocked.

Play could go 9...h6 10 Ne4 (one of the points of this 8 Qf3, 9 Bd3 line)
10...Nd5 and it is reckoned that Black has just about enough play for the
pawn.

However, it is understandable that some players are not altogether

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (4 of 13) [4/10/2001 9:12:14 AM]

convinced by this variation. Instead of Colman’s 8..Rb8, the new move
8...h6!? is gaining credence. Now White cannot answer 9 Bxc6+? Nxc6 10
Qxc6+ because after 10...Bd7 both Queen and Knight are en prise. Maybe
9 b4!? is possible, since after 9...Bxb4 10 Bxc6+ Nxdc6 11 Qxc6+ Bd7 12
Qc4 White threatens both the b4-Bishop and mate on f7. However, Black
has a simple answer in 12...Qe7 13 a3 Ba5 14 Ne4 Nxe4 15 Qxe4 0-0
when White has zero development and Black’s chances look good. This
analysis comes from Heyken and Fette but I imagine Spassky may have
been the first person to discover this.

Van der Wiel-Spassky, Reggio Emilia 1985-86, went instead 9 Ne4 and
after 9....Nd5 White’s pieces do not look well placed.

Personally as White, I prefer to stick to the main line (after 5...Na5 6 Bb5+
c6 7 dxc6 bxc6), namely:

8 Be2

Black invariably kicks the advanced Knight back now by

8...h6

when I play 9 Nf3

I know that Steinitz invented 9 Nh3!? and that Bobby Fischer revived it,
but when I play 4 Ng5 against the Two Knights I usually try to avoid
artificial moves. Only fluid moves that consistently develop pieces, create
threats or challenge the centre really fit into the picture as I see it. (See
Diagram)

I have no experience with 9 Nh3 and
do not presume to know what is
Black’s best line. It’s tempting to chop
off the Knight on h3 at once but that is
probably what White wants you to do.
On the principle that “the threat is
stronger than the execution”, the move
Bxh3 is best postponed. 9...Bd6 is
possible but not as popular as 9...Bc5,
while even 9...Nb7 comes into
consideration. If you want to know
more about this line, study the famous
games Steinitz-Chigorin and

Fischer-Bisguier; the latter is in My Sixty Memorable Games.

The usual move 9 Nf3 threatens the black e-pawn so there follows:

9...e4 10 Ne5

I don’t mind my opponents making these supposedly tempo-gaining
moves like ...c6, ...h6 and ...e4 because all they are really doing is creating
holes and pushing my pieces (especially the Knight) to where they want to
go. Take a look at the diagram. (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (5 of 13) [4/10/2001 9:12:14 AM]

I like my Knight much better on e5
than on e4, where it lands in some of
those foregoing variations. You might
suppose the pawn on e4 is cramping
White but not really. When White
follows up d2-d4 in a few moves from
now, Black has to exchange en passant
because otherwise the anchored Knight
on e5 would become too strong.

Notwithstanding all the coverage I
have given so far to the alternatives up
to this point, in fact nearly all my

games reach this position. When they are over-the-board games (as
opposed to correspondence) the first ten moves are often played with
minimal consumption of clock time by either player.

From this point, I do not intend to cover all the alternatives in much detail.
There are just too many variations and I shall just say a little about the
ones I consider important.

10...Bd6

This is usually played but is not necessarily best.

A lines that should probably not bother White much is 10...Qd4 11 f4 Bc5
12 Rf1, (usually followed by 13 c3 and 14 Qa4, beating Black back). On
the other hand, 10...Qc7 is probably best met by 11 d4 soon transposing to
my main line below when Black captures the pawn en passant and then
plays ....Bd6.

However, Black also has 10...Bc5 which threatens 11...Qd4. Then White
answers 11 c3 (which in turn threatens the fork b2-b4). This is quite an
important alternative to the lines with ...Bd6 because White will soon be
obliged to play c2-c3, which blocks the a1-h8 diagonal. The black Bishop
must quickly return to d6, losing a tempo, but on the other hand White no
longer has the plan (seen in my main line below) of fianchettoing the
Queen’s Bishop. So 10...Bc5 can be seen as a tempo sacrifice to force
White to switch from his optimal plan.

I have encountered this 10...Bc5 line twice, both in postal and
over-the-board play. (See Diagram)

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (6 of 13) [4/10/2001 9:12:14 AM]

In his book Play the Open Games as
Black, GM Emms cites one of these
games as an equalising line for Black:
10...Bc5 11 c3 Qc7 12 f4 Bd6 13 d4
exd3 14 Qxd3! 0-0 15 Nd2! Bxe5 16
fxe5 Qxe5 17 0-0 (Harding-Read, CC
Olympiad 12 1992-93) saying “White
possesses the bishop pair but Black's
lead in development neutralises this".
Neither Read, an ICCF Senior
International Master, nor I thought it
was so clear-cut. The game continued
17...Re8! 18 Bd1 Bg4 19 Nf3 Qc7 20

Nd4 (not forced) 20...c5 and I accepted my opponent’s draw proposal.
Read then write to me: ''In the final position I 'felt' uncomfortable - your
QB was going to be a strong piece while my QN was in limbo. Although
my concrete analysis told me that I was OK I just didn't like my position.”

In my more recent game, played in the Irish league against a 2150
opponent, I played my novelty 15 Nd2 again. Black preferred to play a
pawn down rather than exchange his dark squared Bishop, but I won
without much difficulty.

11 d4

When I was growing up as a chess player, books tended to give both 11 d4
and 11 f4 as of roughly equal value but now I think 11 d4 has clearly come
out top, in conjunction with the following plan to fianchetto the white
Queen’s Bishop.

A recent example of 11 f4 was 11 ...exf3 (Gligoric used to prefer 11...Qc7
here.) 12 Nxf3 00 13 0-0 c5 14 d4 Qc7 15 Nc3 a6 16 Kh1 Bb7 17 d5 (17
Be3 Rad8 gave Black good play in Spassky-Geller, Gothenburg 1955.)
17... Rfe8 18 Nh4 Be5 19 Bf3 (Better is 19 Nf5 Rad8 20 Bxh6 Nxd5!= as
in Estrin-Letic, 1963!) 19...Rad8 20 Re1 Bg3! 21 Rxe8+ Rxe8 22 hxg3
Qxg3 23 Bf4 Qxf4 24 g4 Nc4 25 Ng2 Qg3 26 b3 Ne3 27 Nxe3 Rxe3 28
Bg2 Nxg4 29 Kg1 Qf2+ 30 Kh1 Rh3+ 01 Ledezma Alvarez-Letic,
Loeffler Memorial corr 1995-97.

11 ... exd3

If 11...Qc7 then 12 Bd2 (eyeing the a5-Knight), 12 Nc3 or 12 0-0 are all
possible. Black again faces the dilemma of whether to exchange his
dark-squared Bishop for a Knight in order to recover the gambit pawn.

12 Nxd3 Qc7

Black tries to deter kingside castling, although in some lines White is
willing to give up the h-pawn.

13 b3!

In connection with this move, I am reminded of the interesting observation

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (7 of 13) [4/10/2001 9:12:14 AM]

by GM Rowson in his stimulating new book The Seven Deadly Chess Sins.
He remarks in passing that when the old masters estimated that the Knight
and Bishop were of roughly equal value, this was before the days when the
fianchetto was commonplace. On b2 the Bishop will not only exert
influence but will also be fairly safe from being exchanged by a Knight.
(See Diagram)

In my opinion, Black is now struggling
to justify his gambit. White’s
dark-squared Bishop will be the best
minor piece on the board when it gets
to b2, observing the centre and bearing
down against Black’s King. The
Knight on a5, now denied the vista of
c4, is a liability for Black. It can drop
off if he is careless, and at best it will
have to manoeuvre slowly back to the
centre where it may find itself
opposing a rampaging Bishop in the
endgame.

White’s policy about the extra pawn is, roughly speaking, the following.
Over the next few moves, while White is still completing his development,
he may be willing to allow Black to regain the pawn (usually the f-pawn or
h-pawn). This will, however, come at a price: either many exchanges
draining Black’s position of its dynamism, or in particular the exchange of
the black dark-squared Bishop for a white Knight. If that happens, the
Bishop on b2 generally becomes to strong so that Black prefers not to win
back the pawn.

If you look back at my notes above on the 10...Bc5 line you can see this
principle in action. Mike Read decided to win back his pawn at the
expense of conceding the Bishop pair, but it wasn’t so serious in that case
because my Bishop wasn’t going to b2. My other opponent kept his dark
squared Bishop but then my knight manoeuvre Nb1-d2-f3 consolidated the
extra pawn.

When I first played this line as White, I was always thinking in terms of
trying to work up an attack against the black King, and sometimes it is
possible. In one game, my Queen’s Knight came to e4 and I sacrificed it
on f6 to shatter the black kingside and create a raging attack on his King.
Normally, however, Black will not allow this.

Later I came to prefer the “draining the dynamism” approach. If Black
allows the game to lose its energy, he is left with two weak queenside
pawns against a unified majority and that can be the basis for a strategic
white win.

Another scenario is that Black avoids exchanges and goes for an all-out
attack with material down. In that case, White must be prepared to weather
the storm and win in the endgame or by counter-attack.

You cannot expect me to reveal all my secrets in an article like this so I

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (8 of 13) [4/10/2001 9:12:14 AM]

shall just lightly annotate one game. It is far from being perfect, and Black
probably missed drawing chances. However, the very fact that it is not a
tactical battle will force us to focus on the broad strategic sweep, which is
my aim in this series.

Tim Harding - Pablo Adrian Gritti (Argentina) Loeffler Memorial corr,
1995-99

1 e4 e5 2 Nf3 Nc6 3 Bc4 Nf6 4 Ng5 d5 5 exd5 Na5 6 Bb5+ c6 7 dxc6
bxc6 8 Be2 h6 9 Nf3 e4 10 Ne5 Bd6 11 d4 exd3 12 Nxd3 Qc7 13 b3 0-0
14 Bb2 Nd5

14 ...Ne4 and 14...Re8 are the critical replies according to Emms.

15 Nc3

Often played but this is probably note best. 15 h3 favoured White in
Morozevich-Balashov, Russia Cup, Novgorod 1997.

15...Nf4 16 0-0

In his new book, Hungarian analyst Palkovi suggests 16 Nxf4 Bxf4 17 h3!
Rd8 18 Bd3 c5 19 0-0 c4 20 Nb5! Qc6 21 bxc4 Nxc4 22 Be4!.

16...Nxe2+ 17 Qxe2 Bxh2+ 18 Kh1 Bd6 (See Diagram)

19 Ne4

I have also played 19 Qh5 which is
safer but too drawish. I think there are
fewer unanswered questions there.

19...Qe7 20 f4 f5

My knight sacrifice idea arises in the
variation 20 ...Bf5 21 Nf6+!?
Harding-Garriga Nualart, 1988.
Looking in detail at that game is the
exact opposite of my aim in this series!
(Ultimately I won on time after

misplaying the attack and “achieving” a lost position.)

21 Nc3 Re8

I thought 21 ...Qh4+ 22 Kh1 was the critical line when playing Garriga.

22 Qxe7

Not 22 Rae1?? Qh4+

22 ...Rxe7

If 22 ...Bxe7 I intended 23 Rae1 (23 Ne5 Bf6).

23 Rae1

Black has two Bishops but weaker pawns; my question at the time was,

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (9 of 13) [4/10/2001 9:12:14 AM]

can this be exploited? Ultimately this game showed that it can be.

23...Ba6

If 23 ...Rxe1!? 24 Rxe1 followed by the outpost Ne5 when the Knight is in
no way inferior to his Bishops.

24 Rxe7 Bxe7 25 Re1 Bd6 26 Ne5 c5 27 Nd1 Bb7 28 Bc3 Nc6 29 Ne3

Black is drifting towards a lost endgame as the white pieces find their
optimum squares; now Ne3-c4 threatens. So Black decides to simplify
although he should perhaps have preferred 29 ...Rf8 30 N3c4 Bc7 and
hope to defend against my knight incursions. Probably he did not realise
the danger he was facing in the ending. (See Diagram)

29...Bxe5 30 fxe5 Ne7 31 Rd1 Be4 32
Rd7

The Rook is only temporarily on the
seventh rank but says “I’ll be Back!”.

32...Nc6 33 e6

White heads for a rook ending which
he believes is winnable.

33...Nd4 34 Bxd4 cxd4 35 Nd5

35 Rxd4 Re8 was rejected as he can
regain the pawn and then he has the

superior minor piece. I don't like to play Knight versus Bishop when he
has the correct Bishop for h-pawn.

35 ...Bxd5

If 35...Bxc2 36 Nc7.

36 Rxd5 Re8 37 Rxf5!

After 37 Rxd4 Rxe6 38 c4 Re2 White has no winning chances.

37 ...Rxe6 38 Rd5 (See Diagram)

Here is the strategic picture. All minor
pieces have come off. Black has a 2-1
kingside pawn majority which is never
going to yield a dangerous passed
pawn. If all the queenside pawns die,
the resulting rook ending is a
theoretical draw. Therefore I can play
for a win without risk.

On the plus side, his weak c6-pawn has
become a doomed d4-pawn. His
a7-pawn is also sickly, only defensible
by keeping his Rook passive.

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (10 of 13) [4/10/2001 9:12:14 AM]

38...Re4?!

Black had a difficult decision here. Presumably he should have played 38
...Re2 39 Rd8+ Kh7 40 Rxd4 Rxc2 41 Ra4 Rc7 42 Ra6 although this gives
me chances without risk?

39 a4

Threatening to create an outside passed pawn.

39...Kf7

If 39...Re2 40 Rd8+ Kf7 41 Rxd4 and White has an extra pawn with
winning chances.

40 Rd7+ Kf6

I had expected 40...Ke6 forcing White to decide which pawn to take.

41 Rxa7 Re1+ 42 Kh2 Rc1 43 Rd7

Of course not 43 Rc7?? d3.

43...Ke5

If 43...Rxc2 44 Rxd4 and I should win with some ease. Not 43...Rd1?? 44
c3.

44 Re7+ Kf4?!

If 44...Kd5 45 Re2 (Now my Pawn is protected he must march his King to
c3) 45...Kc5 46 Kg3 Kb4 47 Kf3 Kc3 48 Ke4 Ra1 49 Rf2 gives White
winning chances.

45 Rc7!

This is the point; now ...d3 doesn't work as Rc4+ and b3-pawn defends the
Rook, enabling cxd3. Otherwise I play Rc4 and defend my pawns. Then I
try to advance a-pawn or get my King out. I have a strong passed
queenside pawn probably I will have two or three connected passed pawns
before too long. So now he must be lost. (See Diagram)

45...g5

If 45...h5 then 46 a5 is probably
simplest as I don't have to play Rc4
while his King is on the 4th rank.

46 Rc4 h5 47 Rxd4+ Kf5 48 a5!

The game now speeded up dramatically
as, after four years struggling with the
post, we concluded it by fax within a
few weeks. I am giving up the c-pawn
but pushing my pawns fast.

48...h4

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (11 of 13) [4/10/2001 9:12:14 AM]

If 48...Ke5 then 49 Rc4 Kd5 50 a6 Ra1 51 Ra4 wins rather simply e.g.
51...Rxa4 52 bxa4 Kc6 53 a5 Kc7 54 c4 Kb8 55 c5 Ka7 56 c6 Kxa6 57
Kg3 Ka7 58 Kf3. White mops up the kingside and then brings his King to
d7.

49 Rc4

By this stage there is probably more than one way to win.

49...g4

Black threatens mate in 2 so my reply is forced.

50 g3 h3

If 50...hxg3+ 51 Kxg3 was sent as a conditional; he tries something
trickier.

51 a6 Ra1

51...Re1 52 Rf4+ is also hopeless for him in the long run.

52 Ra4 Rc1 53 a7 Rxc2+ 54 Kg1 Rc1+ 55 Kf2 h2

If 55...Rc2+ 56 Ke3 Rc3+ (56...Rc1 57 a8Q or 56...Rc8 57 a8Q Rxa8 58
Rxa8) 57 Kd4 Rc8 (57...h2 58 a8Q) 58 a8Q Rxa8 59 Rxa8 is an obvious
win with the extra Rook.

56 a8Q h1Q (See Diagram)

After the dry ending, White has the
pleasant task of calculating the forced
mate.

57 Rf4+ Ke5 58 Qe8+ Kd6

Or 58...Kd5 59 Rf5+ Kd6 (59 ...Kd4 60
Qe3#) 60 Qd8+! (quickest) 60...Kc6
(60...Ke6 61 Rf6+ Ke5 62 Qd6+ Ke4
63 Rf4#) 61 Rf6+ Kc5 62 Qd6+ Kb5
63 Qb6#.

59 Qd8+ Kc5

Other mating lines are 59...Kc6 60 Rf6+ Kc5 61 Qd6+ Kb5 62 Qb6#; or
59...Ke6 60 Rf6+ Ke5 61 Qd6+ Ke4 62 Rf4#; or 59...Ke5 60 Qd4+ Ke6 61
Rf6+ Ke7 62 Qd6+ Ke8 63 Rf8#.

60 Rf5+ Kc6

I didn't offer a conditional sequence against this as the mate was easier, so
he played it! I did offer 60...Kb4 61 Qa5+ Kxb3 62 Qb6+.

61 Rf6+ 1-0

Now that the number and length of the variations has reduced, I announced
mate and he resigned.

The Kibitzer

file:///C|/Cafe/Tim/kibb.htm (12 of 13) [4/10/2001 9:12:14 AM]

Don’t get me wrong. I am not claiming that I was winning all along. If you
find a defence for Black somewhere in the middle-game or endgame, it
doesn’t mean my argument fails. There are other ways that White too could
have played along the same strategic lines.

It would be going too far to conclude that in the 5...Na5 Two Knights, Black
loses because of his offside Knight and weak queenside pawns, but that
should be one of your guiding principles when playing White in this
variation!

This concludes my look at the 5...Na5 lines. Next month: the Fritz Variation.

This column is available in

Chess Cafe Reader

format. Click

here

for

more information.

[

The Chess Cafe Home Page

] [

Book Reviews

] [