Math Hysteria - Fun and Games With Mathematics
Math Hysteria - Fun and Games With Mathematics
MATH
HYSTERIA
20 CHAPTER TWO
involved, but Snitchswisher trotted off through the bog to fetch him.
Soon she and the old man - Wishsnitcher Dishpitchersson by name -
reappeared. Suitably primed with silver from Rocknuttersson's purse,
he fished some Tarot cards from his robes and began a divination.
'Below ... the Moon. Above ... the Leaping Cow. To west and east ... '
'The Cat and the Fiddle,' suggested Rocky.
'Yes, but the Cat is inverted, signifying drunkenness ... Below, the
Laughing Hound-'
'Signifying that these entire proceedings are a farce - '
'Signifying merriment. More cards ... the Dish, the Spoon - '
'And the Knife and Fork.'
'No ... the deuce of Forks.' The old man shook his head. 'Which is
strange, since there are no forks in the pack ... Ah! A name ... A spirit
from the future ... an acolyte of 'Big Blue', whatever mystic being that
may be ... Ralph ... Ralph ...'
'That's the hound. They always go 'Ralph! Ralph!' But it's called
'barking', not 'laughing'.'
'No, it is a name ... Ralph ... er ... Grimoire? Grimory? No - Ralph
Gomory, a future numerosophist of great ingenuity ... A three-tined
fork and a four-tined fork, a sigil of enormous power and beauty.
Quick, the charcoal!' The old man drew swift lines on the board
(Figure 3). Then his trance faded.
Rocknuttersson sourly handed ove! a further piece of silver. 'I reck-
on your dad's a few stones short of a henge, Snitchswisher.'
Figure 3
Gomory's slgd creates a chain
that can be fdled by
consecutive dominOes.
DOMINO THEORIES 21
I--
I--
• Figure 4
How to fill in the
dominoes if the two
She sniffed and studied the charcoal lines. 'I am not so sure,
Rockchopper Rocknuttersson. Imagine the two forks are walls. Then a
line of dominoes may be placed between them, in an endless loop. If
two squares are occupied by statues, the loop is cut into two sections.
Possibly just one if the squares are adjacent. If the statues are on squares
of the opposite colour, then each section contains an even number of
squares, so the chain of dominoes can fill it completely. The diagram
represents a proof that no matter which two squares are occupied by
statues - provided only that they are of opposite colour - the remainder
can be covered by dominoes. Indeed it is a constructive proof, showing
exactly how to achieve such a result in any given case' (Figure 4).
Rocky was impressed. 'Snitchswisher, [ apologize to your father
for my scepticism. He has uncovered a remarkable truth.' The old man
muttered something about 'fine words', 'butter', and 'parsnips', and
Rocky handed over another piece of silver to avoid further embarrass-
ment. 'Pnerd! Fetch my scribing-chisel, and the finest slab that is
portable! We shall head the document TENDER FOR RENOVATION OF
QUAGVILLE MARKETPLACE, ROCKHOPPER ROCKNUTTERSSON ROCK RENO-
VATORS, MURKLE MIRE.'
'OK,' said Pnerd. 'Oops, I never told you about the two new statues,
did I?'
Rocky stared at him. 'Two ... new .. .'
'Demagog and Wolligog. The priests decided to cover up some more
cracks.'
22 CHAPTER TW O
A •• •• D
• •
B
• • • • c
• •
Figure 6 Figure 7
Four problematic corner An example of how to handle
arrangements (and 0 block the case A of Figure 6. Each slgd-
placement of a domino to cover the bearing region contains
corner square, but A and Bare preCisely one omitted square of
harmless. each colour.
'Hmmm ... I haven't time to enter into every detail; but I'm pretty sure
you can show that the only occasions on which the board cannot be so
divided are when either the two black squares omitted, or the two
white squares, are in the same comer (Figure 6). In one arrangement, it
is obvious that the comer square is isolated from all the rest, and no
solution can be found . In the other case ... the board can again be divid-
ed into two regions, each containing just one omitted square of each
colour, and each possessing a Gomory sigil of its own (Figure 7). One
region has to have a hole in it, but that doesn't alter the argument. I
believe that a careful analysis will show that it is always possible to
cover the board with dominoes, except when a comer configuration
such as in Figure 6 occurs, for one or other colour.' She shrugged. 'It is
not as elegant a proof as Gomory's, however. Perhaps some future
numerosophist can do better.'
'Anyway: said Rocky, 'it sounds as if we're probably in business.'
He leaped to his feet. 'What we need is somebody to go and check that
the statues haven't cut off one of the comer squares. To be on the safe
side, Pnerd, this time you can also make a map of the positions of the
28 CHAPTER THREE
Figure 8
Can you slide the tables until the
1 2 3 4 5
7 8 9 10 11 12
?•
13 14 15 16 17 18
in the puzzle are, and how to get from one to the other. Mental mazes
that tell you what moves to make, and in what order.'
Dan nodded. Of course. Except .. . 'It's going to be a pretty compli-
cated map, Max. There's an awful lot of positions, and an awful lot of
moves.'
'True. So we'd better find some way to cut them down. Break the
problem into simpler pieces. Hey! Yes, that's it. First of all, let's find out
what we can do easily. Then we can try to kind of string those together.'
'Well, for a start, if you've got a square hole with just the two small-
est tables in it, you can move those tables around pretty freely,' said
Dan. See Figure lOa.
'Yeah, that's the kind of idea. A sort of 'sub-puzzle' where you move
only a few tables around, inside some well-defined boundary.'
(Positions 5-6-7 of Figure 9 use just such a sub-puzzle.) He stopped and
TURNING THE TABLES 31
(a) ~ rn Ed rn BJ rn Figure 10
Some useful manoeuvres.
In each sub-puzzle the
tables can easily be
(b) rearranged without going
outside the boundaries
marked.
"'} "'}
• •
START
"'}
• .
"'}
Figure 11
A partial map of the puzzle. Large diagrams show the plaCing of key tables.
(ross-hatched regions represent sub-puzzles to be solved using the useful manoeuvres of
Figure 10. Lines indICate sequences of moves, sometimes qUite long, that are essentially
'forced' If you know where you want to go. For example the line from START to ( represents
the sequence of 17 moves shown in Figure 9. The small diagrams show how the sub-puzzles
are to be arranged at the start and finish of these sequences. Using this map as a guide,
the puzzle becomes relatively easy to solve.
TURNING THE TABLES 35
Figure 12
?. Three harder sliding block puzzles
Any arrangement of blocks IS
allowed In the cross-hatched
~
region.
Top arrow' the Donkey puzzle
. .----r-..,
Middle Arrow the Century puzzle
Bottom arrow the Century and a
Half puzzle.
The pizzas arrived. Dan's didn't look quite right. Most of the
ingredients were upside down, including the crust. The waitress had
included a whole tuna and tried to set the liquorice on fire.
'Enjoy your puzzle, sir,' she said over her shoulder.
'Send it back,' suggested Max.
'No, no, you heard what she said. I can't resist a challenge, it's a test
of character. The pizza just needs rearranging.' Dan picked up the tuna,
tried to find somewhere to put it while he blew out the liquorice.
Where had the pepperoni gone? Oh, yeah, inside the pineapple. The
plate just wasn't big enough ... He sighed, put the tuna back, and was
on the point of calling the manager to complain that his pizza was too
difficult to solve.
Then straightened his back, squared his shoulders, and reached for
the clipboard.
'What are you doing?' asked Max.
'I can fix it. Just wait till I've made a map of this pizza.'
50 CHAPTER FIVE
••
••••
• Harold
Figure 14
If Harold plus his 61 identICal squares of yeomen can form a single square, what IS the smallest
number of men he can have?
x is at least 1. You may wish to try solving this equation before reading
on. Don't spend too much time on it, though.
According to the theory developed by mathematicians such as
Pierre de Fermat and Leonhard Euler between about 1650 and 1750,
equations of this general kind always have infinitely many solutions,
where 61 can be replaced by any non-square positive integer. If 61
is replaced by a square, then the equation asks for two consecutive
integers to be square, and the only solution is x = 0, y = 1, which is too
trivial to be interesting. The technique for calculating the solutions
involves 'continued fractions': it can be found in most number theory
textbooks and also in Albert H . Beiler's entertaining Recreations in the
Theory of Numbers (see Further Reading). As a warm-up, let's take a
look at the lesser-known Battle of Brighton (1065), where King Harold's
men formed 11 squares, all else being unchanged. Now the equation is
y2 = 11x2 + 1, and a little trial and error reveals the solution x = 3, Y = 10
(Figure 15). That is, 100 = 11 x 9 + 1. The next solution is x = 60, Y = 199,
and there is a general procedure for finding all solutions once you have
found the smallest.
COU NTI NG TH E CATTLE OF TH E SU N 51
•
Figure 15
The Battle of Brighton (1065), where King Harold had to solve the equation y2 = llx2 + 1.
The solution IS x = 3, Y= 10
Trial and error will not solve Dudeney's puzzle, though - well,
maybe on a computer, but not by hand - because the smallest solution
is x = 226,153,980, Y = 1766,319,049. Solutions of the Pell Equation y2 =
Dx 2 + 1 vary wildly with D. The 'difficult' values of D up to 100 - that
is, those that require a value of x larger than 1000 - are D = 29, 46, 53,
58, 61, 67, 73, 76, 85, 86, 89, 94, and 97. By far the worst is 61, so
Dudeney chose wisely. With a bit of effort, you should be able to find
out what happens for D = 60 and D = 62, on either side of Dudeney's
cunning 61: see the answers at the end of this chapter.
Mind you, he could have made the puzzle a lot harder: with D =
1597 the smallest solutions are
x = 13,004,986,088,790,772,250,309,504,643,908,671,520,836,229,100
Y = 519,711,527,755,463,096,224,266,385,375,638,449,943,026,746,249
and D = 9781 is much worse.
The puzzle which Archimedes included in his letter to Eratosthenes
begins 'If thou art diligent and wise, 0 stranger, compute the number
of cattle of the Sun, who once upon a time grazed on the fields of the
Thracian isle of Sicily ... ' In Homer's Odyssey there are 350 cattle of
THE GREAT DRAIN ROBBERY 65
Figure 17
Part of the grounds of
Robbingham Hall. The drain,
which IS straight, passes through
the marked CIrcle.
But where?
fully. Mathematically, we are looking for the shortest curve that meets
every chord of a circle of radius 100 yards - a chord being any straight
line that meets the circle. We must of course include the tangents to the
circle, which meet it at only one point:
'Why, Holmes?'
'Because old Ned said 'a hundred yards or less', implying that a dis-
tance of 100 yards might have occurred, and that is the distance to a
tangent:
'Ah,' said I, impressed by his incisive logic. 'But that is a very com-
plicated problem, for there are many chords to consider. Might we not
simplify it somehow?'
Holmes's head jerked back in surprise. 'An excellent suggestion,
Watson, your best for weeks. It is of course sufficient to consider only
the tangents. A curve that meets all tangents necessarily meets all
chords.'
'Why.?'
76 CHAPTER SEVEN
Figure 12
(a) Sam loyd's Sedan Chair puzzle.
(b) Its solution.
Figure 28
Here I am, as you may see,
a" b'· ab
Airy's Pythagorean
When two triangles on me stand, dissection.
Square of hypothn' is plann'd;
But if I stand on them instead,
The squares of both the sides
are read
a b
\ Z ~ /
/~
/
,,
Figure 29
/
Busschop's hexagon
,
/
/\
/
/
\
/
/ to square.
\
, / ,
\ /
/
l<:'>\ \'/-----"\
(
, /
/
\
, /
/
/
V
/
MONOPOLY REV I SITED 111
- - - Single th row
_ _ _ All throws
o 5 10 15 20 25 30 35
Figure 37 Probabilities of movtng a given number of squares on one turn, taktng rules concerning
doubles into account.
MONOPOLY REVISITED 113
-...
"......
C
10 -
J~oI illin ois
Aven ue
CIl
u
CIl R eadi~ ! Electric
Com pany
B&O
Ra ilroad
'"~
J works
Chance
!
Q.. Ra ilroa
! ~Jai1 Shm/
'-" Pennsylvania \
Go
~
Ra ilroa d
.~ 5 - I Line
:c
It!
/" l
.0
...0
Q..
0
I~ I
1 II~
I I
0 5 10 15 20 25 30 35 40
Square
Figure 38 The steady state probability distribution: squares on the board are numbered 1- 40
and the height of each bar IS the long-term probability of occuPYing that square. See Table 2 (p. 99)
for Bntlsh names.
DIVIDING THE SPOILS 131
proportional ones. You should also bear in mind that along with the
protocol there is a set of strategies, one for each person, which guaran-
tees that they will think they have achieved their goal. These strategies
are not part of the protocol, and are often only tacit, but they are what
guarantees it to be fair. We will often call the people 'players', because
it helps to think of the whole process as a series of moves in a game,
and it is customary to represent the valuables that are to be divided as
a cake.
For two players, say Arthur and Bertha, there is a simple envy-free
allocation protocol: 'I cut, you choose'. Arthur divides the spoils into
two parts; Bertha chooses which of the two parts she wants (Figure 44).
Figure 44
The traditional method of
dIViding a cake between
two people.
Figure 45
19
22 (a) Mor6n's ten-tile
24
rectangle
5
11
•
1
6 3
23 25
17
I 25
SQUARING THE SQUARE 149
rectangle 1
Figure 47
extra
square
~
: =--i -
How to combine two squared
rectangles to get a squared
square, provided the rectangles
r
-r
'-
l
extra
square
have no Illes of common size.
Brooks's rectangle could certainly be tiled in two ways, but since the
same set of tiles was used both times, Brooks's rectangle would not
lead to a squared square. Nevertheless, they hoped that if they under-
stood why Brooks's rectangle could be tiled in two ways, they might get
a useful angle on the problem. On looking at the Smith Diagrams of the
two tilings they realized that they could obtain one diagram from the
other if they 'identified' two nodes - that is, conceptually considered
them as being the same. Moreover, the flow of electricity through the
circuit was not affected by 'short circuiting' the diagram in this manner,
because in this particular instance the two points that were identified
were at the same electrical potential. With some effort they worked out
why this was the case - it was related to symmetries in the Smith
Diagram. From this clue they developed other ways to tinker with
Smith Diagrams, producing different squared rectangles of equal size,
but having fewer tile sizes in common. Eventually this approach paid
off, leading them to a simple squared square formed from 69 tiles. With
further effort. Brooks reduced the number of tiles to 39.
In 1948 T. H. Willcocks reduced the number of tiles further, finding a
squared square with 24 tiles. But his square was not simple. Meanwhile
J. W. Bouwkamp and colleagues were cataloguing all possible squared
rectangles with up to 15 tiles - finding a total of 3663. In 1962 A. W. J.
Duivestijn proved that any simple squared square must contain at least
SQUARING THE SQUARE 151
9 8
15
1 Figure 51
Non-tnvlal squanng
7
of a cylinder.
10
18
18
14
A cylinder can be squared, but this requires at least nine tiles - just
as for rectangles. 'Trivial' squarings just take Moron's rectangles and
join up suitable edges; but there are also two non-trivial nine-tile
squarings. Their tiles are the same sizes as Moron's, but the arrange-
ment is different (Figure 51).
For the cylinder and Mobius band, the edges of the tiles have to be
parallel to those of the surface. However the torus, Klein bottle, and
projective plane do not have any edges, so tiles could conceivably be
set at an angle. In fact, if this is done, a torus can be tiled with just two
square tiles (Figure 52) provided we allow two edges of the same tile to
meet. (As a bonus, there is a proof of Pythagoras's Theorem hidden in
this picture: can you see why?)
Not much seems to be known about tilings of the Klein bottle. Every
tiling of a Mobius band can be glued along its edge (there is only one)
to give a tiling of the Klein bottle, and there are no other ways to tile the
SQUARING THE SQUARE 153
Figure 52
Squanng a torus with just
two tiles - and a hidden
proof of Pythagoras.
(H Int: look at the nght
triangle whose hypotenuse
is the left-hand edge of the
diagram.)
Klein bottle with six or fewer square tiles. Nobody knows whether this
remains true for seven or eight tiles - but it is false for nine.
Virtually nothing is known about tilings of the projective plane. And
what about tiling, say, the surface of a cube? The whole field is wide
open.
PURPOSEFUllY PILING PYRAMIDS 169
building continued all the year round. The actual time might have been
half as long, or twice as long: because of this fundamental uncertainty
about timing it is pointless to enter into fine detail when estimating
other relevant figures.
A certain amount about pyramid building techniques can be
deduced from ancient records and from the layout of the pyramid site
(Figure 57). Pyramids were made from blocks of stone, taken from
nearby quarries whose location is sometimes known. The sole source
of power was human muscles (no water power, for example) aided by
primitive but effective tools such as levers. On the whole, a pyramid
was built in layers from bottom to top - certainly no block could be put
in place until those below had been. Horizontal transport of the blocks
was done by workers dragging wooden sledges (Egyptian carvings
MENKAURE
PYRAMID
~ KHAFRE
PYRAMID
~ -
KHUFU
PYRAMID
Quarry for
ramp material
Khufu
[g]
Harbour
'SPHINX
---.'----.J...~
tI)
~ Mastaba cemetery
o Valley Temple
• Mortuary Temple
- - - causeway
Step, Saqqara
p, Djoser Maidum
I I 1790 Sneferu
r t o 100 200 metres
I I I 1 I 2900
more than a few workers. It turns out that, with the same assumptions,
8300 men then suffice to build Khufu's pyramid. Similar results hold
for other pyramids (Figure 61). Despite the astonishing size of the
pyramids, it seems that manpower was not a problem.
How did the Egyptians lift so many stones into place? A common
theory is that they used huge ramps made of sand, which were later
removed. They certainly used small ramps for various construction
purposes, but it seems unlikely that they built gigantic ramps rising to
the top of the pyramid. It takes a lot of energy to build a ramp - rough-
ly as much as it does to build a complete pyramid - and then the ramp
has to be removed again. Another problem with ramps is that the
workers don't just have to lift the stones against gravity: they have to
lift themselves as they make their way up the ramp. Unless they haul on
ropes, which isn't easy if the ramp is a long one.
A better way would be to keep the movement of workmen as low as
possible. One simple and efficient way to achieve this was suggested to
me a few years ago by Alan Moore, an Englishman who has long been
interested in puzzles about ancient technology. His idea is to use a
190 CHAPTER SEVENTEEN
~
Figure 65
I- I-
Game tree for 4 x 4
yucky choccy
I 1-
,.. I
-
T
Arrows indicate legal
rT II
1
.~
~-
1
I
1
could have been 1-
.- .~ - ~
made instead.
r • II
I
,
I
1
Lll ] =-
I I
•• I I -~
• I[] ':- •I'
•
a lump of soap, and the player who has to eat this square loses. The
solid arrows in Figure 65 shows the moves in the game played with a
4 x 4 bar, and the shaded arrows show all the other moves that could
have been made instead. This entire diagram constitutes the game tree
for 4 x 4 yucky choccy. As we'll shortly see, Too made a bad mistake
and lost a game that should have been won.
A winning strategy is a sequence of moves that forces a win, no
matter what moves the opponent makes. The concept of a strategy
involves not just one game, but all possible games. When you play
chess, most of your planning centres on 'what if' questions. 'If I
advance my pawn, what could his queen do then?' Tactics and strategy
centre around what moves you or your opponent could make in future,
not just the moves that they do make.
There is a neat theory of strategies for 'finite' games - ones that can't
continue forever and in which draws are impossible. It relies on two
simple principles:
SHEDDING A LITTLE DARKNESS 203
Figure 70 An Isosceles nght tnangle (shaded) 'unfolds' by repeated reflection to form a lattICe:
dots show the correspondence between lattice pOints and vertices of the triangle. A light ray such
as ABCD also unfolds, to give a straight line path ABC'D'. Any putative light ray from A back to A
Similarly unfolds to give a straight line such as AN. However, this line must pass through a grey or
white lattice point (here D') so the original ray must hit a vertex and be absorbed.
then the same thing would necessarily happen - implying that no such
path can exist.
To prove this, imagine that we have some path from A to A and
unfold it to get a straight line path from A to some point A' in the
lattice. Because A' folds up to give A, it follows that A' is one of
the black dots in the lattice. Now the black dots are spaced an even
number of lattice units apart in both the horizontal and the vertical
direction - their coordinates are even integers. This implies that some-
where along the line AA' there must be a lattice point that is either
white or grey. This fact is obvious if either the horizontal or vertical
spacing is twice an odd number, because then the midpoint of AN has
at least one coordinate an odd integer and is therefore a grey or white
lattice point. This argument fails if both coordinates are multiples of
SHEDDING A LITTLE DARKNESS 205
Figuren
Angela also cannot see
Bruno's match in rooms
created by reflecting
BRUNO
•
ANGELA
Figure 73
exist. Recall that the curve known as an ellipse has two special points,
known as foci (Figure 73a). It can be proved that any light ray that
passes between the two foci and bounces off the curve will again pass
between the two foci before it next hits the curve. By 'pass between'
the foci we mean 'cross the straight line that joins them'. Bearing this
property in mind it is easy to check that the room in Figure 73b has un-
illuminable regions. Specifically, rays originating in the shaded region
labelled Bruno can never enter the lightly shaded one labelled Angela.
In a letter, Tokarsky sent me many more examples, and some gener-
alizations. Among them are the following. There are polygonal rooms