Professional Documents
Culture Documents
51 Flights of Chess Fancy - Beasley - 2009
51 Flights of Chess Fancy - Beasley - 2009
51 Flights
of
Chess Fancy
and a few other frolics
rdwdw4kd
dp0pdp0p
pdwdwhbd
dwdwdwdw
wdwdwdwd
dPdwdwdw
wdP)P)P)
dwGwdwIw
What was White’s first move
with a bishop?
Contents
Introduction 3
1 Endgame studies 4
3 Problems 29
6 Joke compositions 45
Interlude 53
ISBN 978-0-9555168-1-8
1.1 1.1a
wdwdwdwd wdwdwdwd
dwdwdpiw dwdwdpdw
wdwdwdwd wdwdwdkd
dw0w)Kdw dw0w)wdw
wdPdwdwd wdPdKdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
White to play and win 1 Ke4, after 1...Kg6
Let’s start with something very simple. gives 1.1a but with Black to play.
1 e6 fxe6+ 2 Kxe6 and wins? Yes, but If now 2...Kg5 then 3 e6! fxe6 4 Ke5,
Black will play 1...Kf8, and after 2 exf7 and White will win despite being
Kxf7 he will meet White’s eventual temporarily a pawn down; Black’s king
Kxc5 with ...Kc7 and draw; and if is on precisely the wrong square
White tries 2 Kf6/Ke5 instead, we will (he could draw from either g6 or g4).
have 2...fxe6 3 Kxe6 Ke8, and the Alternatively, 2...f6/f5+ 3 exf6 and
result will be the same. 4-5 Kxc5, or 2...Kg7 3-4 Kxc5.
All right, try 1 Ke4: no, 1...Kg6 1.1a is in fact a position of reciprocal
(see 1.1a) 2 Kd5 Kf5 3 Kxc5 Kxe5, and zugzwang, where Black to play loses
after 4 Kb6 f5 both sides will promote. but White to play cannot win. If to
Hmm. The solution is actually 1 Kf4, avoid it Black plays 1...f6, then 2 Kf5
and only if 1...Kg6 then 2 Ke4. This etc; if 1...Kh6/Kh7 then again 2 e6.
Endgame studies 5
1.6 (by Wallace Ellison and myself) defence of his pawn). This gives 1.6a :
Kdkdwdwd 1.6a
)wdwdwdw
wdwdwdpd Kdkdwdwd
dwdwdwdw )wdwdwdw
wdwdwdwh wdwdwdwd
dwdwdwdw dwdwdw0w
wdwdB)wd wdwdwdwh
dwdwdwdw dwdBdwdw
wdwdw)wd
White to play and hold the draw dwdwdwdw
Without the pawns on g6 and f2, White After 1...g5
would have an easy draw. He would
check the Black king away from c8, The natural move is 2 Be4 penning the
forcing it to move to c7, and then play knight, but Black can temporize by
to keep the knight away from the 2...Kc7, and after 3 f3 Ng2 we have
squares giving on b6. But if Black can 4 B~ Nf4 5 Be4 (as we shall see, if
capture the f-pawn without losing his Black can reach e5 he will win, so
g-pawn, he will win. He will play to a White must guard d3/g6 as well as d5)
square such as e5, forcing the White Ne6 6 B~ Nc5 7 Bb5 (the same motif –
bishop to play to e6 or b5, and then White must defend d3 as well as a4/d7)
make a tempo move with his pawn. and we have reached 1.6b :
The natural opening move is 1 Bd3 –
or should we throw in a check first, 1.6b
1 Ba6+ Kc7 2 Bd3? No, Black will play
2...g5 with an eventual win; a best-play Kdwdwdwd
line is 3 Be4 g4 4 Bd3 Nf3 5 Bf5 Ne5 )wiwdwdw
6 Be6 Nd3 (to meet 7 Bxg4 by 7...Nc5 wdwdwdwd
and 8...Na4 or 8...Nd7) 7 Bd7 (now dBhwdw0w
7...Nc5 can be met by 8 Bb5/Be8, and wdwdwdwd
7...Nxf2 by 8 Bxg4) Nf4 (threatening dwdwdPdw
8...Nd5) 8 Bc6 Ng6 (aiming for c8, wdwdwdwd
which cannot be prevented) 9 B~ Ne7 dwdwdwdw
10 Be6/Bb7 Nc8 11 Bxc8 Kxc8 12 f4
g3 13-14 f6 g1Q 15 f7 Qg2 mate. 2 Be4, after 7 Bb5
So we confirm 1 Bd3, and the natural
reply is 1...g5 (if instead 1...Kc7 then 7...Kc8 (this tempo move has become
2 f4, and Black’s knight is tied to the possible because the knight guards the
Endgame studies 11
squares from which the bishop might Wallace’s original diagram, so this may
check) 8 Bc6 (nothing else is better) not be the precise position that he sent
Nd3 9 Bb7+ (again nothing else is me, but it reproduces the essentials.
better) Kc7 10 B~ Ne5 (Black has
attained his first objective) 11 Bb5/Be6 1.6c
(White’s first priority is to guard c4/d7,
so he must leave his pawn to its fate) wdwdwdwd
Nxf3 12 B~ Ne5 13 Bb5/Be6 g4 14 B~ dw0wdwdw
Nc4/Nd7 and 15...Nb6 mate. wdwdbdwd
Letting the knight out by say 3 Bd3 Hwdwdwdw
is soon seen to be even worse, and wdwdwdwd
2 Be4 is in fact a losing move. Correct dPdwdwdw
is the apparently absurd 2 f4! throwing wdwdwdw0
away the vital pawn. 2...g4 would allow dwdwdKdk
3 f5 with an easy draw (or 3 Bf5+ if
White prefers, since 3...Nxf5 will be White to play and win (intended)
stalemate), so Black must take, 2...gxf4,
and now we play 3 Be4? No, 3...Kc7, Wallace’s intention was 1 b4 Bd5
and White must release the pressure. 2 Kf2 c6 3 Nb7 followed by the play
Correct is 3 Bf5+, playable because the we saw above with reversed colours
capture will again be stalemate, and (after 2 Be4 g5 3 f3 Ng2). This was a
only after 3...Kc7 do we at last play the fine piece of analysis, typical of its
penning move 4 Be4. Now Black is originator, but then it occurred to me to
helpless. 4...Kc8 will be met by 5 Bf5+ wonder: after 1 b4, what happens if
repeating the position we have just had, Black plays the utterly ridiculous move
and 4...f3 allows 5 Bxf3 since the 1...c5? It draws, doesn’t it? White
capture will once more give stalemate. cannot cope with a fleeing c-pawn, so
And if we now ask why 1 Ba6+ Kc7 he must play 2 bxc5, and Black draws
2 Bd3 g5 cannot similarly be followed by the play already seen.
by 3 f4, since 3...gxf4 4 Be4 is drawn, This seemed too good to waste, so
the simpler answer is 3...g4 though I reversed the colours to give 1.6, in
3...Ng2 also wins. which the chance discovery has become
This is an example of “composition” the main line and Wallace’s excellent
by spotting something in a position analysis refutes the apparently natural
which had actually been put on the move 2 Be4, and suggested that we
board by somebody else. During 1995, publish it as a joint study. We sent it to
Wallace Ellison was looking at endings a magazine in Eastern Europe and no
with knight against bishop in which solvers’ comments reached us, but it
the knight needed to win a pawn. 1.6c has always had a good response when
illustrates one of them. I no longer have I have put it up in front of an audience.
12 Fifty-one flights of chess fancy
1.7 (by J. N. Baxter, my version) After 1 Bf1 Bc6 2 h4, the capture
2...gxh3 opens extra lines of attack for
wdwdwdwd White, Black’s bishop cannot play to c6
dbdwdwdw because it is already there, and 2...Bd7
wdwiwdwd is met by 3 Bd3 winning at once. This
dpdpdpdp leaves 2...Be8, and a best-play line is
wdwIw)pd 3 Be2 Bc6 4 Bd1! Bd7 5 Bb3 Be6
)w)wdw)w 6 Bc2 Bc8 7 a4 bxa4 8 Bxa4. When we
wdwdwdB) tried this before, the Black bishop was
dwdwdwdw on e6/f7/g8, and he could play ...Bf7 or
...Bg6. With his bishop on c8, he cannot
White to play and win stop White penetrating to e8, and the
rest will be easy.
I set 1.7 in the British Chess If, after 1 Bf1 Bc6, White plays 2 h3
Magazine with two questions: how instead of 2 h4, 2...Be8 holds the draw.
does White win after 1 Bf1 Bc6 2 h4, White can continue 3 hxg4 hxg4 4 Be2
and why does 1 h4 not work? and then play as before to get his
bishop through to e8, but with no target
e f
wdwdwdwdd a on h5 there is nothing to be gained, and
e b
dwdwdwdw g if instead 3 h4 then 3...Bc6 gives Black
the draw already seen. If White plays
h
wdwiwdwdc
1 h3 in the initial position, 1...Bc6
dpdpdpdp draws. And if Black meets 1 Bf1 with
wdwIw)p) 1...Ba6 instead of 1...Bc6, the answer is
A B
)w)wdw)w not 2 a4 “exploiting” the pin (when
C C
wdwdwdwdD E 2...bxa4 wins for Black) but 2 Bd3 etc.
F G
dwdwdwdw H Baxter had the bishops already on f1
and c6, and the Pc3 on e3. The latter
Take the latter first. Black can meet allowed 1 h4 Be8 2 Bd3 Bd7 3-4 Bb3
1 h4 by the corresponding-square Bg8 5 Kc3 Kc5 (else 6 Kb4) 6 a4, when
defence shown above (if White goes to 6...bxa4 loses and 6...b4+ gives White
A, Black goes to a, and so on), and he an outside passed pawn with all the
must start with 1...Ba8! since he is initiative and perhaps an unwanted win.
already on b7. If White plays to 1.7 removes this unclear line (now, if
b3/c2/d1 and advances the a-pawn, White’s king leaves d4, Black can play
Black, now on g8/e6/f7, can exchange his bishop to d7/h7/g6 and only then his
and meet White’s Bxa4 with ...Bf7 or king to c5, and the draw is not difficult
...Bg6. This keeps the White bishop to show) and adds the elegant refutation
from e8, and it soon becomes clear that 1 h4 Ba8, and though it is only “Baxter,
White will get nowhere. version JDB” I thought it worth doing.
Endgame studies 13
1.9 (after Artur Mandler) when White could now win by 9 Kd5
since 9...a1Q would allow 10 Qh8+).
wdwdwdKd So White must think of something
$wdw)wdw else, and the answer is 1 Kf7 Rf4+
wdwdwdwd 2 Ke6! Re4+ 3 Kd7 Rd4+ 4 Ke8,
0wdwdwdw going right round his pawn and
pdwdrdwd reaching e8 from the left. Now 4...Kd3
dwdwiwdw gives 1.9a with the Black rook on d4
wdwdwdwd instead of f4, and White can play 5 Rd7
dwdwdwdw pinning it and exchanging it off: 5...a3
6 Rxd4+ Kxd4 7 Kd7 (say) a2 8 e8Q
White to play and win a1Q 9 Qh8+. If instead 4...Re4 or
4...Ke2 then 5 Rxa5 and 6 Kf7, with a
The White king will have to hide on e8 shelter for the king after 6...Rf4+ 7 Ke6
sooner or later, but if he goes there Re4+ 8 Re5; if 4...Rd5 then 5 Kf7 Rf5+
straight away, 1 Kf8 Rf4+ 2 Ke8, Black 6 Ke6, and even playing to take off the
will play 2...Kd3 to bring his own king new queen will not help (6...Ke4 7 e8Q
nearer to his pawns : Re5+ 8 Kd7/Kf7 Rxe8 9 Kxe8 and
10 Rxa5).
1.9a This was a spin-off from my
translation of Mandler’s book Studie.
wdwdKdwd I was analysing one of his studies by
$wdw)wdw computer, and a sideline came down to
wdwdwdwd 1.9 with the White king on e6 instead
0wdwdwdw of g8. I expected the computer to play
pdwdw4wd Kd7 or Kf7 indifferently, and was very
dwdkdwdw surprised to see that it had a strong
wdwdwdwd preference for Kd7. But examination
dwdwdwdw soon showed why, and then it was just
a matter of adding the little king walk
1 Kf8, after 2...Kd3 to highlight the distinction.
The fact that each of White’s
If now 3 Kd7 then 3...Rd4+, and if attempts to win from 1.9a demands a
further 4 Ke6 then 4...Re4+ and even different refutation (3 Kd7 Rd4+,
5...Rxe7 is enough to draw; if 3 Kd8 3 Kd8 Re4, 3 Rxa5 Kc2) is a definite
then 3...Re4, and if further 4 Rxa5 bonus (and a nightmare for solvers).
Kc3/Kc2 5 Rxa4 then 5...Rxa4 6 e8Q Usually, a composer has to go to a great
Ra8+; and if 3 Rxa5 then 3...Kc2 4 Kd7 deal of trouble to achieve such a thing.
Rd4+ 5 Ke6 Re4+ 6 Re5 Rxe5+ 7 Kxe5 Here, it turned up completely by
a3 8 e8Q a2 draws (but not 3...Kc3, accident.
Endgame studies 15
1.9 (after Artur Mandler) when White could now win by 9 Kd5
since 9...a1Q would allow 10 Qh8+).
wdwdwdKd So White must think of something
$wdw)wdw else, and the answer is 1 Kf7 Rf4+
wdwdwdwd 2 Ke6! Re4+ 3 Kd7 Rd4+ 4 Ke8,
0wdwdwdw going right round his pawn and
pdwdrdwd reaching e8 from the left. Now 4...Kd3
dwdwiwdw gives 1.9a with the Black rook on d4
wdwdwdwd instead of f4, and White can play 5 Rd7
dwdwdwdw pinning it and exchanging it off: 5...a3
6 Rxd4+ Kxd4 7 Kd7 (say) a2 8 e8Q
White to play and win a1Q 9 Qh8+. If instead 4...Re4 or
4...Ke2 then 5 Rxa5 and 6 Kf7, with a
The White king will have to hide on e8 shelter for the king after 6...Rf4+ 7 Ke6
sooner or later, but if he goes there Re4+ 8 Re5; if 4...Rd5 then 5 Kf7 Rf5+
straight away, 1 Kf8 Rf4+ 2 Ke8, Black 6 Ke6, and even playing to take off the
will play 2...Kd3 to bring his own king new queen will not help (6...Ke4 7 e8Q
nearer to his pawns : Re5+ 8 Kd7/Kf7 Rxe8 9 Kxe8 and
10 Rxa5).
1.9a This was a spin-off from my
translation of Mandler’s book Studie.
wdwdKdwd I was analysing one of his studies by
$wdw)wdw computer, and a sideline came down to
wdwdwdwd 1.9 with the White king on e6 instead
0wdwdwdw of g8. I expected the computer to play
pdwdw4wd Kd7 or Kf7 indifferently, and was very
dwdkdwdw surprised to see that it had a strong
wdwdwdwd preference for Kd7. But examination
dwdwdwdw soon showed why, and then it was just
a matter of adding the little king walk
1 Kf8, after 2...Kd3 to highlight the distinction.
The fact that each of White’s
If now 3 Kd7 then 3...Rd4+, and if attempts to win from 1.9a demands a
further 4 Ke6 then 4...Re4+ and even different refutation (3 Kd7 Rd4+,
5...Rxe7 is enough to draw; if 3 Kd8 3 Kd8 Re4, 3 Rxa5 Kc2) is a definite
then 3...Re4, and if further 4 Rxa5 bonus (and a nightmare for solvers).
Kc3/Kc2 5 Rxa4 then 5...Rxa4 6 e8Q Usually, a composer has to go to a great
Ra8+; and if 3 Rxa5 then 3...Kc2 4 Kd7 deal of trouble to achieve such a thing.
Rd4+ 5 Ke6 Re4+ 6 Re5 Rxe5+ 7 Kxe5 Here, it turned up completely by
a3 8 e8Q a2 draws (but not 3...Kc3, accident.
16 Fifty-one flights of chess fancy
Play in 1.15 starts 1 Qf6+ Kb1 2 Qa1+ where whoever is to move must allow
Kxa1 3 Kc1, which is hackneyed but mate in one. It may be asked if we
avoids starting with the king trapped in cannot reach this in both “right” and
the corner, and we have 1.15a, “wrong” lines without the extra pawns.
All I can say is that I cannot achieve it.
1.15a Suppose, utterly crudely, that we set
whwdwdwd whwdwdwd
dwdwdwdw dwdQdwdw
wdpdwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
wdPdwdwd wdwdwdwd
dwdpdwdw dwdpdwdw
pdpdwdwd pdpdwdwd
iwIwdNdw iwdKdNdw
After 3 Kc1
intending 1 Kc1? Nxd7 2 Nd2 Nc5 and
where Black must play 3...Nd7/Na6 to 1 Kd2! Nxd7 2 Kc1 Nc5 3 Nd2. Firstly,
avoid mate on b3. this can have no history, since Black’s
If White now plays 4 Nd2 still last move was a bad one (if ...c3-c2+,
aiming for b3, Black will reply 4...Nc5, ...Nxd7 would have won, if ...b3xc2+,
and it is White who will be mated. ...d3xc2+ would have won). Secondly,
White must insert 4 c5, and after it isn’t sound anyway, because 1 Kd2
4...Nxc5 5 Nd2 everything works. c1Q+! wins for Black. So the tempo
If instead 4...d2+ then 5 Nxd2 Nxc5 must be lost earlier in the play, and I
6 Kxc2 and again mate next move. cannot come within a mile of success.
Endgame studies 21
The reason that a study like 1.17 is intending 1 Bf1 h2 2 Bxh2 gxh2 etc.
difficult is that the opening sacrifice This turned out to allow unwanted
seems pointless, and not until the solver alternative solutions and today I would
spots the quiet move that follows does think it undesirable anyway, but for
he realise its purpose: 1 Rh4+ Qxh4 some reason the line I looked at first
2 Rg8! This gives 1.17a, was 1 f4 Kxg1 2 f5 Kf2 3 f6 Kxe2
4-5 f8Q g1Q+ 6 Kh7 Qa7+ 7 Kg8 h2 :
1.17a
wdwdw!Kd
wdwdwdRd 1wdwdwdw
dw)wdwdp wdwdwdwd
wdwdwdwd dwdwdwdw
dwdwdwdw wdwdwdwd
wdwdwdw1 dwdwdwdw
dKdwdwdw wdwdkdw0
wdwdwdwd dwdwdwdw
dwdwdwdk
After 2 Rg8 White has only 8 Qe8+, and Black will
escape after either 8...Kf2 9 Qf8+ Kg2
and the rest is automatic: 2...Qh3+ or 8...Kd2 9 Qd8+ Kc2 10 Qc8+ Kb2.
(Black has no other check) 3 Kb4 (as 1.17, like 1.16, was outclassed by
long as White keeps to the b-file, Black others (there is a famous Mitrofanov
will have no check in the middle of the study where a White Ka5 is checked by
board) Qh4+ 4 Kb5 Qh5+ 5 Kb6 a Black Qh5, and White sacrifices his
Qh6+ 6 Kb7, and now Black has no own queen on g5 so as to hide his king
check at all. on a6), but I still have a soft spot for it.
Endgame studies 23
wiwdwdwd wdwdwdwd
dP0wdpdp dwdwdwdw
KdP0wdwd wdwdwdwd
0wdP0wdw dwdwdwdw
r0wdPdw0 wdwdwdwd
1pdw)pdP dwdwdpdB
q4pdw)wd wdwdwdw0
hbHwdwdw dwdwdKHk
White to play and win White cannot win, but...
Two medieval jokes to finish the 1.20 answers the constructional task
chapter. 1.19 was dedicated to the which we saw earlier in the chapter.
memory of the crusading king who was As set, White cannot win, and must
so nicknamed because he was always give stalemate by Nxf3 if he is even to
popping off to the Mediterranean. avoid losing; but take away either the
White can lose a move only in the h1 knight or the bishop, and he has a mate
corner, so 1-17 Ka6 f6 (say) 18-34 Ka6 in two. If the multiple mates in two
h6 35-51 Ka6 h5 52-68 Ka6 f5 (no with the bishop are thought undesirable,
choice now) 69 exf5 and mate at move the addition of an extra pawn on f4
75. Illegal position, of course; Black would remove them, but there seems
must have started with twelve pawns. little point and I prefer it as it is.
Chapter 2
Creating a study from a game position
The previous chapter has included brief notes on how the studies came into being,
but a great deal of detail has been skipped, and in particular we have passed lightly
over most of the false trails with which the practising composer becomes all too
familiar.
In the present chapter, we follow the gestation of a study as it happened.
We take a game position, which had an interesting move both in the play and in the
“might have been”, we distil the essence of it, and we then see what we can do to
make it even richer. On this occasion, the composer was reasonably lucky. All too
often, he isn’t; but if composers were never lucky at all, there would be no studies
for us to enjoy.
A (Tischbierek-Vorotnikov) B
wdwdwdwd wdwdwdwd
dwdwiwdw dwdkdPdw
wdwdw0Kd wdwdwIwd
dwdwdPdw dwdwdPdw
p0p0wdPd w0w0wdwd
dPdPdwdw dPdPdwdw
w)Pdwdwd w)wdwdwd
dwdwdwdw dwdwdw1w
Black to play After 7 Kf6
In February 2002, Hew Dundas sent Leonard asked what better first move
me A, from a game Tischbierek- or moves Black could have played. My
Vorotnikov, Leipzig 1999. He had seen computer thought that both 1...axb3 and
it in Leonard Barden’s Financial Times 1...Kf8 appeared to win, but 1...Kf8
column, and thought I might find it of was clearly the move that he had in
interest. Play continued 1...cxb3 2 cxb3 mind. It therefore occurred to me to try
a3 3 Kg7! a2 4 g5 a1Q 5 gxf6+ Kd7 and create a study in which this move
6 f7 Qg1+ 7 Kf6! (see B) and White was necessary in order to win.
may even be winning despite his initial The first step was to reverse the
disadvantage. colours and set up C :
26 Fifty-one flights of chess fancy
C E
wdwdwdwd wdwdwdw!
dwdwdwdw dwdwdwdw
wdwdpdwd wdwdpdwd
dpdw)wdw dwdw)wdw
wdpdwdw) wdpdwdwd
dk)wdwdw dwdwIwdw
wdwIwdwd wipdwdwd
dwdwdwdw dwdwdwdw
Win by 1 Kc1 only D, after 1 h6 Kb2 ... 4 Ke3 c2
1 Kc1 clearly won; 1 h5 equally clearly (now we have E above) 5 Qh1 c1Q+
didn’t (1...Kb2 2 h6 b4 and both sides (5...c3 6 Qb7+ etc) 6 Qxc1+ Kxc1
will promote). But the refutation of 7 Kd4 Kd2 8 Kxc4 and 9-11 Kxe6 with
1 h5 didn’t include the move ...Kc3, a win for White (which was nonsense,
which played such an important role in but I am telling it as it happened). This
the game, so I moved the pawn forward being so, if we added a pawn on e7 to
to h5 to reinstate it : slow the White king down then 1 h6
ought to fail. This gave F, with which
D the computer was happy.
wdwdwdwd F
dwdwdwdw
wdwdpdwd wdwdwdwd
dpdw)wdP dwdw0wdw
wdpdwdwd wdwdpdwd
dk)wdwdw dpdw)wdP
wdwIwdwd wdpdwdwd
dwdwdwdw dk)wdwdw
wdwIwdwd
Alas, 1 h6 also wins dwdwdwdw
The intention was 1 h6 Kb2! 2 h7 b4 Win by 1 Kc1 only
3 h8Q bxc3+ 4 Ke2 c2 5 Qb8+ Kc3! as
before, but alas, the computer said that This was the sheerest good luck,
1 h6 won anyway. because the reason 1 h6 had worked
However, it occurred to me that this previously had nothing to do with the
was because of lines such as 4 Ke3 c2 line above (Black can play 8...Ke3, and
Creating a study from a game position 27
if 9 Kc5 then 9...Ke4 10 Kd6 Kf5 and it My ambitious idea was to start the
is Black who wins). The line the h-pawn back at h2, to make the Black
computer had actually found from E king run into the well known but
was 5 Qh2 Kb1 6 Qh7 Kb2 7 Qb7+ always pleasant zugzwang trap with
Kc3 8 Qh1 Kb2 9 Qg2 Kb1 10 Qb7+, Kf6/Pf7 against White Pf5/h6,
and the addition of the pawn on e7
had killed this quite fortuitously by wdwdwdwd
blocking the seventh rank. I later dwdwdpdw
moved the pawn from e7 to f7, giving wdwdwiw)
G, and wondered why I had not put it
there to start with.
dwdwdPdw
wdwdwdwd
G dwdwdwdw
wdwdwdwd
wdwdwdwd dwdwdwdw
dwdwdpdw
wdwdpdwd and to sort out the Q-side by exploiting
dpdw)wdP the equally well known win with a king
wdpdwdwd against three passed pawns :
dk)wdwdw
wdwIwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
wdwdwdwd
Win by 1 Kc1 only
dwdwdwdw
G seemed to me to have distilled the
w0p0wdwd
essentials of Tischbierek-Vorotnikov. dwdwdwdw
The key moves were all there (1 h6 wdwdwdwd
Kb2 2 h7 b4 3 h8Q bxc3+ and if 4 Ke2 dwIwdwdw
then 4...c2 5 Qb8+ Kc3 and draws
at least, 1 Kc1 and wins), and the I say “ambitious” because a composer
position, unusually for a study derived who specifies so much detail in the
from a game, was actually simpler and play leaves himself little scope for
more natural than the original. But of constructional finesse; all he can do is
course this was merely a one-move to try the few available settings in turn,
study, and hardly worth publication and hope one will prove sound. In the
even as a constructional exercise. event, I struck lucky with 2.1 overleaf,
It would be very much better if we but there are lines in the analysis which
could have some play after 1 Kc1, so are definitely the computer’s and not
I worked on. mine.
28 Fifty-one flights of chess fancy
3.1 3.3
wdwgwGBd wdwdwdwd
dw0w4w0w dwdwdwdw
wdpdNdPd wdwdNIwd
dw)kHR)w dBdkdwdw
wdw0wdwd wdwdp)wd
dwdpdwdw dwdw)wHw
wdw)wIwd wdwdwdwd
dwdw$wdw dwdwdwdw
White to play and mate in two White to play and mate in five
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdPdk
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdK
wdwdwdwd wdwdwdwd
dwdpdwdw dwdwdwdw
wdNdwdwd wdwdwdwd
dwdwdwHw dwdwdwdw
Losing (Giveaway) Chess White to play and win
White to play and win (a) ordinary chess, (b) Losing Chess
One of the most popular of all chess 4.2 is a two-part study in ordinary and
variants is Losing (Giveaway) Chess, in Losing Chess.
which capturing is compulsory and the
object is to get rid of all your men. The 4.3
king is an ordinary man and may be
captured, and a pawn may promote to rdwdwdwd
king. It is widely played among friends, dwdwdwdw
it is a common choice at end-of-season wdwdwdwd
“fun” meetings, and the endgame phase dwdNdwdw
offers particular subtlety and delight. wdwdwdwd
In The Encyclopedia of Chess Variants, dwdwdwdw
David Pritchard described this as wdwdwdNd
“the most appealing stage of the game, dwdwdwdw
a garden of surprises”, and I think all
who have explored it even superficially Losing (Giveaway) Chess
will endorse his verdict. White to play and win
34 Fifty-one flights of chess fancy
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdPdk
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdK
wdwdwdwd wdwdwdwd
dwdpdwdw dwdwdwdw
wdNdwdwd wdwdwdwd
dwdwdwHw dwdwdwdw
Losing (Giveaway) Chess White to play and win
White to play and win (a) ordinary chess, (b) Losing Chess
The answer to 4.1 is 1 Ne2, offering In ordinary chess, 1 f8R, as known for
Black a second capture. If now 1...dxc2 many years. In Losing Chess, the same
then 2 Nc3 c1B (any other promotion only different: 1 Kh6 Kxh6 2 f8R!
loses at once) 3 Na2 (see the left-hand
diagram below), after which the bishop 4.3
must move away and the knight gives
itself away on the square the bishop has rdwdwdwd
just left. This doesn’t work after dwdwdwdw
1...dxe2, because after 2 Ne3 e1B wdwdwdwd
3 Ng2 Black has 3...Bh4, but White can dwdNdwdw
play 2 Nd4 e1B 3 Ne6 and the knight wdwdwdwd
will have a giveaway wherever the dwdwdwdw
bishop moves (see the right-hand wdwdwdNd
diagram); and this doesn’t work after dwdwdwdw
1...dxc2, because after 2 Nd4 c1B
3 Nc6 Black can escape by 3...Bh6. Losing (Giveaway) Chess
White to play and win
wdwdwdwd
dwdwdwdw 4.1 and 4.2 were discovered by normal
exploration with board and men. 4.3
wdwdNdwd was thrown up by a definitive computer
dwdwdwdw analysis.
wdwdwdwd Black to play would have to allow
dwdwdwdw two immediate giveaways. White to
Ndwdwdwd play starts 1 Ngf4 (or Nge3 by
dwgwgwdw symmetry) Ra1 2 Ng6 Ra2 3 Ne5 :
Variants of chess 35
wdwdwdwd 4.4
dwdwdwdw wdwdwdwd
wdwdwdwd dPdwdwdw
dwdNHwdw wdwdwdwd
wdwdwdwd dwdwdwdw
dwdwdwdw wdwdwdwd
rdwdwdwd 0pdwdwdw
dwdwdwdw wdwdwdwd
dwdwdwdw
Black now has three moves. If 3...Rh2
then 4 Nb6 Rh1 5 Nbc4 Rh8 6 Nb2, Losing (Giveaway) Chess
and we have a left-to-right reflection of White to play and win
4.3 but with Black to play. If 3...Ra1
then 4 Nf6 Ra2 (4...Rb1 is equivalent) 4.5
5 Nfd7, and if 5...Rh2 then 6 Nb6
transposing into the previous line; Kdkdwdwd
if instead 5...Ra1 then 6 Ng4, with dwdwdwdw
another position where Black to play Pdwdwdwd
must allow two immediate giveaways : dndndwdw
wdwdwdwd
wdwdwdwd dwdwdwdw
dwdNdwdw wdwdwdwd
wdwdwdwd $wdwdwdw
dwdwdwdw Optional Replacement Chess
wdwdwdNd White to play and hold the draw
dwdwdwdw
wdwdwdwd 4.4 completes our brief Losing
4wdwdwdw Chess selection. 4.5 features Optional
Replacement Chess, where a player
At the top of the page, this leaves may put a captured man back on the
3...Ra8, met by 4 Ng4 Ra1 5 Ngf6 Ra2 board on any vacant square (a pawn not
6 Ng8 Ra1 7 Nge7 Ra2 (7...Rh1 8 Nb4 on the first or last rank, a bishop only
with a left-to-right reflection of the on a square of the same colour).
position above) 8 Ng6 Ra1 9 Ngf4 Ra8 In ordinary chess, Black would play to
10 Ng2 and we are back at 4.3 but with reach a standard win with a lone knight
Black to play. Moving only a knight, (1 Rc1+ Ndc7+ 2 Rxc7+ Nxc7+ 3 Ka7
White has managed to transfer the Nd5 etc). How can putting a captured
move to his opponent. knight back on the board help White?
36 Fifty-one flights of chess fancy
4.4 Kdkdwdwd
wdwdwdwd hw$wdwdw
dPdwdwdw Pdwdwdwd
wdwdwdwd dndwdwdw
dwdwdwdw wdwdwdwd
wdwdwdwd dwdwdwdw
0pdwdwdw wdwdwdwd
wdwdwdwd dwdwdwdw
dwdwdwdw
Black naturally continues 2...Nxc7+,
Losing (Giveaway) Chess taking the White rook off, and White
White to play and win captures the knight and puts it back a
second time, 3 Kxa7 (Nb6) :
1 b8Q! Not 1 b8R, when 1...a2 2 Rxb3
a1N 3 R~ Nb3 wins for Black; but after wdkdwdwd
1 b8Q and the same, White has 3 Qc2. Iwhwdwdw
This is the simplest Losing Chess Phwdwdwd
position I know where a promotion to
queen is needed in order to win.
dwdwdwdw
wdwdwdwd
4.5 dwdwdwdw
wdwdwdwd
Kdkdwdwd dwdwdwdw
dwdwdwdw
Pdwdwdwd If there were no knight on b6, Black
dndndwdw would win by 3...Nd5. As it is, White
wdwdwdwd threatens to capture this knight for the
dwdwdwdw third time, this time finally taking it off
wdwdwdwd the board, and if Black moves it or
$wdwdwdw defends it he gives stalemate.
White cannot put the knight on b6 at
Optional Replacement Chess move 2 because he would be putting his
White to play and hold the draw own king in check, so he must do it in
two steps. If he tries 2 Rxc7+ (Nb7),
1 Rc1+, and if 1...Nbc7+ then 2 Ka7 hoping for axb7+ next move, Black has
and Black must let the White king out. 2...Nxc7 (Ra7) mate. If he tries 3 Kxa7
1...Ndc7+, therefore, and we have the (Nb7) with the same idea, Black can
first capture and replacement, 2 Rxc7+ reply 3...Nc5 4 Kb6 N7xa6 (Pc4), and
(Na7) : play out the book win with 2N v P.
Variants of chess 37
wdwdwdwd wdwdwdwd
iPdwdwdw dwdwdwdw
wdwdwdwd K$bdwdwi
)wdwdwdw dwdwdwdw
Kdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdr
White to play and win Maximummer
(a) ordinary chess, (b) Cylindrical (a) as set, selfmate in six
Chess, (c) Circular Chess (see below) (b) White P on b6, selfmate in seven
4.8
wdwdwwwdwd wdwdkdwd
iPdwwwdwdw 1q1q1q1q
wdwdwwwdwd Rdwdwdwd
)wdwwwdwdw !wdwdwdw
Kdwdwwwdwd wdwdwdwd
dwdwwwdwdw dwdwdwdw
wdwdwwwdwd wdwdwdwd
dwdwwwdwdw dwdwIwdw
Grasshoppers a7-h7; mate in four
Imagine the two sides made
semicircular and joined so that a1-a8- A grasshopper moves along Q-lines,
h8-h1-a1 forms a continuous ring, but must jump one man and land on the
likewise b1-b8-g8-g1-b1 and so on. square beyond (so Rxa7 can be met by
Men start in their normal positions, and ...Gc7xa7). Black to move could play
promotion is on ranks 1 and 8 as usual. ...Ga7xa5, and if Ra8+ then ...Ga5-d8.
38 Fifty-one flights of chess fancy
come on the other wing: 1 Rh6. Either 4.9 (by Kurt Smulders, my version)
1...Gh5 or 1...Ga4 will allow immediate
mate, so Black must move his king. RdwGwhwd
If 1...Kf8 then 2 Qh5 : dk)wdwdw
wHwdw$pd
wdwdwiwd dwdwdw0w
1q1q1q1q w)wHwdKd
wdwdwdw$ dwdwdwdw
dwdwdwdQ wdwdwdwd
wdwdwdwd dwdwdwdw
dwdwdwdw Neutral pawn c7; mate in two
wdwdwdwd
dwdwIwdw A player may move a neutral man as if
it were his own, or capture it as if it
If now 2...Kg8 then 3 Rxh7, and if were his opponent’s. Neutral pawns
3...Gxh7 then 4 Qe8; if 2...Ke8 then promote to neutral pieces. White can
again 3 Rxh7, and the Gf7 is pinned. play c8R, but c8Q would check his own
If instead 1...Kd8 then 2 Qa2 : king. Smulders’s original is overleaf
and many will prefer it, but the version
wdwiwdwd has a property which I like.
1q1q1q1q
wdwdwdw$ 4.10 (after J. E. H. Creed)
dwdwdwdw r4wdwdwg
wdwdwdwd inGwdwdw
dwdwdwdw wdwdwdwd
Qdwdwdwd Iwdwdwdw
dwdwIwdw wdwdwdwd
dwdwdwdw
Again both king moves can be met by wdwdndad
3 Rxh7, and if 3...Gxh7 then 4 Qg8; dwdwdwdw
and if 2...Gh5 then 3 Rxh5.
If 1 Qh5 then 1...Kd8 (not 1...Ga5, Imitator g2; mate in two
when White plays 2 Ra8+ Gd8 3 Qa5
and Black must allow a mate on d8). An imitator copies each move, and a
If 1 Qa2 then 1...Kf8, and if further move can be made only if the imitator
2 Qh2 hoping for a reflection of the can copy it. 1 Kb6 (Ih3) is legal (Black
play after 1...Kd8 2 Rxh7 then 2...Kg8 cannot play KxK) and check (White
3 Rxa7 Gh1! can). The answer: 1...Bf6 (If1)!
40 Fifty-one flights of chess fancy
RdwGwhwd r4wdwdwg
dk)wdwdw inGwdwdw
wHwdw$pd wdwdwdwd
dwdwdw0w Iwdwdwdw
w)wHwdKd wdwdwdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdndad
dwdwdwdw dwdwdwdw
Neutral pawn c7; mate in two Imitator g2; mate in two
1 Rf7 gives a flight square and makes 1 Bb6 (If1) is not check because the
no threat, and Black has four moves. Black knight blocks the imitator, but it
1...Kxb6 2 c8N (not c8R, when Black restricts Black to moves by this knight
has 2...Rxd8/Rc7). 1...Nd7 2 c8B (not because anything else will bring the
c8Q, when Black can move the queen imitator out into the open. Again Black
away). 1...Ne6 2 c8Q (2 c8B Bd7). has four moves, and again White has
1...Nh7 2 c8R (2 c8N Ne7). Four Black four different mates in reply. Three of
moves, four different promotions. them are king moves: 1...Nc3 (Id2)
Smulders’s original setting 2 Kb4 (Ie1), 1...Nf4+ (Ig3) 2 Kb5
(Ih3), and 1...Ng3 (Ih2) 2 Ka4 (Ih1).
wdBdwdwd This leaves 1...Nd4+ (Ie3), after which
iPHwdwdR no king move works :
wHwdwdwd
4n)wdwdw r4wdwdwg
pdwdwdwd indwdwdw
)wdwdwdw wGwdwdwd
wdwdwdwI Iwdwdwdw
dwdwdwdw wdwhwdwd
dwdwfwdw
yields to 1 Na6, with similar play and wdwdwdwd
an extra line 1...Rxa6 2 Rxb7 where dwdwdwdw
the neutral pawn is captured. Many will
think the version unjustified, but “four The answer is 2 Bxd4 (Ig1), a mating
Black moves, four different mates” is retreat along the line of check quite
a property I like in a two-mover, and it unlike anything that can happen in
seems particularly appropriate here. ordinary chess.
Chapter 5
How did we get here ?
I have always been fascinated by “retro” problems, where the solver is given a
position and has to deduce something about the play which led up to it, and when
I was running the British Chess Magazine problem column I found that my annual
month devoted to such problems attracted as much interest as the rest of the year’s
offerings put together. Never mind that the “games” must have been wholly
unrealistic; it was the logical demonstrations that appealed.
5.1 5.2
rdwdw4kd rdb1kgn4
dp0pdp0p 0p0w0p0w
pdwdwhbd wdw0wdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dPdwdwdw dwdwdPdw
wdP)P)P) P)P)PdP)
dwGwdwIw $NGQIBHR
What was White’s first bishop move? Position after Black’s 8th move
What was the game?
Retros come in many flavours. Some,
like 5.1, pose explicit questions as to 5.3
the nature of the previous play. Some,
like 5.2, require a complete game to a rhb1kgn4
given position (and note that it must be 0p0p0p0p
a game to precisely the move stipulated wdwdwdwd
– it is easy to reach 5.2 after White’s dwdwdwdw
7th move, but this is irrelevant). Some wdwdwdwd
require deductions of other kinds (for dwdwdwdw
example, that a king or one of his rooks P)P)P)P)
has moved, so he cannot castle on that $NGK!BHR
side). And so on...
We met ORC in Chapter 4. In 5.3, Optional Replacement Chess
White’s K and Q have changed places. How soon can we reach this position?
42 Fifty-one flights of chess fancy
5.1 5.2
rdwdw4kd rdb1kgn4
dp0pdp0p 0p0w0p0w
pdwdwhbd wdw0wdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dPdwdwdw dwdwdPdw
wdP)P)P) P)P)PdP)
dwGwdwIw $NGQIBHR
What was White’s first bishop move? Position after Black’s 8th move
What was the game?
At first sight, 5.1 seems innocent
enough, if we ignore the curious fact No, it wasn’t the b8 knight that was
that a knight must have been allowed in captured, it was the g8 knight, and the
to capture the bishop originally on f1 b8 knight is now at g8: 1 f3 h5 2 Kf2
(well, funny things happen in retros). Nh6 3 Kg3 h4+ 4 Kxh4 Ng4+ 5 Kxg4
However, we then notice that Black’s d6+ 6 Kg3 Nd7 7 Kf2 Nf6 8 Ke1 Ng8.
c8 bishop, like White’s f1 bishop, must 5.2 remained unpublished for several
have been captured at home, and hence years, and in the meantime I came
that the bishop at g6 must be a across a statement that a gap of more
promoted pawn. Furthermore, it can than three moves (two by one side, one
only have promoted on b1, so it must by the other) between a uniquely
have captured its way as a pawn from realisable game and the shortest game
e7 across to a2, captured back to b1, to the same position was surely
and then escaped as a bishop, all before impossible. This caused me to produce
White’s b-pawn blocked the way out by a position where the gap was four
advancing to b3. White moves and three Black (from the
If White’s c1 bishop had been still at game array, move White Ng1 to f4, Pf2
home while all this was going on, only to f3, Ph2 off, Ke1 to h2, Black Ng8 to
the a-pawn, the a-rook, and the two e4, Pf7 to f6, giving a position easily
knights would have been available for reached after Black’s 7th move but
the advancing e-pawn to capture: not reachable after White’s 11th only by
enough. So White’s c1 bishop also playing 1 f3 Nf6 2 Kf2 Ne4+ 3 Ke3 f6
must have been captured at home, 4 h4 Kf7 5 Nh3 Kg6 6 h5+ Kxh5 7 Kf4
releasing his queen and his h-rook as Kh6 8 Kg4 Kg6 9 Nf4+ Kf7 10 Kh3
additional fodder for the Black e-pawn, Ke8 11 Kh2).
and White’s first bishop move must This prompted others to explore the
have been b8-a7. field, and my record was soon beaten :
How did we get here? 43
rhb1wgw4 whb1kgw4
0p0p0w0p 0p0w0p0p
wdwdw0wd wdwdwdwd
dwdndwdk dwdpdwdw
wdwdwdwH wdwdwdwd
dwdwdPdw dwdPdwdw
P)P)PdP) P)PdP)P)
$NGQIB$w $wGQIBHR
Position after White’s 11th move Progressive Chess
What was the game? Position after series 5 (a) as set,
(b) White Pd3 on d4
After Black’s 6th move, easy; after Where were the missing men captured?
White’s 11th, only 1-2 Nh4 Nd5 3 f3 f6
4-6 Kg4 Ke5 7 Kh3 Kf4 8 Rg1 Kg5 In Progressive Chess, like Losing Chess
9 Kg3 Kh6! 10 Kf2 Kh5 11 Ke1. This a game which is widely played, White
is a lovely thing, simple, subtle, and plays one move, Black two, White
striking. Far from bewailing the loss of three, and so on (so the moves to date
my short-lived record, I was delighted have been W, BB, WWW, BBBB, and
to see what I had apparently inspired. WWWWW).
5.3 5.5
rhb1kgn4 rdwdkgw4
0p0p0p0p 0p0p0p0p
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdw
P)P)P)P) wdwdwdwd
$NGK!BHR dwGwIwdw
Optional Replacement Chess Maximummer
How soon can we reach this position? White to play and win
5.4 5.5
whb1kgw4 rdwdkgw4
0p0w0p0p 0p0p0p0p
wdwdwdwd wdwdwdwd
dwdpdwdw dwdwdwdw
wdwdwdwd wdwdwdwd
dwdPdwdw dwdwdwdw
P)PdP)P) wdwdwdwd
$wGQIBHR dwGwIwdw
Progressive Chess Maximummer
Position after series 5 (a) as set, White to play and win
(b) White Pd3 on d4
Where were the missing men captured? The essence of the argument is as
follows: (a) Black’s last move must
(a) is straightforward, a possible line of have been with his a-rook, else he
play being 1 Nc3, 2 d5 Nf6, 3 Na4 Nb6 would have had a longer alternative;
Nxa8, 4 Nd7 Nb6 Nxa8 Nb6, 5 d3 Be3 (b) he can never have moved his king in
Bxb6 Be3 Bc1. Minor variations are reply to a check, since this check could
possible, but in each case Black’s Ra8 only have come from a knight, and a
must have been captured at home, his pawn capture would have provided a
Ng8 on b6, and White’s Nb1 on a8. longer alternative; (c) he can never
This fails in (b) because the pawn on have moved his king other than in
d4 blocks the bishop’s path to b6, and reply to a check, nor can he have
something quite different is needed: moved his h-rook, since a knight move
1 Nf3, 2 Nc6 Rb8, 3 Nd4 Nxc6 Nxb8, or a pawn-two would have provided a
4 d5 Nf6 Nd7 Nxb8, 5 d3 d4 Nd2 Nf3 longer alternative. So Black cannot
Ng1. Again minor variations are play 0-0-0, but he can play 0-0.
possible, but in each case Black’s Ra8 The winning procedure is now clear
must have been captured on b8, his (force ...0-0, force ...Kh8, mate him).
Nb8 on c6, and White’s Ng1 on b8. The simplest sequence appears to be
Black’s original Ng8 is now at b8, 1 Bh6 Rd8 2 Bxg7 Ra8 3 Bd4 Rd8
and White’s Nb1 at g1; every capture 4 Bxa7 Ra8 5 Bd4 Ra1+ 6 Bxa1 Bh6
has occurred on a different square. 7 Be5 Bc1 8 Bxc7 Bh6 9 Bf4 0-0
And why does this fail in (a)? 10 Bd6 Bc1 11-16 Kxb7 Bc1
Because White cannot lose a move 17-18 Kxd7 Bc1 19 Kxe7 Bh6 20 Kf6
during series 5; he can play Pd2 to d3 Bc1 21 Bxf8 Bh6 22 Bxh6 Kh8
and Nb1 to g1 in four moves, six, or 23 Ke7 f5 24 Kf8 f4 25 Bg7, and of
any larger number, but not in five. course there are many alternatives.
Chapter 6
Joke compositions
By its very nature, there is no precise definition of a joke composition. If there
were, it would not be a joke. Sometimes the joke is merely in the presentation,
sometimes, indicated by “(perhaps)”, the normal rules are to be bent a little ...
6.1 6.3
wdwdw4wI kdwdwdwd
dwiw)wdw dw!wdwdw
wdwHwGq0 wdwdwdwd
HwdQdBdP dwdwdwdw
wdwdwdwd wdwdwdwd
dwdwdwdw dwdwdwdp
wdwdwdwd wdwdwdw)
dRdwdwdw dwdwdwdK
White to play and win (perhaps) Fuddled men; mate in two
6.5 6.6
wdwdwdwd rdwdwdw4
dwdwdw)Q dp0wdp0p
wdwdw$wd pdndwhqd
dwdwdw$w gkdbdwdR
wdwdwdwd wdPdwdwd
dBGwdwdw )K0wdwdw
wdwIwdwd wdw)wGwd
dwdwdwdw dwdwdwdw
Kriegspiel Cylinder Chess
Black is down to a bare king Should White resign?
White hears “Black has moved”; can
he now mate in one? 6.7 (version by Noam Elkies)
rdwdwdwI rdwdrdkd
dwdwdw)P dpdqdpZp
wdwdw!wd pdndw!pd
dwdwgwdw dwdpZwGw
wdwdwdwd wdw)wdwd
dwdwdwdw dPZwdwdw
wdwdwdwd PZwdw)P)
dwdwdwdk $wdwdRIw
White to play and save the game White to play and win
(perhaps)
6.10b, left
In 6.9, you are playing with an ancient 6.10c, right
and venerable chess set in an ancient
and venerable club and you appear to wdwdwdkd
be hopelessly lost, but there is dwdwdw)w
(perhaps) a way out of the morass... wdwdwZwd
dwdwZKdw
As everyone knows, the Loch Ness wdwZwdwd
Monster has two properties: (a) it dkZwdwdw
undoubtedly exists; (b) nobody has w1wdwdwd
actually seen it. In Loch Ness Chess, dKdwdwdw
the long black diagonal from b2 to g7
represents the dark and mysterious 6.10b: Is White mated?
loch, and as long as a man on the loch 6.10c: White to play and win
is observed by another man (of either
colour), it is perfectly safe. However, 6.10d
as soon as it is no longer observed,
the monster appears and swallows it. wdwdwdwd
A pawn “observes” the squares it can dwdwdwZw
capture to. From the game array, play wdk)wZwd
1 d4 (safe, White’s queen observes d4) !wdwZw)w
e5 (also safe, White’s Pd4 observes e5), wdwZwdwd
and if 2 dxe5 the monster swallows the dwIwdwdw
now unobserved pawn on e5. wZwdwdwd
In 6.10a to 6.10d, the task is to solve dwdwdwdB
the given problem both in ordinary
chess and in Loch Ness Chess. What was the last move?
50 Fifty-one flights of chess fancy
rdwdwdw4 wiwdwdwd
dp0wdp0p dw0Pdpdw
pdndwhqd KdBdw)wd
gkdbdwdR dwdwdwdw
wdPdwdwd wdwdwdwd
)K0wdwdw dwdwdwdw
wdw)wGwd wdwdwdwd
dwdwdwdw dwdwdwdw
Cylinder Chess; should White resign? Stalemate in one (perhaps)
(a) as set, (b) add Black rook on c8
First thoughts: no, of course he
shouldn’t, he has just given mate. 6.8
Second thoughts: where did the
mating pawn come from? Not from d3, wdwdwdwd
which would have involved two dwdwdwdw
captures and Black has only lost one w)wdwdwd
man. So it must have come from c2, dwdwdwdw
and Black must be in the middle of wdwdwdpd
taking it en passant. The completion of dwdwdwdp
this move will leave White mated (the pdPdwdw$
Black queen is pinning the rook on h5), dwdwiwdK
so the answer is “yes”.
Correct answer: No. It is not polite to 1 b7 a1R intending 2...0-0-0 mate;
resign before your opponent has now what?
finished moving.
Echo Black’s promotion by 2 b8R, and
6.7 (top of next column). As set, now you can meet 2...0-0-0 by playing
promote to king; with a Black rook 3 RxR en passant.
added on c8, promote to a Black king. A few years ago, there was a vogue
The original version of this was for problems incorporating each of
submitted as a joke entry to a tourney the “odd” moves of modern chess
for stalemate problems which were (underpromotion, castling, en passant
unorthodox in some way, and the capture). Well, what those who take
judges awarded it a “Dishonourable themselves seriously can treat as a
Mention” of which I was immensely serious exercise, those of us who are
proud. Noam Elkies subsequently less serious can use as the medium for a
improved the construction. joke ...
Joke compositions 51
6.9 (version by Noam Elkies) move on g7. This doesn’t work in Loch
Ness chess because the bishop’s move
rdwdwdwI leaves the queen unobserved, but now
dwdwdw)P there is no need for White to bring up a
wdwdw!wd second man to guard g7; he can play
dwdwgwdw 1 Qg7 mate at once. The logic is the
wdwdwdwd same as that which we saw after 1 d4
dwdwdwdw e5 2 dxe5 (e5 off) from the game array;
wdwdwdwd the queen is safe from the monster
dwdwdwdk since she is observed by the Black king,
but if Black takes the queen he will
White to play and save the game be unobserved and the monster will
(perhaps) swallow him.
dwdwd wdwd*dwd
wdwdw dwdwdwdw
dwdwd wdwdwdwd
wdwdw dwdwdwdw
HNHNH wdwdwdwd
dwdwdwdw
7.1 On a 5x5 board, White has five wdwdwdwd
policemen who start in a straight line at dwdwdwHw
the bottom, and Black has a burglar
who may start anywhere. All men move 7.3 (after George Jelliss) On an empty
as knights, but there is no capturing. board, put a White knight on its home
Can White play to trap the burglar, or square g1. White aims to reach Black’s
can the burglar run for ever? palace, e8, but after each move Black
can place a mine on any empty square,
dwdwdwd and this square will henceforth be
barred to the knight. Can the knight
NdwdwdN reach the palace, or will it eventually
dwdwdwd have to step on a mine?
wdwdwdw
dwdwdwd 8.1 Given a currency with coins of
wdwdwdw value 1, 2, 5, 10, 20, 50, and 100,
dwdNdwd obtain four sums of money, each
consisting of exactly two coins and
7.2 This time there are only three each being twice as large as its
policemen on a 7x7 board, but they can predecessor.
move simultaneously. The policemen
start from the position given above, and 8.2 Given the same currency, obtain a
the burglar is on a dark square away sum of money which can be realised by
from a corner. The burglar moves first; a single coin, by two coins, by three, by
can the policemen catch him? four, or by five.
54 Fifty-one flights of chess fancy (and a few other frolics)
2 3 4 5
13
6 7 8
1
9 10 11 12
2 3 4
13
5 6 7 8 9
7.1 (Burglar and five policemen) gives a position we shall see again :
7.2 (Burglar and three policemen) holds out longest by returning to d4,
and White plays to a4/g4 plus a square
This time there are only three which threatens d4, say c2 :
policemen on a 7x7 board, but they can
move simultaneously. The policemen dwdwdwd
start from the position given below, wdwdwdw
dwdwdwd
dwdwdwd NdwhwdN
NdwdwdN dwdwdwd
dwdwdwd wdNdwdw
wdwdwdw dwdwdwd
dwdwdwd
wdwdwdw There are now two cases. Suppose
dwdNdwd first that Black keeps away from the
side policemen, say by going to c6. The
and the burglar is on a dark square policemen play to b6/e5/d4 :
away from a corner. The burglar moves
first; can the policemen catch him? dwdwdwd
If he starts in the centre, he can be wHndwdw
caught, but from any other non-corner
square he can run for ever. He moves to
dwdwHwd
one of the squares shown below :
wdwHwdw
dwdwdwd
wdwdwdw
dwdndwd dwdwdwd
whwdwhw
dwhnhwd There follows b4 (holds out longest),
ndndndn d5/d3/c2 :
dwhnhwd
whwdwhw
dwdndwd dwdwdwd
wdwdwdw
dwdNdwd
Now he always has four options, and
the police can only block three of them.
whwdwdw
However, if he starts in the centre, he
dwdNdwd
cannot reach this group. His first move wdNdwdw
can threaten only one square in the dwdwdwd
group, c5, e5, c3, or e3, and White goes
to c5, e5, and c3 or e3 as needed. Black c6 (holds out longest), e5/b4/d4 :
Three knight puzzles 57
8.1 (once, twice, four times ...) 8.3 (11 coins, < 2 duds, 5 weighings)
To get four sums of money from coins Given eleven coins of which at most
of value 1, 2, 5, 10, 20, 50, and 100, two are duds, and given that (a) a dud is
each consisting of two coins and each either heavy or light, (b) two heavy
being twice as large as its predecessor, duds or two light duds balance each
set other, and (c) a heavy dud and a light
dud balance two true coins, the duds if
15 = 5 + 10 any can be identified in five weighings
30 = 10 + 20 by balancing
60 = 10 + 50
120 = 20 + 100 1, 3, 4 against 2, 5, 6
1, 4, 2 against 3, 7, 8
8.2 (one potato, two potatoes ...) 1, 2, 3 against 4, 9, 10
5, 7, 9 against 6, 8, 10
To obtain a sum of money which can be
realised by one, two, three, four, or five and setting up a fifth weighing in the
of the above coins, consider light of the results. It may be verified
that each of the 81 possible sets of
20 results from these four weighings can
10 + 10 arise in precisely three ways.
10 + 5 + 5 This was a fortunate arithmetical
5+5+5+5 accident. If we have n coins, there are
10 + 5 + 2 + 2 + 1 2n2 + 1 possibilities, and w weighings
give 3w different sets of results.
This was the only solution with the 2.112 + 1 = 35, and a weighing of three
English coinage as it was some years coins out of eleven against three gives
ago, but now the existence of a £2 coin 81 = 34 possibilities whether the result
provides an alternative. is left down, right down, or balance.
Chapter 9
Three thirteen-hole pegboard puzzles
I have done little work on “full-size” peg solitaire boards since completing The Ins
and Outs of Peg Solitaire in 1985, and this little can be found in issue 28 (2003) of
George Jelliss’s The Games and Puzzles Journal and in a paper “New problems on
old solitaire boards” which George Bell and I wrote for the colloquium on Board
Games Studies at Oxford in 2005. The latter is due to be published in Volume 8 of
Journal of Board Games Studies, and in the meantime George has posted a copy
on the web.
However, during 1987 and 1988 I examined some smaller boards, and found
some very attractive problems on them. A small selection appears below. Most of
these problems have not been published before, at least not by me, though it is
quite possible that somebody else rediscovered and published them in the interim
(or even discovered them before I did).
It should perhaps be stressed that a solitaire problem, unlike a chess problem,
rarely has a unique solution. Even if the jumps required are uniquely determined,
the player normally has a measure of latitude in the order in which he makes them.
The solutions that follow are therefore specimens only.
1 1
2 3 4 2 3 4
5 6
7 5 6 7 8 9
8 9
10 11 12 10 11 12
13 13
Vacate 7, play to finish at 7: 1-7, 11-3, Vacate 1, mark 13, and play to leave
10-1-7, 12-1, 7-2-4-7-10-12-7. With its the marked peg at 13: 5-1, 12-2, 1-7,
final six-sweep and two in the previous 11-3, 9-1-7, 2-12, 13-5-7-9-13.
play, the final survivor has jumped over Vacate 7, play to interchange 5 and
eight of the eleven pegs to be removed. 9: 1-7, 5-1, 12-2, 4-12, 13-7, 10-4,
Play to interchange 1 and 13: 2-7, 2-10, 9-13-5, 1-9.
13-2, 9-5, 1-7-13, 8-3, 12-1, 2-4-7-1. Vacate 2, play to interchange 1 and
Play to interchange 3 and 11: 2-7, 5: 12-2, 9-7, 6-8, 13-7, 3-11, 1-9-7, 5-1,
4-2, 13-4, 11-5, 3-11, 10-1-12-7, 5-9-3. 7-13-5.
Chapter 10
Three bridge frolics
10.1 3NT from any side, low heart. Declarer discards a spade
against any defence from dummy and ducks a spade, setting
up seven spade tricks to go with the ace
Consider a suit distributed as follows: of diamonds and the heart already in
Q 10 9 the bag, and the defence can take at
most three more tricks before letting
– KJ
him in to make them.
A8765432 Similar play occurs if the defenders
This yields seven tricks only if there is take either or both of their aces before
a side entry to the long hand, because if playing a low heart or club, though
the ace is played on the first or second declarer must be careful to answer the
round then the third round will be won lead of a low heart by ducking a spade
in the short hand, and if the ace is held and a low club by ducking a diamond;
up until the third round the defence will and if the defenders broach spades or
take the first two tricks. However, if a diamonds before taking their four
preliminary discard can be made from tricks, declarer will make an overtrick.
the short hand, a first-round duck will This appeared in The Games and
set up the rest of the suit. Puzzles Journal in 1988. Fourteen years
Now consider a deal consisting of later, I opened my morning copy of
four such suits : The Guardian, and there in the bridge
S Q 10 9 column was this very deal with the
H – name “Thomas Andrews” attached to it.
D A8765432 Apparently he had performed a
C KJ systematic analysis by computer of all
rotationally symmetric deals, had
S – S KJ
spotted this one as being of particular
H A8765432 H Q 10 9
interest, and had published it on his
D KJ D –
web site as the new discovery he
C Q 10 9 C A8765432
genuinely believed it to be. I might add
S A8765432 that both he and The Guardian behaved
H KJ impeccably, and unhesitatingly and
D Q 10 9 indeed handsomely acknowledged the
C – prior discovery as soon as they became
Suppose the defence starts by leading a aware of it.
62 Fifty-one flights of chess fancy (and a few other frolics)
10.3 Did the Rabbit play too well? The Rabbit passes the first diamond,
and takes the second. He has a vague
One of the Rueful Rabbit’s rare failures feeling that there is a long diamond suit
occurred in the celebrated match out somewhere, and his idea, if neither
between the Leprechauns and the the hearts nor the clubs break, is to
Gremlins (Bridge in the Menagerie, throw someone in on the third round
1979 Faber paperback edition, pages and obtain a ruff and discard. He then
67-84). tests the hearts and the clubs, ending in
S 986 hand. In fact this makes no difference,
H 9876 because if the second suit breaks he can
D 83 reach the long card by crossing in
C 9874 spades, but it is a matter of elementary
Rabbit technique to block the run of his
S Q J 10 7 S A432
suits if it is possible to do so. At trick
H A43 H KQ52
eight, he leads the ace of clubs from
D A94 D Q
hand, and this is what the Hog can see :
C AQ5 C K632
S A43
S K5
H 5
H J 10
C A led D –
D K J 10 7 6 5 2
C 63
C J 10
The Gremlin, North, opened with a S K5
psychic One Spade, the Hog, South, H –
responded Two Diamonds, and the D 10 7 6 5
Rabbit, West, ended up in 3NT. “The C –
Gremlin led a diamond and the Rabbit, The Hog has already had to discard one
relying on the spade finesse for his diamond on the third heart, and he can
ninth trick, went three down.” Three see that if he discards another he will be
down seems excessive – even the thrown in on the third round and forced
Rabbit would surely have tested both to lead away from his K5 in spades.
the hearts and the clubs before risking He therefore bares his king of spades, a
the finesse – but in fact the Rabbit routine brilliancy seen a thousand times
would seem to have incurred a well- in the annals of the Griffins, and waits
deserved penalty for having tried to for the finesse.
play properly. Had he performed with But the Rabbit, flustered by the
his usual incompetence, he would have failure of the clubs to break, leads the
made two overtricks. ace of spades from the wrong hand, and
Consider the following line of play. that is that.
Sources and technical notes
1.1 diagrammes 125, April-June 1998. follow, but White’s h-pawn will go, and
diagrammes, founded in 1973, is a French can he in fact win the resulting ending?
composition magazine which suddenly took After 2 Kxd5 Kf6 3 Nc4 Ne7+ 4 Ke4
wing in the late 1980s and for half a dozen (I don’t think anything else is better) Ng6
years was quite the best of the composition 5 Nd6/Na5 Nxh4 6 Nxb7 Nf5 my computer
magazines which were written in languages initially rated the position as +1.5 to White,
that I could read. I was its endgame study but the deeper it analysed the less convinced
columnist for ten years from 1993, but it became, and by the time it got to about
I don’t say it was a good magazine because 25-ply it seemed to have conceded that
I used to write for it; I was quietly proud to White could not force a win; Black’s
write for it because I thought it a good h-pawn was too great a distraction. In any
magazine. case, Black’s defences outside the main line
must be refuted relatively simply if a study
1.2 EG 34, November 1973. is to be aesthetically satisfying, and a
sideline needing analysis as deep as this
1.3 diagrammes 149, April-June 2004. would put the setting out of court even if a
The bishop-and-pawn study had appeared in win in it were eventually to be found.
the British Chess Magazine in 1947, and I
had quoted it in a special number of British 1.4 diagrammes 126, July-September
Endgame Study News devoted to British 1998. No significant backward extension
compositions from the period 1937-49. appears possible without adding extra
Unfortunately I didn’t notice the unwanted material. Black’s last move could obviously
line Bc4 until after the magazine had gone have been h5-h6, but g2 is a bad square for
to press, and I was not greatly surprised White’s knight; wherever it might have
when readers (Noam Elkies was the first) come from, unless its move was a capture it
started writing in to draw attention to the would have had a better alternative. 3 Nd2+
oversight. Noam also noticed, as I did, that is given by the computer as an alternative
1.3a was reciprocal zugzwang, but the winning move, but it is a blind alley; Black
extension back to 1.3 was my own. plays 3...Kf4 threatening to bring his knight
It may be asked whether an exchange on into play by 4...Ng3, White must play 4 Nf1
c4 cannot be incorporated into the play, to prevent this, and 4...Ke4 repeats.
thus avoiding starting in a position where My first attempt to produce a study by
Black’s c-pawn is already offside. All I can examining the lists of reciprocal zugzwangs
say is that I have been unable to do so other in EG ended in disaster. I searched the
than by crudely putting a Black piece on c4 zugzwangs with B + N v N for positions
and letting White capture it straight away. with a similar relationship between the
For a time, I thought it might be possible to knights and the Black king, again with a
move the White king back to f4 and the view to having a knight promotion by Black
Black pawn to d5, and to add knights on b2 in the preceding play, and came up with
and a5, with intention 1 Ke5 (1 Nd3 Nc6) White Ka2, Bb4, Nd6 (3), Black Ka4, Pd3
Nc4+ 2 Nxc4 dxc4 etc, but 1...Nc6+ proved (2), win not by 1 Bc5? d2 2 Ne4 d1N but by
unexpectedly troublesome. It concedes the 1 Ba3! d2 Ne4 d1N 3 Bc5. I published this
d-pawn at once and the b-pawn will soon in diagrammes, and the next post brought a
Sources and technical notes 65
puzzled letter from Harold van der Heijden: the computer’s discovery as the main line,
“Dear John, what is going on, this had been there was really no other way of describing
published by Simkhovich in 1940!” And the result. None of these studies appears
indeed it had been; he, long before the here, nor do one or two other joint studies
advent of computers, had discovered the where “my” contribution was discovered for
same position of reciprocal zugzwang, and me by the computer, but in the present case
had worked out exactly the same way of I found the key move for myself and felt
exploiting it (64, 1940). As a non-computer that my contribution had been a significant
production, this was a classic, the play after one.
the reciprocal zugzwang being by no means
trivial, and it remains among my favourites 1.7 British Chess Magazine, November
even though it embarrassed me by totally 2008. Baxter’s original took 2nd Prize in the
anticipating a discovery which I had thought 1961 New Statesman Tourney.
was my own.
1.8 British Endgame Study News,
1.5 The Problemist, January 2005. September 2004.
The Problemist is the magazine of the
British Chess Problem Society, and since 1.9 diagrammes 139, October-December
1968 it has included a column for original 2001. It gave my solvers a lot of trouble.
endgame studies (edited by Adam Sobey One, having played to 1.9a, sent me one of
until 2000). In truth, problem magazines White’s plausible tries, and I said No, Black
provide an environment very far from ideal plays such-and-such; then he sent me the
for endgame studies, but The Problemist second try, and again I said No; then he sent
gave me a platform for my early efforts, and the third try, and I said a third No. At that
I still try to send it something respectable point, I put him out of his misery.
from time to time.
1.10 diagrammes 130, July-September
1.6 Moravskoslezský šach, 1996 (as by 1999.
“WDE and JDB”). Moravskoslezský šach
was a short-lived magazine which circulated 1.11 Original setting in The Problemist,
in the Eastern half of the Czech Republic. March 1981; version by Christopher Jones,
Several further studies appeared in my British Endgame Study News, September
diagrammes column as by “So-and-so and 2000. BESN is a quarterly magazine which
JDB”. This often meant only that I had does not normally take originals unless they
tested a composer’s contribution by are incidental to an article or have been
computer and that it had disclosed a flaw derived from a study already published in it,
for which there was an obvious fix, and but it regularly reprints British originals
wherever possible I tried to persuade the which have appeared elsewhere, and it also
composer that I had done nothing and that carries articles describing how a certain
he would have found and fixed the fault study came into being. 1.11 featured in one
for himself had he also had access to a of these, and Christopher wrote in to
computer for testing (a significant amount suggest the alternative setting.
of what appears in print has been silently The just-in-time Black wins in the lines
influenced by editorial input of this kind). 3 Nb4 Bxd7 and 4 Nf6+ Ke6 were rare
But sometimes the composer insisted, and examples of a composer having good luck.
when the computer-discovered bust was a The composer has no control over sidelines
win or draw for Black and the “fix” like this; either they work or they don’t, and
consisted of reversing the colours and using all too often they don’t.
66 Fifty-one flights of chess fancy
1.12 The Problemist, July 1990. Black’s the difficulties that can arise. Much as I love
answer to the various win-a-piece lines is to endgame studies, I consider them unsuitable
try to win or exchange off White’s last for really serious competitive solving,
pawn. If 1 Ne5+ Kg5 2 Nxc6, taking off where money or prizes are at stake and
Black’s bishop, Black has 2...Ne3, with people will argue over the slightest detail.
3 Be2 Nf5 4 g4 Ng3 5 Bd1/Bf3 h5 or 3 Bf3 For such purposes, the precision of
Nf1 4 g4 Nh2. Some other lines allow “mate within n moves” is essential, and
White to avoid the exchange of pawns by I am fully in agreement with those who say
answering ...h4 by g4, but with Black’s that the WCSC should be openly and
bishop bearing down on the promotion unashamedly a problem-solving event and
square his h-pawn then becomes the endgame study round should be
unpleasantly strong. No line is particularly dropped.
difficult, but there are quite a lot of them,
and I was quietly relieved when the 1.15 The Problemist, July 2007.
computer endorsed my original verdict. The White pawn on c5 appears to be
necessary in order to give White a tempo
1.13 World Chess Solving Championship, move, but can we not manage without the
1994. Gurvich (1st Prize, Tourney of the Black pawn on c6? On a 9x9 board, yes,
Dagestan Committee of Physical Culture because the Black knight can start at e9, and
and Sport, 1952) had White Kb8, Rh5, Bg8 a version of the study in this form appeared
(3), Black Kf8, Be3, Ng1, Pg6 (4), draw by in the German composition magazine
1 Rh8 Kg7 2 Bh7 g5 3 Bf5 Kxh8 4 Bg4 etc. feenschach in 1973. On an 8x8 board, an
I dealt with the problem of side variations unobstructed dark-square knight which can
by announcing in advance that each of the reach b3 in three moves can also reach e2,
studies to be used in the championship had a and this is fatal for White.
clear main line, and that everybody giving it I discovered the present setting (or
in full would receive full points. something very similar) in the 1980s or late
1970s, but I omitted to write it down,
1.14 World Chess Solving Championship, and when I came to send it off to an editor
1996. Tronov (Shakhmatny vestnik 1913, I was unable to reconstruct it. I rediscovered
corrected in the British Chess Magazine in it two or three years ago. We may note that
1995) had White Kf2, Pg5/b4/e4/b3/d3 (6), the interference between the Black knight
Black Ke5, Bd1, Pa6/d6/e6/g6/h4 (7), draw and pawn is mutual; the pawn blocks the
by 1 b5 axb5 2 Ke3 d6 3 d4+ Kd6 4 e5+ and knight at move 3, the knight returns the
5 b4. compliment at the end.
As regards the occasional appearance of
such finishes over the board, one leading 1.16 The Problemist, January 1971. Why
solver, who was an international master as do I say that the Meyer study outclasses it?
a player and therefore entitled to be taken Because the starting position is more
seriously, said that in his opinion the draw natural, the White pawn starts right back on
after 3...Ba2 was obvious, and that a Black the second rank, and the White king moves
player trying to win in real life would play into position during the play. This last can
3...Bxd3. A possible continuation is now be achieved in 1.16a only by introducing
4 Kxd3 e4+ 5 Kc3 Ke5 6 K~ Kf6 7 Kc3 some extraneous motivation, such as
Kg5 8 Kd4 Kxg4 9 Kxe4 Kh4 10 Kd5 g5 making the king’s move capture something.
11 e4 etc, all White’s moves apart from the With the king on g7 instead of g8 in 1.16a,
sixth having been forced. I don’t think he the move 2 Bg2 would indeed fail, but the
found many supporters, but it does illustrate unwanted move 2 f5 would work instead.
Sources and technical notes 67
1.17 The Problemist, January 1972. 1.19 diagrammes, special number 22,
Mitrofanov (1st Prize, Rustaveli April-June 1998, as “after T. R. Dawson”.
Memorial Tourney, 1967) had White Ka5, Dawson had produced an orthodox problem
Re4, Pa6/g6/b5/d5/h5 (7), Black Ka7, Bd6, featuring similar out-and-back king play
Ne5/f3, Ph2 (5), with intention 1 b6+ Ka8 (Fairy Chess Review, February 1950, White
2 Re1 Nxe1 3 g7 h1Q 4 g8Q+ Bb8 5 a7 Kh1, Pb7/c6/c4/e3/e2/g2 (7), Black Ka7,
Nc6+ 6 dxc6 Qxh5+ 7 Qg5!! Qxg5+ 8 Ka6 Pc7/e7/g7/b6/g6/c5/e4/g3 (9), mate in “not
Bxa7 9 c7 and a remarkable winning less than 60” by 1-9 Ka6 10-16 Ka6 etc),
position with 2P v Q + B + N. 2...Nc4+ but in truth a closer predecessor was Kling,
gives Black a difficult draw, but moving the quoted in Alexandre’s 1846 collection
other knight from f3 to g2 is generally (original source not known to me), White
accepted as fixing it Kc6, Bh1, Nd6/b5, Pa6/f5/d4/g4/e3/h3 (10),
My original intention was to have the Black Kb8, Pa7/b6/f6/g5/h4 (6), mate in 10
rook arriving at g8 by underpromotion, the moving only the king (1-3 Kf3 4-6 Kf3
pawn on h7 being absent and promotion to 7-9 Kc6 10 Kc7). The “Gare de Lyon” joke,
queen being met by ...Qg3+ and stalemate. one of the classic bilingual puns, is due to
However, this created difficulties at the end, 1066 And All That.
when Black could play ...Qh7 pinning the
pawn. I did eventually get it to work, but 1.20 British Chess Magazine, December
only at the cost of some extra material, and 1994.
the result was not elegant. John Roycroft
mentioned my efforts to David Gurgenidze 2.1 Correspondence Chess, Summer
at a meeting in Tbilisi in 1975, and 2002. The Barden column was in “today’s”
Gurgenidze set up a brilliant double-file Financial Times (communication from Hew
version with a b-pawn in which White dated 16 February 2002). Apart from the
promoted to a rook on f8 and then moved duals in the main line, which occur in all
his king up the a-file if Black checked on long-range studies using this mechanism,
the g-file and up the b-file if Black checked Black can invert the moves ...f6 and ...c2 in
on the h-file (EG, 1976). Unfortunately the reply to 1 h4. However, if he plays ...f6
there also had to be a White pawn on c6, earlier, say 3 h6 b4 4 h7 f6, White can
and this created a dual Ka4/Kb4 if Black play 5 cxb4 c3+ 6 Kd3 c2 7 h8Q c1Q
played ...Qh3+. The idea would work 8 Qxf6+ with a win (Marc Bourzutschky’s
perfectly on a 10x10 board, though at least provisional database for Q + 2P v Q + P
one extra pawn appears to be needed in the confirms, saying that White can promote or
top right-hand corner to give the White king convert to a winning 6-man position within
a haven if Black simply lets White’s b-pawn 16 moves). Marc’s provisional database
promote and plays for perpetual check. assumes promotion only to queen, but this is
The version used in the 1994 WCSC had not a position in which Black is likely to
White Kg4, Rg8, Bh8, Pg7/a6 (5), Black have a drawing underpromotion resource.
Ka1, Qa3, Pg6/g5 (4), win by 1 Rb8 Qa4+
2 Kxg5 Qa5+ 3 Kxg6 Qxa6+ 4 Kg5 Qa5+ 3.1 British Chess Magazine, January
5 Kg4 Qa4+ 6 Kg3 Qa3+ 7 Kg2 Qa2+ 1976. 3.1a was used in the Belgian national
8 Kg1 Qa7+ 9 Kh1. It is of course greatly solving championship in either 1991 or
inferior to the original, but the final twist 1992, and was subsequently published in
8...Qa7+ 9 Kh1 caught one or two. The Hindu, 20 June 1992. I have moved
3.1 one file to the right, and reflected 3.1a
1.18 British Endgame Study News, left to right, to bring the positions into
September 2003. alignment.
68 Fifty-one flights of chess fancy
3.2 Used in the French national solving generations as little more than convenient
championship, 1993, and subsequently examples of what not to do; and because
published in the British Chess Magazine, their administration and judging soaks up
June 1995. vast amounts of effort, the WCCT
particularly so, and if devoted instead to
3.3 The Independent, 4 May 1987. The composition this would produce far more of
only point of possible originality lay in the value than is generated by the tournament.)
hesitant retreat of the White king from f6 to
f7 to g7. The piloting of the Black king by 3.5 World Chess Solving Championship,
making him capture knights had been done 1994. A long problem with two variations
many times before. was needed to complete the selfmate round,
Another example in which an apparently and having the manuscript of JiKí Jelínek’s
simple single-line problem created havoc in book on echoes in Bohemian selfmates for
a solving competition was given by a translation, I was able to look for a long-
problem by David Fawcett which appeared range echo which hadn’t been done before.
in the Gentleman’s Journal around 1870. “Bohemian”, in this context, refers to
(This particular Gentleman’s Journal was a multiple-variation problems in which the
weekly paper for boys which ran from 1869 character of the mates is pre-eminent (each
to 1872, and Fawcett appears to have been square in the mated king’s field is either
one of its schoolboy readers.) This had blocked by a friendly man or guarded by a
White Ke4, Bc2, Nd5/d4, Pf3/a2/e2 (7), single enemy man, no square being doubly
Black Kc5, Pf4 (2), mate in six by 1 Ke5 guarded nor guarded and blocked, and every
Kc4 2 Bf5 Kc5 3 Bh3 Kc4 4 Kd6 Kxd4 enemy man left on the board, apart possibly
5 Bf1 Kc4 6 e3. I set it in five-move form from the king and one or more pawns, takes
(White Ke5, Black Kc4) in the solving part in the mate). Selfmates of this kind in
competition at a meeting of the British which the king is mated on squares as far
Chess Problem Society in 1997, and even apart as h5 and h2 are relatively rare.
in this simplified form it defeated nine
competitors out of thirteen. 4.1 The Problemist, January 1997. If
instead 1 Nc~ (say Nb4) d2 2 N~ (either)
3.4 5th World Chess Composition then 2...d1B, and while White can now give
Tournament, 1992-95. I am a disbeliever in one knight away Black will soon give his
composition tourneys in general and an bishop to the other; if 1 Ne1 d2 2 Ne2/Nh3
emphatic disbeliever in the WCCT, but our then 2...d1N.
captain that year had done me several Evseev (2nd Prize, phénix 1992) had
favours at times not always convenient to White Nb8 (1), Black Pd5/c3/e3 (3), win by
himself and I felt I owed him something. 1 Nc6 and either (a) 1...d4 2 Nxd4 etc, or
In the event, 3.4 gained us a couple of (b) 1...c2 2 Nd4 and either 2...e2 3 Nxc2 or
points, but it was really far too light for this 2...c1B 3 Ne2 with several more N v B
tournament, and Michael and I both felt that wins, or (c) 1...e2 2 Ne5 d4 3 Nc4. This was
it would have given more pleasure in an both earlier than mine and more profound,
ordinary solving column. but I have been told that my simpler setting
(Why am I a disbeliever in composition also has its appeal.
tourneys? Because I believe that we should
compose to entertain, not to gain points or 4.2 International problemists’ meeting,
prizes; because the leading places seem too Bournemouth 1989. John Roycroft set a
often to go to complicated heavyweights, light-hearted tourney for “twin” studies,
which will be regarded by subsequent with the rider that more account would be
Sources and technical notes 69
taken of the nature of the twinning 4.5 British Chess Magazine, December
mechanism than of any subtlety in the play. 1992 (as part of my annual off-beat solving
I submitted 4.2 as a joke entry, and he gave competition). The refutation of 2 Rxc7+
it the prize for sheer cheek. (Nb7) by 2...Nxc7 (Ra7) mate was due to
Graeme Oswald, one of my solvers; I had
4.3 Variant Chess 27, Spring 1998. VC given 2...Nxc7+ (R off) followed by a
was founded by George Jelliss in 1990 as 2N v P win (which is sound enough but
one of two successors to his magazine takes longer). According to the computer,
Chessics (1976-87). best play after 3 Kxa7 (Nb7) Nc5 4 Kb6
The classic “lose a move with a knight” N7xa6 (Pc4) results in mate at move 46.
study in ordinary chess is of course that by This, like 1.6 and 1.12, was a chance
H. A. Adamson, Chess Amateur, January discovery. I was looking for a variation on
1924 (not “1923” as sometimes quoted), the classic mate with a lone knight, and
White Kb6, Nc2, Pc7/h6/e5 (5), Black Kc8, suddenly noticed the bust.
Bd5, Pb7/e7/h7/e6 (6), win by 1 Na3 Bc6
2-3 Nc3 Bd3 and now 4-5 Nc5 Bd5 6-7 Nf4 4.6 As a triplet, probably not before The
Bf5 8 Ne2 Bd3 (to meet 9 Nd4 by 9...Bc4) Classified Encyclopedia of Chess Variants
9 Nc3! returning to the position after (2007), page 221, though the failure of 1 a6
3...Bd3 but with Black to play. Black now in Circular Chess had been noted in British
has nothing better than 9...Bc4, and after Endgame Study News, special number 24,
10 Ne4 White will soon win a pawn. March 2001. The Classified Encyclopedia
As regards the genesis of the present was the second edition of David Pritchard’s
study, I had programmed my computer to Encyclopedia of Chess Variants (1994),
generate a definitive database of three-man completed by myself at the request of his
endings without pawns, and on playing widow. Circular Chess is played by a group
through the longest win with 2N v R centred on the Tap and Spile, Hungate,
I noticed that we seemed to likely have a Lincoln, who have held an annual World
“lose a move” manoeuvre somewhere in the Championship since 1996. They play to
ending. So indeed it proved. slightly different rules (a1 light, K-side on
White’s left and Black’s right, no capturing
4.4 Variant Chess 34 (nominally “Winter en passant), but the essentials are the same.
1999” but not actually published until well
into 2000). Its originality is in some doubt, 4.7 International problemists’ meeting,
because Laurent Bartholdi had produced a Bournemouth 1989 (produced for a solving
definitive database of three-man endings competition). The competition problems,
with pawns a couple of years before, and excluding a few found to be unsound, were
had he or anyone else trawled this database subsequently published in The Problemist
for positions in which P = Q was the only in November 1989. Unlike the other twin
move to win this position would have been problems in this book, where a small change
thrown up. However, there is a difference in the stipulation produces a radical change
between constructing a computer database in the solution, this is merely two different
and singling out a position within it as settings of the same idea. I am quite unable
worthy of particular attention, and the to decide which I prefer.
publication of 4.4 in VC 34 did not result in
any claim of prior discovery; nor, so far as I 4.8 International problemists’ meeting,
know, has any such claim been made since. Bournemouth 1989 (solving competition).
Other promotions clearly don’t win, and in This problem succeeded in outwitting its
fact P = N and P = K both lose. own composer. Several solvers claimed
70 Fifty-one flights of chess fancy
1 Qa2 etc, and I could neither remember nor be told that it had been done already, but as
recreate the refutation and conceded a cook. with 4.4 nobody has brought a prior
appearance to my notice. “White Ng1 to f4,
4.9 Inserted in a correction slip to Some etc” appeared in The Problemist in March
flights of chess fancy, 1989. Smulders’s 4.9a 1993, and 5.2a in Tsume-shogi Paradise
appeared in Europe Échecs in 1970. 1995. Gerd Wilts, at the cost of a more
This was a sad story. In 1972, not then artificial position, subsequently matched
being aware of the Smulders, I produced an Nagano’s numerical difference but with a
all-promotion setting in which three of the unique game in each part: U.S. Problem
promotions took place on one square and Bulletin, 1997, White d2/h2/d1/e1/g1/h1
the fourth on another. When the authors of empty, Kf2, Qg4, Re6, Ne2, Pd5/f4/g3 (15),
the Oxford Companion wanted an example Black b8/e8/g8 empty, Kf7, Nh5, Pf6 (15),
of neutral men for their first edition (1984), play 1-2 d5 Nd4 and now (a) 3 f4 Nxe2
they consulted me, and being still unaware 4 Kf2 Ng3 5 hxg3 Nf6 6 Rh6 Nh5 7 Re6 f6
of the Smulders I offered them mine. 8 Qg4 Kf7 9 Ne2 or (b) 3 Kd2 Nxe2 4 Ke3
Having become aware of the Smulders, Ng3 5 hxg3 Nf6 6 Rh6 Nh5 7 Re6 f6 8 Qg4
I urged them very strongly to use it as the Kf7 9 Qh3 Kg6 10 Ne2 Kg5 11 f4+ Kh6
example in their second edition (1992), but 12 Kf2 Kg6 13 Qg4+ Kf7.
they told me they preferred to put in an
example of an entirely different kind. 5.3 British Chess Magazine, December
The result is that the Smulders has not 1992. The reason for the stipulation “find
had anything like the exposure it deserved, the shortest game” was to try and sort out
though I did my small best to put the matter the top end of a solving competition (I was
right by using it to exemplify neutral men in afraid that if I stated the length required
the Classified Encyclopedia. in advance I would get a large number of
all-correct solutions), but people have told
4.10 International problemists’ meeting, me they think it was unfair and I have some
Bournemouth 1989 (solving competition), sympathy. The problem has indeed been
with knights instead of rooks on a8/b8. My quoted with the stipulation “Position after
reasoning at the time was that it was a Black’s 4th move, find the game”, and
problem about bishops and knights and the I have seen no reason to object.
imitator wasn’t part of a part of a normal set
anyway, so why restrict ourselves to the 5.4 Variant Chess 30, Winter 1998.
normal two knights and bring rooks into it?
But experience has shown that the version 5.5 The Problemist, November 1978.
with rooks is much easier to set up on a
board to show to friends, and today I think 6.1 EBUR, September 1996. EBUR was
I would have decided differently. a Dutch-Flemish endgame study magazine.
According to the Oxford Companion, the
5.1 The Problemist, January 1972. The pioneering setting of “dummy pawn to
original setting had no knight on f6, when draw” was by Kling, and Dawson, in
White’s first move still had to have been Ciassa’s Fairy Tales (1947), published a
b8-a7 but could have been a capture. position in which White had to do it three
times running. There is also a problem by
5.2 Used in the French retro solving Sam Loyd in which such a promotion is the
championship, 1998, and subsequently only way to force mate in three. I therefore
published in phénix (a French composition remain surprised that nobody had set the
magazine), September 1998. I expected to task “dummy pawn promotion to win”
Sources and technical notes 71
before, because once the idea is conceived 6.9 Chess Braintwisters, 1999. Again my
it is not difficult to realise, but yet again original setting was faulty.
nobody has brought a prior appearance to
my attention. 6.10 Variant Chess 56, February 2008, an
earlier setting of 6.10d having appeared in
6.2 Originally sent to be pinned up at the The Problemist in November 1989.
international problemists’ 1995 meeting as One reader was kind enough to describe
a good-will message, with promotion to 6.10d as one of my best efforts: “pointed,
GQRBN implicitly assumed and stipulation beautifully set, and not too difficult either”.
“Shortest mate” (there were complications “Pointed”, yes; I always try to produce
in part (b) after 1 Kf1 Ga8 2 bxa8Q+ into something with point, and since I try to
which I did not feel like going). Paul Byway realise it as simply as possible “not too
quoted it as a problem in Grasshopper Chess difficult” is a natural consequence. But
in Variant Chess 20, Summer 1996, which “beautifully set” is in the lap of the gods.
enabled the stipulation to be simplified to We may call ourselves “composers”, but in
“Win”: a great improvement, in my opinion. truth we are merely discoverers; all possible
“compositions” within a game were created
6.3 British Chess Magazine, July 1987. when the rules of the game were laid down,
and all we are doing is digging them out.
6.4 British Chess Magazine, April 1992. Sometimes we are lucky.
My inspiration for “it’s not chess, it’s
draughts” was a problem by Dunsany in 7.1 According to David Pritchard, this
Fairy Chess Review in August 1948. The was a game invented by T. Sturgeon and
draughts position itself is of course ancient; patented in 1890. His burglar started on c3
for example, William Payne’s Introduction and moved first, and there were six
to the Game of Draughts (1756) has White policemen on a1, c1, e1, a5, c5, and e5. The
kings d4/f2 (2), Black king c1, man b4 (2), demonstration that only five policemen
play 1 d4-c3 b4^d2 2 f2-e1. were needed appeared in Variant Chess 53,
October 2006.
6.5 Chess Braintwisters (subsequently
reissued as Outrageous Chess Problems), 7.2 Variant Chess 53, October 2006.
Burt Hochberg, 1999.
7.3 The Games and Puzzles Journal 30,
6.6 Blue Danube Joke Tourney, 1993 (an December 2003 (with the knight on c1
event held at the international problemists’ instead of g1). George Jelliss, who was the
meeting that year). magazine’s editor, attributed it to myself
without qualification, but it owed a lot to a
6.7 British Chess Problem Society problem by him in The Problemist in
weekend, Oxford, 1993, reported in The November 1989 and to his exposition of
Problemist, May 1993. Neither “promotion chessboard hypercubes in up to six
to king” nor “promotion to a Black piece” dimensions in Chessics 13 (January-June
was a new idea, though the combination in 1982) and 14 (July-September 1982).
this form may well have been. When quoting it in diagrammes I gave its
authorship as “GPJ, version JDB” on the
6.8 Original to this book. My original grounds that all I had done was make a
setting was faulty, and Harold van der cosmetic alteration to the presentation, but
Heijden suggested a fix. I hope I have when I quoted it again with this attribution
remembered it correctly. in Variant Chess 45, George demurred, and
72 Fifty-one flights of chess fancy
on reflection I think “JDB after GPJ” is which was mentioned in connection with
more accurate. My contribution may have 3.3 above), and the David Fawcett who
been merely cosmetic, but the cosmetics of composed the problem mentioned there was
a problem are important. surely the same as the “D. Fawcett” who
The pattern in the diagram is that of a made at least one excellent contribution to
two-dimensional projection of a four- Bizalion’s solitaire columns.
dimensional hypercube, and the problem
provides an illustration of how a figure 9.2 A puzzle embodying this board and
which at first sight is quite abstruse can turn giving two of these problems with others
up perfectly naturally. was on sale in London in the late 1980s or
early 1990s under the title “The Crystal
8.1 The Games and Puzzles Journal 30, Palace Wheel Puzzle”. I have seen only a
December 2003, with a decorative text. 8.2 photocopy of an eight-page descriptive
the same. booklet (A6 size) which contains neither
solutions nor acknowledgement, but it it
8.3 Colson News Volume 4 Number 3, possible that there was a separate leaflet or
August 1989. Colson News was a magazine booklet containing these. All the problems
edited by Cedric Smith, and although were in fact supplied by myself.
primarily devoted to the idea of “two-way
numbers” (the digits 9, 8, 7, 6, and 9.3 Original to this book.
sometimes 5 are replaced by upside-down
digits representing –1, –2, and so on) it 10.1 The Games and Puzzles Journal 4,
carried a fair bit of recreational material. March-April 1988. I was inspired by a hand
The next issue carried not only my official from Bridge in the Menagerie, where the
solution but a tour-de-force version in verse Hog bluffs his opponents into continuing
by Blanche Descartes, which according to clubs against 3NT, unblocks an obstructive
Richard Guy was a nom-de-plume adopted AK of diamonds from dummy, and runs off
by Smith, Leonard Brooks, Arthur Stone, six diamonds from hand to land an
and Bill Tutte. These were the four who first impossible contract.
solved the problem of dissecting a square
into smaller squares no two of which were 10.2 The deal itself was published in
the same size. Richard gives the formal Bridge World in 2004, but the editor
reference for this as “Duke Math. J., 1940”, rewrote the text to take account of what
but I think most of us know it through the would be assumed by the magazine’s
“as it happened” story in one of Martin readership and to conform to its house style
Gardner’s early books. (which was fair enough) and then printed
a version greatly reduced even from that
9.1 Original to this book. The general which I had approved. This is the first time
idea of using marked pegs, and stipulating that my original text has been published.
that these and no others are to be the pegs The gimmick of overlooking the opening
left on the board at the end, seems to me to bid and imagining that partner’s Two Club
enrich small-board solitaires even more than overcall was a Two Club opener had of
it enriches ordinary solitaire. The first writer course already been exploited by the Rabbit,
to have suggested their use appears to have with markedly better results.
been C. Bizalion in the Gentleman’s
Journal Recreation Supplement in 1870-71 10.3 Original to this book as far as
(this was a monthly puzzle and pastime publication is concerned, though I have sent
supplement to the Gentleman’s Journal it informally to various like-minded friends.
The scene is a cafe table, anywhere from Vancouver to Vladivostok.
Around it, a group of chessplaying friends; on it, such refreshment as their
taste and the time of day may dictate, and a board and men. One of the
group puts a few men on the board, and turns it so as to face the others:
“White to play and win,” he says, or “White to play and mate in four,”
or perhaps he asks some question such as “What was the game to this
position?”
This is the word of chess fantasy. It is not the accident-dominated world
of normal play, where the occurrence of a brilliancy depends not just on
the winner’s insight and experience but also on the loser’s inadvertent
co-operation in making the right blunder to create the opportunity; rather,
it is the chess equivalent of the novel or short story, where the characters
have been placed just where they need to be and every man on the board is
there for a purpose. And it is a world of pure enjoyment, untarnished by the
sorrow of defeat (except for the poor old Black king, who always seems to
get the worst of it).
So forget about rating points, give yourself a rest from trying to find a
new twist in your next opponent’s favourite opening, take out a bottle and a
glass, settle yourself down in your favourite armchair, and start reading.
You will find endgame studies, problems that have been set in national or
international solving competitions, studies and problems in variant forms of
chess, “how did we get here” problems, jokes and curiosities ranging from
Thomas à Becket to the Loch Ness Monster, and a small selection of
puzzles and oddities from fields other than chess. This is chess for fun,
chess for pleasure; chess as it should be.
ISBN 978-0-9555168-1-8
51 Flights
of
Chess Fancy
rdwdw4kd
dp0pdp0p
pdwdwhbd
dwdwdwdw
wdwdwdwd
dPdwdwdw
wdP)P)P)
dwGwdwIw