Vous êtes sur la page 1sur 4

5.

8 A BRIEF BUDGET OF DISSECTION CURIUSHIES


We conclude the chapter with a small collection of planar and spatial dissection
curiosities. It is not to be regarded as a proper part of our general textual d
evelopment, but as a browsing ground for any interested reader. The examples. wh
ile illustrating some of the fascination of this odd geometrical art. may appeal
to some reader's recreational instincts. A number of unsolved problems are stat
ed.
5.8.1 Dudeney's dissection of an equilateral triangle into a square. Henry Ernes
t Dudeney (1857-1931), unquestionably England's foremost inventor of puzzles, wa
s spectacularly adept in the art of geometrical dissection. His best-known disco
very in this field is his four-piece dissection of an equilateral triangle into
a square. A five-piece solution of this problem (see the end of section 5.3) had
been known for some time, and it was generally believed t at this could not be
bettered. Dudeney's discovery was made about 1902, and has since become famous i
n dissection literature. Figure 5.8a explains the dissection; a proof of its cor
rectness is left to the reader. The segments AD, DB, BE, EC, FG are all equal to
half the side of the triangle; EF is equal to the side of the equivalent square
; DJ and GK are each perpendicular to EF.
If the four pieces 1, 2, 3, 4 are successively hinged to one another at the poin
ts D, E, G, then, holding piece I fixed and swinging the connected set of pieces
4-3-2 counterclockwise, the equilateral triangle is neatly carried into the squ
are. A set of four connected tables has been built based upon this fact; swingin
g the tables in-one direction causes the tops to fit together into a single equi
lateral triangular table, and swinging them in the other direction causes the to
ps to fit together into a single square table.
L.V. Lyons has used Dudeney's construction to obtain a dissection of the plane i
nto a mosaic of interlocking equilateral triangles and squares, as pictured in F
igure 5.8b.
5.8.2 Dissection of a regular octagon into a square. Dudeney bettered several lo
ng-established records in polygonal dissection. He was the first, for example, t
o dissect a regular pentagon into a square with only six pieces (see Problem 26,
section 5.3), and to dissect a given square into three equal squares with only
six pieces. Ven.- curious along this line is a statement made by Alice Dudeney (
Mrs. H. E. Dudeney) in the preface to Puzzles and Curious Problems, by H. E. Dud
eney, published in 1932, after the author had died. The book was reprinted in 19
36. 1941, and 1948, revised by James Travers. The statement is: - It is remarkab
le that the regular octagon can be cut into as few as four pieces to form a corr
esponding square." The implication is that such a four-piece dissection was anot
her of Dudeney's triumphs. But no one has since been able to find such a solutio
n. A beautiful five-piece dissection, illustrated in Figure 5.8c, was published
by Travers in 1933. it is strange that Travers, who revised the book in question
, made no correction of or comment on Alice Dudeney's unverifiable statement. It
is difficult to believe that a four-piece solution of the problem is possible.
5.8.3 The minimal dissection problem. The minimal dissection problem is that of
determining the least number of pieces needed to dissect a given polygon into an
other given equivalent polygon. Only very small inroads have been made into this
problem. W. B. Carver and Alfred Tarski showed that if it is the minimum number
of pieces needed to dissect a rectangle of dimensions a and b, a â ¥ b, into a square
, then
where x = (a/b)1/2, and E(y) means the least integer not less than y. This resul
t was improved upon when E. E. Moise showed that n â ¤ E(x) + 1 and Tarski, in a later
paper, showed that n â ¥ E(x). It follows then that it = E(x) or E (x) + 1. In partic
ular, the reader might like to try to answer the following two questions:
(a) Is there a three-piece dissection of a 2 x 15 rectangle into a square?
(Here E(x) = 3.)
(b) Is there a five-piece dissection of a 3 x 64 rectangle into a square?
(Here E(x) = 5.)
that a 32 x 33 rectangle can be dissected into nine squares, no two of which are
equal. This raised the question of whether a square can be dissected into a fin
ite number of squares. no two of which are equal. It was felt that this latter p
roblem was impossible. but such turned out not to be the case. The first publish
ed example of a square dissected into unequal squares appeared in 1939; the diss
ection was given by R. Sprague of Berlin, and contained 55 sub
squares. In 1940, R. C. Brooks, C. A. B. Smith. A. H. Stone, and W. T. Tutte, in
a joint paper, published a dissection containing only 26 pieces. These men inge
niously established a connection between the problem of dissection and certain p
roperties of currents in electrical networks. In 1948, T. H. Willcocks published
the 24-plece dissection of a square into unequal squares pictured in Figure 5.8
e. and this dissection is May the record so far as least number of pieces is con
cerned.
8.5 Perfect dissections. A dissection is said to be perfect if all the pieces si
milar but unequal: the number of pieces is called the order of the dissection. A
dissection is called finite or infinite according as the order of the dissectio
n is finite or infinite. The problem of â squaringâ the square is that of finding a fini
e perfect dissection of a square into squares. A great deal of work has been don
e in recent years in connection with perfect dissections. We here list some of t
he results found, and conclude with a couple of questions that still remain unan
swered.
(a) If a rectangle can be finitely dissected perfectly into squares, then th
e sides of the rectangle are commensurable.
(b) If the sides of a rectangle are commensurable, then the rectangle can in
an infinity of ways be finitely dissected perfectly into squares.
(c) There is an infinite perfect dissection of a rectangle into squares.
(d) There is no perfect dissection of a rectangle into squares of order less
than 9, and exactly two of order 9, (Proved by H. Reichardt and H. Toepken in 1
940, One of the two possible dissections is pictured in Figure 5.8d.)
(e) There is no finite perfect dissection of an equilateral triangle into eq
uilateral triangles. (Proved by W. T. Tutte in 1948.)
(f) There is no finite perfect dissection of a rectangular parallelepiped in
to cubes.
There is a proof of this that is so pretty and simple that we sketch it here. Su
ppose there is a finite perfect dissection of a rectangular parallelepiped P int
o cubes. Then the bottom, face of P is a "squared" rectangle. Within this "squar
ed" rectangle there is a smallest square. which clearly cannot be along an edge
of the rectangle. This means that the smallest cube A resting on the bottom base
of P is surrounded by larger cubes. On top of cube A, smaller cubes must rest.
forming a "squared" square on the top face of A. Within this "squared" square th
ere is a smallest square, giving rise to a cube B that is the smallest cube rest
ing on the top of cube A. Continuing the argument. there must be a smallest cube
Cresting on top of cube B, and so on ad infinitum. But this is impossible.
A perfect dissection of a rectangle into squares is said to be simple if the dis
section does not contain within it a perfect dissection of a smaller rectangle.
Note that the perfect dissection illustrated in Figure 5.8d is simple, -whereas
that illustrated in Figure 5.8e is not simple.
(g) The least known order to-date of a simple perfect dissection of a square
into squares is 37. (Found by T. H. Willcocks in 1959.)
(h) There is a simple perfect dissection (of order 69) of a square into squa
res in which no four subsquares share a common vertex and no square other than t
he four corner squares is bisected by a diagonal of the complete 6gure. (Found b
y W. T. Tutte.)
(i) What is the smallest possible order for a perfect dissection of a square
into squares?
(j) Does there exist a simple perfect dissection of a 1 x 2 rectangle into s
quares?
5.8.6 Dissecting - a checkerboard. The problem of enumerating the number of ways
in which an n x n checkerboard can be cut, with cuts made only along edges of s
quares of the board, into two congruent connected parts, seems not to be easy.
A solution is known for a 6 x 6 board.
5.8.7 Dissection into dominoes. We call a 1 x 2 rectangle a domino. One an ask a
number of interesting questions, such as:
(a) In how many ways can a 2 x n rectangle be dissected into dominoes? (See
Problem E 1470, The American Mathematical Monthly, Jan. 1962.)
(b) In how many ways can 2k squares be cut from an 8 x 8 checkerboard so tha
t the remaining part can be dissected into dominoes ?

5.8.8 Regular tessellations. A Planar regular tessellation is a dissection of th


e Euclidean plane into regular polygons. If the polygons are all confluent, then
they must have 3, 4, or 6 sides. More interesting is the situation where two or
more sizes of the same kind of polygon are admitted, or where two or more diffe
rent kinds of polygons are admitted. The literature on regular tessellations is
very large, and the subject becomes particularly interesting in the Lobachevskia
n non-Euclidean plane, where one can form tessellations of congruent regular n-g
ons with p of them about each vertex for any p such that I n + I p < 1/2. Solid
regular tessellations have also ken extensively discussed.
5.8.9 The translation restriction. H. Lindgren, of Australia. has considered dis
sections wherein one polygon is cut into pieces which can, by translations alone
, be rearranged to form a second polygon. Such a dissection may be called a tran
slation dissection. Examples of Lindgren's achievements along this line are:
(a) A 4-piece translation dissection of a quadrilateral into any equivalent
quadrilateral having the same angles.
(b) A 10-piece translation dissection of a regular nine-sided polygon into a
n equilateral triangle.
5.8.10 Parallelogramic dissection of a parpolygon. A parpolygon is a polygon of
an even number of sides in which opposite sides are equal and parallel. It is a
nice geometric application of mathematical induction to prove that: Every parpol
ygon of 2n sides can be dissected into n(n - 1)/2 parallelograms.
The theorem is certainly true for n = 2. Suppose it is true for n = k, and let P
k+1 be any parpolygon of 2(k + 1) sides. Then Pk+1 can be dissected (as indicate
d in Figure 5.8f) into k parallelograms and a parpolygon Pk of 2k sides. By our
supposition. Pk can be dissected into k(k - 1)/2 parallelograms. It follows that
P,..,, can be dissected into
k + k(k - 1)/2 = (k - 1)k/2
parallelograms. The theorem now follows by mathematical induction.
It is interesting to note that if the parpolygon is equilateral, then the parall
elograms become rhombuses of the same length side. A regularâ 2n-gon is an example of
an equilateral parpolygon.
5.8.11 The Fibonacci numbers and a dissection puzzle. There is a familiar geomet
rical paradox. pictured in Figure 5.8g, where a square, subdivided like a checke
rboard into 64 small unit squares. is cut into four pieces which can apparently
be reassembled to form a rectangle containing 65 unit squares. The explanation o
f where the additional unit square comes from in the fact that the edges of the
four pieces which seem to fie along a diagonal of the rectangle really do not do
so, but form a very flat parallelogram of exactly one square unit of area. A nu
mber of mathematicians, among them Lewis Carroll of Alice in Wonderland fame. ha
ve considered the problem of generalizing this paradox. A neat generalization is
furnished by the so-called Fibonacci numbers, defined by f1= f2 = l, fi+1 + fi
(i= 2, 3, â ¦). It can be shown that
fn+1 fn+1 = fn2 + (1-n)n,
whence a square of side f, can be dissected, as in Figure 5.8h. into four
pieces which can be reassembled to almost form an fn-1 x fn-1 rectangle.
If we take n = 6. we obtain the original dissection puzzle.
5.8.12 The law of cosines. Figure 5.8i indicates a simple proof. utilizing

dissection notions. of the law of cosines:


c2 = a2+ b2 â 2ab cos C
where a. b. c are the sides of any triangle and C is the angle Opposite side c.
5.8.13 An equilateral triangle problem. Some geometrical problems can be neatly
solved with dissection methods. This is illustrated here, and in the next two it
ems. Consider the problem. Find the area of an equilateral triangle if a point P
within the triangle is at distances 3, 4, 5 from the three Vertices.
Referring to Figure 5.8j, cut the triangle into three pieces along the solid

tines from P and then arrange these pieces around the triangle as indicated to f
orm a hexagon of twice the area of the triangle. Cut this hexagon along the soli
d and dashed lines from P to form three equilateral triangles and three right tr
iangles, all of known sides. The desired area is now readily computed to be
[ -(32-42+52)/4+ 18]1/2
5.8.14 An area problem. If A', B', C' (see Figure 5.8k,) are the first
trisection points or the sides BC, CA, AB of a given triangle ABC show that the
cevians AA'.. BB', CC' are the side lines of a triangle whose area is one-sevent
h the area of the given triangle.
A very simple dissection proof is indicated in Figure 5.8k2

Vous aimerez peut-être aussi