Crux

Mathematicorum
Published by the Canadian Mathematical Society.

http://crux.math.ca/

The Back Files

The CMS is pleased to offer free access to its back file of all
issues of Crux as a service for the greater mathematical
community in Canada and beyond.

Journal title history:

➢ The first 32 issues, from Vol. 1, No. 1 (March 1975) to
Vol. 4, No.2 (February 1978) were published under the
name EUREKA.
➢ Issues from Vol. 4, No. 3 (March 1978) to Vol. 22, No.
8 (December 1996) were published under the name
Crux Mathematicorum.
➢ Issues from Vol 23., No. 1 (February 1997) to Vol. 37,
No. 8 (December 2011) were published under the
name Crux Mathematicorum with Mathematical
Mayhem.
➢ Issues since Vol. 38, No. 1 (January 2012) are
published under the name Crux Mathematicorum.
 CRUX MATHEMATICORUM
March
Volume 20 #3 mars
1994

CONTENTS I TABLE PES MATIERES

On the Distribution of Zero Points of the Real-valued

1 - 2J"=1at/'(x + a,) tasakazu Nihei 61

The Olympiad Corner: No. 153 \E. Woodrow 64

Problems: 1921-1930 73

Solutions: 1826, 1828, 1831-1840 75

Canadian Mathematical Society Soci&6 math&natique du Canada

< $ >
 Founding Editors / R£dacteurs-fondateurs: Leopold Sauve, Frederick G.B. Maskell
Editors-in-Chief / Re*dacteurs en chef: G.W. Sands, R.E. Woodrow
Managing Editor / R£dacteur-g£rant: G.P. Wright

EDITORIAL BOARD / CONSEIL DE REDACTION

G.W. Sands (Calgary)
R.E. Woodrow (Calgary)
G.P. Wright (Ottawa) - Managing Editor
R. De Peiza (East York Board of Education)
C. Fisher (Regina)
R. Guy (Calgary)
D. Hanson (Regina)
A. Liu (Alberta)
R. Nowakowski (Dalhousie)
J. Totten (Cariboo College)
E. Wang (Wilfrid Laurier)

GENERAL INFORMATION RENSEIGNEMENTS GENERAUX

Crux Mathematicorum is a problem-solving journal at the Crux Mathematicorum est une publication de resolution de
senior secondary and university undergraduate levels for problemes de niveau secondaire et de premier cycle
those who practice or teach mathematics. Its purpose is universitaire. Bien que principalement de nature educative,
primarily educational but it serves also those who read it for elle sert aussi a ceux qui la lisent pour des raisons
professional, cultural or recreational reasons. professionnelles, culturelles ou recreative.

Problem proposals, solutions and short notes intended for Les propositions de problemes, solutions et courts articles
publication should be sent to the appropriate member of the a publier doivent etre envoyes au membre approprie du
Editorial Board as detailed on the inside back cover. conseil de redaction tel qu'indique sur la couverture arriere.

SUBSCRIPTION INFORMATION RENSEIGNEMENTS SUR L'ABONNEMENT

Crux Mathematicorum is published monthly (except July Crux Mathematicorum est publie mensuellement (saufjuillet
and August). The subscription rates for ten issues are et aout). Les tarifs d'abonnement pour dix numeros
detailed on the inside back cover. Information on Crux figurent sur la couverture arriere. On peut egalement y
Mathematicorum back issues is also provided on the inside retrouver de plus amples renseignements sur les volumes
back cover. anterieurs de Crux Mathematicorum.

ACKNOWLEDGEMENTS REMERCBEMENTS

The support of the Department of Mathematics & Statistics Nous rendons hommage a Pappui du departement de
of the University of Calgary and of the Department of mathematique et statistique de l'Universite de Calgary et du
Mathematics of the University of Ottawa is gratefully departement de mathematiques de PUniversite d'Ottawa.
acknowledged.

Canadian Mathematical Society / Societe mathematique du Canada 1994

Printed at / imprime a: Synergy Print and Copy, Ottawa.

ISSN 0705-0348 Publication Mail Registration No. 5432

 61

ON THE DISTRIBUTION OF ZERO POINTS

OF THE REAL-VALUED FUNCTION
n
a
1- 2 i/(X + °>i)
1=1
Masakazu Nihei

1. I N T R O D U C T I O N Let / ( * ) be a polynomial with real coefficients. If a is

a real number such that f(a) = 0, then a is called a "real zero point" (in what follows we
will call it simply a zero point). Alternatively, a is called a real root or real solution of the
equation f(x) = 0.
In general, we can not expect to obtain, in concrete form, the solutions of an
equation of fifth degree or more (see [4; p. 218] for example). In view of this fact, it would
be a highly intriguing proposition to find intervals that will contain one and only one of
the real roots of a given equation (i.e., separation of real roots).
To begin with, we will confine our consideration to the separation of real roots of a
particular cubic equation expressed in a comparatively pretty form as follows:

(x + 2)(x + 3) + 2(x + l)(x + 3) + 3(s + l)(x + 2) = (x + l)(x + 2)(x + 3). (1)

That is, we seek the real numbers x for which equation (1) holds. Let

f(x) = {x + l)(x + 2){x + 3) - (x + 2)(x + 3) - 2{x + l)(x + 3) - 3(x + l)(x + 2);

then the function f(x) is continuous on the open interval I I = (—oo, oo). Since /(—3) =
—6 < 0 and /(—2) = 2 > 0, f(x) has at least one zero point on the open interval (—3, —2)
by virtue of the Intermediate Value Theorem (see [1] or [2]). Likewise, we can see that
f(x) has at least one zero point on the open interval (—2, —1).
We can also see that f(x) has one and only one positive zero point, once we have
taken into account the fact that / ( 3 ) = —18 < 0, / ( 4 ) = 8 > 0 and that (1) is a cubic
equation.
In this note, we will generalize the above-mentioned problem. It is of special interest
to determine as precisely as possible the interval containing positive zero points.

2* RESULTS Since the equation (1) does not have —1, —2, —3 as its roots,
division of both sides of (1) by (x + l)(x + 2)(x + 3) will not bring about a change when
studying how its roots are separated. This means that it suffices to study how the zero
points of the real-valued function

(2)
'w^-TTT-^-dh
are distributed in order to study how the roots of the equation (1) are separated.
The following result shows the separation of the roots of the equation (1) generalized
in the form of (2).
62

THEOREM: Let 0 < ax < a2 < • • • < a n , n > 2, £ ? = 1 «• = *• Then the

real-valued function
n _

has one zero point on each of the open intervals (—at-, — at__i) (i = 2 , . . . , n), and further,
one positive zero point on (s — an,s(n — l ) / n ) .
Now, we will prepare three lemmas for our proof of this theorem, the first of which
is well-known, i.e., arithmetic means dominate geometric means, and geometric means
dominate harmonic means.
Lemma 1: For any n positive numbers ai, a 2 ,. 5 U"nj

«i + a2 H h an n
n
n 1/ai + l/a 2 H + l/an
Equality holds in each inequahty if and only if
Proof. As the proof of the inequahty on the left side is well known, see [3; p. 52-59]
for example, we will prove only the inequality on the right side. By the arithmetic-mean-
geometric-mean inequality, we obtain

1/ Qi + 1/fla + - ' + l/an >^ J 1 1

n "-" Y a i a 2 •" • #n ^/aia2 • * • an
Hence we have the desired result. D
L e m m a 2: Keeping x > 0 fixed, we let a i , . . . , a n vary under the condition that
ai ^ 1_ a ^ = 5 ( a i < . . . < a n ). In this situation, f(x) becomes minimum precisely when
ax = • • • = an = s/n.
Proof. Let ^(a;) = YZ=i x/(ai + #)• Then f(x) = #(:r) + 1 — n. In fact,

g{x) + l-n = ±(-JL--i)+l=±-f£- + l=f{z).

Therefore all we have to do is to show that g(x) becomes minimum when ax = • •
Because the arithmetic mean dominates the harmonic mean (Lemma 1), we have

\ I ^ =. :
n2
: —
xn2
:
* ( •
i^a,i + x ~~ (ai H h a n + nx)/x s + nx
which is the value of g(x) when a^ = • • • = a n , and equality holds if and only if a\ = • • • =
a n . Therefore #(#) is larger than the right hand member, provided that a\ < • • • < an.
This completes the proof of Lemma 2. Q
 63

L e m m a 3 : The sum of the roots of a monic polynomial xn+an-ixn~1+- • -+a1x+a0

is - a n - ! .
The proof is left to the reader. (A high school student might want to go over this
point with his or her teacher.)
P r o o f of t h e T h e o r e m : f(x) is monotone increasing and continuous at points
other t h a n x = —at- (i = 1, • • •, rc), and f(x) varies from —oo to oo on (—at-, —a t _i), where
i = 2, • • •, n. Moreover, / ( 0 ) = 1 — n < 0, and since f(x) —» 1 as x --» +oo, the function
f(x) has a zero point on each of the open intervals (—af-, —a f -i) and (0, oo).
We are now left with the problem of determining the positive zero point A. Let x
be the fixed value x = s(n — l ) / n . By Lemma 2, the minimum value for f(x) occurs with
a x = a 2 = • • • = a n ? which in this case yields f(s(n — l)/n) = 0 (each at- = s/n). If not all
of the a,- coincide, then f(s(n — l ) / n ) > 0, again by Lemma 2, and the positive zero point
A is strictly less t h a n s(n — l ) / n .
It remains to be proved that A > s — an. To do this, we multiply f(x) by n£=iO c + a«)
to obtain a polynomial <jr(x) of degree n. Observe that q(a) = 0 if and only if f(a) = 0,
and so q has roots c*i, • • •, a n _i, A. Order the roots so that a± < • • • < a n _ i < 0 < A, where
—an_t*+i < af- < — an_t- for each i = 1,2, • • •, n — 1. In the polynomial g(x), the coefficient
on the term x71"1 is
(ai + h a n ) - (ai + • • • + a n ) = 0
and so the sum of the roots of q(x) is zero (by Lemma 3). Thus,
n—1 n—1

Hence, our theorem is proved. O

C o r o l l a r y : f(x) = 1 — Y2=i^/(x + 0 ^ a s a z e r o P°i n1 t o n each of (—£,—i + 1)
(i = 2, • • •, n), and just one such point on (n(n — l ) / 2 , (n 2 — l ) / 2 ) .

3. FURTHER PROBLEMS Finally, let us propose two problems pertaining

to the theorem stated above.
P r o b l e m 1: Can we find an interval containing the positive zero point of the
function f(x) (in the theorem stated above) more precisely (by elementary methods)?
P r o b l e m 2: Investigate how the zero-points of the function

/W = i - Exr r ar («.->0A->0)
Si + «
are distributed.

A c k n o w l e d g e m e n t : The author would like to thank Mr. Isao Ashiba, the editor
and the referee for their helpful suggestions.
64

References:
[1] E.W. Swokowski, Functions and Graphs, 3rd edition, Prindle, Weber & Schmidt,
Boston, Massachusetts, 1980.
[2] G.H. Hardy, A Course of Pure Mathematics, Cambridge University Press, 1967.
[3] E. Beckenbach and R. Bellman, An Introduction to Inequalities, Random House and
the L.W. Singer Company, 1961.
[4] R.C. Thompson, Elementary Modern Algebra, Scott, Foresmanand Company, Glenview,
Illinois, 1974.

Fujishiro High School

Fujishiro, Ibaraki
300-15, Japan

THE OLYMPIAD CORNER

No. 153
R.E. WOODROW

All communications about this column should be sent to Professor R.E. Woodrow,
Department of Mathematics and Statistics, The University of Calgary, Calgary, Alberta,
Canada, T2N 1N4*

The Olympiad Contest we give this issue is the 9th Balkan Mathematical Olympiad,
under the auspices of the Greek Mathematical Society. My thanks go to Georg Gunther,
Sir Wilfred Grenfell College, Corner Brook, who collected this contest, and many others,
when he was team leader at the Moscow I.M.O.

9th BALKAN MATHEMATICAL OLYMPIAD

Athens, Hellas, May 4-9, 1992

1 . (Bulgaria). Let m and n be positive integers and

A(m, n) = m 34n+6 - m3*n - m 5 + ra3.

Find every n such that A(m, n) is divisible by 1992 for every ra.
2 . (Cyprus). Prove that for each positive integer n

(2n2 + 3n + l)n>6n(n\)2.
 65

3 . (Greece). Let ABC be a triangle and let D, E, F be points on sides B<7, C-A,
A 5 respectively (different from A, 5 , C). If AFDE is inscribable in a circle show that:

A(DEF) (EF\2
(ABC) - UDJ *
4* (Romania). For every integer n > 3 find the minimum positive integer /(n)
such that every subset of the set A = { 1 , 2 , 3 , . . . , n} which contains f(n) elements contains
elements x^y,z £ A which are pairwise relatively prime.
* * *

As a pre-Olympiad contest this month we give the problems of Part II of the Alberta
High School Mathematics Competition for 1994. My thanks go to the chairman of the
contest board, Alvin Baragar of The University of Alberta, for furnishing a copy.

ALBERTA H I G H SCHOOL M A T H E M A T I C S C O M P E T I T I O N , P A R T I I
February 8, 1994

1 . Find all polynomials P(x) that satisfy the equation

P(x2) + 2x2 + lOx = 2xP(x + 1) + 3.

2» An isosceles triangle is called an amoeba if it can be divided into two isosceles

triangles by a straight cut. How many different (i.e., not similar) amoebas are there?
3* (a) Show that there is a positive integer n so that the interval

((n+i^i) '(n+i^)2)
contains an integer JV.
(b) Find the smallest integer JV which is contained in such an interval for some n.
4 . ABCDE is a convex pentagon in the plane. Through each vertex draw a
straight line which cuts the pentagon into two parts of the same area. Prove that for some
vertex, the line through it must intersect the "opposite side" of the pentagon. (Here the
opposite side to vertex A is the side CD, the opposite side to B is DE, and so on.)
5 . Let a, &, c be real numbers. Their pairwise sums a + b^h+c and c+a are written
on three round cards and their pairwise products afe, be and ca are written on three square
cards. We call (a, 6, c) a tadpole if we can form three pairs of cards, each consisting of one
round card and one square card with the same number on both. An example of a tadpole
is (0,0,0).
(a) Find all possible tadpoles of the form (a, a, a).
66

(b) Prove that there is a tadpole that is not of the form (a, a, a). (You do not have to find
the actual values of a, 6, and c.)
* * *

We next start the solutions readers have submitted to problems from the 1993
numbers of the Corner. The first solutions are to some of the problems of the 1991-92
A.H.SM.C. Part JJ [1993: 4].
1 . The Committee to Halt Excessive Amount of Photocopying (CHEAP) is itself
accused of over-expenditure in photocopying, even though it never makes more than one
copy of anything. The new committee set up to investigate this accusation makes, for
each of its 13 members, a photocopy of everything CHEAP has photocopied, so that it
can study whether the expenditure has been justified. Each committee is charged 7 cents
per page for the first 2000 pages and 5 cents per page thereafter. It turns out that the
photocopying expenditure of the new committee is 10 times that of CHEAP. How many
photocopies did CHEAP make? Find all possible solutions.
Solution by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia.
Let Ai, A2 be, respectively, the number of photocopies made by CHEAP and the
new committee, and Pi, P2 be the respective costs (in cents). We have

A2 = 13AX (1)

P2 = lOPi (2)
There are three cases:
i) Ax > 2000.
Here Px = 2000 • 7 + (Ax - 2000) • 5 = 5AX + 4000, P 2 = 2000 • 7 + (A2 - 2000) • 5 =
65AX +4000, by (1). Now, (2) gives

65Ai + 4000 = 50AX + 40000 =* Ax = 2400.

ii) Ai < 2000 and A 2 > 2000.

Here Px = 7Ai, P 2 = 2000 + (A2 - 2000) • 5 = 65AX + 4000 so

65Ai + 4000 = 70A! =4> A x = 800.

iii) Ax < 2000, and A2 < 2000.

Here P 2 = 7A2 = 7 • 13AX = 13 • 7Ai = 13Pa > 10Pi giving no solutions.
Therefore CHEAP made either 800 or 2400 copies.
2 . The base of a tub is a square with sides of length 1 metre. It contains water 3
centimetres deep. A heavy rectangular block is placed in the tub three times. Each time,
the face that rests on the bottom of the tub has a different area. When this is done, the
water in the tub ends up being 4 centimetres, 5 centimetres and 6 centimetres deep. Find
the dimensions of the block.
 67

Solutions by John Morvay, Springfield, Missouri; and by Panos E. Tsaoussoglou,

Athens, Greece.
Let x < y < z be the length of the sides of the block. From the data Axy + 100 x
100 x 3 = 100 x 100 x 4 so xy = 25 x 102. Also 5xz + 100 x 100 x 3 = 100 x 100 x 5
so xz = 40 x 102. Finally 6yz + 100 x 100 x 3 = 100 x 100 x 6 so yz = 50 x 102. Thus
{xyzf = 25 x 40 x 50 x 106 = 5 x 1010. Thus xyz = 10 5 V^ This gives z = 40\/5,
y = 25^5, and x = 20\/5.

4» Suppose x, y and z are real numbers which satisfy the equation ax + by + cz = 0,

where a, b and c are given positive numbers.
(a) Prove that x2 + y2 + z2 > 2xy + 2yz + 2xz.
(b) Determine when equality holds in (a).
Solution by Panos E. Tsaoussoglou, Athens, Greece.
Now cz = — (ax + by).
(a) It is sufficient to prove

2 2 _ (ax + by)2 2xy - 2(x + y)(ax + by)

x ~r y — o ^ "j c ^> v.
c2 c
This gives
(a + cfx2 + 2[(a + c)(b + c) - 2c2]xj/ + (c+ bfy2 > 0.
Because (a + c)2 > 0 it is sufficient to show

A = [(a + c)(b + c) - 2c2]2 - (a + cf(c + b)2 < 0.

But

A = [(a + c)(b + c) - 2c2 - (a + c)(c + i)][(a + c)(b + c) - 2c2 + (a + c)(c + 6)]

= -2c 2 (a6 + ac + be) • 2 < 0.

(b) In case of equality

x2 + y2 + z2 - 2x(y + z)-2yz =Q (1)

then
x 2 - 2x(y + z) + (y- z)2 = 0.
Now Ai = (t/ + z) 2 - (y - z)2 = 4yz > 0.
So t/ and z are both nonpositive or both nonnegative. By symmetry all three of
a:, y, z are nonpositive or nonnegative and we may assume the latter. Factoring (1) gives
(v^+ ^/y + v^)(%/y + >fi- v^)(v^ + V* -y/v)(>fi + Vy - v ^ ) = o- This holds if
y ^ + y ^ = ^/J, o r % / J + v / ^ = v ^ o r V y + V ^ = : V^- ®* course if we also use the
condition ax + by + cz = 0, a, 6, c € R + we see that the only solution i s # = y = z = 0
since x^y^z >Q oi x,y,z < 0.
5* ABCD is a square piece of paper with sides of length 1 metre. A quarter-circle
is drawn from B to D with centre A. The piece of paper is folded along EF, with E on
68

AB and F on AD, so that A falls on the quarter-circle. Determine the maximum and
minimum areas that the triangle AEF could have.
Solution by Seung-Jin Bang, Albany, California; and by Panos E. Tsaoussoglou,
Athens, Greece. We give Tsaoussoglou's solution.
Let the perpendicular from A to EF meet
EF at P. Since folding along EF puts A on the
circle, AP is half the radius. Let EP = M, FP =
iV, A2£ = x and A F = y.
Then M2 = x2 - \, N2 = y2 - \ and the
area of AEF is given by

1(^-1/4 + ^-1/4).
Also a:2 + y2 = (M + N)2 = x2 - 1/4 + y2 - 1/4 + 2^/z 2 - 1/4^/z/2 - 1/4 giving
yx2 — 1/4AA/ 2 — 1/4 = 1/4. Applying the Arithmetic-Geometric mean inequality we get
that the minimum area is l / 2 y 1/4 = 1/4. This occurs when Jx2 — 1/4 = Jy2 — 1/4, i.e.
x = y and Jx2 — 1/4 = 1/4 so x = y = \/2/2.
For the maximum area take x = 1 (or y = 1). Then \/3/2 • y y2 — 1/4 = 1/4 and
y = \/3/3. In both cases x = 1, y = -\/3/3, a: = \/3/3, y = 1 the area is \/3/6.

The next solutions are to problems of the 1991 British Mathematical Olympiad
[1993: 4-5].
1 . Prove that the number
3 n + 2 x 17"
where n is a non-negative integer, is never a perfect square.
Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; Seung-
Jin Bang, Albany, California; J. Brenner, Palo Alto, California; Pavlos Maragoudakis,
Pireas, Greece; Stewart Metchette, Culver City, California; Waldemar Pompe, student,
University of Warsaw, Poland; Bob Prielipp, University of Wisconsin-Oshkosh; Dale
Shoults, student, The University of Calgary; D.J. Smeenk, Zaltbommel, The Netherlands;
Panos E. Tsaoussoglou, Athens, Greece; and Chris Wildhagen, Rotterdam, The Nether-
lands. We give Pompe's solution.
If n = 0, we have 3° + 2 x 17° = 3, and it is not a perfect square. If n > 1, we have:
1 7 = 1 mod 8 3 2 = 1 mod 8
172 = l m o d 8 and 32* = 1 mod 8
2 x 17n = 2 m o d 8 3 2 * + 1 = 3 mod 8
Hence
_ J 3 mod 8, if n is even
3 n + 2 x IT
-i 5 mod 8, if n is odd.
 69

But, if k is a positive integer

' 0 mod 8, if k is divisible by 4

*3 = < 4 mod 8, if fc is even but not divisible by 4
lmod8 if k is odd.

Therefore 3 n + 2 x 17 n cannot be a perfect square.

2« Find all positive integers k such that the polynomial x2**1 + x + 1 is divisible
by the polynomial xk + x + 1. For each such k specify the integers n such that xn + x + 1
is divisible by xk + x + 1.
Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by
Seung-Jin Bang, Albany, California; by Pavlos Maragoudakisf Pireas, Greece; by Dale
Shoults, student, The University of Calgary; and by Chris Wildhagen, Rotterdam, The
Netherlands. We give Bang's solution to a slightly more general problem.
We determine the ordered pairs (n, k) such that xn + x + 1 = (xk + x + l)p{x),
where p(x) is a nonzero polynomial in x.
It is clear that n > k and n = k is a trivial possibility. So now we assume n > &,
and let u be a root of the equation xk + x + 1 = 0 . Then u? ^ 0 and UJU + a? + 1 = 0 giving
un_uk = ^k ^n-k _ i ) = o. This implies that un~k = 1.
Since \u\n~"k = \u>n~k\ = 1 we have |u>| = 1 and from 1 = |u?|* = \cvk\ = \UJ + 1 | we
obtain that the real part of a;, Rea? = —1/2. Now we have k > 2 and that UJ = (—l+<s/3i)/2
or 44? = (—1 — y/3i)/2 where i = <*/—l. If a? is a multiple root of xk + x + 1 = 0 then
fcc^"1 + 1 = 0 and o^"""1 = —1/& contradicting Jc*^& x | = 1. It follows that a; is a simple
root. Since there are only two possibilities for or, & = 2. Now let n = / mod 3 where
0 < I < 3. From u;n + a; + 1 = <J + u + 1 = 0 we see that 1 = 2, and n = 2 mod 3.
It follows t h a t (n, fc) = (fc, fe) or (2m + 2,2), rn a positive integer.
Answer (i) k = 2, (ii) (n, fc) = (fc, fc) or (3m + 2,2).
3* ABCD is a quadrilateral inscribed in a circle of radius r. The diagonals AC,
BD meet at E. Prove that if AC is perpendicular to BD then

EA2 + EB2 + EC2 + ED2 = 4r 2 . (*)

Is it true that if (*) holds then AC is perpendicular to BD1 Give a reason for your answer.
Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by
Seung-Jin Bang, Albany, California; by Pavlos Maragoudakis, Pireas, Greece; by Dale
Shoults, student, The University of Calgary; by D.J. Smeenk, Zaltbommel, The Nether-
lands; and by Edward T.H. Wang, Wilfrid Laurier University, Waterloo, Ontario. We
give Ardila's solution.
Suppose AC ± BD. By Pythagoras

EA2 + EB2 + EC2 + ED2 = ~ ( A B 2 + BC2 + CD2 + DA2).

Let O be the circumcentre of ABCD. By the law of cosines

70

AD2 = AO2 + OD2 - 2AO • OD cos LAOD = 2r 2 - 2r 2 cos ZA0£>, etc.

So we have

EA2 + EB2 + EC2 + ££> 2 = 4r 2 - r 2 (cos ZAOD + cos LDOC + cos ZCOB + cos LBOA).

But ZAOZ> + LBOC =AD + BC= 2(LACD + LBDC) = 2(90°) = 180°. So cos ZA0I> +
cos LBOC = cos LDOC + cos ZBOA = 0 and (*) follows.
AC may not be perpendicular to BD even when (*) holds. For example, let ABCD
be a rectangle that is not a square. Then E is the centre of the circle and EA = EB =
EC = ED = r, so (*) holds, but AC / BD.
4 . Find, with proof, the minimum value of (x + y)(y + z) where x, y, z are positive
real numbers satisfying the condition xyz(x + y + z) = 1.
Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by
Seung-Jin Bang, Albany, California; by Pavlos Maragoudakis, Pireas, Greece; by Waldemar
Pompe, student, University of Warsaw, Poland; by Henry Ricardo, Tappan, New York; by
Dale Shoults, student, The University of Calgary; by Panos E. Tsaoussoglou, Athens,
Greece; and by Edward TM. Wang, Wilfrid Laurier University, Waterloo, Ontario, We
give Pompe's two solutions.
S o l u t i o n L Consider a triangle ABC with sides a = y + z,b = x + z, c = x + y
respectively. (Such a triangle of course exists because a + b> c, etc.) Then s = x + y + z,
5 — a = x, 5 — 6 = y, and s — c = z. According to Heron's formula

Area = ys(s — a)(s — b)(s — c) = yjxyz(x + y + z) = 1.

But on the other hand

Area = - a c s i n B = ~(x + y)(y + z) sinl?.

Thus
(x + y)(y + z) = ^->2.
smi)
Equality holds, for example, when x = z = 1 and y = \ / 2 — 1. Therefore 2 is the desired
minimum value.
Solution I I . We will use the Arithmetic Mean-Geometric Mean inequality. We
have

Equality holds, for example, when x = z = 1 and y = y/2 — 1, giving 2 as the minimum
value.
5 . Find the number of permutations (arrangements) PiiP2,P3,P4iP5<>P6 of 1, 2, 3,
4, 5, 6 with the property: for no integer n, 1 < n < 5, do |?i,2?2? • • • ,Pn form a permutation
of 1 , 2 , . . . , n .
 71

Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by

Dale Shoults, student, The University of Calgary; by Edward T.H. Wang, Wilfrid Laurier
University, Waterloo, Ontario; and by Chris Wildhagen, Rotterdam, The Netherlands. We
give Wildhagen's solution.
Let N be any integer greater than 1. For 1 < n < N, let an be the number of
permutations P i , P 2 , • • • ? Pn of 1,2,...,ra with the property:
For no integer k with I < k <n — l d o P i P 2 . . . P* form a permutation of 1 , 2 , . . . , fc.
Then it is clear that the following recurrence relation holds:

M = X>n(JV-n)!
n=l

Since a\ = 1 it follows that a2 = 1, a 3 = 3, a 4 = 13, a 5 = 71 and a6 = 461.

6W Show that if x and y are positive integers such that x2 + y2 — x is divisible by

2xy then # is a perfect square.

Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by

Seung-Jin Bang, Albany, California; by Stewart Metchette, Culver City, California; by
Waldemar Pompe, student, University of Warsaw, Poland; by Dale Shoults, student, The
University of Calgary; by Panos E. Tsaoussoglou, Athens, Greece; and by Chris Wildhagen,
Rotterdam, The Netherlands. We first give Ardila's solution.
Let x2 + y2 — x = k(2xy), so y2 = x(2ky + 1 — x). Now let x = arc2, where a is
square free. It follows that for some integer m

2ky + 1 — x = am2 and y2 = a2n2m2.

Since a is square free a\y. Write y = ab. Then 1 — x = am2 — 2ky = am2 — 2kab =
a(rn 2 — 2kb) and as x = an 2 so 1 = a(n 2 + m 2 — 2kb). Thus a | l so x = n 2 , as we wished
to prove.

[Editor's note. Next we give Wildhagen's generalization and solution]

We show that 2xy can be replaced by xy. Suppose that xy\N where JV = x2+y2—x,
and i , y £ IN. Let p be any prime factor of x, and let pk\\x, (i.e. pk\x and p f c + 1 / # ) . We
want to show that k is even. Let pl\\y. Clearly I > 1. Then / + / | | x y , hence / + / | i V .
The obvious relations Pk\\x2 — x and p&'fl|iV" imply j?fc(|y2 so k = 21, an even number.

7® A ladder of length I rests against a vertical wall. Suppose that there is a rung on
the ladder which has the same distance d from both the wall and the (horizontal) ground.
Find explicitly, in terms of I and d, the height h from the ground that the ladder reaches
up the wall.

Solutions by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia; by

Seung-Jin Bang, Albany, California; by Dale Shoults, student, The University of Calgary,
Alberta; and by Chris Wildhagen, Rotterdam, The Netherlands. We use Shoults'solution.
72

For each possible pair of values of / and d, both

hx and h2 are possibilities for h. They are distinct unless
the angle the ladder makes with the wall is 45°.
Since the area of the triangle B plus the area of
triangle T plus the area of the square S equal the area
of the large triangle

d(h2 — d) d{hx — d) a hih2

+ + d =
—2 —2 —
This simplifies to give hih2 = d(hi + h2) or hi + h2 = hih2/d.
Squaring 1 and rearranging gives a)
hl + hl={-!^-2(h1h2). (2)

Applying Pythagoras to the large right triangle, I2 = hi + h\. Combining with (2) gives

,2=(M^_2(M2) (3)

Using the quadratic formula on (3) to solve for h\h2 gives h\h2 = d? + dy/d2 + Z2, (4)
the positive sign is needed because y/d? +12 > d and the negative sign would make
hih2 < 0. Set m = d + y/d2 + I2. We have hih2 = dm. (5)
Combining (1) and (5) we have hi + hm. Multiplying by h2, substituting dm for hih2 and
rearranging gives h\ — mh2 + dm = 0.
Solving this for h2, replacing m in terms of d and /, and simplifying gives the two
possible values of h.

h = I \d+ Vd2 + I2 ± yjl2 - 2d2 - 2dVd2 + l2\ .

* * *

Next some misprints that crept into his solutions and were spotted by Seung-Jin
Bang, Albany, California.
In [1993: 6], 23rd line (solution of #5) uf(n + 2) - 2 n + 3 /(n + 1) = .. .* should read
"f(n + 2) - 22n+3f(n + 1) = ..."
In [1993: 9] 16th line (solution of #10) "the number of elements in X is m" should
read "the number of elements in X is 4m".
* * *

We conclude this number of the Corner with an elegant alternative solution to a

problem discussed in the January 1993 number of the Corner.
8 . [1991: 197; 1993: 8-9] Proposed by Ireland.
Let ABC be a triangle and t the line through C parallel to the side AB. Let the
internal bisector of the angle at A meet the side BC at D and the line t at E. Let the
 73

internal bisector of the angle at B meet the side AC at F and the line i at G. If GF = DE
prove that AC = BC.
Solution by K.R.S. Sastry, Addis Ababa, Ethiopia.
If AC ^ BC then suppose AC > BC.
Then 0 > or, /3/2 > or/2 and referring to the figure,
we have AI > BI, IE > IG, whence ID > IF.
Thus AI + ID > BI + IF, or AD > BF. (1)
Now AABD and AECD are similar, so

BD c _AD
DC b~ DE'
Also AAFB ~ ACFG giving AF/FC = c/a =
BF/FG. Thus
AD FG__c £
D£ X
£F " 6 X
c or B F =
6 < 1

This yields AD < BF, contradicting (1). The supposition AC < BC also leads to a
contradiction. We conclude AC = BC.
Remark. This is a variant of the famous Steiner-Lehmus (also infamous! — for the
inumerable wrong proofs) theorem: If two internal angle bisectors of a triangle are equal,
then the triangle is isosceles.

That concludes this number. Send me your Olympiad and pre-Olympiad contests
as well as your nice solutions.

PROBLEMS
Problem proposals and solutions should be sent to B. Sands, Department of
Mathematics and Statistics, University of Calgary, Calgary, Alberta, Canada T2N IN4.
Proposals should, whenever possible, be accompanied by a solution, references, and other
insights which are likely to be of help to the editor. An asterisk (*) after a number indicates
a problem submitted without a solution.
Original problems are particularly sought. But other interesting problems may also
be acceptable provided they are not too well known and references are given as to their
provenance. Ordinarily, if the originator of a problem can be located, it should not be
submitted by somebody else without permission.
To facilitate their consideration, your solutions, typewritten or neatly handwritten
on signed, separate sheets, should preferably be mailed to the editor before O c t o b e r 1,
1994, although solutions received after that date will also be considered until the time when
a solution is published.
74

1921. Proposed by Toshio Seimiya, Kawasaki, Japan.

D and E are points on sides AB and AC of a triangle ABC such that DE\\BC, and
P is an interior point of AADE. PB and PC meet DE at i*1 and G respectively. Let 0%
and O2 be the circumcenters of APDG and APFE respectively. Prove that AP ± O1O2.
1922. Proposed by Marcin E. Kuczma, Warszawa, Poland.
The function / is defined on nonnegative integers by: /(0) = 0 and

f(2n + 1) = 2/(n) for n > 0, /(2n) = 2/(n) + 1 for n > 1.

(a) Let g(n) = / ( / ( n ) ) . Show that </(n — #(n)) = 0 for all n > 0.
(b) For any n > 1, let r(n) be the least integer r such that / r ( n ) = 0 (where
/*(») = / ( / ( » ) ) , / 3 ( " ) = / ( / 2 ( n ) ) , etc.). Compute

liminf —7-7 .

1923. Proposed by K.R.S. Sastry, Addis Ababa, Ethiopia.

In triangle ABC, cevians AD, BE, CF are equal and concur at point P. Prove that

PA + PB + PC = 2(PD + PE + PF).

1924. Proposed by Jisho Kotani, Akita, Japan.

A large sphere of radius 1 and a smaller sphere of radius r < 1 overlap so that their
intersection is a circle of radius r, i.e., a great circle of the small sphere. Find r so that
the volume inside the small sphere and outside the large sphere is as large as possible.
1925. Proposed by Ignotus, Godella, Spain.
Let n be a fc-digit positive integer and let v(n) be the set of k "right-truncations"
of n: for example, v(1994) = {1994,199,19,1}. Show that there are infinitely many n
such that v(n) is a complete set of residues mod k. (This problem was inspired by Crux
problems 1884 and 1886 [1993: 264, 265].)
1 9 2 6 * Proposed by Waldemar Pompe, student, University of Warsaw, Poland.
On sides BC, CA, AB of A ABC are chosen points Ax, Bi, G\ respectively, such
that AA\B\C\ is equilateral. Let o\, 02, 03 and 0\, 0<i, O3 be respectively the incircles
and incentres of triangles AC\Bi, BA1C1, CB\A\. If 0\C\ = O2C1, show that
(a) BtOs = Btd and Ax02 = At03;
(b) three external common tangents to the pairs of circles oi,o 2 ; 02,03; o3,oi,
different from the sides of A ABC, have a common point.
1 9 2 7 . Proposed by Rolf Kline, Edmonton, Alberta.
Suppose that, for three consecutive years, a certain provincial government reduces
what it spends annually on education. The percentage decreases year by year are a, b and c
percent, where a, b, c are positive integers in arithmetic progression. Suppose also that the
amounts (in dollars) the government spends on education during these same three years
are three positive integers in harmonic progression. Find a, b and c.
 75

1 9 2 8 * Proposed by Herbert Gulicher, Westfalische Wilhelms-Universitat, Miinster,

Germany.
In the tetrahedron A1A2A&A4, not necessarily regular, let a,- be the triangular face
opposite vertex At- (i = 1,2,3,4). Let Q be any point in the interior of at and P a point
on the segment A±Q. For i = 2,3,4 let Bi be the point where the plane through P parallel
to at- meets the edge AiA{. Prove that

AXB2 AtB3 ; AtB4 = AtP

AtA2 A1A3 A1A4 AXQ '

1929* Proposed by David Doster, Choate Rosemary Hall, Wallingford,

Connecticut.
Define a sequence a i , a2, #3, • . . by a\ = 6 and

an+l = [^n + jVan - 12J

for all n > 1, where [ J denotes the greatest integer function. Prove that an = 1 mod 10
for all n > 2.
1930. Proposed by Vaclav Konecny, Ferris State University, Big Rapids,
Michigan.
T\ is an isosceles triangle with circumcircle K. Let T2 be another isosceles triangle
inscribed in K whose base is one of the equal sides of T\ and which overlaps the interior of
T\. Similarly create isosceles triangles T 3 from T 2 , T 4 from T3, and so on. Do the triangles
Tn approach an equilateral triangle as n —» 00?

SOLUTIONS
No problem is ever permanently closed. The editor will always be pleased to consider
for publication new solutions or new insights on past problems.

1 8 2 6 . [1993: 78] Proposed by P. Penning, Delft, The Netherlands.

(a) In a box we put two marbles, one black and one white. We choose one marble
at random. If it is white, we put it back in the box, add an extra white marble to the
box, shake the box, and draw again, continuing to replace the marble along with an extra
white marble every time a white marble is chosen, until the black marble is chosen and
the game ends. W h a t is the average number of marbles chosen?
(b) W h a t is the average number of marbles chosen if we add an extra white marble
only after every second white marble that is chosen?
76

Solution by Marcin E. Kuczma, Warszawa, Poland,

Let the random variable X denote the number of marbles drawn, and let E(X)
denote the expected value of X. Then in case (a),

iMX-»)-I|?
2 34
- n1 J
n+1 n(n +1) '
while in case (b),

\ / o O O O A ,
223344 jfe k fc + 1 fc2(fc + l)
and
_ / T r 07N 112233 Jb-lJb-1 k 1 1
Pr{X = 2k) = 2 2 3 3------
44 Jb k k + lk + 1 k(k + l)2
Hence in case (a),
E(X) = £ ) nPr(X = ») = f ) - i - = oo,
n l
n=l n=l "*"

while in case (b),

oo oo
E(X) = ^2(2k - l)Pr(X = 2k - 1) + £ 2&Pr(X = 2Jfe)
fc=i Jt=i
2k 1 2 3 2
=V ~ +V = V* ( _ 1"\ + V
£i *2(*+1) h (*+1)2 £i w + 1 ) *v ei (*+1)2
£jU fc + J jfe*2 tAk + 1)2

using the known fact that X J £ = I ( 1 / ^ 2 ) = ?r2/6. [J^difor^ no£e: for example, see the recent
article "6/7T2" by Gareth Jones, in the December 1993 Mathematics Magazine, pages 290-
298, especially §5.]
Also solved by FEDERICO ARDILA, student, Colegio San Carlos, Bogota,
Colombia; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, TJX.; JORDI DOU,
Barcelona, Spain; KEITH EKBLAW, Walla Walla Community College, Walla Walla,
Washington (with only an approximate value given for part (h)); HANS ENGELHAUPT,
Franz-Ludwig-Gymnasium, Bamberg, Germany; RICHARD I. HESS, Rancho Palos Verdes,
California; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; BEATRIZ
MARGOLIS, Paris, France; R.P. SEALY, Mount Allison University, Sackville, New
Brunswick; and the proposer.
Both Hess and Janous proved the more general result that if an extra white marble
is added after every kth white marble is drawn, then E(X) = 1 + ((k) where ((z) =
J2^L 1 (l/n 2 ) is the Riemann Zeta function.
 77

* * * * 3fC

1 8 2 8 . [1993: 78] Proposed by T. W. O. Richards, Cheddar Gorge, Great Britain.

In the last century, the English mathematician Arthur Cayley introduced a per-
mutation problem, loosely based on the card game Treize, which he called M o u s e t r a p .
Suppose t h a t the numbers 1 , 2 , . . . , n are written on n cards, one on each card. After
shuffling (permuting) the cards, start counting the deck from the top card down. If the
number on the card does not equal the count, then put that card at the bottom of the deck
and continue counting. H the two are equal then put the card aside and start counting
again from 1.
Let's say the game is w o n if all the cards have been put aside. In this case, form
a new deck with the cards in the order in which they were set aside and play a new game
with this deck. For example, if we start with n = 5 cards in the order 25143, we win:
25143 -» 3251 -» 3251 - • 513 -> 513 -> 51 - * 51 -» 1
and the new deck is 42351, which wins again:
42351 -> 3514 -» 351 - * 13 -» 3
but now our deck, 24513, puts aside no cards at all. Is there an arrangement (using more
cards, if necessary) which will give you three or more consecutive wins?
Solution by Richard L Hess, Rancho Palos Verdes, California,
W i t h one or two cards there is the repetitive arrangement:

1-*1-*1... ; 12->12->12... .

W i t h six cards, there is just one arrangement:

165342 -> 132564 -* 125346 -> 136524,.

With eight cards, again there is just one arrangement:

52173846 -> 21463578 -> 72135648 -» 21435867.

With nine cards, there are 8 arrangements:

157362948; 157392486; 231765489; 469523718;

523149768; 543987261; 684523719; 756482319.

Also solved by CHARLES ASHBACHER, Cedar Rapids, Iowa; P. PENNING, Delft,

The Netherlands; A.N. T WOORD, Eindhoven University of Technology, Eindhoven, The
Netherlands; and the proposers. There was one incorrect solution sent in, probably due to
misunderstanding the problem.
No theoretical approach was found either by the solvers or by the proposers. All
the solutions were done by trial and error or by computer, 't Woord poses the question:
is there some n such that there are no permutations (on at least 6 elements) that give n
successive wins? Penning found the following frequencies for the number of arrangements
that gave 0,1,2 or 3 wins for n = 3 , 4 , 5 , 6 , 7 , 8 ;
78

n 0 wins 1 win 2 wins 3 wins total

~F 4~ 2~ 6~
4 18 4 2 24
5 105 14 1 120
6 636 72 11 1 720
7 4710 316 14 5040
8 38508 1730 81 1 40320

1 8 3 1 * . [1993: 112] Proposed by Walther Janous, Ursulinengymnasium, Innsbruck,

Austria.
Let a;, y, z be any real numbers and let A be an odd positive integer. Prove or
disprove that
x(x + y)x + y(y + z)x + z(z + x)x > 0.
Solution by Marcin E. Kuczma, Warszawa, Poland.
The inequality is trivial for A = 1, easy for A = 3, difficult for A = 5, false for A > 7.
Counterexample for A > 7:
x = —2, y = —1, z = 5.
The expression takes value 7 • 3A — 4A, which is negative for A > 7.
Now the proof for A = 3 and A = 5.
The nonsingular linear substitution u = y + zJv = z + x,w = x + y transforms the
claimed inequality into
(t; + w — u)wx + (w + u — v)ux + (u + v — w)vx > 0. (1)
Since the expression in (1) is cyclic and invariant under (u, t;, w) ^ (—u, —v, —to), there is
no loss of generality in assuming u > 0, v > 0.
Case w > 0. Let us regroup the terms in (1) as follows:
(u A+1 + uvx - uxv) + (v A+1 + vwx - vAw) + (w A+1 + timA - w A u).
Either we have uxv < izA+1 or uxv < uvx (according as u > v or u < v). Hence the first
expression in parentheses is nonnegative. So are the other two, by cyclicity.
[Incidentally, we have shown that the proposed inequality is valid for every exponent
A > 1 when all three sums u = y + z,v = z + xJw = x + y are nonnegative (or nonpositive,
with A an odd integer). Value 0 is attained when all three expressions in parentheses are
zero; and this is easily seen to be the case only for u = v = w = 0, i.e., x = y = z = 0.]
Case w < 0. We will show that strict inequality holds in (1) (for A = 3 and A = 5).
By homogeneity, it is enough to consider w = —1. Denote the resulting polynomial by
F(u,v) resp. G(u, v) for A = 5 resp. A = 3:
F(u,v) = v6 + \u + l)vs - (u5 + l)v + (u6 - u5 + u + 1),
G(u, v) = u4 - (v + l)u3 + (v3 + l)u + (v4 + v3 - v + 1)
(the specific grouping of terms in either case has its aim).
 79

We first consider G(u,v). Since d2G/du2 = 6u(2u — v — 1), we see t h a t , for any
fixed v > 0, the first order derivative dG/du is a decreasing function of u G [0, (v + l)/2]
and an increasing function of u G [(v + l ) / 2 , oo). Thus dG/du (regarded as a function of
u, for v fixed) assumes at u = (u + l ) / 2 its (strict) minimum value

Consequently dG/du > 0 for u G [0, oo) and hence (7(u, v) is increasing in u. So

nt \^nrt\ \ 4, 3 , i / > v3-v > 0 fori;>l,

G(u, v) > G(0, t;) = t r + t r — v + 1 < '
l>-v + l>0 for0<v<l.
Thus G(Uj v) > 0 for u, t; > 0 is proved.
To deal with F ( u , v), we will need a lemma.
Lemma. Let <p(t) be a differentiable convex function in an interval J and let
a, /? G J be such that a < 0,

p'(a) < 0 < ^ ) , V(a) + (/? - a ) ^ ( a ) > 0. (2)

Then y(t) > 0 in J .

Proof of the lemma. From the first condition of (2) we infer t h a t the global minimum
of (p over J occurs at some point between a and /?. The second condition of (2) says that
the tangent to the graph of tp at a is above the t-axis at /?, hence also at every point of
[a, /?] (the slope <pf(a) being negative). Every point of the graph of <p lies above or on that
tangent; the lemma results. D
Now fix u > 0 arbitrarily and note that F ( u , t;) is a convex function of variable
v G J = [0, oo). Thus, to prove F ( u , v) > 0, it will be enough to find a and ft (depending
on u) so as to satisfy the conditions (2) of the lemma (for tp{v) = F(u, v)). Our choice will
be the following:

if 0 < i * < 9 / 8 , take a = 1/2, 0 = 1; if u > 9/8, take a = u / 2 , /3 = 3u/5.

So we must verify

Standard calculation brings these six inequalities to the polynomial form:

9
-16u5 + 5 u - 8 < 0 < - u 5
+ 5u + 10, 6 4 w 6 - 1 2 8 u 5 + 76u + 19 > 0
for 0 < u < - , (3)
o
9
-8u 5 +5ti 4 -16 < 0 < 358u 5 +2025u 4 -3125, 159u 6 -30Otx 5 +l28^+320 > 0 for u > - . (4)
80

In (3), the second and the third inequalities are true at u = 0 and at u = 9/8; and
since the functions involved are concave in [0,9/8], these two inequalities are satisfied in
that interval.
The first inequality in (3) and the first two inequalities in (4) hold trivially.
To prove the third inequality of (4), we again use the lemma. Denoting the poly-
nomial 159u 6 -300u 5 +128u+320 by tp(u) we compute cp"(u) = 30u3(159u - 200); so <p(u)
is convex in the interval [200/159, oo). Choosing a = 3/2 and ft = 8/5 we can easily verify
conditions (2). Therefore ip{u) > 0 for u > 200/159; in particular, <£>(200/159) > 0. And
since y>(0) > 0 and (p is concave in [0,200/159], we conclude y>{u) > 0 for all u > 0. This
seems to complete the proof of F(u, v) > 0 for u, v > 0.
Thus, for A = 3 and A = 5 the proposed inequality is true for every triple of real
numbers x, y, z, with equality only for x = y = z = 0.
Counterexamples for A > 7 were also found by CHRISTOPHER J. BRADLEY,
Clifton College, Bristol, U.K.; RICHARD I. HESS, Rancho Palos Verdes, California; and
A. N. 'T WOORD, Eindhoven University of Technology, Eindhoven, The Netherlands.
There was one incorrect solution sent in.

1 8 3 2 . [1993: 112] Proposed by K.R.S. Sastry, Addis Ababa, Ethiopia.

An old unsolved problem is: "is it possible that a box can have its sides, face
diagonals, and space diagonal all of integer lengths, i.e., are there positive integers a, 6, c
such that
a2 + 62, 62 + c2, c2 + a2, and a2 + b2 + c2
are all perfect squares?" What if we replace the squares by triangular numbers? For n a
positive integer, let tn = n(n + l)/2 be the nth triangular number,
(a) Find positive integers a,b, c such that

ta + hi h + tCj tc + ta

are all triangular numbers.

(b)* Is there such a solution so that ta + £& + tc is also a triangular number?
Solution to both parts by Richard I. Hess, Rancho Palos Verdes, California.
A small computer search found the cases in the table below, where ta + <^, ta + tCJ
U + tc and ta + th + tc are all triangular.

a b c ta+U ta+tc h+h ta + h + te

11 14 14 171 = t 18 171 = *18 210 = <2o 276 = t 23
230 741 870 301476 = *7?6 405450 = igoo 653796 = t1143 680361 = tU66
609 779 923 489555 = t989 612171 = tuos 730236 = *i208 915981 = *i353
714 798 989 574056 = tun 744810 = *i22o 808356 = 11271 1063611 = fuss
1224 1716 3219 2222886 = £2108 5932290 = *3444 6655776 = £3648 7405476 = *3848
 81

Both parts also solved by CHARLES ASHBACHER, Cedar Rapids, Iowa; HANS
ENGELHAUPT, Franz-Ludwig-Gymnasium, Bamberg, Germany; JOHN G. HEUVER,
Grande Prairie Composite H.S., Grande Prairie, Alberta; and R.P. SEALY, Mount
Allison University, Sackville, New Brunswick. Part (a) solved by FEDERICO ARDILA,
student, Colegio San Carlos, Bogota, Colombia; CHRISTOPHER J. BRADLEY, Clifton
College, Bristol, U.K.; J.A. MCCALLUM, Medicine Hat, Alberta; P. PENNING, Delft,
The Netherlands; CHRIS WILDHAGEN, Rotterdam, The Netherlands; and the proposer.
One incorrect solution was received.
Can anyone find an infinite family of solutions?

1 8 3 3 . [1993: 112] Proposed by Toshio Seimiya, Kawasaki, Japan.

E and F are points on sides BC and AD, respectively, of a quadrilateral ABCD.
Let P = AE n BF and Q = CF (~l DE. Prove that E and F divide BC and AD (or BC
and DA) in the same ratio if and only if
[FPQ] [EQP]
[EDA] [FBC] '
where [XYZ] denotes the area of triangle XYZ.
Solution by Federico Ardila, student, Colegio San Carlos, Bogota, Colombia.
Let « i = LEAD, a2 = lEDA,
fa = LFBC, fa = IFCB, ji = IEPF,
72 = LEQF as in the figure. Applying the
law of sines we obtain
FP AF FQ FD
sin«i sin7i sin 0:2 sin 72
(1)
EP EB EQ EC
sin/?i sin 71 sin fa sin 72
^•^•^A
Letting F = LPFQ and E = LPEQ, C D

[FPQ] [EQP] FP •FQ sin F EP •EQ sin E

(2)
[EDA] [FBC] ED • ISA sinE FB • F C s i n F '
Using (1) together with
ED AD EA FB BC FC
sin ot\ sin E sina2 ' sin/? 2 sinF sin/?i
we obtain that (2) is equivalent to
AF sin c*i FD sin a 2 EB sin fa EC sin fa
sinF sinE
sm7i sin 72 — sm7i sin 72
AD sin a i AD sin a 2 BC sin fa BC sin fa
sinE •sinF
sinJE sinE sinF sinF
82

Letting AF/AD = A and BE/BC = fi and cancelling the sines we see that this equality-
is equivalent to A(l — A) = fi(l — fi), or
(JI - A)(JI + A - 1) = /x2 - A2 - fi + A = 0,
and this occurs if and only if /z = A or \i = 1 — A, i.e.,
AF^_BE_ AF _ CE
r
AD " BC ° A£> " BC '
which we may interpret as E and F dividing BC and AD respectively in the same ratio,
as we wished to prove.
Also solved by MARGIN E. KUCZMA, Warszawa, Poland; and the proposer.

1 8 3 4 . [1993: 113] Proposed by Marcin E. Kuczma, Warszawa, Poland.

Given positive numbers A, G and H, show that they are respectively the arithmetic,
geometric and harmonic means of some three positive numbers x, y, z if and only if
A3 G3 , ^ 3 / A\2

Solution by Hans Engelhaupt, Franz-Ludwig-Gymnasium, Bamberg, Germany,

If
1 1 1 3
x + y + z = 3A, xyz = G3 , and -H h - = 77 , (1)
x y z H
then
3G3
#2/ + 3/2 + *s = - = - ,
so (1) holds if and only \ix,y,z are the solutions of

There are only positive solutions of (2), because A,G,H > 0. Rewrite (2) as

(v-A)3+(^-ZA*yv-A) + A(^-3AA+As-G* = 0-1

then with the substitution v = u + A one gets

AU solutions of the equation u3 — pu + q = 0 are real numbers if and only if the condition
q2/A < p3/27 holds, so it follows that (1) is equivalent to
2 / s~i*\ 3

H^-D-(*-?)*•
 83

which (on multiplying out) is equivalent to

A3 G3 , ^ 3 / A\2

Also solved by FEDERICO ARDILA, student, Colegio San Carlos, Bogota,

Colombia; CHRISTOPHER J. BRADLEY, Clifton College, Bristol, U.K.; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; MURRAY S. KLAMKIN, Univer-
sity of Alberta; KEE-WAILAU, Hong Kong; FRANCISCO LUIS ROCHA PIMENTEL,
Fortaleza, Brazil; and the proposer,
Klamkin points out problem 4248 of School Science and Mathematics, solution on
pages 335-336 of the November 1991 issue. This problem is to show that

('-DM-&) ('-£)•
which is equivalent to (3).

1 8 3 5 . [1993: 113] Proposed by Joaquin Gomez Rey, LB. Luis Bunuel, Alcorcon,
Madrid, Spain.
Evaluate
1
for n = 1,2,3,... .
s^Ofc )
Solution by Richard Mcintosh, University of Regina.
For any function / ( # ) , define the difference operator A by Af(x) = f(x +1) — f(x).
Then it can be found in every book on finite differences that

An/(*) = EM)""* (" )/(* + *), n = 0,1,2,

[here as usual Anf(x) = A(An"1f(x)) is defined recursively, -—Ed.]. Iif(x) is a polynomial

of degree n — 1, then

o=A"/(o) = E(-ir fc Q/w.

In particular, for

_ frik - l \ _ (nk - l)(nk - 2) • • • (nk - n + 1)

« * >- - \• n- - \") -- (»-l)l
we get

£<-)-'(:) (!-;)=»•
84

Therefore
t(_l)(_.2)...(-n + l)
1
tf-^ffld-iO-^K-- !)-^ (n-1)!
= 1.

Also solved by FEDERICO ARDILA, student, Colegio San Carlos, Bogota,

Colombia; FRANCISCO BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain;
WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria; A. N. T WOORD, Eind-
hoven University of Technology, Eindhoven, The Netherlands; CHRIS WILDHAGEN, Rot-
terdam, The Netherlands; and the proposer. One other reader sent in the correct solution
without proof.
Both Ardila and Bellot point out that the result in the proposed problem is a special
case of more general identities due to H. W. Gould. These identities can be found in the
American Mathematical Monthly 85 (1978), 450-465 and 63 (1956), 84-91, respectively,
't Woord's and Wildhagen's solutions are essentially the same as Solution I above. The
proposer's solution is a combinatorial argument using inclusion-exclusion.

1 8 3 6 . [1993: 113] Proposed by Jisho Kotani, Akita, Japan.

Let ABCD be a quadrilateral inscribed in a circle T, and let ACnBD = P. Assume
that the center of T does not lie on AC or BD. Draw circles with diameters AB, BC,
CD, DA, and let the areas of the moon-shaped regions inside these circles and outside T
be Fi, F 2 , F3, F 4 . Mi, M 2 , M 3 , M 4 are the midpoints of the sides of ABCD, and # 1 , .ff2,
#3, H4 are the feet of the perpendiculars from P to the sides of ABCD. Prove that, if
F\ + F 2 + F3 + F 4 = area(.AI?C7D), then M l 9 Af2, M 3 , M 4 , Hi, H2, H3, H± are concyclic.
Solution by Waldemar Pompe, student, University of Warsaw, Poland.
Lemma. Let K and L be two points in the plane. The locus of points P such that

{KPf + (LP)2 = constant

is a circle with centre M, the midpoint of KL.

The lemma follows from the theorem of Pythagoras. [See, for example, Locus 10 on page 14
of Nathan Altshiller Court, College Geometry, or Theorem 97 on page 68 of R.A. Johnson,
Advanced Euclidean Geometry. —Ed.]
Let a, b, c, d be the sides of the given quadrilateral (with a = DA, etc.), and let R
and O be the radius and centre of T. We have

Fl+F,+F3+Fl+rf -i(f)%f(|)*+f(f)'+f(D"+ -*ABcm

and from the given condition we get

a2 + b2 + c2 + d? = 8R2.
 85

We shall prove that the diagonals of ABCD are perpendicular. Let the segment
CC be the diameter of F; because of the assumption that the diagonals do not pass
through 0 , C ^ A. Since CDC and CBC are both right angles, we get

(CD)2 + (CD)2 + (BC)2 + (CBf == 8R2 = a2 + h2 + c2 + rf2,

or (CD)2 + (CB)2 = a2 + ft2. According to the lemma, points A and C lie on a circle
centred at the midpoint M of BD (where M ^ 0 by assumption). This implies (by sym-
metry in OM) that a = DA = B C . Therefore LDCA = ZC'CB. Since LCDB = ZCC'B
we get that the triangles DPC and CBC are similar, which implies (since LCBC = 90°)
that AC and P D are perpendicular.
Now our problem is to prove that if diagonals AC and BD are perpendicular, then
Mi, M 2 , M 3 , M 4 , J?i, H2j H3, JI4 are concyclic. Obviously M1M2M3M4 is a rectangle and
has a circumcircle with diameters M1M3 and M2M±, so

LHtPB = Z P ^ B = ZJ3IX7 = ZM3PA

showing that j?i, P , M3 lie on a line. [This is the theorem of Brahmagupta, e.g., see
Theorem 276 on page 137 of Court's College Geometry, or Theorem 3.23 on page 59 of
Coxeter & Greitzer, Geometry Revisited. See also the similar result pointed out by Jordi
Don on [1991: 53]!—Ed] Therefore IM3H1M1 = 90° and hence H\ lies on the circumcircle
of M1M2M3M4. The same argument shows that H2, £T3, H4 lie on that circle too, which
was to be shown.
Also solved hy FEDERICO ARDILA, student, Colegio San Carlos, Bogota,
Colombia; FRANCISCO BELLOT ROSADO, LB. Emilio Ferrari, Valladolid, Spain;
JORDI DOU, Barcelona, Spain; P. PENNING, Delft, The Netherlands; D.J. SMEENK,
Zalihommel, The Netherlands; and the proposer.
The proposer adds that the argument can be reversed: If M\, M2, M3, M4 are
concyclic then Pi + F2 + P 3 + F4 = area(AJBCD).
Dou points out that the second part of the proof (the final paragraph above, start-
ing with AC ± BD) provides the solution to Crux 1866 [1993: 203]. Bellot notes the
relationship of our problem to number 3 from the 1991 British Math. Olympiad [1993: 5].
He gives two references that investigate further properties of "orthodiagonal quadrilaterals":
[1] Jordan Tabov, Simple properties of the orthodiagonal quadrilaterals, Mathematics and
Informatics, 1:1 (February 1991), 1-5.
[2] Agnis Andzans, On the inscribed orthodiagonal quadrilaterals, Mathematics and
Informatics, 3:1 (1993), 6-8.
* * * # #
86

1 8 3 7 . [1993: 113] Proposed by Andy Liu, University of Alberta.

A function / : H —> H + is said to be strictly log-convex if

/(*•>/(*,)> (/(^p)) 2
for all #i 5 #2 £ IRj with equality if and only if #i = #2- / is said to be strictly log-concave
if the inequality is reversed.
(a) Prove that if / and g are strictly log-convex functions, then so is / + g.
(b)* Does the same conclusion hold for strictly log-concave functions?
Combination of solutions of Marcin E. Kuczma, Warszawa, Poland; and A.N.
't Woord, Eindhoven University of Technology, Eindhoven, The Netherlands.
(a) If / and g are strictly log-convex, then for any real numbers #i, x 2 ,

2
= [/(^)]%2/(fi±f2)g(fi±ii) + [ J (fi±^)]
< f(xi)f(x2) + 2y/f(x1)f(x2)g(xi)g(x2) + g{xt)g(x2)
< f(xi)f(x2) + f(xx)g{x2) + f(x2)g(xi) + g{x{)g(x2)
= [f(x1)+g(x1)][f(x2)+g(x2)]
= (f + g)(xi)-(f + g)(x2),

where the second inequality follows by the A.M.-G.M. inequality. Equality holds if and
only if xi = x2.
(b) The answer is NO. For example,

f(x) = e~x2+2x and g(x) = e^ 2 " 2 * (=/(-*))

are strictly log-concave, because l n / ( x ) and hig(x) are strictly concave. Yet

( / + *)(1) • ( / + 9){-l) = (/(I) + / ( - I ) ) 2 = (e + e- 3 ) 2 > 4 = [(/ + </)(0)]2,

showing that / + g is not log-concave.

Part (a) also solved by FEDERICO ARDILA, student, Colegio San Carlos,
Bogota, Colombia; WALTHER JANOUS, Ursulinengymnasium, Innsbruck, Austria;
MURRAY S. KLAMKIN, University of Alberta; WALDEMAR POMPE, student, Uni-
versity of Warsaw, Poland; CHRIS WILDHAGEN, Rotterdam, The Netherlands; and the
proposer.
Only Kuczma and Jt Woord solved part (b), and their counterexamples were virtually
the same.
Klamkin notes that the result of part (a) is known; see for example Theorem F,
pages 18-19 of Roberts and Varberg, Convex Functions, Academic Press, 1973.
 87

* * * # *

1 8 3 8 . [1993: 113] Proposed by Stoyan Kapralov and Uiya Bluskov, Technical

University, Gabrovo, Bulgaria.
Find all sequences ai < a2 < • • • < an of positive integers such that

at + a2 + • • • + an = 26, a\ + a\ + • • • + a\ = 62, a\ + a* + - -. + a?n = 164.

Solution by Leonardo P. Pastor, Cordoba, Argentina,

Let S = (a,-), 1 < i < n, be a sequence of positive integers that satisfies the three
equations. Note that for each at- in the sequence, we have 1 < af- < 5. Otherwise the left
hand side of the third equation would be greater than 164.
Let Xj j / , z, w and t be the non-negative integers representing the number of l's,
2's, 3?s, 45s and 5's in the sequence, respectively. Therefore, we should have

x + 2y + Sz + 4w + 5t = 26,

x + 4y + 9z + 16w + 25t = 62,

x + Sy + 27z + Uw + 125t = 164.
From this it is clear that the number of 5's, i.e. t, is 0 or 1 and that if t = 1 then w = 0.
Also, we have that w < 2. Solving this system in terms of w and t we find

x = 5 — 4K; — 15i,

y = 3 + 6w + 20f,
z = 5 - 4w - lOt.
Thus the only solutions are (£, WjX,y,z) = (0,0,5,3,5), (£, u;, x, t/, z) = (0,1,1,9,1),
(t,ti;,a:,jf,z) = ( 0 , 2 , - 3 , 1 5 , - 3 ) and (t,iy,x,t/,z) = (1,0,-10,23,-5). Clearly the last
two do not solve the problem. Therefore the only sequences which satisfy the conditions
of the problem are

(1,1,1,1,1,2,2,2,3,3,3,3,3) and (1,2,2,2,2,2,2,2,2,2,3,4).

4/50 solved by HL. ABBOTT, University of Alberta; CHARLES ASHBACHER,

Cedar Rapids, Iowa; SAM BAETHGE, Science Academy, Austin, Texas; MARGHERITA
BARILE, student, Universitdt Essen, Germany; CHRISTOPHER J. BRADLEY, Clifton
College, Bristol, U.K.; MIGUEL ANGEL CABEZON OCHOA, Logrono, Spain; TIM
CROSS, Wolverley High School, Kidderminster, U.K.; KEITH EKBLAW, Walla Walla
Community College, Walla Walla, Washington; HANS ENGELHAUPT, Franz-Ludwig-
Gymnasium, Bamberg, Germany; F. J. FLANIGAN, San Jose State University, San Jose,
California; RICHARD I. HESS, Rancho Pahs Verdes, California; WALTHER JANOUS,
Ursulinengymnasium, Innsbruck, Austria; DAG JONSSON, Uppsala, Sweden; KEE-WAI
LAU, Hong Kong; J. A. MCCALLUM, Medicine Hat, Alberta; GOTTFRIED PERZ,
88

Pestalozzigymnasium, Graz, Austria; FRANCISCO LUIZ ROCHA PIMENTEL, Forta-

leza, Brazil; WALDEMAR POMPE, student, University of Warsaw, Poland; R. P. SEALY,
Mount Allison University, Sackville, New Brunswick; A. N. 9T WOORD, Eindhoven Uni-
versity of Technology, Eindhoven, The Netherlands; CHRIS WILDHAGEN, Rotterdam,
The Netherlands; KENNETH M. WILKE, Topeka, Kansas; PAUL YIU, Florida Atlantic
University, Boca Raton; and the proposers. Two other readers found only one of the two
solutions.

1 8 3 9 . [1993: 113] Proposed by N. Kildonan, Winnipeg, Manitoba.

Notice that
122 = l l 2 + l = 1 2 2 - 2 2 ,
i.e., the (base 10) integer N = 122 can be partitioned into two parts (1 and 22), so that
the first part is the difference between N and the greatest square less than JV, and the
second part is the difference between N and the least square greater than JV. Find another
positive integer with this property.
I. Solution by Sam Baethge, Science Academy, Austin, Texas.
Partition the number into two parts "a" and "6" and express the number as
a(10k) + b. Then

n2 + a = a(10k) + b and a(10*) + b = (n + l ) 2 - b.

If we eliminate 6 from this system we have

a(10k) + 2(n2 + a- a(10k)) = (n + l) 2

or
( n - l ) 2 = a ( 1 0 * - 2 ) + 2,
which has solutions for a = 1 and k any positive even integer. The given example is for
k = 2 and n = 11. Others are:
k = 4, n = 101 and 10202 = 1012 + 1 = 1022 - 0202,
k = 6, n = 1001 and 1002002 = 10012 + 1 = 10022 - 002002,
k = 8, n = 10001 and 100020002 = 100012 + 1 = 100022 - 00020002,

etc.
II. Solution by P. Penning, Delft, The Netherlands.
Write N = 100a + b with 0 < 6 < 100. Then

100a + b = x2 + a = (x + l) 2 - 6,

and so
2z + l = a + 6 and x2 = 99a + 6. (1)
 89

Introduce
2m = x — 1 and a = 2c + 1
[for note from (1) that
98a = x2 - 2x - 1 = (x - I ) 2 - 2, (2)
from which it is easy to see that x and a must both be odd.— Ed.]. From (2) we get
49(2c+l) = 2 m 2 - l o r

49c = m 2 - 25 = (m + 5)(m - 5),

which implies [since 49 and 10 are relatively prime]

m = 49t; ± 5 (3)

where t; is an integer. There are only two solutions within the constraint 0 < h < 100,
namely:
v = 0? m = 5, c = 0, a = 19 a: = 11, 6 = 22
which leads to the given example 122; and

v = 2, m = 93, c = 176, a = 353, x = 187, fe = 22

which leads to the example

35322 = 187 2 + 353 = 1882 - 22. (4)

[Editor's note. For instance, eliminating a in (1) yields

x2 - 198x - 99 = - 9 8 6 ,

so 0 < b < 100 means that

- 9 8 0 0 < x2 - 198x - 99 < 0,

or
100 < (x - 99) 2 < 9900,
or
10 < \x - 99| < 100,
so 0 < x < 89 or 109 < x < 199. Now (3) and x = 2m + 1 give

x = 98t; + 1 ± 10,

and the only integer solutions are t; = 0, x = 11, and v = 2, x == 187, as claimed.]
J4ISO so/verf 6y FEDERICO ARDILA, student, Colegio San Carlos, Bogota,
Colombia; CHARLES ASHBACHER, Cedar Rapids, Iowa; CHRISTOPHER BRADLEY,
Clifton College, Bristol, U.K.; MIGUEL ANGEL CABEZON OCHOA, Logrono, Spain;
90

TIM CROSS, Wolverley High School, Kidderminster, U.K.; KEITH EKBLAW, Walla
Walla Community College, Walla Walla, Washington; HANS ENGELHAUPT, Franz-
Ludwig-Gymnasium, Bamberg, Germany; RICHARD K. GUY, University of Calgary;
RICHARD I. HESS, Rancho Palos Verdes, California; JOHN HEUVER, Grande Prairie
Composite H.S., Grande Prairie, Alberta; WALTHER JANOUS, Ursulinengymnasium,
Innsbruck, Austria; J.A. MCCALLUM, Medicine Hat, Alberta; WALDEMAR POMPE,
student, University of Warsaw, Poland; A.N. 'T WOORD, Eindhoven University of Tech-
nology, Eindhoven, The Netherlands; and the proposer.
Engelhaupt, Guy, Hess and the proposer found both solutions I and II. The other
solvers were nearly evenly divided between those with Solution I (usually finding the infinite
family) and those with Solution II.
Guy and Hess give other solutions too, e.g.

180125042 = 134212 + 1801 = 134222 - 25042

and
395930202 = 198972 + 39593 = 198982 - 0202.
This last solution with (4) hints at another infinite family(?). Guy's solution was in fact
a complete analysis of the problem, finding "all" solutions in some sense.

1 8 4 0 . [1993: 114] Proposed by Jun-hua Huang, The 4th Middle School of Nanxian,
Hunan, China.
Let AABC be an acute triangle with area F and circumcenter O. The distances
from O to BC, CA, AB are denoted da, cfc, dc respectively. AAiBiCi (with sides a\, &i,
ci) is inscribed in AABC, with Ai G BC etc. Prove that

daa\ + dhbi + dcc\ > F.

Comment by Murray S. Klamkin, University of Alberta.

Since da = 12 cos A, etc., where R is the circumradius, the inequality can be rewrit-
ten as
. - „ ~. F acosA + bcosB + ccosC
a\ cos A + 6i cos B + c\ cos G > — = —— —,
R 2
and in this form it is the same inequality as proposed by G. Tsintsifas in Problem E2968
in the American Mathematical Monthly, solution by O.P. Lossers on pages 361-362 of the
May 1985 issue.
Also solved by FEDERICO ARDILA, student, Colegio San Carlos, Bogota,
Colombia; MARCIN E. KUCZMA, Warszawa, Poland; and the proposer.
The proposer's proof is similar to Lossers' in the Monthly, and the solutions of
Ardila and Kuczma appear longer.
 EDITOWAL BOARD / CONSEIL DE REDACTION
Dr. G.W. Sands, Editor-in-Chief Department of Mathematics and Statistics, University of Calgary,
J Calgary, Alberta T2N 1N4
Dr. R.E. Woodrow, Editor-in-Chief Department of Mathematics and Statistics, University of Calgary,
Calgary, Alberta T2N 1N4 |
Dr. G.P. Wright, Managing Editor Department of Mathematics, University of Ottawa,
Ottawa, Ontario KIN 6N5 |
Dr. R. DePeiza East York Board of Education, 840 Coxwell Avenue,
1 East York, Ontario M4C 2V3 |
Dr. C. Fisher Department of Mathematics, University of Regina,
Regina, Saskatchewan S4S 0A2

Dr. R. Guy Department of Mathematics and Statistics, University of Calgary,

Calgary, Alberta T2N 1N4 |

Dr. D. Hanson (Articles) Department of Mathematics, University of Regina,

Regina, Saskatchewan S4S 0A2

Dr. A. Liu (Book Reviews) Department of Mathematics, University of Alberta,

Edmonton, Alberta T6G 2G1 |

Dr. R. Nowakowski Department of Math., Stats and Computing Science, Dalhousie

University, Halifax, Nova Scotia B3H 3J5

Dr. J. Totten Department of Math. & Stat., Cariboo College, Kamloops,

British Columbia V2C 5N3 |

Dr. E. Wang Department of Mathematics, Wilfrid Laurier University

Waterloo, Ontario N2L 3C5 |

Short articles intended for publication should be sent to Dr. Hanson, contest problem sets and solutions to Olympiad Corner
problems should be sent to Dr. Woodrow and other problems and solutions to Dr. Sands.

BACK ISSUES AND SUBSCRIPTION INFORMATION

INFORMATION SUR L'ABONNEMENT ET LES NUMEROS PRECEDENTS

1994 (Volume 20) SUBSCRIPTIONS / ABONNEMENTS 1994 (volume 20)

(10 issues per volume / 10 num^ros par volume)
Regular / R6gulier $46.00
CMS Members / membres de la SMC $23.00

These subscription rates include shipping and handling

Les frais de manutentions sont inclus dans ces tarifs
CANADIAN ADDRESSES ADD 7% GST/ ADRESSES CANADIENNES, AJOUTER 7% DE TPS

BACK ISSUES / NUMEROS ANTERIEURS

Volume: $50.00 Issue / Num6ro: $ 6.00

BOUND VOLUMES / VOLUMES RELIES

(available / disponible: 1&2, 3, 7, 8, 9): $10.00 each/chacun

Canadian addresses, please add 7% GST (#R118 833 979).

Les adresses canadiennes, veuillez ajouter 7% de TPS (#R118 833 979).
 CANADIAN MATHEMATICAL SOCIETY
SOCIETE MATHEMATIQUE DU CANADA
577 King Edward
P.O. Box 450, Station A
Ottawa, Ontario KIN 6N5 CANADA
Tel: (613) 564-2223, Fax: (613) 565-1539

ORDER FORM / BON DE COMMANDE

CRUX MATHEMATICORUM

Subscription / Abonnement _ X $46.00 d)

Back issues / Numgros arvterieurs X $ 6.00 = J2)

X $50.00 __ _(3)

Bound volumes / Volumes reltes _ __ x $10.00 _(4)

( 1 + 2 + 3 + 4)

Canadian addresses must add GST: 7% of total of lines 1 to 4 {# R118 833 979)
979)
Les adresses canadiennes: veuillez ajouter 7% de TPS sur la somme
des lignes 1 a 4 (# R118 833 979)

TOTAL $

• Cheque enclosed / cheque inclus • Invoice / Facture

(micro-encoded cheques only
cheques encodes seulement)

• Please charge / Portez a mon compte • VISA • MasterCard

Account no. / Numero de compte Expiry date / date d'expiration

Signature

MAILING ADDRESS / ADRESSE POSTALE

|| Name / Nom: |

I Address / Adresse: ||

J City, Province / Ville, Province: ||

|| Postal Code, Country / Code postal, pays: