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.
& & f t & * * f t f t & f t f t f t f t f t f t f t f t f t f t f t & f t f t f t f t £ f t f t f t f t f t f t f t &
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 ^ ^

ISSN 0705 - 0348

CRUX MATHEMATICORUI^

Vol.7, No. 7
August - September 1981

Sponsored by
Carlaton-Ottawa Mathematics Association Mathematique d'Ottawa-Carleton
Publie par le College Algonquin
The assistance of the publisher and the support of the Canadian Mathematical Olympiad
Committee, the Carleton University Mathematics Department,, the Ottawa Valley Education Liaison
Council, and the University of Ottawa Mathematics Department are gratefully acknowledged.
*****ft**ft*ft*****ft********4ft*ft*ft*********ft***ft*ft*

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

undergraduate levels for those who practise or teach mathematics. Its purpose is primarily
educational, but it serves also those who read it for professional, cultural, or recreational
reasons.
It is pxiblished monthly (except July and August) , The yearly subscription rate for ten
issues is $12.00. Back issues? $1*20 each. Sound volumes with index: Vols. 1&2 (combined),
$12.00f Vols. 3-6, $12.00 each. Cheques and money orders, payable to CRUX MATHEMATICORUM
(in US funds from outside Canada) , should be sent to the managing editor.
All communications about the content of the magazine (articles, problems, solutions, etc.)
should be 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, Architecture Department, Algonquin College, 281 Echo Drive, Ottawa,
Ontario,. KlS 1N3.
Managing Editor: F.G.B. Haskell, Mathematics Department, Algonquin College, 200 Lees Ave.,
Ottawa, Ontario, KlS 0C5,
Typist-compositor: Lillian Marlow,

* * *
CONTENTS

De 1 'enseignement des math£matiques Germaine de Sta£l 196

The Puzzle Corner Alan Wayne 198, 219
Three More Proofs of Routh's Theorem M.S. Klamkin and A. Liu 199
Problems - Problernes . ........ 203
206
Solutions
The Olympiad Corner: 27 M.S. Klamkin 220

Dan J. Eustice 1931 - 1981 22M-

- 195 -
 - 196 -

DE L'ENSEIGNEMENT DES P1ATHEMATIQUES

GERMAINE DE STAEL
Ce qui suit est un extrait du Chapitre XVIII de la Premiere Partie de
De L'Allemagne, par Madame de Stael, nee Anne-Louise-Germaine Necker (1766-1817).
L'ouvrage parut d'abord a Londres (en francais) en 1813, sa publication en France
ayant ete interdite par Napoleon. Le texte a ete tire de 1*edition en deux volumes de
Garnier-Flammarion (1968).

L'etude des langues, qui fait la base de 1'instruction en Allemagne, est beaucoup
plus favorable aux progr&s des faculty dans 1'enfance, que celle des mathematiques
ou des sciences physiques. Pascal, ce grand ggom£tre, dont la pens£e profonde
planait sur la science dont il s'occupait spgcialement, comme sur toutes les autres,
a reconnu lui-meme les defauts inseparables des esprits formes d'abord par les
mathgmatiques: cette etude, dans le premier a*ge, n'exerce que le m£canisme de
1'intelligence; les enfants que 1'on occupe de si bonne heure 5 calculer perdent
toute cette s$ve de 1'imagination, alors si belle et si fgconde, et n'acquierent point
a" la place une justesse d'esprit transcendante: car l'arithm£tique et I'algSbre se
bornent a" nous apprendre de mi lie maniSres des propositions toujours identiques.
Les probl§mes de la vie sont plus compliques; aucun n'est positif, aucun n'est
absolu: il faut deviner, il faut choisir, 3 1'aide d'apergus et de suppositions
qui n'ont aucun rapport avec la marche infaillible du calcul.
Les verit£s demontr£es ne conduisent point aux ve>ite*s probables, les seules
qui servent de guide dans les affaires, comme dans les arts, comme dans la socie"te\
II y a sans doute un point oQ les math&natiques e11es-m§mes exigent cette puissance
lumineuse de 1'invention sans laquelle on ne peut p£n£trer dans les secrets de la
nature: au sommet de la pens£e 1'imagination d'Homdre et celle de Newton semblent
se rSunlr, mais combien d'enfants sans g£nie pour les mathgmatiques ne consacrent-
ils pas tout leur temps a cette science! On n'exerce chez eux qu'une seule faculty,
tandis qu'il faut de*velopper tout l'§tre moral dans une £poque ou T o n peut si
facilement deYanger l'ame comme le corps, en ne fortifiant qu'une partie.

Rien n'est moins applicable a la vie qu'un raisonnement mathgmatique. Une

proposition en fait de chiffres est decidement fausse ou vraie; sous tous les autres
rapports le vrai se mele avec le faux d'une telle manieYe, que souvent 1'instinct
peut seul nous decider entre les motifs divers, quelquefois aussi puissants d'un cote
que de 1'autre. L'gtude des mathematiques, habituant a" la certitude, irrite contre
toutes les opinions oppos£es a la notre; tandis que ce qu'il y a de plus important
pour la conduite de ce monde, c'est d'apprendre les autres, c'est-5-dire de
 - 197 -
concevoir tout ce qui les ported penser et a sentir autrement que nous. Les
mathematiques induisent I ne tenir compte que de C G qui est prouve*; tandis que
les venue's primitives, celles que le sentiment et le genie saisissent, ne sont pas
susceptibles de demonstration.
Enfin les mathematiques, soumettant tout au calcul, inspirent trop de respect
pour la force; et cette energie sublime* qui ne compte pour rien les obstacles et
se platt dans les sacrifices, s'accorde difficilement avec le genre de raison que
developpent les combinaisons algebriques.
II me semble done que, pour lfavantage de la morale, aussi bien que pour celui
de 1'esprit, il vaut mieux placer 1*etude des mathematiques dans son temps, et comme
une portion de 1 * instruction totale, mais non en faire la base de Teducation, et
par consequent le principe determinant du caractere et de 1'ame.
Parmi les systemes d'education, il en est aussi qui conseillent de commencer
1'enseignement par les sciences naturelles; elles ne sont dans l'enfance qu'un
simple divertissement; ce sont des hochets savants qui accoutument a s'amuser avec
methode et a* etudier superficiellement. On s'est imagine qu'il fall ait, autant qu'on
le pouvait, eviter de la peine aux enfants, changer en deiassement toutes leurs
etudes, leur donner de bonne beure des collections d'histoire naturelle pour jouets,
des experiences de physique pour spectacle. II me semble que cela est aussi un
systeme errone. S'il etait possible qu'un enfant apprft bien quelque chose en
s'amusant, je regretterais encore pour lui le developpement d'une faculte, 1'attention,
faculte qui est beaucoup plus essentielle qu'une connaissance de plus. Je sais qu'on
me dira que les mathematiques rendent particulierement applique; mais elles
n'habituent pas a" rassembler, apprecier, concentrer: Tattention qu'elles exigent
est pour ainsi dire en ligne droite: 1'esprit humain agit en mathematiques comme un
ressort qui suit une direction toujours la meme.
Ljeducation faite en s'amusant disperse la pensee; la peine en tout genre est
B
un des grands secrets de la nature: Tesprit de 1 enfant doit s'accouturner aux
efforts de l'etude, comme notre ame a la souffrance. Le perfectionnement du premier
age tient au travail, comme le perfectionnement du second a* la douleur: il est a"
souhaiter sans doute que les parents et la destinee n'abusent pas trop de ce double
secret; mais il nly a d'important a" toutes les epoques de la vie que ce qui agit
sur le centre meme de 1'existence, et 1'on considere trop souvent 1'etre moral en
detail. Vous enseignerez avec des tableaux, avec des cartes, une quantite de choses
a" votre enfant, mais vous ne lui apprendrez pas a" apprendre; et 1'habitude de
s'amuser, que vous dirigez sur les sciences, suivra bientOt un autre cours quand
l'enfant ne sera plus dans votre dependance.
 - 198 -
Ce n'est done pas sans raison que l'etude des langues anciennes et modernes a
ete la base de tous les etablissements d'education qui ont forme les hommes les
plus capables en Europe: le sens d'une phrase dans une langue etrangeYe est a la
fols un probl&me grammatical et intellectuel; ce probleme est tout a fait
proportionne a Tintelligence de 1'enfant: d'abord il n'entend que les mots, puis
il s'ei&ve jusqu'a la conception de la phrase, et bientot apr&s le charme de
Texpression, sa force, son harmonie, tout ce qui se trouve dans le langage de
Thomme, se fait sentir par degres a 1'enfant qui traduit. II s'essaie tout seul
avec les difficulty que lui presentent deux langues a la fois, il s'introduit dans
les idees successivement, compare et combine divers genres d'analogies et de
vraisemblances; et l'activite spontanea de 1'esprit, la seule qui developpe vraiment
la faculte de penser, est vivement excitee par cette etude. Le nombre des facultes
qu'elle fait mouvoir a la fois lui donne Tavantage sur tout autre travail, et T o n
est trop heureux d'employer la rnemoire flexible de 1'enfant a retenir un genre de
connaissances, sans lequel il serait borne toute sa vie au cercle de sa propre
nation, cercle etroit comme tout ce qui est exclusif.
L'etude de la grammaire exige la meme suite et la meme force d'attention que
les mathematiques, mais elle tient de beaucoup plus pr£s a* la pensee. La grammaire
lie les idees T u n e a l'autre, comme le calcul enchatne les chiffres; la logique
grammaticale est aussi precise que celle de TalgSbre, et cependant elle s'applique
a tout ce qu'il y a de vivant dans notre esprit: les mots sont en m§me temps des
chiffres et des images; ils sont esclaves et libres, soumis a la discipline de la
syntaxe, et tout-puissants par leur signification naturelle; ainsi T o n trouve dans
la metaphysique de la grammaire, Texactitude du raisonnement et 1'independance
de la pensee reunies ensemble; tout a passe par les mots et tout s'y retrouve
quand on sait les examiner: les langues sont inepuisables pour 1'enfant comme pour
Thomme, et chacun en peut tirer tout ce dont il a besoin.
* * *

THE PUZZLE CORNER

Puzzle No. 1: Transposals (8)

Compute the size of ONEs in space

By TWO each height and half each base.
A THREE will do, in any case.
ALAN WAYNE

* fc fe
 - 199 -
THREE MORE PROOFS OF ROUTH'S THEOREM
M.S. KLAMKIN and A. LIU
The sides BC,CA,AB of a triangle ABC are divided at D,E,F, respectively, in
the ratios

BD : DC = r : 1, CE : EA = s i 1, AF : FB = t : 1,
and the intersections of the cevians AD,BE9CF are the vertices of triangle GHK, as
shown in Figure 1. With square brackets denoting areas of triangles, it is a known
result of Routh that

[GHK] (rst - l ) 2
(1)
TABUI (st+s+l)(tr+t+1) (rs+r+1)

A number of different proofs of this elementary

but interesting result appear in the literature:
e.g.9 Routh Cl] (not available to us), Coxeter
C2] ? Kay [3], Melzak O ] , Niven [5]; and the re-
sult appears as an exercise in Pedoe [63, with a
hint for a solution. Coxeter's proof is via bary-
centric coordinates, and although not short, it is
valid for external as well as internal division of
the sides (i.e., r3sst need not be positive). Kay's B
proof is synthetic but rather long. Melzak uses areas Figure 1
and then, to obtain certain ratios of segments, he
orthogonally projects the triangle onto an isosceles right triangle. (Routh's re-
sult is affine invariant; consequently, it suffices to show that it is valid for
any particular triangle.) Niven8s proof uses coordinate geometry; it could have
been simplified somewhat if he had started with an isosceles right triangle. The
proofs of Kay, Melzak, and Niven assume that r,s,t are nonnegative, but analogous
proofs can easily be devised when some of pss9t are negative.
Here we present three proofs different from those we have found in the literature.
The first one is vectorial, the second is "reverse" vectorial, and the third is a
rather simple synthetic proof. Although we have not come across the vectorial proof
before, by virtue of its directness it may already be somewhere in the literature.
The remaining two proofs appear to be new.

Vectorial Proof*
Let A denote the vector from an origin outside the plane of triangle ABC to
the point A, etc. It then follows that
 200
(2)

= C-y{(C-B) - ™ < A - B ) } ,
-»- -*- ->•
where # and ?/ are unknown scalars. Since A,B,C are linearly independent, we can
equate corresponding coefficients of A,B,C, giving
x _ ty rx
r+1 " t+13 r+1 1-y

(equations which are not independent). Solving for #, we find

_ t(r+l)
tr+t+l *

and thus, from (2),

(*r+*+l)H = A + *B + trC.
•¥ + -+
-*--»--). -> -v -»-
Then by symmetry (cyclical interchange of A,B,C; r,5,t; and G,H,K) we obtain
(rs+r+l)K = B + rC + rsA,

(st+s+-l)(j = C + st + stB.

Since the area of triangle GHK is given by

2CGHK] = |(H-G) x (K-G)l = |(HXK") -t-(KxG) + (GxH) |, (3)

with a similar expression for [ABC], we have

[GHK] = |(HxK)+ (KxG)f (GxH)j
W
[ABC] KBxC) + (CxA) + (AxB)l'

On carrying out the calculations for (3), it will be found that W reduces to (1)
(or to its absolute value if r>s9t are arbitrary real).

"Reverse" Vectorial Proof.

This derivation was suggested by Problem 5 in
the Fifth Wisconsin Mathematical, Engineering and
Science Talent Search (March 1979), where it was
given that r r = # = 3 = l i n Figure 2, and it
was required to show that [ABC] = 7[A , B'C'].
Here we have

B'C = a \ C'A' = b\ A'B' = o%

and

AB* = xa\ B C = yb\ CA' = zo\

u
Figure 2
 201
It follows immediately that

[ABC 3 = z/rAC'A*] = t/(l+.T)rA'B,C,]5 etc.,

and then that

TABC1
FfligiQ; n = 3(1+*/) + #(1+3) + y(l+x) +1

= (a?+l)Q/+l)(j3+l) -xyz. (5)

We must now determine x9y9z in terms of r,s9t.
First observe that, again with the origin of vectors outside the plane of tri-
angle ABC, we have

A = B' + aKB'-C) = 0A'+ (l+ar)B'- xC

B = C' + z/CC'-A') = -i/X' + oB' + (l+*/)C* (6)

C = A' + 3(A'-B') = (l+z)P- sB' + oC\

Now if, in solving (6) for A',B',C° by Cramer's Rule, we find that B' = fc(aA+eB+YC),
then

D = fc'CpB+yC) and r = y/3, etc.

Solving (6) for B', we get

0 A -a:
-*- ->-
B* = fe -y B i+z/
1+2 C 0 1

Thus e = #(1+3), y = xys so r = t//(l+s) and, by symmetry, s = s/(l-kc), £ = ar/(l+z/).

Solving for a?, we get

t(rs+r+l) A tr+t+l
x
" 1-rst 9 * + 1 ~ 1-rst ;
then, again by symmetry,
•Kst+g+1) rs+r+1
y 1-rst y+1 1-rst
sitr+t+1) st+s+1
s -f J =

Finally, after simplification (5) is found to be equivalent to

[A'B'C'3 (rst-1)2 .
[ABC] (st+s+l)(trit+l){rs+r+l)
 - 202 -
Synthetic Proof.
We ignore for now the cevian BE of Figure 1, and
join B to H, the intersection of cevians AD and CF,
as shown in Figure 3. We have

CCAH] = [CAD] - CCHO]

= (CABD] - [HBD])/r

= (CAFH] + CBHF])/r

= CAFH](1+*)/tr.
Since also Figure 3

tCABC]
CCAF] = [CAH] + [AFH] = CAFH](*r+t+l)/tr,
1+t
it follows that
rAFH1 . , , rt 2 [ABC]

and so

CCAH] = t[ABC3
tr+t+l
Then, by symmetry from Figure l,

CABK] =
rCABC] and [BCG] = aCABC]
rs t-r+1 st+s+1
Finally,

[GHK]
OTcJ" rs+r+1 st+s+l tr+t+l
which reduces to (l) after combining terms and simplifying. 0
The last two proofs can easily be adapted to take care of arbitrary real values
of r,s9t.
Niven [5] had noted that the denominator in formula (l) vanishes if and only
if two or more of the cevians AD,BE,CF are parallel. One can view this case as one
for which the area of triangle GHK is unbounded. Also, if rst = l and one factor of
the denominator in (1) vanishes, then the other two factors of the denominator also
vanish. This case occurs when the three cevians are parallel (or are concurrent at
infinity).
REFERENCES
1. E.J. Routh, A Treatise on Analytical Statics3 with Numerous Examples, Vol. 1
(2nd Ed.), Cambridge University Press, London, 1896, p.82.
 - 203 -
2. H.S.M. Coxeter, Introduction to Geometry, Second Edition, John Wiley & Sons,
Hew York* 1969, pp. 211, 219-220.
3. David C. Kay, College Geometry, Holt, Rinehart and Winston, New York, 1969,
pp. 205-207.
M-. Z.A. Melzak, Companion to Concrete Mathematics, John Wiley & Sons, New York,
1973, pp. 7-9.
5. Ivan Niven, "A New Proof of Routh's Theorem," Mathematics Magazine, 49 (1976)
25-27.
6. D. Pedoe, A Course of Geometry for Colleges and Universities, Cambridge
University Press, 1970, p. 55, Exercise 11.8.
Department of Mathematics, University of Alberta, Edmonton, Alberta, Canada T6G 2G1.
* * *

PROBLEMS--PROBLEMES
Problem proposals and solutions should be sent to the editor, whose address
appears on the front- page of this issue. Proposals should, whenever possible, be
accompanied bg 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 be-
fore January 1, 1982, although solutions received after that date will also be con-
sidered until the time when a solution is published.

657a Proposed by Ngo Tan, student, J.F. Kennedy U.S., Bronx, N.I.
A quadrilateral ABCD is circumscribed about a circle y and
y n AB,BC,CD,DA = E9FSG,H9 respectively.
Let AC n BD = I and
AH = AE = a, BE = BF = b, CF = CG = o, DG = DH = d.
Prove that
IA a a . IB b
TC = a and
TD = d
658 f Proposed by Charles W. Trigg, San Diego, California.
In the decimal system, do there exist consecutive squares that have the
same square digit sum?
 - 201 -
659. Proposed by Leon Bankoff, Los Angeles, California,
If the line joining the incenter I and the circumcenter 0 of a triangle
ABC is parallel to side BC, it is known from an earlier problem in this journal
(CrUX 388 [1979 : 201]) that
o2 _ (2fl-r)2(i?+r) .
8
F?
In the same configuration, cosB + cos C = 1, and the internal bisector of angle A
is perpendicular to the line joining I to the orthocenter H (Mathematics Magazine,
Problem 758, 43 (November 1970) 285-286).
Prove the following additional properties, which were listed but not proved in
the last-mentioned reference:
(a) If the internal bisector of angle A meets the circumcircle in P, show
that AI/AP = cos A.
(b) The circumcircle of triangle AIH is equal to the incircle of triangle ABC.
(c) A M P = 2ffr = vfc-AI-BI-CI .
2 + s1r,2 = J; c o s 2 + cos2
(d) sin ! | f |= I *
(e)' tan^2 = f ^ -
R+r

6601 Proposed by Leon Bankoff, Los Angeles, California,

Show that, in a triangle ABC with semiperimeter s, the line joining the
circumcenter and the incenter is parallel to BC if and only if

s = (DL + EM + FN) t a n | ,

where L,M,N bisect the arcs BC,CA,AB, respectively, of the circumcircle and D,E,F
bisect the sides BC,CA,AB, respectively, of the triangle.

661 • Antrag von Alan Wayne, Holiday, Florida.

Zu finden: die einzige entsprechende Losung der jeder der folgenden
unabhangigen Additionen, welche die Sechs als Basis ihrer Aufzahlung haben.

(a) EINS (b) ZWEI

EINS , ZWEi .
ZWEI VIER
662« Proposed by Kaidy Tan, Fukien Teachers' University, Foochow, Fukien, China.
An isosceles triangle has vertex A and base BC. Through a point F on AB,
a perpendicular to AB is drawn to meet AC in E and BC produced in D. With square
brackets denoting area, prove synthetically that
[AFE1 = 2CCDE] <=> AF = CD.
 - 205 -
6631 Proposed by Allan Wm. Johnston Jr., Washington, D.C.
Prove that every fifth-order pandiagonal magic square can be written in
the form

1 EfF_I
B
A

C+H-I
B+G-I

D+E-I
H

A+F-I
C+D-I

A+D-I F+G-I I BhC-I E+H-I

C B+E-I A+H-I D+G-I F

G+H-I D CiF-t E A+B-I

664 • Proposed by George Tsintsifas, Thessaloniki, Greece.

An isosceles trapezoid ABCD9 with parallel bases AB and DC 9 is inscribed
in a circle of diameter AB. Prove that AC > (AB + D O / 2 .
665 • Proposed by Jack Garfunkel, Queens College, Flushing, N.Y.
If A,B»C,D are the interior angles of a convex quadrilateral ABCD, prove
that

/ i r c o s ^ Ecotf

(where the four-term sum on each side is cyclic over A,B,C,D), with equality if and
only if ABCD is a rectangle.
6661 Proposed by J.T. Groenman, Arrihem, The Netherlands.
The symmedians issued from vertices A,B,C of triangle ABC meet the opposite
sides in D,E,F, respectively. Through D,E9F9 lines d9e9f are drawn perpendicular to
BC,CA,AB, respectively. Prove that d9e9f are concurrent if and only if triangle ABC
is isosceles,
6671 Proposed by Dan Sokolowsky, California State University at Los Angeles.
A plane is determined by a line D and a point F not on D. Let C denote
the conic consisting of all those points P in the plane for which PF/PQ = r, where
PQ is the distance from P to D and r > o is a given real number.
Given a line L in the plane, show how to determine by elementary means the
intersections (if any) of L and C.
6681 Propsed by Michael W, Ecker, Pennsylvania State University, Worthington
Scranton Campus.
For any natural number n9 let s{n) be the sum of the proper divisors of n (i.e.,
s{n) - a(n) -n, where a(n) has its usual number-theoretic meaning of sum of all posi-
tive divisors of n). A set E of natural numbers is said to be imperfectly amicable
 - 206 -
if eia) = s(b) for all a9b e E, i.e., if s{E) = {k} for some nonnegative integer k.
An imperfectly amicable set is said to be maximal if none of its proper supersets
is imperfectly amicable. Find all infinite maximal imperfectly amicable sets of
natural numbers and prove there are no others.
ft & s'«

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.

5611 [1980: 211] Proposed by Alan Wayne3 Pasco-Hernando Community College, New
Port Rickeys Florida.
Det gjjrfr ikke noget at vaere fjrfdt i Andegaarden, naar man kun har ligget i et
Svaneaeg!
HANS CHRISTIAN ANDERSEN
(It doesn't matter if you are born in a duckyard, if you come from a swan's
egg!)
But usually, in a duckyard,

PEEP x PEEP = DUCKLING.

Restore the digits in this alphametical multiplication.

I. Solution by the proposer.

This is an "ideal" alphametic, where appear all ten digits, whose sum is 45 = o
(all congruences, here and later, are modulo 9 ) . If P + E E X9 where o < x < 8, then
(PEEP) 2 + PEEP = DUCKLING + PEEP
implies that (2ar)2 + 7x = x and xi^x+l) = o. Thus x - o or 2 and, since clearly P > 3 ,
we must have P + E = 9 or 11. Observing that P * 5 or 6 (otherwise P = G) and E * o
(otherwise DU = NG), we find the only possibilities for PEEP to be
3663, 4554, 7227, 8:118, 3883, 4774, 7447, 8338, 9229.

Trying to keep direct calculations to a minimum, in the spirit of the game, we

eliminate the four underlined possibilities because in each case there is a duplica-
tion of digits between EP and the last two digits of (EP) 2 . Squaring the remaining
five possibilities shows that the unique answer is

4554 x 4554 = 20738916. D

Some time ago, I spent almost a month in Copenhagen, and saw performances of
Den Grimme Aelling (The Ugly Duckling). In Denmark the ducks do not go "QuackI Quack!"
They go "Rap! Rap!"
 - 207 -
II. Solution by Clayton W, Dodges University of Maine at Orono.
Vir must haven dar PEEP > 3162 and dar DUCKLING < 99980001. Als0 dar P * i 5 5, 6 s
Or 0. Dan vir searchen mit dar programmable calculator, s0 har ligget dar unique
Solution 45542 = 21738916 tsicl.

Also solved by JAMES BOWE, Erskine College, Due West, South Carolina; ALLAN WM.
JOHNSON JR., Washington, D.C.; FRIEND H. KIERSTEAD, JR., Cuyahoga Falls, Ohio; FRED
A. MILLER, Elkins, West Virginia; HERMAN NYON, Paramaribo, Surinam? BOB PRIELIPP,
University of Wisconsin-Oshkosh; ROBERT TRANQUILLE, College de Maisonneuve, Montreal,
Quebec; CHARLES W. TRIGG, San Diego, California; and KENNETH M, WILKE, Topeka, Kansas.

Editor's comment. Two solvers gently chided the editor for publishing this pro-
blem. (To preserve their anonymity, their names have been omitted from the above list
of solvers.) One of them (who sent in only an answer) wrote: "...it was a bad alpha-
metic because its complete solution entailed a quite undue amount of mere numerical
trial." The other wrote: "P can obviously not be 0, 1, 5, 6 as these would give an
identical value for G. P is likewise not 2 since the square root of an 8-digit
number must be greater than 3000. From that point on there is little scope for
reasoning...Therefore, this is a problem which few afficiandoes Lsicl of the form
will regard as good."
The editor refers readers to his remarks following the solution of Crux 551
[1981 : 1811. Whenever their solution entails "a quite undue amount of mere numerical
trial", our two solvers, and other "afficiandoes" of the form, might consider the
possibility that their solution, and not the problem, is at fault. Certainly, in
this problem there was plenty of "scope for reasoning", as our solution I shows,
and the unavoidable residue of numerical trial was very modest indeed.
Our second solver's "programmable calculator" laid an egg. It could apparently
not be "programmed" to lay square eggs. Rap! Rap! Wanna buy a duck?
* * >'*

562 • C1980: 2111 Proposed by Dan Sokolowsky, [now at ] California State Uni-
versity at Los Angeles.
Given is a circle y with center 0 and diameter of length d, two distinct points
P and Q not col linear with 0, and a segment of length I, where o<l<d. Construct
(specify when possible) a circle through P and Q which meets y in points C and D
such that chord CD has length I.
Solution by Kesiraju Satyanarayanas Gagan Mahal Colony, Hyderabad3 India (re-
vised by the editor).
We do not assume that 0,P,Q are not collinear, since this condition has no
bearing on the existence of solutions, as will become apparent. We first dispose
 - 208 -
of a trivial case. Suppose P and Q are both on y. If IPQI = I* any circle through
P and Q is a solution; and if jPQ| # l9 then y itself is, in a sense, the only solu-
tion. We now assume that P and Q are not both on y.
The circle 8 with centre 0 and radius \M*-lz is easily constructed. Its radius
is OM, where M is the midpoint of any chord of y of length I. A chord of y has length
I if and only if it is tangent to 3. Let a be any circle through P and Q, and let p
be the radical axis of y and a. (For ease of construction, it is best to take for a
a circle which intersects y, for then p is simply the line of support of their common
chord.) We must now consider two cases.
i) If p is parallel to PQ (in which case this relation holds for any choice of
a), there are two tangents to B parallel to p. If either one of them intersects Y
in C and D, then PQCD is an isosceles trapezoid, hence cyclic, and its circumcircle
is an answer to our problem. Thus there are two solutions in this case, unless l=d
when the two solutions coalesce into one.
ii) If p n PQ = R (in which case R is independent of the choice of a since it
is the radical centre of y and any two circles through P and Q ) , suppose a tangent
to B through R meets y in C and D. The powers of R with respect to circles y and a
are equal; hence RP-RQ = RC-RD, and the points P,Q,C,D lie on a circle which is a
solution to our problem. Thus there are two solutions if R is outside B, one if R
is on B, and none if R is inside B.

Also solved by DONALD AITKEN, Northern Alberta Institute of Technology, Edmonton,

Alberta; JORDI DOU, Escola Tecnica Superior Arquitectura de Barcelona, Spain^ J.T.
GROENMAN, Arnhem, The Netherlands; GEORGE TSINTSIFAS, Thessaloniki, Greece; and the
proposer.
* * *
5631 C1980: 211] Proposed by Mioliael W, Ecker, Pennsylvania State Universitys
Worthington Scranton Campus.
For n a positive integer, let (a 1 ,a 2s ...,a ) and (bi9bl9m..9b ) be two permuta-
tions (not necessarily distinct)of (i,2,...,«). Find sharp upper and lower bounds for

axbx +a2b2+ ... +«„&„•

I. Solution by Paul R. Beesaeks Carleton Universitys Ottawa, Ontario,

This is a special case of a famous theorem on rearrangements of any two sequences
(a\9...9a ) and (b\9...9b ) of real numbers [1]. By renumbering we may assume that

a\ < a2 <, ... < a and b\ < bo < ... < b .

1 l
* n * n
The theorem then says that if TT denotes any permutation of the set {l,2,...,n}, we
have
 - 209 -
a b +a b
ln 2 n-l+-'+anbl * E a b
k *(k) ' aibl +
^2+ --• +«nV (D

(All summations, here and later, are for k = l,2,...,n.) As noted in [1], if one
interprets thefc.'sas weights and the a.'s as distances from one end of a rod at
which the weights are to be suspended, then (1) states that the maximum first moment
with respect to that end of the rod occurs when the heaviest weights are suspended
furthest from that end, and the minimum moment when the lightest weights are so
suspended. Although this intuitive justification of (l) requires positive a.'s and
&. 8 s, (1) is actually valid for all real a.3b..
In the given case, with a,~b«- k3 (1) gives
E k(n-k+l) £ Efcir(fc)< E k2
or
tt(ft+l)(n+2) < E fa(k) < n(n+l)(2n+l) (2)
6 6
For completeness we give a proof of (2) which is essentially the proof given
of (1) in [1] for real a.3b.. To obtain the upper bound in (2), suppose TT is a
permutation of the set {1,2,,.,,??} for which i < 3 but -n(i) > TT(J) for some pair
( i 3 j ) . Let TT! be the permutation obtained from -n by interchanging -n(i) and -n(j)s
so that ir'ii) ~ ir(j)9 TT'(J) = ir(t), and u*(k) = ir(fc) for all k * i3j. Then
A' = Efcir'(fc)> E kv{k) = A

since
4' -4 = iftU) + J * U ) -i-n(i) -Jir(j) = (j-i){*(i) ' ir(j )) > 0.
This shows that the sums 4 = Efcir(fc)can be increased if a pair (i,j) of the above
kind exists, so the largest value of A occurs when no such pair (i,j) exists. Hence
the upper bound in (2) follows, and a similar proof works for the lower bound, as
well as for (1).
II. Solution by the St. Olaf College Problem Solving Group, Northfield, Minnesota.
Let a = {au...9a ) and b = ( b u . . . 3 b n ) . By the Cauchy-Schwarz inequality,
2
<z-2> < Hall ||2?|| = l + 22 + ... + nz = •g , (3)

with equality if and only if a =fc^andthe required maximum is the right member of
(3). Now let
a% = (n+l-aj, n+l-a2* •••.» w+l-a );
then
2>-a' = E bk(n+l-ak) = (w+1) E ^ - Z>-a =n(n|1} - Z>-a,
 - 210 -
where the summations are for k = i,2,.,.,n. The b which minimizes this is the b
which maximizes b*a9 namely b = a. Hence the required minimum is

n n% n(n+l)2 n(m-l)(2n+l) _ n(n+l)(n+2)

a
2 "* 6 ~ 6

III. Solution by V.N. Marty, "Pennsylvania State University, Capitol Campus,

Middletown, PennsyIvania.
We take a statistical approach. Here we have, with summations for i = l,2,...,n,
to. = th. = "Mil, Ia2 a E ^2 = n(n+l)(2n+l)
^ v 2 -v ^ 6
and, with the usual notation for the mean,
Z(a.-a)* = E( K^)2 = «0^1I.

The correlation coefficient of (au...9a ) and (bu.. .5Z> ) is given by

£ ( a . -a)(b. -b) Za.b. - (Za.)(Zb.)/n

i>{E(a. -a) z }{I(fc. -b)zl n(n2~l)/12

or
Za.b. ~ n(n + l) 2 /'+
i- ^ .
p =
n(nz - D / 1 2
With this value of p, the known relation -1 < p ^ 1 is equivalent to
nin -H)(n + 2) < ra ^ < n(n + l)(2n + 1)
6 i i 6
The upper bound is attained when b . = a. for i = i,2,...,w and the lower bound is
attained when b . - n + 1-a. fort = i,29. ,„n,
Also solved by JORDI DOU, Escola Tecnica Superior Arquitectura de Bracelona,
Spain? ERNEST W. FOX, Marianopolis College, Montreal, Quebec; ALLAN WM. JOHNSON
JR., Washington, D.C.; FRIEND H. KIERSTEAD, JR., Cuyahoga Falls, Ohio; KESIRAJU
SATYANARAYANA, Gagan Mahal Colony, Hyderabad, India;' ROBERT A. STUMP, Hopewell,
Virginia; GEORGE TSINTSIFAS, Thessaloniki, Greece; and the proposer. Comments
were received from M.S. KLAMKIN, University of Alberta; and L.F. MEYERS, The Ohio
State University.

Editor's comment.
Fox and Klamkin also gave reference [11 and Meyers referred to a related pro-
blem in [2]. Murty formulated another related problem which will appear in the next
issue of this journal.
REFERENCES
1. G.H. Hardy, J.E. Littlewood, G. P61ya, I n e q u a l i t i e s , Cambridge University
Press, 1952, p. 261, Theorem 368.
2. David Singmaster, Solution to Problem 1059 (proposed by David E. Daykin),
Mathematics Magazine, 53 (March 1980) 115-116.
 - 211 -
5641 C1980: 212] Proposed by A. Liu, University of Alberta.
Raymond M. Smullyan's delightful IJhat is the name of this book? is now
available in paperback form from Prentice-Hall. On page 141 is the following Pro-
blem 155:
There are four doors X, Yf Z, W leading out of the Middle Sanctum. At least
one of them leads to the Inner Sanctum. If you enter a wrong door, you will be de-
voured by a fierce dragon.
Well, there were eight priests A, B, Cf D, E, F, G, H, each of whom is either
a knight or a knave. (Knights always tell the truth and knaves always lie.) They
made the following statements to the philosopher:
A: X is a good door.
B: At least one of the doors Y, Z is good.
Ct A and B are both knights.
D: X and Y are both good doors.
Es X and Z are both good doors.
F: Either D or E is a knight.
Gz If C is a knight, so is F.
H; If G and I are both knights, so is A.
Smullyan's problem: Which door should the philosopher choose?
My problem: The philosopher lacked concentration. All he heard was the first
statement U ' s ) and the last statement (#as) plus two fragments:
C: A and ... are both knights.
G: If C is a knight, ... .
Prove that he had heard enough to make a decision.
Solution by Clayton W. Dodge3 University of Maine at Orono,
Given that these four comments only are heard simplifies the problem, since the
extraneous material is not present. If H is lying, then G and H are both knights
and A is not. This is a contradiction, so H is a knight. Hence, if G is also a
knight, then A must be a knight. If G is not a knight, then £'s statement is false,
which can only occur if C is a knight (and the conclusion to G's statement is false).
Hence C is a knight and A is a knight. Therefore, in either case, A is a knight, so
X is a good door.
Also solved by RICHARD A. GIBBS, Fort Lewis College, Durango, Colorado; LEROY F.
MEYERS, The Ohio State University; BOB PRIELIPP, University of Wisconsin-Oshkosh;
and the proposer.
* & *
565. [1980: 212] Proposed by Jack Garfunkel, Flushings N.Y.
In an acute-angled triangle ABC, the altitude issued from vertex A CB,C]
meets the internal bisector of angle B [C,A] at P CQ,R]. Prove that
AP • BQ-CR = AI - BI • CI,

where I is the incenter of triangle ABC.

 - 212 -
I. Solution by George Tsintsifas, Thessalonikis Greece.
Applying the law of sines to triangle API, we get
AP AI
s7nW = sin API ' (1)

Now
/AIP = 90° ± | and /API = 9 0 ° ? | ,

the choice of upper or lower signs depending on the relative magnitude of angle A
compared with angles B and C. In any case (l) becomes
AP_ = _M , .
K}
cos(C/2) cos (B/2)
By cyclic permutation of A,B,C and P,Q,R5 we obtain from (2)
BQ BI . CR CI
cos (A/2) cos (C/2) cos (B/2) cos (A/2) '
and AP * BQ • CR = AI • BI • CI follows immediately.
II. Solution by Jordi Dous Escola Tecnica Superior Arquitectura de Barcelona^
Spain.
Let P' CQ'jR'] be the intersection of the altitude from A [B»C] and the internal
bisector of angle C [A,B]. The bisectors of angles B and C are antiparallels with
respect to the angle formed by the altitude from A and the bisector of angle A. It
follows that AI 2 = AP • AP', and similarly we have BI 2 = BQ • BQ' and CI 2 = CR • CR'.
Hence
(AI-BI-CI) 2 = (AP-BQ-CR) • (AP1-BQ'-CR') .
Now triangles ACR and ABQ' are similar, so CR/BQ* = AC/AB, and similarly AP/CR' = BA/BC
and BQ/AP' = CB/CA. Thus
AP-BQ-CR ,
AP'-BQ'-CR' = 1 '
and it follows that
AI-BI-CI = AP-BQ-CR = AP'-BQ'-CR' . (3)
Also solved by W.J. BLUNDONf Memorial University of Newfoundland; KESIRAJU
SATYANARAYANA, Gagan Mahal Colony, Hyderabad, India; CLAYTON W. DODGE, University
of Maine at Orono; J.T. GROENMAN, Arnhem, The Netherlands; NGO TAN, student, J.F.
Kennedy H.S., Bronx, N.Y.; G.C. GIRI, Midnapore College, West Bengal, India; and
the proposer.

Editor's comment.
Not one solver questioned the necessity for the condition that the given tri-
 - 213 -
angle be acute-angled, and not one used this condition explicitly in his proof. The
editor would like to see one good reason why the result (3) should not be true for
every triangle.
* & *

566. C1980: 212] Proposed by Kenneth S. Williams, Carleton Universitys Ottawa.

Characterize those positive integers which can be expressed as a square
plus the product of two consecutive integers.
Examples: 24 = 2 2 + 4 • 5, 637 = 25 2 f 3 • 4.

Solution by Viktors Linis, University of Ottawa.

If the positive integer N is of the form
N = a2+b(b+l), (1)
where a and b are nonnegative integers not both zero, then 4/7+1 = (2a)2+(2£+l)2. Con-
versely, if an odd integer greater than l is a sum of two squares, it must be of the
form (2a 1 ) 2 +(2&*+l) 2 = HN+l9 where a' and bl are nonnegative integers not both zero,
from which N = (a' ) 2 +&'(£>' + 1 ) . Therefore a positive integer N is of the form (1)
If and only if HN+l is a sum of two squares.
It is known that a positive integer is representable as a sum of two squares
if and only if it can be written in the form n\n<i9 where the square-free positive
integer n2 has no prime factor of the form Mm+3. (Four proofs of this theorem can
be found in Hardy and Wright [13.) We conclude that N is of the form (l) if and
only if w+l satisfies the italicized condition, that is, if and only if all prime
factors of the form Um+3 have even exponents in the canonical factorization of w + 1 .
Also solved by PAUL R„ BEESACK, Carleton University, Ottawa, Ontario? W.J.
BLUNDON, Memorial University of Newfoundland; J.T. GROENMAN, Arnhem, The Nether-
lands; J.A.H. HUNTER, Toronto, Ontario? ALLAN WM. JOHNSON JR., Washington, D.C. ;
J.A» McCALLUM, Medicine Hat, Alberta? FRED A. MILLER, Elkins, West Virginia? HERMAN
NYON, Paramaribo, Surinam? BOB PRIELIPP, University of Wisconsin-Oshkosh? KENNETH
M. WILKE, Topeka, Kansas? and the proposer.

Editor's comment.
Additional references [23 and [33 for the italicized theorem were given by Prie-
lipp and Johnson. Johnson noted that the identities

N = N2 -N(N-l) and N = N(N+1) - N2

show that ewery positive integer can be represented as w2-x(x+l) and as y(y+l)-z2.
The result is trivial but nonetheless surprising in the present context.
REFERENCES
1. G.H. Hardy and E.M. Wright, An Introduction to the Theory of Numbers, Fourth
 2M -
Edition (I960) or Fifth Edition (1979), Cambridge University Press, pp. 299-302.
2. W. Sierpifiski, Elementary Theory of Numbers, Hafner, New York, 1964, p.361.
3. J.V. Uspensky and M.A. Heaslet, Elementary Number Theory, McGraw-Hill, New
York, 1939, pp. 336-3«+0.
,'« ft *

567. C1980: 212] Proposed by George Tsintsifas, Thessaloniki, Greece.

1
A moving equilateral triangle has its vertices A , B', C on the sides
BC, CA, AB, respectively, of a fixed triangle ABC. The regular tetrahedron M-A'B'C
has its vertex M always on the same side of the plane ABC. Find the locus of M.
Solution by the proposer.
We first find the locus of the center of the moving triangle A ' B ' C . It is easy
to see that, for any given triangle A'B'C, the circumcircles of triangles A B ' C ,
BC'A', and CA'B' are concurrent in a point
0 (see figure). Since

/BOC = /A+/0BA + /0CA

= / / W / O A ' C + /0A'B'
= /A + 60°

and similarly /COA = /B + 60° and

/AOB = [Z + 60°, it follows that point
0 is fixed for all triangles A ' B ' C .
Furthermore, /B'OC = TT -A,

/C'OA' = ir-B,
/A'OB* = ir-C,
and so angles B ' O C , COA', A'OB' are
constant for all triangles A ' B ' C . Thus
the point 0 is the center of similarity of the moving triangle A ' B ' C . The pedal
triangle A 0 B 0 C 0 of the point 0 with respect to triangle ABC is also equilateral. It
is now easily shown that the similarity s(0,fc,e), with center 0, ratio k = 0A'/0A0,
and angle e = /A 0 0A', transforms triangle A 0 B 0 C 0 into triangle A ' B ' C . If Io and V
are the centers of triangles A 0 B 0 C 0 and A ' B ' C , respectively, then S(0,k9Q) also
transforms I 0 into I', from which it follows that

/ p i 0 r = ZPA 0 A' = ir/2.

We conclude that the locus of the center I1 of the moving triangle A'B'C lies on
the line through I 0 perpendicular to O I Q .
 - 215 -
We now consider the vertex M of the regular tetrahedron M - A ' B ' C Its distance
from the plane of triangle ABC is MI' = * / 27?B , C or, since B'C = pOA 8 , where p is
constant, MI' = Jz/SpOb'. Let A 0 A' = x and 0A 0 = d\ then MI 1 = ^I/Sp^id2 or, if
we set y = MI',

y2 = &Hx2+d2). (1)

This means that the point M moves on one branch of the hyperbola (1) which lies in
a plane perpendicular to that of triangle ABC. The trace of this hyperbola in the
plane of triangle ABC is the line I0I' perpendicular to 0I 0 .
We omit the proof of the converse.

Also solved by JORDT DOU, Escola Tecnica Superior Arquitectura de Barcelona,

Spain; and KESXRAJU SATYANARAYANA, Gagan Mahal Colony, Hyderabad, India.
it it it

5681 C1980: 213] Proposed by Allan Wm. Johnson Jr.s Washington, B.C.
The six distinct decimal digits composing a given integer can be shuffled
to form two three-digit factors whose product is the given integer. If one factor's
digits are an arithmetic progression and the other factor is divisible by 99, what
is the six-digit integer?

I. Solution de Hippolyte Charles3 Waterloos Quebec.

Soient ABC et DEF les deux facteurs recherches, dont le premier est celui qui
est divisible par 99. A priori, tous les multiples 99ras m = 2,3,...,9, doivent etre
considers comme valeurs possibles de ABC9 car tous sont composes de chiffres dis-
tincts (et B = 9 dans chacun d'eux). Or on a (toutes les congruences etant modulo 9)

A+B+C+D+E+F; 0 et A+B+ C= 0 ,
done aussi D + E 4 F = 0. Si {D,E,F} = {a~dsa3a+d}9 il resulte 3a = 0, d'oQ 3a = 9 ou 18
et a = 3 ou 6. Ainsi {3,6} n (D,E,F) * 0, et T o n peut gliminer 99-4 = 396 et
99«7 = 693 des valeurs possibles de ABC. Le tableau qui suit donne, pour les valeurs
de ABC qui restent, les progressions arithm£tiques {a-dsa3a^d} pour lesquelles
3a = 9 ou 18 qui n'ont pas de chiffre commun avec ABC.

ABC {P,E,F} ABC

198 {2,3,4} {5,6,7} {0,3,6} 891
297 {1,3,5} {4,6,8} {0,3,6} 792
495 {0,3,6} 594

Les six permutations de chaque progression {D,E,F} permettent de former six nombres
DEF distincts. On multiplie chacun d8eux par les valeurs correspondantes de ABC et
 - 216 -
on ne retient que les produits dont les chiffres forment une permutation de
{A,B,C,D,E,F}. Tous les produits sauf un sont e*limings, et la rgponse unique est

891-432 = 384912.
II. Comment by the proposer.
This problem was inspired by the following citation in Dickson [1]: "D. Biddle
applied congruences to find numbers like 15 and 93 whose product 1395 has the same
digits as the factors." Using a computer, I sought products having six distinct
digits that can be shuffled into three-digit factors. There exist products where
both factors have digits in arithmetic progression:

420-678 = 284760, 321-678 = 217638, 321-975 = 312975.

There also exist six-digit examples of the form ABODEF = CBADEF:
651-240 = 156240, 251-608 = 152608, 281-650 = 182650.
Also solved by JAMES BOWE, Erskine College, Due West, South Carolina; CLAYTON
W. DODGE, University of Maine at Orono; FRIEND H. KIERSTEAD, JR., Cuyahoga Falls,
Ohio; J.A. McCALLUM, Medicine Hat, Alberta; HERMAN NYON, Paramaribo, Surinam; DONALD
P. SKOW, Griffin & Brand, Inc., McAllen, Texas; ROBERT TRANQUILLE, College de Maison-
neuve, Montreal, Quebec; CHARLES W. TRIGG, San Diego, California; KENNETH M. WILKE,
Topeka, Kansas; and the proposer.

REFERENCE
1. Leonard Eugene Dickson, History of the Theory of Numbers, Volume 1, Chelsea,
New York, 1971, p. 463.
5*? & *

C1980: 213] Proposed by Charles W. Trigg, San Diego,

569, California.
Using Euclidean geometry, show that the planes perpendicular to a space
diagonal of a cube at its trisection points contain the vertices of the cube not on
that diagonal. (A solution of this problem using analytic geometry appears as No. 3197
on page 469 of the May 1969 issue of School Science and Mathematics.)

I. Solution by Leroy F. Meyers, The Ohio State University.

Let P and Q be opposite ends of a space diagonal in a cube of edge length a,
let A be one of the three vertices adjacent to P, and let T be the foot of the per-
pendicular from A to PQ. Since
AP 2 + AQ 2 = a2 + (a/2) 2 = (a/3) 2 = PQ 2 ,

we see that triangle PAQ has a right angle at A, so that

a2 = AP2 = PT-PQ s PT-a/3 ;
hence
PT = a//3 = ~PQ
 - 217 -
and T is a trisector of PQ. Thus the plane perpendicular to PQ at T passes through
A. Similar results hold for the other vertices.

II. Comment by M.S. Klamkin, University of Alberta.

It is known that the corresponding result is valid for an n-dimentional cube,
that is, we can replace "cube" by "n-cube" and "trisection points" by "rc-section
points" in the given statement.
An analytic proof is immediate. Let the vertices of the n-cube be (±i,±i,...,±i)
arid consider the diagonal determined by (-l,-l,„..,-l) and (i,i,...,i). The projection
of a vertex {x^x^* --*xn) onto
this diagonal is the point (<z,a,. ,.,a), where
Z(x.-a) = o. The values of a for the vertices
(1,1,...,1,-1), (1,1,...,1,-1,-1), (1,1,...,1,-1,-1,-1), etc.,

are

n-2 n-h n-6

n > n > n > "' J

which establishes the desired result. Q

The above is a Euclidean proof, as the proposal requires (even if it is not a
synthetic proof, as the proposer probably had in mind).

Also solved by JAMES BOWE, Erskine College, Due West, South Carolina; CLAYTON
W. DODGE, University of Maine at Orono? J„T» GROENMAN, Arnhem, The Netherlands;
SAHIB RAM MANDAN, Bombay, India? FRED A. MILLER, Elkins, West Virginia; MSc in
Mathematical Education Problem-Solving Class, Polytechnic of the South Bank, London,
England, reported by DAVID SINGMASTER (two solutions); KESIRAJTJ SATYANARAYANA, Gagan
Mahal Colony, Hyderabad, India; GEORGE TSINTSIFAS, Thessaloniki, Greece? and the
proposer (two solutions).
k * *

570, [1980: 213] Proposed by V.N. Murty, Pennsylvania State University, Capitol
Campus, Middletoion, Pennsylvania.
If x,y3z > o, show that

x + y
,. TxtyHx + z) * *>
cyclic

with equality i f and only i f x = y = z.

I, Solution by the proposer.

The i d e n t i t y
Z2x2(y+z)2 + Zyz(y-z)2 = (y + z)(z + x)(x ^ y)(x + y + z)
is easily established by noting that its left member is a symmetric and homogeneous
polynomial of degree 4 in x3y,z which vanishes when y = ~z. It follows that
 - 218 -
2c 2 (j/+g) = lyzjy-z)1
E {x+y)(x+z) ' y
(y^z)(z^x)(x+y)
<» x + y + 3 j ,

with equality i f and only i f x=y=z.

I I . Comment by W.J. Blundon, Memorial University of Newfoundland,

I f we set y + z = a9 z + x = ib, a? + y - c, i t is easy to show that xsy3z > o i f
and only i f a,b3o are the sides of a triangle, and then the proposed inequality is
equivalent to

Z2a2(s-a)2 £ abcs3

where s is the semi perimeter, with equality i f and only i f the triangle is equilateral.

Also solved by W.J. BLUNDON, Memorial University of Newfoundland? G.C* GIRI,

Midnapore College, West Bengal, India? J.T. GROENMAN, Arnhem, The Netherlands? M.S.
KLAMKIN, University of Alberta? J.A. McCALLUM, Medicine Hat, Alberta? NGO TAN,
student, J.F. Kennedy H.S., Bronx, N.Y.? BOB PRIELIPP, University of Wisconsin-
Oshkosh? KESIRAJU SATYANARAYANA, Gagan Mahal Colony, Hyderabad, India? and GEORGE
TSINTSIFAS, Thessaloniki, Greece.
ft ft *

5711 C1980: 251] Proposed by Sidney Kravitz, Dover, New Jersey.

The party conventions are over and, come November, Americans will have to

ELECT
REAGAN
0R_ .
CARTER
Members of the U.S. House of Representatives are also entitled to find the unique
answer to this problem. Indeed, because of John Anderson, they may have to.

Solution by Kenneth M. Wilke, Topekas Kansas.

We first observe that

E > 5, T + N = 10, C = R + 1, C+A + 0+l = E (mod 10) . (1)

Suppose E = g; then A = 8 and there is no value for L since L + A = R < 10 requires

R = 8 or g while L + A ;> lo requires A = 9. Now suppose E = 8. If A = 6, then R = 7
(since L + A = R < 10) and hence C = 8 = E. Thus we must have A = 7, and it follows
that C + 0 = 10. Since {C,0} = {1,9} requires R = 8 or R = o, both impossible, and
C = 6 requires R = 5 and L = 7 or 8, both impossible, we must therefore have C = 4
and 0 = 6 . The remaining values R = 3, L = 5, G = 2, T = l, and N = 9 fall out
immediately, yielding the solution
 - 219 -
858M-1
387279
63 .
H73183

This is the only solution, because similar analyses for the other possible values
of E [the details are omitted] yield no other solutions.

Also solved by J.A.H. HUNTER, Toronto, Ontario; ALLAN WM. JOHNSON JR., Washington,
D„C? J„A. McCALLUMr Medicine Hatf Alberta? HERMAN NYON, Paramaribo, Surinam? SAMSOE
PROBLEM GROUP, Silkeborg, Denmark; CHARLES W. TRIGG, San Diego, California? and the
proposer.

Editor's comment.
The editor had no difficulty in deciding which solution to feature, because
Wilke's was the only satisfactory one received. One solver exhibited a fine sense
of bipartisanship by coming up with two distinct answers; but an election recount
showed that one was incorrect. Two solvers gave little more than the obvious re-
lations (1) and then, hey presto, out came the correct answer, like Minerva springing
full-blown from the brow of Jupiter. The remaining four solvers, perhaps reluctant
to expose the untidiness of their thought processes, gave just the correct answer.
Alphametics (or cryptarithms, as some people still prefer to call them) appear
in several journals. One journal prints bushels of them in eyery issue and then
gives only an answer for each. We take the view that the answer itself is of no
mathematical interest. The only thing that counts is the ingenuity displayed in
arriving at the answer. Nevertheless it is still all right for solvers to send in
only an answer (we like long lists of solvers), especially if they are not too proud
of the way they obtained it. They are assured of a place among the also-rans, and
of an occasional Bronx cheer from the editor.
* * *

THE PUZZLE CORNER

Puzzle No. 2: Word Deletion (8; 3,5)

A mathematics buff may get
The residue at any pole,
FIRST find the SECOND of a set,
And knows the meaning of a WHOLE.

ALAN WAYNE

Answer to Puzzle No. l [1981: 198]: Triangle, relating, integral.

A & *
 - 220 -

THE OLYMPIAD CORNER: 27

M.S. KLAMKIN

The Twenty-Second International Mathematical Olympiad (1981).

There was no International Mathematical Olympiad (I.M.O.) in 1980 since no
country had offered to be the host nation. This was the first break in the con-
tinuity of the yearly-held competition since it was started by Romania in 1959.
This year, the I.M.O. was held at Rutgers University, New Brunswick, New Jersey
and at Georgetown University, Washington, D.C. The competition and the attendant
social and cultural activities lasted from July 8 to July 19. This was the first
time it was held outside Europe. There were 185 students competing, from 27 coun-
tries, record numbers both of students and countries. The countries participating
for the first time were Australia, Canada, Colombia, Luxembourg, Mexico, Tunisia,
and Venezuela.
The results of the competition (see listing on following page) are announced
officially only for individual members of the teams, a maximum of 42 points being
awarded for a perfect paper (7 points for each of the 6 problems). (This is a
change from the traditional method of scoring, where the six problems were assigned,
before the competition, weights varying from 5 to 8 points, totaling 40, and the
contestants were not informed of the weight of each problem.) However, team standings
are always compiled unofficially by adding up the scores. In this unofficial team
standing, the U.S.A. team was first with a total score of 314, beating out the West
German team by the very narrow margin of 2 points; the United Kingdom team was
third with 301. The Canadian team was seventh with a score of 249. This is an
auspicious start for Canada since a team which enters for the first time rarely
does so well.
The Canadian team consisted of:
David Ash, Ontario First Prize Award
George Gonthier, Quebec " " "
John Chew, Ontario Second Prize Award
Gary Timar, Ontario " "
Julian West, British Columbia Third Prize Award
Arthur Baragar, Alberta
David Bernier, Quebec
John Bowmen, Alberta

The Canadian team was coached by G.J. Butler, University of Alberta and
E.J. Barbeau, University of Toronto.
 - 221 -
INDIVIDUAL SCORES (PERFECT SCORE M2)

Contestant No. 1 2 3 4 5 6 7 8 Total Rank

Australia 22 13 12 10 17 12 8 28 122
Austria 42 41 41 36 37 42 21 30 290 4
Belgium 37 24 25 38 6 3 3 3 139
Brazil 42 23 21 14 17 33 21 21 172
Bulgaria 32 42 40 34 28 37 33 41 287 5
Canada 42 25 23 16 35 "2 37 29 249 7
Colombia 14 13 13 11 1 13 14 14 93
Cuba ! 25 34 22 12 19 9 14 6 141
Czechoslovakia 38 38 40 42 32
Finland 27 41 17 10 31 9 38 33 206
France 29 15 42 26 8 29 42 18 i 209 10
Greece 24 11 13 13 15 2 12 14 104
I Hungary 41 42 41 40
Israel 42 32 25 31 16 29
Luxembourg 42
Mexico 3 4 2 2 1
Netherlands 25 25 37 36 27 16 17 36 219 9
Poland 35 18 41 36 24 28 35 42 259 6
Romania ! 36 35 33 32
Soviet Union 42 33 36 35 42 42
Sweden 22 20 17 32 32 28 38 18 207
Tunisia 14 18
United Kingdom 42 38 41 36 27 37 38 42 301 3
United States 42 42 42 39 42 35 39 33 314 1
Venezuela 6 3 9 4 14 4 23 1 64
West Germany 41 39 37 42 42 41 28 42 312 2
Yugoslavia 26 31 42 32 23 37 35 20 246 8

As can be seen from the listing of scores, some countries did not send a
full 8-member team. It is advisable for a country without a strong tradition in
national olympiads to send a small group the first time to gain the experience of
an international olympiad. In some cases travel costs may be a factor affecting
the size of a team. The Soviet Union, which has won the I.M.O. eight times since
1959, inexplicably sent only a 6-student team. This was all the more surprising
 - 222 -
since, as noted in this column several times in the past, the Soviet Union has a
very strong tradition in national and regional olympiads.
The U.S.A. team consisted of:
N.D. Elkies, New York First Prize Award
• B.N. Fisher, New York
J.D. Primer, New Jersey " " "
B.R. Hunt, Maryland
G.N. Patruno, New York Second Prize Award
R.A. Stong, Virginia
J.R. Roche, Minnesota Third Prize Award
D.S. Yuen, Illinois
The U.S.A. team was coached by the author and Andy Liu, University of Alberta.
Many participants thought that this year's problems were too easy. There
were 26 perfect scores, many more than ever before. The problems are given below,
so the reader can judge for himself. Solutions to these problems (along with
those of the Tenth U.S.A. Mathematical Olympiad given earlier in this column
[1981: m i ] ) will appear later this year in a booklet, Olympiads for 19819 to be
compiled by Samuel L. Greitzer and obtainable from
Dr. Walter E. Mientka,
Executive Director,
M2VA Committee on U.S. Contests,
917 Oldfather Hall,
University of Nebraska,
Lincoln, Nebraska 68588.

See the appended references for information about past national and international
mathematical olympiads.

XXII INTERNATIONAL MATHEMATICAL OLYMPIAD

July 13, 1981
Time: 4 1/2 hours
Each problem will receive equal weight

1, P is a point inside a given triangle ABC. D,E,F are the feet of the
perpendiculars from P to the lines BC, CA, AB, respectively. Find all
P for which
BC + CA+ AB.
PD" W PF
is least.
 - 223 -
2„ Let l <; r < n and consider all r-element subsets of the set {1,2,... ,n}.
Also consider the smallest number in each of these subsets and let F{n,r)
denote the arithmetic mean of these smallest numbers. Prove that

3, Determine the maximum value of m2 + n2 if m and n are integers satisfying

1 <, m9n £ 1981 and (n2 -rnn-m2)2 = 1.

July 14f 1981

Times 4 1/2 hours
Each problem will receive equal weight

4i (a) f r ° r which values of n > 2 is there a set of n consecutive positive

integers such that the largest number in the set is a divisor of the
least common multiple of the remaining n - 1 numbers?
(b) For which values of n > 2 is there exactly one set having the stated
property?

5, Three congruent circles have a common point 0 and lie inside a given tri-
angle. Each circle touches a pair of sides of the triangle. Prove that
the incenter and the circumcenter of the triangle and the point 0 are col linear.

6, The function f(x,y) satisfies

(1) f(Osy) = y + 1,
(2) /(ar+1,0) = f(x,l),
(3) f(x+l,y+l) = f(x9f{ac+l,y))
for all nonnegative integers x9y. Determine /C+,1981).

Readers may be interested to know how the olympiad problems are arrived at.
Each participating country is requested to submit 5 problems with solutions to the
host country (which submits no problems) prior to the olympiad. The host country
filters down these loo to 150 problems to a reasonable number (19 this year), re-
jecting those that are too easy, well known, textbook problems etc. Soon after
their arrival in the host country, the leaders of each team congregate in a loca-
tion appreciably distant from the teams and deputy leaders to select the problems
for the examination. They then have no further contact with their teams until
after the examination has taken place. After a long discussion of the problems
and their solutions, the problems to be given at the olympiad are chosen demo-
cratically by an open vote. I and the delegates from the Soviet Union, Sweden,
and Yugoslavia, at least, had voted for some of the harder problems which were
 - 224 -
not used. (I shall give the 13 unused "filtered" problems in this column next
month.) However, many of the countries participating for the first time did not
vote for the apparently harder problems and they won out. Perhaps this judgment
is unfair, because it is not easy to judge, for any given problem, how difficult
it will appear to the competing students. Even though I thought that this olympiad
was on the easy side, compared to previous ones, I did not expect, and I think the
other team leaders did not expect, to see so many perfect scores,
I hope that next year, when the I.M.O, is expected to be held in Hungary,
the team leaders will not lean too sharply in the other direction and make the
examination too difficult.

REFERENCES
1. Samuel L. Greitzer, International Mathematical Olympiads 1959-1977,
Mathematical Association of America, Washington, D.C., 1978.
2. N.B. Wassiljew, A.A. Jegorow, and R. Zelinka, Aufgaben von Mathematik-
olympiaden in der UdSSE und in der CSSR, Volk und Wissen Volkseigener Verlag,
Berlin, 1965.
3. W. Engel and U. Pirl, Aufgaben mit Ldsungen aus Olympiaden Junger
Mathematiker dev DDR, I, IX, Volk und Wissen Volkseigener Verlag, Berlin, 1975.
1. V. Skvortsov, "Mathematical Olympiads", Socialist Mathematics Education,
Ed. F.J. Swetz, Burgundy Press, Southhampton, Pennsylvania, 1978, pp. 351-370.
5. M.S. Klamkin, "A Bibliography of Mathematical Competitions" in The
Oli'mpiad Corner: 8, this journal, 5 (1979) 220-227.

Editor's 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.
& * *'«

DAN J. EUSTICE
1931 - 1981

We regret to announce that Professor Dan Eustice of The Ohio

State University died unexpectedly on August 5, 1981 at the age of
50. He had been for several years Problem Editor for Mathematics
Magazine and had in the past contributed to this journal. Surviving
him are a wife and three children.
We extend our deepest sympathy to the bereaved family.