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.
 ft * ft * & * ft & * ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft ft & ft ft

CRUX M A T H E M A T I C O R U M

V o l . 1 1 , No* 3

March 1985
Sponsored by
Carleton-Ottawa Mathematics Association Mathematique dsOttawa-€arl@to»
Publie par le College Algonquin, Ottawa
The assistance of the publisher and the support of the Canadian Mathematical Society*
the Carle ton University Department ©f Mathematics and Statistics , the University of Ottawa
Department of Mathematics, and the endorsement of the Ottawa Valley Education Liaison Council
are gratefully acknowledged.

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

CRUX MATHEMATICORtM is a problem-solving journal at the senior secondary and university

undergraduate levels for those who practise or teach mathematics. Its purpose is prisiarily
educational, but it serves also those who read it for professional, cultural, or recreational
reasons.
It is published monthly (except July and August) . The yearly subscription rate for ten
issues is $25 in Canada* $28 (or US$21) elsewhere. Back issues? each $2*50 in Canada*
$3.00 (or US$2«Sa) elsewhere. Bound volumes with index? Vols., 1&2 (combined) and each of
Vols, 3-10, $19 in Canada, $20 (or US$15) elsewhere* Cheques and money orders, payable to
CRUX MATHEMATICORUM, should be sent to the Managing Editor.
All communications about the content (articles, problems, solutions, etc,) should b®
sent to the Editor* All changes of address and inquiries about subscriptions and back issues
should be sent to the Managing Editor*
Editors Leo Sauve, Algonquin College, 140 Main Street, Ottawa, Ontario, Canada K1S 1C2.
Managing Editors Kenneth S. Williams, Algonquin College, 203 Lees Ave., Ottawa, Ontario,
Canada K1S 0C5.
Assistant to the Editors: Mrs. Siu Kuen Lam.

Second Class Mail Registration No* 5432. Return Postage Guaranteed,

CONTENTS

The Olympiad Corner: 63 . . . • . . . . . . M.S. Klamkin 70

A Call (from Australia) for Articles . . . . . . . . . . 81
Problems - Problfemes: 1021-1030 . . . . . . . . . . . . 82
Solutions: 8 5 7 , 901-911, 915-917 . . . . . ...... 84
Crux Mathematicorum: A Call for Editors . . ...... 100

_ 69 -
 - 70 -
THE OLYMPIAD CORNER: 63
M.S. KLAMKIN

The 1985 International Mathematics Olympiad will be held in Finland from June
29 to July 11. Each invited country can send a team of up to six students. For
further information, write to: The IMO-85 Organizing Committee, Department of
Mathematics, University of Helsinki, Hallituskata 15, SF-ooioo, Helsinki, Finland.
I now extend the list, begun last month, of problems proposed by various par-
ticipating countries (but unused) in past International Mathematics Olympiads. (I
am grateful to L.M. Glasser for translating the problems from Mongolia and the
U.S.S.R.) As usual, I solicit from all readers elegant solutions to these problems.
26. Proposed by Australia.
Let xi9x2>* .. 9 x denote n real numbers lying in the interval [0,1]. Show
that there is a number x e [0,1] such that
n
1 1
n.LA % 2

27 i Proposed by Australia.
Let ABCD be a convex quadrilateral, and let A ^ B ^ O ^ D x be the circum-
centers of triangles BCD, CDA, DAB, and ABC, respectively.
(i) Prove that either all of A l9 Bx,C l5 Dx coincide in one point, or else they
are all distinct. Assuming the latter case, show that A x and Cx are on opposite
sides of line B ^ , and that Bx and dl are on opposite sides of line A ^ . (This
establishes the convexity of quadrilateral AiBjCiDi.)
(ii) Let A 2 ,B 2s C2,D2 be the circumcenters of triangles B ^ D x , CxDxA l9 D ^ B x ,
and AxBxCx, respectively. Show that quadrilateral A2B2C2D2 is similar to quadri-
lateral ABCD.
28 1 Proposed by Belgium.
Determine all integers x such that
Xk + X3 + X2 + X + 1

is a perfect square.
29 i Proposed by Belgium.
Determine all integer solutions (x,y) to the Diophantine equation
. x3 - y3 - 2xy + 8.
 - 71 -
30i Proposed by Brazil.
A box containsp white balls and q black bal1s^ and beside the box lies a
large pile of black balls. Two balls chosen at random (with equal likelihood) §pe
taken out of the box. If they are of the same color* a b U c k ball from the pile
is put into the box; otherwise, the white ball is put back Into the box. The pro-
cedure is repeated until the last two balls are removed from the box and one U s t
ball is put tn. What Is the probability that this last ball Is white?

31, Proposed by Bulgaria.

Prove that, for every natural number n9 the binomial coefficient ( )
divides the least common multiple of the numbers 1,2,3,... ,2«.

32g Proposed by Bulgaria.

A regular n-gonal truncated pyramid with base areas Si and Si and lateral
surface area S is circumscribed about a sphere. Let A be the area of the polygon
whose vertices are the points of tangency of the lateral faces of the truncated
pyramid with the sphere. Prove that

AS = ^iS 2 C0S 2 ~.

33 • Proposed by Bulgaria.
Given are a circle r and a line I tangent to it at B. From a point A on r,
a line AP i I is constructed, with P e L If the point M is symmetric to P with
respect to AB, determine the locus of M as A ranges on r.

34, Proposed by Canada.

Determine the permutation a = (alsa2,...,a ) of (i,2,...,n) which maximizes

Q - a\a^ + a>2a3 + •«* + a a

I*
and also the permutation a which minimizes Q.

35i Proposed by Canada.

You are given an algebraic system with an addition and a multiplication
for which all the laws of ordinary arithmetic are valid except commutativity of
multiplication. Show that

-1 -1 -1 -1
(a + ab a) + (a + b) -a
-1 -1 ~1
where x is that element for which x x = xx = e, the multiplicative identity.
36 i Proposed by Czechoslovakia.
Let

S = {-===== I rn9n p o s i t i v e i n t e g e r s } .
 - 72 -
Show that for each (x9y) e S*S with x < y there exists z e S such that x < z < y.
37 g Proposed by Finland.
Four circles C9Ci3C29C3 and a line I are given, all in the same plane. The
circles C l 5 C 2 ,C 3 are all distinct, each touches the other two and touches also C and
I . If the radius of C is 1, determine the distance between its center and I.
3 88 Propos ed by Franc e.
Let the numbers be all positive and let
v
ls'k ~ v^i^2* • »w, , fc = 1,2,...,n.
Prove that
n
K
k=± fc=i *
39 , Proposed by Mongolia.
A country has n cities, any two of which are connected by a railroad. A
railroad worker has to travel on each line exactly once. If at any stop there is
a city he must reach but cannot (having already traveled on the line to or from that
city), then he can fly. What is the smallest number of plane tickets he must buy?
40 . Proposed by Poland
Let r be a unit circle with center 0, and let Pl5P2,...,P be points of r
such that n

OPi + 0P 2 + ... + OP = 0.

Prove that P2Q + P2Q + ... + P Q > n for all points Q.

41s Proposed by Poland.
A convex figure F lies inside a circle with center 0. The angle subtended
by F from every point of the circle is 90°. Prove that 0 is a center of symmetry
Of F.

42 i Proposed by Rumania.
I f ( 1 + X + X2 + X3 + X^) = a 0 + CLiX + . . . + # 1 9 8 ^ 1 9 8
\
(i) determine the g.c.d. of the coefficients a 3 , a 8 , a 1 3 , ..., &1983'
( H ) show that 1 0 3 4 7 > a 9 9 2 > 10 3L+0 .
43 i Proposed by Spain.
Solve the equation
tan22r + 2 tan 2x tan 3x - 1 = 0.
 - 73 -
44. Proposed by Sweden.
Let n be a positive integer having at least two distinct prime factors.
Show that there is a permutation (al,a2^ * * ,a ) of (i,2 9 . .. ,n) such that
n 2ua,
J k c o s - — - = o.

451 Proposed by the U.S.A.

Three roots of the equation
x1* - px^ + qx2 - FX +s - 0

are tan A 9 tanB, tan C 9 where A 9 B 9 C are the angles of a triangle. Determine the
fourth root as a function of (only) p , q3 r, and s .
46 , Proposed by the U«S»S»B*
Let (p..) be a given m*n matrix with real entries, and let
n m
A. = Jp . . and 5. = Ip... (1)

We say that a real number is "rounded off" if it is an integer or 5 if not an integer^

when it is replaced by one of its two nearest neighboring integers. Show that the
p 9 * s A . 9 and B . can be rounded off so that (l) still remains valid.
47. Proposed by the U»S»S«R«
In the Martian language any finite ordered set of Latin letters is a word.
The "Martian Word" editorial office issues a many-volume dictionary of the Martian
language5 in which the entries are numbered consecutively in alphabetical order.
The first volume contains all the one-letter words9 the second volume all the two-
letter words, e t c M and the numbering of the words in each successive volumes con-
tinues the numbering in the preceding one. Determine the word whose number is the
sum of the numbers of the words
Prague, Olympiad, Mathematics.
48 • Proposed hy the U.S.S.R.
Let 0 be the center of the axis of a right circular cylinder; let A and
B be diametrically opposite points in the boundary of its upper base; and let C be
a boundary point of its lower base which does not lie in the plane OAB. Show that
/BOC + /COA + /AOB = 2TT.
49 • Proposed by the U.S.S.R.
Let xi 3 x 2 ,...,# be numbers such that 1 > xi > x 2 ^ * * , > x > 0, If
 - 74 -
0 < a < 1, prove that
,A JCL A a <a-l a a-± a
(1 + x-i + xo + . . . + x ) < 1 + x-] + 2 xo + .. . + n # .

50, Proposed by Vietnam.

Given are the function Fix) = ax2 + bx + e and GGr) = c# 2 + &cc + a, where

|F(0)| < 1, \F(l)\ < l, and |F(-l)| < 1.

Prove that, for \x\ < 1,

(i) |F(a?)| < 5/4 ;
(ii) |G(a?)| < 2.
*

I now present solutions to some problems proposed earlier in this column.

1, [1984-: 107] From the 1983 Chinese Mathematics Olympiad.

Show that

Arcsin# + Arccosx = -, -1 < x < 1.

Solution by Gali Salvatore, Perkins3 Quebec.

Let a = Arcsinx and 3 = Arccosx, so that sin a = cos 3 = x. Now

-1 < x < 0 —> -| < a < 0, |<3<7T

and
0 < x < 1 => 0 < a < |, 0 < 3 < ^;

and in each case cos a = s i n g = Jl~-x2. In each case a l s o , 0 < a+3 ^ TT and

sin (a+3) = sin a cos 3 + cos a sin 3 = x2 + (l-x2) = l,

and therefore a+3 = TT/2, as required.

2, [1984: 1071 From t/ze I£53 Chinese Mathematics Olympiad.

A function / defined on the interval [ 0 , 1 ] satisfies

/ ( 0 ) = / ( ! ) and 0 < xl*x2 < 1 => \f(x2)-f(xl)\ < \x2-x1\.

Prove t h a t 0 - xl,x2 - l —> \f(x2)-f( xl)\ < |.

Solution by Michael W. Ecker3 University of Scranton3 Pennsylvania.

We give what i s e s s e n t i a l l y a geometric proof. We assume without l o s s of
g e n e r a l i t y t h a t f(o) = f(±) = o. Letting xx = 0 and 1 s u c c e s s i v e l y gives \f(x2)\<x2
for x2 * 0 and |/Gr 2 )l < 3 -ar2 for x2 * 1. Thus the graph of y = fix), 0 < # < 1,
 - 75 -
lies inside the square shown in
the figure.
aA)
The desired result is tri-
vially true if x2 = a?l9 so we
assume that x2 * x±. Then the
hypothesis

f(x2) - f(Xl)
< 1
x2 X-]

implies that the slope of the

line joining any two distinct
points P1{x1,f(xi)) and P2(x2,f(x2))
on the graph of / is numerically
less than l. If this line meets the
boundary of the square in the points
Pi and P£5 with abscissas x{ and x2% a,-D
respectively^ then

\f(x2) ~ f(Xl)\ < \f{xl2) - f(x[)\ <

3, C1984: 107] From the 1983 Chinese Mathematics Olympiad

For the quadrilateral ABCD of the figure.,
the following proportion holds:

[ABD] : [BCD] : [ABC] 4 l,

where the brackets denote area. If the points M e AC

and N e CD are such that AM : AC = CN : CD and B 5 M 5 N
are collinear9 prove that M and N are the midpoints
of AC and CD 9 respectively.

Soluifcm by Mike Molloy3 student^ Osgoode

ToDnshfp High School^ Ontario.
Let AM/AC = r. Only one value of p is
possible9 since rotating BMN around B increases
one of AM/AC 5 CN/CD and decreases the other. Thus if r = 1/2 is a permissible value
of is it will be the only one. Let Mf and N1 be the midpoints of AC and CD9 respec-
tively. Then

BM'xBN' J(BA+BC)x(BC+BD)
4

= J(BAxBC + B~AXB"D + B"CxB"b)

 - 76 -
|(BAxBC + 3BAxBC - 4BAxBC) (1)

o,
where in line (1) we have used the given area ratios.
Thus B J M ' J N 1 are collinear and r = 1/2.

1, [1984: 141] From the 1981 Leningrad High School Olympiad.

Find all integers x for which \^x2 - \2x - 27| is a prime number.

Solution by Paul Wagner3 Chicago3 Illinois,

2
Since fix) = ^x - \2x - 27 = (2x + 3)(2a? - 9 ) , a necessary condition for
\f(x)\ to be a prime is that |2^+3| = 1 or \2x -9| = 1 , that is, x = - 2 , -1, 4,
or 5. Conversely, \f(x)\ is a prime for each of these values of x.

2, [1984: 141] From the 1981 Leningrad High School Olympiad.

2 2
Find (a+b)/(a-b) if a > b > 0 and a +b = 6ab.

Solution by Mark Kantrowitz3 student, Maimonides School, Brookline3 Massachusetts.

2 2 2
From a +b = 6afc? we get (a+b) = 8a& and (a-£) 2 = 4a&. Then, since a > b > 0 9

a + b
•2.

38 [1984: 141] FP(?tf? t/ze ^^<9I Leningrad High School Olympiad.

From a point M outside an angle with
vertex A two straight line segments are drawn,
one of which cuts off on the sides of the
angle two congruent segments AB and AC, and
the other intersects these sides at the
points D and E, respectively. Prove that
|BD|/|CE| - |MD|/|ME|.

Solution by Mark Kantrowitz3 student,

Maimonides School, Brookline3 Massachusetts.
Referring to the figure, we apply the
theorem of Menelaus and obtain

AC-EM-DB = CE-MD-BA,

and the desired result follows from the fact that IABI = I AC I
 - 77 -
L\, [1984-: 14-1] From the 1981 Leningrad High School Olympiad,
x u u T*
Given that x + yJ = xr + y , where a: and y are natural numbers, prove that
x = y.

Solution by K. Seymour3 Toronto3 Ontario.

The same conclusion holds if x and y are positive real numbers. We suppose that
x * y and obtain a contradiction. We further assume without loss of generality that
x > y. Then
x y y x y , x-y „N y, x-y „N
x + y^ = xr + y <=> xu(x ^ - 1) = 2/^(2/ d - 1),
and the contradiction arises from ^ > yy and ^ " ^ > i/X~ye

5, C1984: 141] FPOTZ t/ze IZPtfl Leningrad High School Olympiad.

Show that in order for the diagonals of a quadrilateral to be perpendicular
it is necessary and sufficient that the midlines of the quadrilateral be congruent.
(A midline is a line segment connecting the midpoints of opposite sides.)

Solution by K. Seymour3 Toronto3 Ontario.

If A 9 B 9 C 9 D are the consecutive vertices of a quadrilateral and all vectors have
a common origins then

the midlines are congruent <= ,M-£4, ,M-M,

=
<=> [—^—' - ' — 2 — ' °

<=> (A - C)®(B - D) = 0
<=> the diagonals are perpendicular.

5S [1984: 141] From the 1981 Leningrad High School Olympiad.

Show that the two common tangents to the circle x +y2 2
= 2 and the parabola
2
y = x /s are perpendicular.

Solution by Mark Kantrowitz3 student3 Maimonides School3 Brookline3 Massachusetts.

Let y = mx+b be the equation of a common tangent. Since this line meets the
2
circle in only one point, the discriminant of (mx+b) + x2 - 2 must vanish ; and
similarly for the parabola the discriminant of mx+b - x2/S must vanish. These two
discriminant equations are

2m2 = b2 - 2 and 2m2 = -2>,

from which b - -2, then m = ±1 and the two common tangents are perpendicular.
 - 78 -
7, [1984: 142] From the 1981 Leningrad High School Olympiad.
The angle bisectors AD, BE, CF of a triangle ABC concur in the incenter I.
If
£1 - §1 _ CI
ID " IE " IF'
prove that triangle ABC is equilateral.

Solution by K. Seymour, Toronto, Ontario.

Since I is equidistant from the sides of the triangle, the equality of the given
ratios implies that the three altitudes of the triangle are equal, and hence that the
triangle is equilateral.
•T.

8, [1984: 142] From the 1981 Leningrad High School Olympiad.

25 17
Which number is larger, 4 8 or 344 ?

Solution by John Morvay3 Dallas, Texas.

Since
4825 = (72 - l ) 2 5 < 750 and 344 1 7 = (7 3 + I ) 1 7 > 7 5 1 ,

it follows that 4 8 2 5 < 344 1 7 .

98 [1984: 142] From the 1981 Leningrad High School Olympiad.

Prove that the positive root of the equation

a?te+l)te+2)...(a?+1980)(a:+1981) = 1

is less than 1/1981! .

Solution by John Morvay, Dallas, Texas.

That the equation has exactly one positive root follows from Descartes' rule of
signs. This positive root must be less than 1/1981!, for when x = 1/1981! is sub-
stituted into the left side of the equation,

x 1 9 8 2 + ... + 1981!#,

where all coefficients are positive, the result is greater than l.

10, L1984: 14 2] From the 1981 Leningrad High School Olympiad.

Find all primes a and b such that a + b is also a prime.

Solution by Mark Kantrowitz, student, Maimonides School, Brookline, Massachu-

setts.
Either a or Z> must be 2, say a. Then b is an odd prime. Now
 79 -
b = 6n+l => 8 + (6tt+l) 5 n + 2 = 0 (mod 3)
and
2? = 6n-l => 8 + (6n-l)6n = 0 (mod 3),
Thus the only possibility left is b = 6n+3 with n - 0 9 and

2 3 + 3 4 = 89, a prime.

H i [1984: 142] From the 1981 Leningrad High School Olympiad.

Let o < x < TT/6. Show that for all natural numbers n
sin x + tan 2 ^ + sin3^c + . . . + tan x< 1.4.
Solution by John Morvay3 Dallas3 Texas.
Since s i n x < 1/2 and t a n ^ < l / / 3 9 the sum of the f i n i t e series is less than
1 1
/ 1 1 1 v /l 1 1 X 2 3 7 < 1.4,
( + + + + ( + + + +
2 F i ^ -'J i 3"2 i3 ••«) = — T — 1 =6
4 3

12 1 [1984: 142] Frc?/7? the 1981 Leningrad High School Olympiad.

Find the lengths of the edges of a rectangular parallelepiped with a square
base if they are natural numbers and the total surface area is numerically equal to
the sum of all the edge lengths.

Solution by H. Kaye3 Brooklyn3 N.I.

There are eight edges of the two square bases each of length a9 say 9 and four
l a t e r a l edges each of length b, say. Then 8a+4& = 2a2+4a&9 from which
2b = a(a + 2b - 4 ) . (1)
It follows from (1) t h a t a is even* say a = 2a\ and then
b = a(a% + b - 2) = na9 say.
Now, from (1),
2n + 4 A 3
a - 2n + 1 = 1 + 2n+l"

Finally* « = 1, and a = Z? = 2.

13. [1984: 142] From the 1981 Leningrad High School Olympiad.
In the convex quadrilateral ABCD9 /BAC = /CBD and /ACD = /BDA. Show that
2 2 2
IACI = IBCI f lADI .
 - 80 -
Solution by M.S.K.
With the angles 0,a,3 as shown in the
figure, we note that /ABC = /ADC = 6 and then,
by the law of sines,

AC BC AD ,
sin 0 " sina " sin 3 " A> Sdy
"

From [ABCD] = [DBC] + [DAB], where the square

brackets denote area, we obtain

~BD»ACsin0 = ~BD*BCsina + | B D « A D s i n 3 ,

from which
fr AC r BC n AD
AC*-r- = DBC»-=- + AAD*-y->

and |AC|2 = |BC|2 + |AD|2 follows,

\\\% [1984: 142] From the 1981 Leningrad High School Olympiad.
Jb+c jc+a Ji+b ^a+b+c+1
Show that 2"'" + 2"'"- + 2"'" < T + 1 for all a9b9c > 0.

Solution by Mark Kantrowitz3 student^ Maimonides School^ Brookline^ Massa-

chusetts.
The given inequality is equivalent to

F(x9y9z) - 2 + xyz - x - y - z > 0 9

^a+b+c •a
obtained by dividing throughout by 2°""'^ and then setting x = 2 ,y = 2 , and
z = 2 , so that 0 < x9y9z < l.
Since F(x9y9z) is linear in each variable, it takes on its greatest and least
values at the ends of the intervals, that is, at the corners of the cube with ver-
tices (o,o,o), (i,o,o), ..., (1,1,0), ..., (1,1,1). Since this least value is
zero, F(x9y9z) > 0 for all points (x9y9z) in the interior of the cube.

J^, [1984: 142] From the 1981 Leningrad High School Olympiad.
Show that, for all positive x and y9

x^ + z/3 < x 3.^ + y3.^jl

Solution by M.S.K.
More generally, for x,y^m,n > 0 consider
 81 -
m . m _ m n]x_ m@ nj^o
x +y < x +
fy ^ Ix S
Since the case x = z/ is immediate, we can assume without loss of generality that
y = Xx with X > 1. The inequality then reduces to

(x « + l/» _ l ) ( x -l/„ . 1} ^ 0§

and this is clearly true. •

A multivariate generalization is given by

p p m I a?. )l/n

where m3n*xe > 0 and p is a p o s i t i v e i n t e g e r . We can s i m p l i f y the inequality., by

l e t t i n g x. - a.9 to

r
dido*• .a ) a. £ ) a* .

This follows by applying the A.M.-GoM. i n e q u a l i t y and Chebyshev's i n e q u a l i t y :

- mn _ v „ mn _ wn+p
la. Ear. Za. La . r
nae ^J^ < _£._£_ ^ _ 1 _ 8
^ P P P P

In closing,I would ask all readers submitting solutions to any of the problems
in this or in past columns to clearly identify the problems by stating their num-
bers and the place (year and page number) where they originally appeared.
Editorss note. All communications about this column should be sent to Pro-
fessor M*S« KLamkin, Department of Mathematics, University of Alberta, Edmonton,
Alberta, Canada T6G 2G1.

A CALL (FROM AUSTRALIA) FOR ARTICLES

The Mathematical Scientist? published by the Australian Mathematical Society,

is an international journal devoted to the interface between mathematics and the
real world; in other words, it is concerned with mathematical modelling as applied
to the physical, biological, and social sciences. Have you done some interesting
work of this kind and not yet published it? If so, please send your article to any
member of the following subset of the Editorial Boards
Prof* J.M. Gani, statistics Dept., University of Kentucky^ Lexington^, KY 40506;
Dr« H„ Ockendon, Somerville College, Oxfordf OX26HD, England;
Prof. B.C. Rennie, James Cook University, N, Queensland 4811f Australia,
To save time we ask just for a clear typescript^ the conversion into ready-for-
printer format is left until after refereeing and acceptance.
 - 82 -
PROBLEF!S--PROBLE^ES
Problem proposals and solutions should be sent to the editor, whose address
appears on the front page of this issue. Proposals shouldf 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 his permission.
To facilitate their consideration, your solutions, typewritten or neatly hand-
written on signed, separate sheets, should preferably be mailed to the editor before
October 1, 1985, although solutions received after that date will also be considered
until the time when a solution is published.

10211 Proposed by Allan Wm. Johnson3 Washington^ B.C.

In the etymological decimal addition

SERGE
DE
NlMES
DENIM

maximize NIMES (where 1 = 1 ) , the city in southern France that gave its name to
denim cloth.

1022 i Propose par Armel Mercier3 Universite du QuSbec d: Chicoutimi*

Soient k un entier positif et i un entier verifiant 0 < i < k« Montrer
que
J (J +1)/ k
M-I)
l — i 'CV' " V -V)* <
= n {cf _ 1)s
3K
3=0 q J=l
k k
oQ [ .] designe le coefficient binomial de Gauss dgfini par [n] = l et

et q est une variable r^elle arbitraire.

1023 B From a Trinity College3 Cambridge3 examination paper dated June 73 1901.
Show that, for n = 1,2,3,.,,,

n
v 2 3TT 3 1
Y Arctan r-o- = A r c t a n — Arctan ~—•-.
^ k2 *+ n n+1
102^8 Proposed by William Tunstall Pedoe3 student3 The High School of Dundee3
Scotland*
Prove that an odd number which is a perfect square cannot be perfect.
 - 83 -
1025, Proposed by Peter Messer3 M»Dej Mequon3 Wisconsin.
A paper square ABCD is folded so that A
vertex C falls on AB and side CD is divided into
two segments of lengths I and m% as shown in the
figure* Find the minimum value of the ratio l/m.

1026, Proposed by Stanley Eabinowitz3 digi-

tal Equipment Corp*3 Nashua3 New
Hampshire.
D, E5 and F are points on sides BC, CA9 and
AB3 respectively,, of triangle ABC9 and ADS BE, and CF concur at point H. If H is
the incenter of triangle DEF9 prove that H is the orthocenter of triangle ABC.
(This is the converse of a well-known property of the orthocenter,)

1027, Proposed by M.S. Klamkin3 University of Alberta.

Determine all quadruples (a9b9c9d) of nonzero integers satisfying the
Diophantine equation

abcd(- + r + - + J) = (a + b + c + d)2
a b e d

and such t h a t a2 + b2 + e2 + d2 i s a prime.

1028, Proposed by David Singmaster3 Polytechnic of the South Barik3 London3

England.
Students learning modular arithmetic are pleasantly perplexed by

4-5 = 6 (mod 7) .

Solve the following (and possibly other) generalizations:

(a) a(a+l) = a+2 (mod a+3).
(b) a(a+l) = a+2 (mod 777), where m is not necessarily a+3.
(c) a(a+l) = (a+2)(ai-3) (mod m).

1029, Proposed by Farshad Khorrami3 student3 The Ohio State University.

Find necessary and sufficient conditions on the complex numbers a and b
so that each root of
z2 + az + b = 0

has absolute value less than 1 .

1030» Proposed by J.T. Groenman3 Arnhem3 The Netherlands.

Given are two obtuse t r i a n g l e s with sides a$b,c and p,q3r9 the longest
sides of each being c and r3 respectively. Prove that

ap + bq < or.
 - 84 -
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.

857, [1983: 179; 1984: 304; 1985: 20] Proposed by Leroy F6 Meyers^ The Ohio
State University.
(a) Given three positive integers, show how to determine algebraically (rather
than by a search) the row (if any) of Pascal's triangle in which these integers
occur as consecutive entries.
(b) Given two positive integers* can one similarly determine the row (if any)
in which they occur as consecutive entries?
(c)" The positive integer k occurs in the row of Pascal's triangle beginning
with l, k$ ... . For which integers is this the only row in which it occurs?

III. Comment on part (c) by David Singmaster3 Polytechnic of the South Bank3
London^ England.
In 1971, I examined N(a)9 the number of times the integer a occurs in Pascalfs
triangle [l]. I showed that N(a) = O(loga) and conjectured that N(a) was bounded.
The proposer's part (c)" asks to characterize those a for which N(a) = 2. Abbotts
Erdos, and Hanson [2] examined this and showed that the normal and average orders
of N(a) are both 2. (See [4] for definitions of normal and average orders.) The
normal order assertion implies that N(a) = 2 for almost all integers (in the usual
sense of density). Specificallys if g(x) is the number of integers a < x such
that N(a) > 25 then they showed easily that g(x) - OUx),
In [3], I reported on my search for repeated binomial coefficients. There are
infinitely many a with N(a) > 69 obtained by solving

n
(n+1) = ( )

leading to
F F - 1 k = F F - 1
*2i+2r2i+3 ±9 K
*2i 2i+3
where the F. are the Fibonacci numbers9 beginning with F 0 = 0. The only other non-
h8
trivial repeated binomial coefficients up to 2 are the following:

120 = (\6) = (10)

v
3 '
210 - (21) , (10)
V
v 2 i 4 ;

1540 = (56) = (22)

v
V2 I 3 '
7140 - v( 122 °)' = V(
36
3 '
)
 - 85 -
11628 = (^3) = (^)

2H310 = (2f) = {lnQ)

3003 = (728) = (1/) = ( « )

Note that the last equality is the case i = 1 of the above general solution. Note
also that #(3003) = 8, and 3003 is the only known a with N(a) > 8.
Perhaps some reader can investigate this question further.
REFERENCES

1. David Singmaster, "How often does an integer occur as a binomial coeffi-

cient?", American Mathematical Monthly9 78 (1971) 385-386.
2. H.L. Abbott9 P. Erdos, and D. Hanson, "On the number of times an integer
occurs as a binomial coefficient", American Mathematical Monthly, 81 (1974) 256-261,
3* David Singmaster, "Repeated binomial coefficients and Fibonacci numbers",
Fibonacci Quarterly, 13 (1975) 295-298.
4. G.H. Hardy and E.M. Wright, An Introduction to the Theory of Numbers>
Oxford, I960, pp. 2639 356,
* ft *

901« C1984- 18] Proposed by Charles W9 Trigg3 San Diego3 California.

In
BOXER = HITS,

the X doubles as a multiplication sign. Find (a) the fewest HITS, and (b) the most
HITS, that the BOXER can deliver.
I. Solution by Edwin M. Klein3 University of Wisconsin-Whitewater.
Here are the 26 (with B0 > ER) computer-generated solutions to the alphametic,
listed in order of increasing HITS, so solutions 1 and 26 are the answers to parts
(a) and (b), respectively.
1. 46x23 = 1058 10. 58x37 = 2146
2. 57x24 = 1368 11. 56x39 = 2184
3. 49x32 = 1568 12. 65x38 = 2470
4, 69x23 = 1587 13. 65x48 = 3120
5. 42x38 = 1596 14. 82x45 = 3690
6. 52x34 = 1768 15. 64x58 = 3712
7. 58x34 = 1972 16. 85x46 = 3910
8. 52x38 = 1976 17. 68x59 = 4012
9, 54x39 = 2106 18. 76x53 = 4028
 - 86 -
19. 79x52 = 4108 23. 74x68 = 5032
20. 78x59 = 4602 24. 87x62 = 5394
21. 92x53 = 4876 25. 85x74 = 6290
22. 86x57 = 4902 26. 95x72 = 6840

II. Comment by Stewart Metchettes Culver City3 California.

Of the 26 solutions [as listed in solution I ] , only No. 4 has an odd number of
HITS (one left hook was not followed by a right one); Nos. 2, 6, 10, and 15 involve
the consecutive digits l to 8; Nos. 21 and 24 involve the consecutive digits 2 to 9;
and the only digit used in all solutions is 5.

Also solved by CLAYTON W. DODGE, University of Maine at Orono; RICHARD I. HESS,

Rancho Palos Verdes, California; ALLAN WM. JOHNSON JR., Washington, D.C; FRIEND H»
KIERSTEAD, JR., Cuyahoga Falls, Ohio; J.A. McCALLUM, Medicine Hat, Alberta? STEWART
METCHETTE, Culver City, California; GLEN E. MILLS, Pensacola Junior College, Florida;
KENNETH M8 WILKE, Topeka, Kansas; and the proposer.

Editorfs comment.
This is the sort of problem that a computer can solve at the flick of a bit,
and most solvers availed themselves of this fact (or at least sent in only answers,
however obtained). The few who chose to slug it out by brute force for the edifi-
cation of the editor managed to find few labour-saving finesses. They won the match,
but they ended up exhausted on the mat, almost too weak to claim the fruits of
victory.
* * a

902 s C198U: 18] Proposed by J. Chris Fisher,, Universite Libre de Bruxelles3

Belgique (now at University of Regina9 Saskatchewan).
(a) For any point P on a side of a given triangle, define Q to be that point
on the triangle for which PQ bisects the area. What is the locus of the midpoint
of PQ?
(b) Like the curve in part (a), the locus of the midpoints of the perimeter-
bisecting chords of a triangle (see Crux 674 [1982- 2563) has an orientation that
is opposite to that of the given triangle. Is this a general principle? More
precisely, given a triangle and a family of chords joining P(t) to Q(t), where (i)
Pit) and Q(t) move counterclockwise about the triangle as t increases and (ii)
Pit) * Q(t) for any t, does the midpoint of PQ always trace a curve that is clock-
wise-oriented?

Solution by Jordi Dou3 Barcelonas Spain.

(a) It is well known and easy to show that the variable lines which form with
the sides of a fixed angle a triangle of constant area are tangent to a hyperbola
whose asymptotes are the sides of the angle. Moreover, for each line the point of
 - 87
contact is precisely the midpoint M of
the segment PQ cut off by the sides of
the angle* The desired locus is there-
fore a curvilinear triangle having three
cusps. As shown in the figure, it is
formed by arcs of the three hyperbolas
having pairs of sides of the given
triangle ABC as asymptotes, each arc
being tangent to two of the medians
AD, BE, and CF of the triangle. The
three cusps are the midpoints of
these medians.
As P traverses the perimeter
of ABC once in the counterclockwise sense, the midpoint M of PQ traverses its locus
twice in the clockwise sense, once as P traces the path AFBD (and Q traces the path
DCEA), and once more as P and Q complete their circuit of the perimeter of ABC. D
Let s be the area of the given triangle and a the area of the region bounded
by the curvilinear locus. Since affine transformations preserve area ratios, the
ratio o/S is constant for all triangles. If ABC is an isosceles right triangle
with AB = AC = l, it is easy to show by elementary calculus (we leave the details
to the reader) that a = (In 8 - 2)/8. Since S = 1/2 in this case, we therefore have

a In 8 2 1
s ~ 50
for all triangles.
(b) As stated in the proposal, this question is trivial and the answer is no.
It suffices to have P(t) sufficiently close to Q(t) for all t to ensure that the
midpoint M of PQ moves in the same sense as P and Q 9 In each of the two cited
examples, "PQ bisects the area" and "PQ bisects the perimeter", the correspondence
between P and Q is an involution, and perhaps the proposer tacitly assumed this
condition. But even with this additional condition the answer is still no. We
give three examples which shed some light on this matter.
With the same notation as in the figure, we assume that when P is at A,B,D,
then Q is at D,C,A, respectively. Let p and q be the distances, measured along the
perimeter in the counterclockwise sense, from A to P and Q, respectively, where

0 < p < AB => AB+BD < q < AB+BC

and
AB < p < AB+BD => AB+BC < q < AB+BC+CA.
 - 88 -
Finally, let q = <|>(p)» where <J> is increasing and involutoric. As P goes from A to
B to D (and 0 goes from D to C to A ) , the midpoint M of PQ traces a closed curve y
that goes from the midpoint M 0 of AD to D and back to M 0 .
Example 1. If dq/dp is strictly increasing for all p , then the curve y is
traced in the same sense as P and Q.
Example 2, If dq/dp is constant for all p , then y consists of the segment M 0 D
traced in both senses.
Example 3. If dq/dp is strictly decreasing for all p 9 then the curve y is
traced in the sense opposite to that of P and Q.

Also solved by J.TS GROENMAN, Arnhem, The Netherlands (part (a) only); RICHARD
I. HESS, Rancho Palos Verdes, California; WALTHER JANOUS, Ursulinengymnasium, Inns-
bruck , Austria; and the proposer.

903 i C198M-: 19] Proposed by Stanley Rabinowitz3 Digital Equipment Corp«3

Nashua3 New Hampshire.
Let ABC be an acute-angled triangle with circumcenter 0 and orthocenter H.
(a) Prove that an ellipse with foci 0 and H can be inscribed in the triangle.
(b) Show how to construct^ with straightedge and compass, the points L,M,N
where this ellipse is tangent to the sides BC,CA,AB, respectively, of the triangle*
(c) Prove that AL,BM,CN are concurrent.

I. Solution to part (a) by the proposer.

More generally, it Is known [l]
that a conic is determined when three
tangents and a focus F are given,
and then the other focus is the iso-
gonal conjugate of F with respect to
the triangle formed by the three
tangents. So, for every triangle
and every point F, there is a conic
inscribed in the triangle, having F
as one focus. The nature of this
conic depends upon the position of
F. According to [l], the conic is
a parabola if F lies on the circum-
circle of the triangle (the isogonal
(e)
conjugate of F is then a point at
infinity), and it is an ellipse (e)
Figure 1
or a hyperbola (h) according to the
 - 89 -
region in the plane in which F lies, as shown in Figure l.
In our problems if we take for the focus F the orthocenter H s which lies in-
side the acute-angled triangle, then the inscribed conic is an ellipse, and its
other focus is the circumcenter 0, the isogonal conjugate of H e

Figure 2

II. Solution to parts (b) and (o) by Leon Bankoff3 Los Angeles3 California.
(b) One of the standard properties of an ellipse is that any tangent makes
equal angles with the focal radii at the point of contact. Accordingly if, as
shown in Figure 2, H 2 is the reflection of H in side BC, then 0H X n BC = L, the
point of contact of the ellipse with side BC. The straightedge and compass con-
struction of L is therefore obvious- Similarly, 0H 2 n CA = M and 0H 3 n AB = N §
where H 2 and H 3 are the reflections of H in CA and AB f respectively.
 - 90 -
(c) As noted earlier in this journal [1977: 114], the concurrency of AL 3 BM f CN
is an immediate consequence of Brianchon's Theorem, when ABC is considered as a
degenerate hexagon. D
Figure 2 illustrates the relationship between the circumcircle, the nine-point
circle, and the ellipse with foci H and 0 inscribed in acute triangle ABC*
The nine-point circle turns out to be the auxiliary circle of the inscribed
ellipse, whose major axis PQ is therefore equal to the circumradius of the triangle.
The reflections of H in the sides of the triangle lie on the circumcirole, thus
suggesting an alternate method for locating the points L,M,N. The nine-point circle,
centered at N f , the midpoint of OH, with radius half that of the circumcircle, can
be drawn, and it passes through the feet of the three altitudes of ABC; these alti-
tudes can then be extended to H 1 ,H 2 s H 3 on the circumcircle and, as before, we have
OHj n BC = L, etc.
Also solved by LEON BANKOFF, Los Angeles, California (also part (a)); JORDI DOU,
Barcelona, Spain? ROLAND H. EDDYf Memorial University of Newfoundland; J9T* GROENMAN,
Arnhem, The Netherlands; DAN PEDOE, University of Minnesota; BASIL C. RENNXE, James
Cook University of North Queensland, Australia; and the proposer (also parts (b) and
(c)).

REFERENCE

1. Luigi Cremona9 Elements of Projective Geometry^ Dover9 New York, 19605 pp.
260-261.
& it &

904s E1984: 19] Proposed by George Tsintsifas3 Thessaloniki^ Greece.

Let M be any point in the plane of a given triangle ABC* The cevians
AM3BM,CM intersect the lines BC 3 CA S AB in A'.B'jC, respectively. Find the locus of
the points M such that
[MCB S ] + [MAC 1 ] + [MBA 1 ] = [MC ? B] + [MA f C] + [MB f A] 9 (1)
where the square brackets denote the signed area of a triangle.
Solution by W.J. Blundon3 Memorial University of Newfoundland,
In ureal coordinates^ the coordinates of the vertices of ABC as triangle of
reference, and of an arbitrary point M in the plane, are
A(l,o,o), B(0 9 1 9 0), C(0,0,l), and W(x,y9z) where x+y+z = l.
Thus the coordinates of A1 are (o,fc,l-fc) for some k9 and the collinearity of A 9 M S A S
gives
j1 0 0
\x y z = 09 whence k = -^-.
0 k l-k y+z
 - 91 -
With similar results for B' and C . we have

A' = £^<°.y.*>. B' 1 1

z+x ' * * x+y ^
We now assume without loss of generality that [ABC] = l . Then (see the a r t i c l e by
0. Bottema in t h i s journal [1982: 228-231])

X y z x y z
s f
[MCB ] = — 0 0 1 [MC B] = x y 0
z+x s+a? x+y a?+j/'
X 0 z 0 1 0
With these and similar r e s u l t s , equation ( l ) in the proposal becomes

xy yz zx _ zx xy yz
z+x x+y y+z ~ x+y y+z z+x9

an equation equivalent to (y-z)(z-x)(x-y)(x+y+z) = 0, o r , since x+y+z = l , to

(y - 2 ) ( s - ^ ) ( x - j/) = 0.

Thus M(a:,y,s) lies on the locus if and only if at least two of x9y9z are equal The
required locus is therefore

I n m n n9

where l9m3n are the lines containing the medians through A,B,C, respectively.

Also solved by JORDI DOU* Barcelona, Spain? J*T. GROENMAN, Arnhemf The Nether-
lands; WALTHER JANOUS, Ursulinengymnasium* Innsbruck, Austria; DAN PEDOE, University
of Minnesota? and the proposer*
* * *

905» [1984: 19] Proposed by J.T. Groenmans Arnhem3 The Netherlands*

Let ABC be a triangle that is not right-angled at B or C. Let D be the
foot of the perpendicular from A upon BC, and let M and N be the feet of the per-
pendiculars from D upon AB and AC, respectively,
(a) Prove that, if /A = 90°, then /BMC = /BNC. (This problem is given with-
out proof in M.N. Aref and William Wernick, Problems and Solutions in Euclidean
Geometry, Dover, New York, 19689 p6 95, Ex. 8,)
(b) Prove or disprove the converse of part (a).

Solution by D0J„ Smeenk3 Zaltbommel3 The Netherlands.

We show that /BMC = /BNC whether or not /A = 90°, and this implies that part
(a) is true and its converse is false.
The proof is simple. From

AD2 = AM*AB = AN»AC,

it follows that quadrilateral BMNC is cyclic, and hence that /BMC = /BNC0
 - 92 -
Also solved by JORDI DOU# Barcelona, Spain? RICHARD I. HESS, Rancho Palos
Verdes, California; WALTHER JANOUS^ Ursulinengymnasium, Innsbruck, Austria; and
the proposer*
ft ft ft

ft
906 i [1984: 19] Proposed by E.J. Eckevt and P.d. Vestergaard3 Institute of
Electronic Systemss Aalborg University Centre3 Denmark*
Let P denote the set of rational points on the unit circle c, that is s the
set of all points (r/t9 s/t) where r9s9t are integerss t > 0 9 and r2+s2 = t2B It
is known that P is dense in C« Let T be the subset of P for which t is prime* Is
T dense in CI
Editorfs comment.
No solution was received for this problem, which therefore remains open. For
a proof that P is dense in C9 see Problem 109 [1976: 81].
ft ft ft

9071 [1984: 19] Proposed by Kenneth S9 Williams^ Carleton University3 Ottawa.

The four consecutive positive integers 76 9 77, 78 9 79 are such that no
one of them is expressible as the sum of two squares. Prove that there are infi-
nitely many such quadruples of consecutive integers.

Solution by Leroy F* Meyers3 The Ohio State University.

More generally, for each positive integer k it is possible to find infinitely
many sets of k consecutive positive integers none of which is a sum of two squares
of integers. It is known [i] that a positive integer n is not the sum of two
squares of integers just when there is a prime p = 3 (mod M-) which appears with an
odd exponent in the canonical factorization of n into a product of powers of dis-
tinct primes.
Let qi9q2>'.. »#T, be k distinct primes* with each q. = 3 (mod 4 ) . (It is always
possible to find k such primes since there are infinitely many of thems by Dirich-
letfs Theorem.) By the Chinese Remainder Theorem.* find a positive integer x such
that, for all j = 1,2,... 97<9

x = q . - j (mod q2).
<J V

Then x-ij i s d i v i s i b l e by q. but not by q2 ( s i n c e q. > j i f t h e primes a r e l i s t e d in

d d «7

increasing order), and so x+j is not the sum of two squares of integers. Hence
each of the k positive integers

is not expressible as the sum of two squares of integers5 and all positive integers
congruent to these modulo (<?i^2- • °<7T, ) 2 are
likewise not so expressible.
 - 93 -
For k = l,2 5 9 . 9 9 7 9 the smallest fc-tuples of consecutive positive integers that
are not sums of two squares of integers are
3
6 7
21 22 23
21 22 23 2*4
75 76 77 78 79
91 92 93 9694 95

186 187 188 189 190 191 192

Also solved by CLAYTON W. DODGE, University of Maine at Orono; F. DAVID HAMMER,
Palo Alto, California; RICHARD I. HESS, Rancho Palos Verdes, California? FRIEND H e
KIERSTEAD, JR., Cuyahoga Falls, Ohio; EDWIN M 8 KLEIN, University of Wisconsin-
Whitewater; BOB PRIELIPP, University of Wisconsin-Oshkosh; KENNETH M* WILKE, Topeka,
Kansas; and the proposer. A comment was received from J.Te GROENMAN, Arnhem, The
Netherlands.

REFERENCE

1. Ivan Niven and Herbert S* Zuckermanf An Introduction to the Theory of Num-

bers (Third Edition)9 John Wiley & Sons, New Yorks 1972, p* 117.

908, C'1984: 19] Proposed by M.S. Klamkin3 University of Alberta.

Determine the maximum value of
P = sinaA • sinBB • sin Y C s
where A*B,C are the angles of a triangle and a s B 9 y ^re given positive numbers.
Solution by Walther Janous3 Ursulinengymnasium3 Innsbruak3 Austria.
The function P(A f B f C) is continuous and nonnegative over the compact set
A + B + C=7TS 0 < A9B,C < TT9 (1)
and vanishes just on its boundary* Consequently^ P attains a maximum value at some
Interior point of the region (l), where 0 < ASB,C < TT. We use the method of La-
grange multipliers to find this maximum value.
Let
F(A S B*C S A) = P(A,B*C) - A(A+B+C-7r)8
Then, since dF/dX = o9 the maximum value of F will occur when
K - Pa cos A aF _ Pgcos B 3P __ Pycos C
A A U)
8A " sin A " " °* 9B ' sin B " * dC ~ sin C ~A "Ue
u

We show that cos A cos B cos C ^ 0, If cos A = 0 S for example, then X = 0, so also
cos B = cos C = 0 and A+B+C = 37r/2f a contradiction. Now, from (2), X * 0 and
 - 94 -
iPx tan A tan B tan C t$n A + tan B + tan C -
(_) = —-— = . = -z—. —. . =x say.

Since tan A + tan B + tan C = tan A tan B tan C, we therefore have

k(a + 3 + y) = k3a3Y*

so k2 = (a+3+Y)/a3Y anc
I

tan^A = a ( a ! B + T ) , tan2B = 3(a + B + Y ) ) tan2C = lio+gj^. (3)

BY Ya a
B
F i n a l l y , from sin A = A e m ^ ^ e t c . , we obtain

s i n A = J"5tf^\» sinB = ^ C f S f g r , sinC = J 5 M E L

T(a+3)(a+y) f(3+Y>(B+a) J(Y+OO(Y+3)
and so

P f g(g+B+Y) »a/2
T / . /f eB(a+g+Y)
/ ^ a ^ ^f/ 2 „ri wY(a+g+y)
« ^ e ^ iY/2
\'
max \(a+B)(a+Y)) * \(3+Y)(B+a)/ UY+a)(Y+3)J

In particular, for a = B = Y = l w e get P = 3/3/8. This is item 2.7 in the

max
l!
Bottema Bible81 [ 1 ] .
Also solved by FRIEND H. KIERSTEAD, JR., Cuyahoga Falls, Ohio; N»Y9 WONG, Hong
Kong; and the proposer.

Editorfs comment.
The proposer noted that a closely related problem appears without solution in
Hardy [2], where it is credited to Math, Tripos. 1935. This problem asked only to
show that (3) holds when p is a maximum.

REFERENCES

1. 0. Bottema et al., Geometric Inequalities* Wolters-Noordhoff, Groningen,

1969, p. 19.
2. G.H. Hardy, A Course of Pure Mathematics, Cambridge University Press, 1958,
p. 311.

909 i [1984: 201 Proposed by Stan Wagon3 Smith College3 Northampton, Massa-
chusetts.
For which positive integers n is it true that, whenever an Integer's decimal
expansion contains only zeros and ones, with exactly n ones, then the integer is
not a perfect square?

Comment by Stanley Rabinowitz, Digital Equipment Corp«3 Nashua, New Hampshire*

All positive integers are congruent to the sum of their digits modulo 9, and
all perfect squares are congruent to 0, 1, 4, or 7 modulo 9. So an integer con-
 - 95 -
taining only zeros and n ones cannot be a perfect square if n f. o 9 l, 4, or 7
(mod 9 ) . This eliminates many from consideration, and strong computer evidence
suggests that nearly all the remaining ones can be eliminated as well. In fact,
we conjecture that the only perfect squares containing only zeros and ones are those
k
of the form 100 , where k is a nonnegative integer, for which n = l.
It is easy to find all r-digit numbers whose squares end with r zeros or ones*
For example, for r = 1 only o 9 l, and 9 have this property. To check all 2-digit
candidates, we need only check those 2-digit numbers that end in 0, l, or 9, In
each case, there are only 10 numbers to check. Proceeding in a similar manner9 once
we have found all fc-digit numbers whose squares end with k zeros or ones, we can
quickly check for (fe+i)-digit numbers by tacking on each of the digits from 0 to 9
to the left of the fc-digit numbers^ and testing the squares of the results.
Using this algorithm, we have tested all numbers up to 10 36 „ Other than the
k
trivial ones of the form 100 , we did not find a single number of up to 36 digits
that was a perfect square and consisted entirely of zeros and ones* As we keep
adding digits on the left, it becomes even more unlikely that such a perfect square
will be found.
Some near misses:

4251 2 = 18071001
24499 2 = 600201001
124499 2 = 15500001001
2S1249 2 = 79101000001
375501 2 = 141001001001
425501 2 = 181051101001
781249 2 = 610350000001
10257492 = 1052161011001
145262492 = 211011910010001
167687512 = 281191010100001
389757492 = 1519109010111001
434742512 = 1890010500011001
1552807492 = 24112111010001001
2M49499992 = 60000502010100001
2847807492 = 81100075001001001
100503755012 = 101010047711101001001
316242187512 = 1000091211611100000001
3165297807492 = 100191102101010011001001
5000000000000000012 = 250000000000000001000000000000000001

Comments were received from FRIEND H. K1ERSTEAD, JR.^ Cuyahoga Falls, Ohio;
and the proposer.
 - 96 -
910 i C1984: 20] Proposed by 0. Bottema3 Del ft 3 The Netherlands*
Determine the locus of the centers of the conies through the incenter
and the three excenters of a given triangle.

I. Solution by Dan Pedoe3 University of Minnesota.

Solutions are provided by Problem 14 in Dure!! [1] and Problems 80*8 and 80.9
in Pedoe [2].
The incenter is the orthocenter of the triangle formed by the three excenters,
so that all conies through the four points are rectangular hyperbolas, and the locus
of centers for these conies is the nine-point circle of the triangle formed by any
three of the four points* and this is the circumcircle of the original triangle*

II. Comment by Roland H« Eddy3 Memorial University of Newfoundland.

More generally, Smith [3] uses trilinear coordinates to find the locus of the
centers of the conies through four arbitrary points. When the four points are the
incenter and the three excenters, the locus turns out to be the circumcircle of
the given triangle.
The actual proposal is given in Sommerville [4].

Also solved by W*Je BLUNDON, Memorial University of Newfoundland; JORDI DOU,

Barcelona, Spain; J,T8 GROENMAN, Arnhem, The Netherlands; R.C. LYNESS, Southwold,
Suffolk, England; D.J8 SMEENKf Zaltbommelf The Netherlands; and the proposer.

Editorfs comment.
A related problem stated and proved in Milne [5]: The locus of the foci of all
parabolas touching the three sides of a triangle is the circumcircle of the triangle,

REFERENCES

1, C.V* Durell, Projective Geometry, Macmillan, London, 1945, p, 203.

2e Ds Pedoe, A Course of Geometry for Colleges and Universities, Cambridge
University Press, New York, 1970, p. 355*
3, Charles Smith, An Elementary Treatise on Conic Sections^ Macmillan, London,
1927, p. 366*
48 D.M.Y. Sommerville, Analytical Conies* G. Bell and Sons, London, 1961,
p. 180.
5, W«Pe Milne, Homogeneous Coordinates9 Edward Arnold & Co., London, 192*4,
p. 118.
it it it

9111 [1984: 53.1 Proposed by Allan Wm. Johnson Jr,3 Washington3 D«C&
Solve the f o l l o w i n g synonymical alphametic \n the smallest possible base:

HARD
SHARP.
HARSH
 - 97 -
Solution by Friend E9 Kierstead3 Jr«3 Cuyahoga Falls9 Ohio.
From the column sums we have

D + P = H + bi9
2R = S + bj - i9
2A = R + bk - j 9

2H = A + bm - k9
S = H - m9

where b > 6 is the base and i9j3k9m are the carries. It is immediately obvious
that 777 = 1, so H = S+lo Then it is easily shown that

7S = (2>-2)(j+2fc+iO - I.

if i = 0, then either b = 9 (mod 7) or j =fe= l. If j =fc= l, then R = A = H

=fc-l,and there is therefore no solution* If & = 9 , then

S = j + 2k + 4, R = 5j +fe+ 2, A = 2j + 5fc + 1.

The only solution is with j = l,fc= o which, with H = S+l 9 gives

S = 5, H = 69 R = 7, A = 3, {D9 P} = {2, 4 h

If i = l , then £> * 99 nor can j = k = l . There are no solutions for fc < l l .

Therefore the smallest possible base is 9, and the only solutions in t h a t base
are
6372 6374
56374 and 56372.
63756 63756

Also solved by FRANK P. BATTLES, Massachusetts Maritime Academy, Buzzards Bay;

CLAYTON Wa DODGE, University of Maine at Orono; RICHARD I. HESS, Rancho Palos Verdes,
California? ANDY LIU, University of Alberta? J«A„ McCALLUM, Medicine Hat, Alberta;
GLEN E. MILLS, Pensacola Junior College, Florida; STANLEY RABINOWITZ, Digital Equip-
ment Corp., Nashua, New Hampshire; KENNETH M. WILKE, Topeka, Kansas; ANNELIESE ZIM-
MERMANN, Bonn, West Germany; and the proposer.
A A &
5C ** *%

915, [1984: 54] Proposed by Jack Garfunkel3 Flushings N.Y.

If x+y+z+w = 180°, prove or disprove that

sin (x+y) + sin (y+z) + sin (z+w) + sin (w+x) z sin 2x + sin 2y + sin 2z + sin 2w,

with equality just when x = y = s = w.

Solution by M.S. Klamkin3 University of Alberta.

The proposal as stated is easily disprovede For x = 270° 9 y = -90° 9 and
2 = w = 0° s the left side of the inequality is negative and the right side zero. How-
ever, the tnequsl ity may hold if some of x%y9z9w are negative, even if x+y+zw * 180°,
 - 98 -
and, as we will see from the generalization proved below* it always holds whenever
each of the four terms on the left is nonnegative, that iss whenever

2&:1TT < x+y < (2^ 1 +1)TT 9 2k2n £ y+z £ (2k2+l)°n9

2fc3TT < Z+W < (2fc3+l)7T9 2fei5.Tr < W+X < ( 2 ^ + 1 )TT,

where ki3k2^k39k^ are integers. In particular* it always holds if x^y^z^w > 0 and
x+y+z+w = 180°, with equality when x = y = z = w.
Let the angles x», i = l s 2 98 ..9n (with x A = xx) be such that

^ ^ ^+l ^

where the k. are integers. Then

sin (xi+x2) + sin (^2+^3) + ... + sin (x +xi) > sin 2a: 1 + sin 2x2 + ... + sin 2x . (

The proof is simple:

n
2 \ sin2j?a = )> (sin 2 ^ + sin 2x2) -2 1 sin (xi+x2) cos (xi~x2)
i=l cyclic cyclic

< 2 ][ sin tei+^2)9

cyclic

and (l) follows, with equality when each x.-x. is an integral multiple of 2TT. D
If all x. > 0 and xi+x2+.*.+x = 180°9 there is the following geometric inter-
u TL

pretation of (1). Let p = A1A2...A be a positively oriented convex polygon in-

scribed in a circle with center 0, and let /A.OA* A = 2a?.. Also, let Pl be the
associated convex polygon whose vertices A! are the midpoints of the arcs Ax.A. .
U Is Is T-L

Then (l) is equivalent to the statement

area of P1 > area of P.

This inequality appears in the authorss paper [l], and no doubt has also appeared
much earlier. It also follows immediately that

perimeter of P1 > perimeter of P.

Also solved by W»J. BLUNDON, Memorial University of Newfoundland; CLAYTON W a

DODGE, University of Maine at Orono; J*T. GROENMAN, Arnhemf The Netherlands; VEDULA
N. MURTYf Pennsylvania State University, Capitol Campus; and ESTHER SZEKERES, Turra-
murra^ New South Wales, Australia,, Comments (pointing out that the problem as
stated was incorrect) were received from RICHARD I. HESSf Rancho Palos Verdesf Cali-
fornia; FLORENTIN SMARANDACHE, Lycee Sidi El Hassan Lyoussi, Sefrou, Maroc; and
JORDAN B a TABOV, Sofia, Bulgaria.

REFERENCE

l. MS. K1amkinf "On Some Geometric Inequal itiesJis Mathematics Teacher, 60

(1967) 323-328.
 . 99 -
916 8 [ 1 9 8 4 : 54] Proposed by J.T. Groenman3 Arnhem3 The Netherlands.
Find 1im s if
n-*°° n
71
1

Solution by Frank P6 Battles3 Massachusetts Maritime Academy3 Buzzard9s Bay;

and Vedula Ne Murty3 Pennsylvania State University3 Capitol Campus (independently),
It is stated and proved in Hobson [1] that

1 e 1
a n — -= £ - cote.
I — ttan
fc=l 2 2
S e t t i n g 0 = TT/2 9 we then o b t a i n

Tim S = -.

Also solved by WaJ« BLUNDON, Memorial University of Newfoundland; CURTIS COOPER,

Central Missouri State University at Warrensburg; RICHARD I. HESS, Rancho Palos
Verdes, California (partial solution); WALTHER JANOUS, Ursulinengymnasium, Innsbruck,
Austria; FRIEND H» KIERSTEAD, JR., Cuyahoga Falls, Ohio; M.S* KLAMKIN, University of
Alberta; KEE-WAI LAU, Hong Kong; RICHARD PARRIS, Phillips Exeter Academy, New Hamp-
shire; BOB PRIELIPP, University of Wisconsin-Oshkosh; STANLEY RABINOWITZ, Digital
Equipment Corp*, Nashua, New Hampshire; and the proposer.

REFERENCE

1. E.W. Hobson, A Treatise on Plane and Advanced Trigonometry* Dover, New Yorks
1957s pa 129.
A A A
«r% *t ?t>

917 i [198^ 54] Proposed by Rick Moorhouse3 University of Toronto.

How can the eight vertices of a cube be divided into two sets of four
forming two directly congruent tetrahedra such that the four vertices of egch tetra-
hedron lie in the planes of the four faces of the other?
Solution by the proposer.
Let the eight vertices be named so that three of
the faces are ABCD\ CD 8 C 8 D, C f DA f B' (and consequently
ABCDA'B'C'D1 is a Hamiltonian path), as shown in the
figure. Then the two congruent "orthoschemes11 ABCD
and A'B'C'D' are the required "Mobius tetrahedra8\
Also solved by CLAYTON W. DODGE, University of
Maine at Orono; JORDI DOU, Barcelona, Spain; RICHARD
I. HESS/ Rancho Palos Verdes, California; ANDY LIU,
University of Alberta; RICHARD PARRIS, Phillips Exeter
Academy, New Hampshire; and STANLEY RABINOWITZ, Digital Equipment Corp a , Nashua, New
Hampshire.
 - 100 -
Editor?s comment9
An orthoscheme is the name given by Schlafli to a tetrahedron whose four faces
are all right triangles Cl9 p* 156]. "Mobius tetrahedra1' refers to the following
theorem of Mobius [1, p. 258]: If the four vertices of one tetrahedron lie respec-
tively in the four face planes of another^ while three vertices of the second lie
in three face planes of the first3 then the remaining vertex of the second lies in
the remaining face plane of the first.

REFERENCE

1. H*SSM. Coxeter, Introduction to Geometry, Wiley, New York, 1969.

& & &

CRUX MATHEMATICORUM
A Call for Editors

The mathematical community has been saddened to learn of the recent death of
Fred Maskell, who served as Managing Editor of Crux Mathematicorum for many years.
The Canadian Mathematical Society extends sympathy to Mr. Maskell's family and many
friends while acknowledging with sincere thanks the tremendous contribution both he
and Editor Leo Sauve have made to mathematics through their service with Crux.
Crux Mathematicorum seems now to have reached a crossroads in its existence
and the C„M»S9 has been asked to assume responsibility for Its publication. This Is
a unique problem-solving journal with an international reputation devoted to mathe-
matics at all levels and it would well complement the existing research publications
of the C«M.S9 The level of financing required is considerable, however. In addi-
tion, the Society is reluctant to begin a new volume year without the assurance that
it will have editors for years to come. Editors of the Canadian Journal of Mathe-
matics and the Canadian Mathematical Bulletin now serve five-year terms.
Editing a publication such as Crux Mathematicorum requires special and diverse
talents. An editor must have the inclination and ability to compile articles, pro-
blems and letters into a monthly issue of interest to a wide audience of mathematics
educators. Besides the amalgamation of submitted solutions to the same problem,
editorial comments and historical information are often necessary. In the future
operation of the journal, It is envisaged that word processing facilities at the
editorss institution would be virtually essential, thereby enabling the monthly dis-
patching of camera-ready copy to a managing editor!s office in Ottawa.
Suggestions concerning the future of Crux Mathematicorum and/or applications
for the editorship (commencing January 1986) should be sent as soon as possible to:
Professor Marcel Deruaz, Chairman,
CMS Ad Hoc Committee on Crux Mathematicorum,
Department of Mathematics,
University of Ottawa,
Ottawa, Ontario,
KIN 9B4
fc & &