Crux

Mathematicorum
Published by the Canadian Mathematical Society.

http://crux.math.ca/

The Back Files

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

Journal title history:

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

Vol. 12, No. 4

April 1986
Published by the Canadian Mathematical Society/
Publie par la Societe Mathematique du Canada
The support of the University of Calgary Department of Mathematics and
Statistics is gratefully acknowledged,
* * *
CRUX MATHEMATICORUM is a problem-solving journal at the senior secondary
and university undergraduate levels for those who practise or teach mathe-
matics. Its purpose is primarily 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 subscrip-
tion rate for ten issues is $22*50 for members of the Canadian Mathematical
Society and $25 for nonmembers* Back issues: $2.75 each. Bound volumes with
index: Vols, 1 & 2 (combined) and each of Vols* 3-10: $20• All prices
quoted are in Canadian dollars. Cheques and money orders, payable to CRUX
MATHEMATICORUM, should be sent to the Managing Editor,
All communications about the content of the journal should be sent to the
Editor. All changes of address and inquiries about subscriptions and back
issues should be sent to the Managing Editor.
Founding Editors: Leo Sauve, Frederick G.B. Maskell.
Editor: G.W. Sands, Department of Mathematics and Statistics, University
of Calgary, 2500 University Drive N.W., Calgary, Alberta, Canada, T2N 1N4.
Managing Editor: Dr, Kenneth S. Williams, Canadian Mathematical Society,
577 King Edward Avenue, Ottawa, Ontario, Canada, KIN 6N5*
ISSN 0705 - 0348.
Second Class Mail Registration No. 5432, Return Postage Guaranteed,
* * *
CONTENTS
The Olympiad Corner: 74 M.S. Klamkin 67
Problems: 1116, 1131-1140 77
Solutions: 999, 1001-1009 80

- 66 -
 - 67 -

THE OLYMPIAD CORNER: 74

M.S. KLAMKIN

All communications about this column should be sent to M.S. Klamkin,

Department of Mathematics, University of Alberta, Edmonton, Alberta, Canada,
T6G 2G1.
-t-K

This month's problem set consists of the 4 Annual American Invitational

Mathematics Examination (AIME). I am grateful to Professor Walter Mientka for
these problems. For a description of the AIME, see [1983: 170]. The answers
to these problems will be given next month. Any questions or comments about
this AIME should be addressed to Professor George Berzsenyi, Department of
Mathematics, Lamar University, Beaumont, Texas 77710.

4 Annual American Invitational Mathematics Examination

March 18, 1986 — Time: 3 hours
A 12
1. What is the sum of the solutions of the equation Jx
7 - Vx
2. Evaluate the product
(JS + 76" + JT) (JS + J£ ~ JT) (J5 - J5 + JT)(-J5 + JG + JT).
Z. If tanx + tany = 25 and cotx + coty = 30, what is tan(x + y)?
4. Determine 3x4 4- 2x5 if x x , x 2 , x 3 , x 4 and x 5 satisfy the system of
equations given below:
2xt + x2 + x3 + xA + x5 = 6
xA + 2x2 + x 3 + xA + x 5 = 12
xx + x 2 + 2x3 + xA + xs = 24
xx + x 2 + x 3 + 2xA + x 5 = 48
xx + x2 + x 3 + xA + 2xs = 96.
5. What is the largest positive integer n for which n3 + 100 is
divisible by n + 10?
6. The pages of a book are numbered 1 through n. When the page
numbers of the book were added, one of the page numbers was
mistakenly added twice, resulting in the incorrect sum of 1986. What was the
number of the page that was added twice?
7_. The increasing sequence 1, 3, 4, 9, 10, 12, 13, ... consists of all
those positive integers which are powers of 3 or sums of distinct
 - 68 -

powers of 3. Find the 100 term of this sequence (where 1 is the i S term, 3
is the 2 termf and so on).
8. Let S be the sum of the base 10 logarithms of ail of the proper
divisors of 1,000,000, (By a proper divisor of a natural number we
mean a positive integral divisor other than 1 and the number itself.) What is
the integer nearest to S?
9* In AABC shown below, AB = 425, BC = 450 and CA = 510. Moreover, P
is an interior point chosen so that the segments DE9 Fa and Hi are
each of length d, contain P, and are parallel to the sides AB9 BC and CA,
respectively• Find of.

A
10. In a parlor game the "magician" asks one of the participants to
think of a three-digit number (abc), where a, b and c represent
digits in base 10 in the order indicated. Then the magician asks this person
to form the numbers (&cb), (bac), (bca), (cab) and (cba), to add these five
numbers, and to reveal their sum, N« If told the value of N, the magician can
identify the original number, (abc). Play the role of the magician and
determine (abc) if N = 3194.
11. The polynomial 1 - x + x 2 - x 3 + ... + xl* - x 1 7 may be written in
the form aQ + aty + a2y2 + a 3 y 3 + ... + a l s y 1 6 + al7y17, where
y = x + 1 and the a.'s are constants. Find the value of a2 ,

12. Let the sum of a set of numbers be the sum of its elements. Let S
be a set of positive integers, none greater than 15. Suppose no
two disjoint subsets of S have the same sum. What is the largest sum a set S
with these properties can have?
13* In a sequence of coin tosses one can keep a record of the number of
instances when a tail is immediately followed by a head, a head is
immediately followed by a head, etc. We denote these by TH9 HE, etc. For
example, in the sequence HHTTHHEHTHHTTTT of 15 coin tosses we observe that
there are five HE, three HT, two TE and four IT subsequences. How many
different sequences of 15 coin tosses will contain exactly two EE, three ET9
four TE and five TT subsequences?
 - 69 -

14. The shortest distances between an interior diagonal of a

rectangular parallelepiped (box), P, and the edges it does not meet
are 2JE, 30/VT3" and 15/JTU. Determine the volume of P.
15. Let AABC be a right triangle in the xy-plane with the right angle
at C. Given that the length of the hypotenuse AB is 60, and that
the medians through A and B lie along the lines y = x + 3 and y = 2x + 4,
respectively, find the area of AABC.
*

I now give solutions to some previous problems,

VI ALL-RUSSIAN MATHEMATICAL OLYMPIAD 1979-80 [1982: 72]

VIII Grade

J,. A group of tourists decided to sit in a set of buses in such a way

that each bus would contain the same number of tourists. At first
they tried to sit 22 on each bus, but it turned out that one tourist was left
over, But then one bus left empty, and the tourists were able to divide
themselves equally among the remaining buses. If each bus holds fewer than 33
people, how many buses and how many tourists were there (originally)?

Solution.
Let n and b denote the number of tourists and buses, respectively. Then
226 = n - 1 and k(h - 1) = n where h < 33. Eliminating n,
, _ k + 1 _ 1 . 23
l +
° ~ k - 22 ~ k - 22 #
Thus (since b is an integer), k = 23, 5 = 24 and n = 529.

2. Along a segment AB, 2n points are chosen which are symmetric in

pairs with respect to the midpoint of the segment. Any n of these
points are colored blue, and the rest are colored red. Prove that the sum of
the distances from the red points to A is equal to the sum of the distances
from the blue points to B,

Solution.
We coordinatize the points on a line. Let A = -tf, B = K. Also, let the
red points be flp R2,, . ., R and the blue points be B p B2,..., B . Then the

sum of the distances from the red points to i is I(i?. + K), and the sum of the

distances from the blue points to B is l(K - B.). These two sums are equal
 - 70 -

since by centrosymmetry 2(/? + B ) , the sum of the coordinates of all 2n

points, is zero.

3. In a regular hexagon ARCDRF, points M and K two the midpoints of (1)

and DE, respectively, and L is the intersection of segments AM and
BK* Prove that the area of triangle ABL is equal to that of quadrilateral
MDKL. Also, find the measure of the angle between lines AM and BK*
Solution. A^
o
If we rotate AM 60 about
the center of the circle it goes
into BK by rotational symmetry.
Thus lALB = 60°, Also,
Area ABCM = Area BCDK. Thus
Area ABL = Area MDKL.
4. If {x} denotes the fractional part of x (e.g., {7/5} = 2/5), how
many distinct numbers are there in the sequence
{1V1980}, {2V1980}, {3V1980}, ... f {19802/1980}?

Solution.
See [1984: 290].

5, From a point M on the circumcircle of a triangle ABC,

perpendiculars MN and MK are drawn to lines AB and AC,
respectively. For which point M will NK be longest?

Solution.
ANMK is cyclic with circumdiameter
AM. Also, INAK is a fixed angle.
Consequently, AW will be a maximum when
AM is a maximum, i.e. M is antipodal
(diametrically opposite) to A.

IX Grade

1. Can the natural numbers from 1 to 30 be arranged in a 5 x 6

rectangular array in such a way that (a) all columns have the same
sum and (b) all rows have the same sum?

Solution.
The sum of all the numbers is 30(1 + 30)/2 465. If the sum of each
 - 71 -

column is C and the sum of each row is #, then 6C = 5/? = 465. Clearly, this
is impossible for (a) the column sums. However, it is possible for (b) the
row sums, as indicated in the following array:
1 6 11 20 25 30
2 7 12 19 24 29
3 8 13 18 23 28
4 9 14 17 22 27
5 10 15 16 21 26

2. For which natural numbers n i s 2 8 + 2 1 1

+2 a perfect square?
[This problem also appeared in the 1981 Hungarian Mathematical
Olympiad. See [1981: 267; 1982: 46],]

3. Each vertex of a convex (2n + l)-~gon is colored with one of three

different colors. No two adjacent vertices are colored the same.
Prove that the polygon can be partitioned by nonintersecting diagonals into a
set of triangles each of which has its three vertices of different colors.
[Diagonals which meet at a vertex are considered nonintersecting.]

Solution by Graham Denham, student, Old Scona Academic High School,

Edmonton, Alterta.
We prove a more general result on a convex m-gon where m need not be odd,
with the additional condition that each colour appears at least once. For m
odd, this condition is redundant.
We use induction on w. For m = 3, the result is trivial. Suppose the
result holds for 3, 4,..., © - 1. Now consider anffl-gon.We have two cases.
Case 1. One of the colours appears only once. Then the other two
colours must appear alternately on the other vertices. If we draw all the
diagonals from the vertex with the unique colour, a desired triangulation is
obtained. Note that the induction hypothesis is not needed in this case.
Case 2. Each colour appears at least twice. We then claim that there
exist three consecutive vertices with each a different colour. Otherwise, the
colours of the vertices must alternate all the way around the perimeter in
just two colours, precluding any appearance of the third colour. Now draw the
diagonal which cuts off the triangle defined by those three vertices. The
remaining (m - l)-gon still satisfies the induction hypothesis since all three
colours are still present by the hypothesis that each colour appears at least
twice. Thus by induction the desired triangulation of the a?-gon can be
 - 72 -

accompl i shed.

4. In expressing the fraction m/n as a decimalf where m and 12 are

natural numbers and n < 100, a student found, at a certain place
after the decimal point* the sequence of digits 167. Show that the student
must have made an error.

Solution.
The solution here is essentially the same as that of Andy Liu for problem
#6 [1985: 141] in which 501 replaces the sequence 167, As there, it suffices
to find the smallest value of n for which m/n = 0.167,** for some m < n. The
six sequences {{k 4- l)/6*}f {(* + l)/(6Jt + 1)}, {(k + l)/(6* + 2)},
{(k + l)/(6i -l- 3)}, {(k + l)/(6* + 4)}, and {(* + l)/(6Jt + 5)} are all
decreasing in *. Also, each member of each sequence is > 1/6 = 0.1666,.. and
each sequence approaches 1/6 in the limit. Since
22/130 = 0.169...
22/131 = 0.167...
21/125 = 0.168...
the desired minimum value of « is 131 and so the student must have made an
error.

5. Equilateral triangles ABC and A'B'C are drawn in a plane (both

sets of vertices being labeled clockwise). The midpoints of
segments BC and B'C coincide. Find
(a) the angle between the lines AA' and BB';
(b) the ratio AA'/BB'.

Solution.
Although this problem can be solved by synthetic geometry, we will use
complex numbers to demonstrate their utility in certain types of geometry
problems?. We also do the same for the subsequent problem #5, Grade X.
Consider the Argand diagram with the midpoint M of EC as the origin, MB
as the positive real axis, and MA as the imaginary axis. Also, we can assume
BC = 2. Then the complex number representations for Af B, C are iJS, -1, and
1, respectively. If the angle between BC and B'C is P, then A', B', and C
ip ip
are e iJS, -e 2/>
, and e f respectively. Since A - A* = i-/F(l - <?ip) ®^d
B - B' = -<1 - eip)f
o
(a) the angle between AA' and BB' is 90 ,
(b) the ratio AA'/BB' = 7?.
 - 73 -

X Grade

1. For each vertex of a tetrahedron, the point symmetric to that

vertex with respect to the centroid of the opposite face is chosen.
Find the ratio of the volume of the tetrahedron whose vertices are these new
points to that of the original tetrahedron.

Solution.
Our solution is vectorial since this is a particularly good
representation for this type of problem• Let A, B, C, D denote vectors from a
common origin to the vertices A, B, C, D of the tetrahedron. The centroid of
the face opposite to A is given by (B * C + D)/3. The point symmetric to A
with respect to the latter centroid is given by
A' = (B + C + D)/3 + ((B + C + D)/3 - A)
or A' = 2(S - A)/3 - A where S = A + B + C + D . The other three points are
obtained by symmetric interchange of the letters. Since
A' - B' = -5(A - B)/3, etc.,
the desired ratio is (5/3)3. Similarly, the ratio of volumes for the
analogous problem for an /^-dimensional simplex is ((n + 2)/n) .

2. The map of a city has the shape of a convex polygon. Each diagonal
of the polygon is a street, and the intersections of the diagonals
are intersections of the streets (but the vertices of the polygon are not
considered to be intersections of streets). Streetcar lines go through the
city. Each line goes from one end of a street to the other end, and has stops
at each intersection as well as at the endpoints. At each intersection only
two streets cross, and a streetcar runs along at least one of them. Show that
one can transfer from any intersection to any other, making no more than two
transfers. (A transfer may be made whenever two streetcar lines have a common
stop.)

Solution by Andy Liu, University of Alberta, Edmonton, Alberta.

Suppose one wants to go from intersection D to intersection W. At least
one street through D, say AC, is serviced by a streetcar. Similarly, a
streetcar runs along XZ through W. If AC coincides with XZ, no transfer is
necessary. If AC intersects XZ, one transfer will suffice. If neither of
these cases holds, let B be any vertex between A and C on the opposite side of
W and let Y be any vertex between X and Z on the opposite side of D. B and Y
exist since A and C cannot be adjacent nor can X and Z. We can assume that AY
 - 74 -

and BZ determine a point of intersection, as otherwise we can replace BZ with

BX. Now either AY or BZ is serviced by a streetcar* Since each shares a
common stop with both AC and XZ, one can get from D to W by making no more
than two transfers,

cL Consider the 2k numbers

2 1 - 1, 2* - 1, ... , 22k - 1,
where k > 1. Show that at least one of them is a multiple of 2k + 1.

Solution by Andy Liu, University of Alberta, Edmonton, Alberta.

2k
Consider 2 1 , 2 2 s ..., 2 . None of these are congruent (mod 2k + 1) to 0
since (2,2* 4- 1) = 1. If no two of them are congruent (mod 21 + 1) to each
otherf then for some x between 1 and 2* inclusive, 2 = 1 (mod 2 1 + 1 ) , and
x i /
hence 2 - 1 will be a multiple of 2k + 1. Thus suppose that 2 = 2J
(mod 2k -h 1) where 1 < i < j < 2k. Then, 2 J ~ i s 1 (mod 2* + 1). The same
conclusion now follows as in the first case if we set x = j - i.

4. HR being the set of all real numbers, find all functions F: IR —• IR

which satisfy
pF(a) + (1 - p)F(b) > F(pa + (1 - p)b)
for all a, 5, p € IR.

Solution by Andy Liu, University of Alberta, Edmonton, Alberta.

If (ufF(iz)) and ( F , F ( F ) ) are any two points on the graph of y = F(x),
then by the hypothesis, all the points of the graph lie on or below the line
joining the two points. We now show by an indirect proof that all the points
of the graph are on a straight line, i.e., F is a linear function. Assume
there are three points of the graph which do not lie on a straight line. This
gives a contradiction since at least one of the points is above the line
joining the other two points.

5. The squares ABCD, A^^^^^, and A2B2C2D2 are coplanar (and their
vertices are labeled counterclockwise). Vertices A and Ax
coincide, and so do vertices C and C2. Show that DlD2 1 BMf where M is the
midpoint of BXB2, and that DkD2 = 2Bflf.

Solution.
As in the previous problem #5, we use a complex number representation.
Let A, B9 C, D, Blf and Bz be given by
A = 1, S = i, C = -1, D = -i, Bt = z, and B 2 = w.
Then
 - 75 -

Ax = 1,
Bx = z,
C x = B! + i ^ - 1) = z + i(z - 1),
Dx = C x + i(C1 - B J = 1 + i(z - 1),
C2 = -1,
Z?2 = C a + i(C2 - B 2 ) = -1 - i(l + w)i
tf = (Bx + B2)/2 = (z + v)/2.
Then Z7i - £?2 = 2 + i (z + w) and -2i(B - M) = 2 + i(z + w). Thus DXDZ is
perpendicular to BM and DXD2 = 2B^f.

Since I have had some queries concerning problems from the 1985 Dutch
Mathematical Olympiad - First Round [1986: 2] (the answers were given last
month [1986: 49])? I now give solutions to some of these problems. I am
grateful to Andy Liu for them.
o
B2. On one of the sides of an angle A of 60 , a point P is given such
that AP = 1 0 . On the other side of the angle a point Q is chosen
2
in such a way that AP + AQ2 + PQ 2 is a minimum. Compute AQ*
Solution.
Let R be a point on AQ extended such that AQ = QR. Then using the
2 2 2 2
formula for the length of a median (im = 26 + 2c - a ), we obtain

AQ2 + PQ2 = (AP2 + PR2)/2. Since AP is constant, AP2 + AQ2 + PQ2 is a minimum
if and only if P/? is a minimum. This occurs when PR is perpendicular to AR.
Since iPAR = 60°, A/? = 5 and AQ = 5/2.

B3. A rectangular block with edge lengths a, bf c has a volume

numerically equal to its total surface area. Furthermore, it is
given that a, h9 c are integers with a < b < c. Determine all possible
triples (a,6,c) satisfying these conditions.

Solution*
We have 2(ab + 6c + ca) = a6c. If a < 2, the left-hand side exceeds the
right-hand side. If a > 6, it's the other way around, since the left-hand
side is in any case at most 66c. Hence a = 3, 4 or 5.
For a = 3,
c = 66/(6 - 6) = 6 + b ^6g .

For 6 = 7 , 8, 9, and 10, c = 42, 24, 18, and 15 respectively. For 6 = 11, c
 - 76 -

is non-integral. For 6 > 12, c < 12s contradicting b < c.

For a = 4,
c = 45/(6 - 4) = 4 4- piL-j. .

For 6 = 5 and 6, c = 20 and 12, respectively. For 6 = 7, c is non-integral.

For 6 > 8 f c < 8, contradicting 6 < c.
When a = 5,
c = iob/(3b - io) = 2° + 3(3*°; 1U).

For 6 = 6, c is non-integral. For 6 > 7, c < 7, contradicting 6 < c*

Hence (3,7,42), (3,8,24), (3,9,18), (3,10,15), (4,5,20), and (4,6,12) are
the only triples satisfying the hypotheses,

B4« a, 6, and c are positive integers of 2, 3, and 5 digits,

respectively, all digits being less than 9, The digits of c are
distinct. Furthermore, a6 = c and adding 1 to each digit does not affect the
truth of this equation. Determine a, 6, and e.

Solution.
We have a6 = c and (a -I- 11) (6 4- 111) = c + 11111 • Eliminating c, we
obtain Ilia + 116 = 9890, The only solutions (a,6) satisfying 10 < a < 88 and
100 < 6 < 888 are (12,778), (23,667), (34,556), (45,445), (56,334), (67,223)
and (78,112) • However, only a = 56 and 6 = 334 yields a value of c satisfying
the hypotheses.

C2, The terms a of a sequence of positive integers satisfy

Compute a7 if it is given that a 6 = 144,

Solution.
Let at = x, a 2 = y, and a3 - z* Then
a 4 = z(y 4- x ) ,
a s = z(y 4- x)(z -l- y)
and
a 6 = 144 = zz(y + x)(z + y)(y + x + 1).
Note that y 4- x and y + x + 1 are two consecutive positive integral divisors
of 144, with y + x > 2. Hence, y 4- x = 2 f 3 or 8, The first case leads to a
non-integral value of z. The other two cases lead t o x = 2 , y = l , z = 2 and
x = 7 , y = z = l , and both give a7 = 3456,
 - 77 -

C3. A carpenter saws from a block a polyhedron with 30 vertices and 18

faces• The faces are 5 quadrangles, 6 pentagons, and 7 hexagons.
How many interior diagonals does it have? (An interior diagonal connects two
vertices not in the same face.)

Solution.
A = 435 segments joining two distinct vertices. Of these,

+ + == e on e
^\?\ ®\?\ ^? ^^ ^ ^ ^ a c e s » with each edge of the polyhedron
being counted twice. By Euler's formula, the polyhedron has 30 + 18 - 2 = 46
edges. Hence the number of space diagonals is 435 - 195 + 46 = 286.
*

Finally, here is a correction to problem #10 in my November 1985 column

[1985: 272]. In the last line, change "n-1?" to M(n-1)!?M.
* * *

P R O B L E M S

'Pia&tem proposals and solutions should le sent to the editor, whose

address appears on the §ront page o§ this Issue. Proposals should, whenever
possllle, le accompanied If a solution, references, ami other lnslg,hts which
are tc&elf to le o§ help to the editor. An astercsfc (%) after a numler
Indicates a prollem sulmltted without a solution.

Original prollems are particularly, bought. "Met other Interesting,

prollems maf also le acceptable provided thef are not too melt known and
references are q,lven as to their provenance. Ordinarily, l§ the originator o§
a prollem can le located, it should not le submitted If somelodf else without
his or her permission.

"Jo ^eeL&cta-te their consideration, four solutions, typewritten or neatly,

handwritten on signed, separate sheets, should preferallf le malted to the
editor tefore November i, 1986, although solutions received a§ter that date
wilt also le considered until the time when a solution c^, punished.

1116(b) [1986: 27] (Corrected) Proposed by David Grabiner, Claremont

High School, CIaremont, Calitornia.

Let g(n) be the second smallest positive integer which is not a

factor of n. Continue the series g(n)9 g(g(n)), g(g(g{n))),... until you
reach 3. What is the maximum length of the series?

1131. Proposed by M.S. Klamkin, University of Alberta, Edmonton, Alberta.

Let AlA2A3 be a triangle with sides alf a 2 , a 3 labelled as usual,

 - 78 -

and let P be a point in or out of the plane of the triangle. It is a known

result that if Rlf R2, R3 are the distances from P to the respective vertices
Alf A2t A3$ then alRlt a2R2, a3J?3 satisfy the triangle inequalityf i.e.
alRl + a2R2 + a3R3 > 2a .R. , I 1,2,3. (1)

For the a i.R.i to form a non-obtuse triangle, we would have to satisfy

ax2Rg2 * a22R22 + a 3 2 / ? 3 2 > 2a. 2 i?. 2

which, however, need not be true. Show that nevertheless

ax2Rx2 + a 2 2 i ? a 2 + a 3 2 i ? 3 2 > V5a. 2 j?. 2

which is a stronger inequality than (1).

1132. Proposed by J*T* Groenman, Arnhem, The Netherlands

A triangle AJSC has

circumcircle Klt with centre 0 and
radius R, and inscribed circle Kz,
with centre r and radius r. A third
circle K3 of centre T and radius rx
has the property that there is a
quadrilateral ABlClDl which is both
inscribed in K% and circumscribed about
&3. Find rl in terms of R and r.

1133. Proposed by Stanley Rabinowitz, Digital Equipment Corp., Nashua,

New Hampshire.

The incircle of triangle ABC touches sides BC and AC at points #

and E respectively. If AD = BE, prove that the triangle is isosceles.

1134. Proposed by Loren C. iMrson, St. Olaf College, Northfield,

Minnesota.

Let n be a positive integer, and consider the set of all /?~tuples

{ax,a2,...,a ) of integers for which

(i) 0 < ak < a 2 < < a < r; , and

- i3

(ii) ax + a 2 + .-fa = 0 (mod /?)

Prove that the integers 0,1,2,...,n-l each occur the same number of times as
coordinates of elements in this set.
 - 79 -

1135. Proposed by Jack Garfunkel, Flushing, N.Y.

(A variation of an old problem, dedicated to Dr. Leon Bankoff .)

(a) Given equilateral triangles ABC and A'B'C in the same plane, both
labeled counterclockwise, prove that triangle MXM2M3 is equilateral, where Mx,
M2, M3 are the midpoints of AA', BB', CC respectively.
(b) Given similar triangles ABC and A'B'C in the same plane, prove
that triangle filtl2Ml3 is similar to triangle ABC, where ttlf M2, M3 are as in
(a).

1136. Pressed by George Tsintsifas, Thessaloniki, Greece*

Let AW be a triangle and D, E, F points on BC, CA, AB

respectively. Denote by G p G2, G 3 the centroids of triangles AEF, BDF, CUE
respectively. Prove that
\r r r 1 - 2[ABC] + [DBF]
IU1U2U3\ - g
where [tf] stands for the area of the figure M..

1137. Proposed by Walther Janous, Ursulinengymnasium, Innsbruck, Austria.

Prove or disprove the triangle inequality

1 + 1 + 1 > 375"
m m, m ~ s
a b c
where m , zn, , m are the medians of a triangle and s is its semiperimeter.

1138. Proposed by L.J. Upton, Mississauga, Ontario.

You are given four discs A, B, C, D, of identical appearance, but

weighing 1, 2, 3, and 4 units not necessarily respectively. Determine the
weights of the discs in four weighings on a 2-tray balance (no extra weights
supplied).

1139. Proposed by Clark Kimberling, University of Evansville, Evansville,

Indiana.

Let ABC be a triangle and let A', B', C be the touch points of the
nine-point circle with the A-excircle, B-excircle, and C-excircle,
respectively. Prove that AA', BB', CC concur in a point F', and that F' is
collinear with the centers of the incircle and nine-point circle.

1140. Proposed by Jordi Dou, Barcelona, Spain.

Given triangle ABC, construct a circle which cuts (extended) lines

BC, CA, AB in pairs of points A' and A'', B' and B'', C and C ' respectively
 - 80 -

such that angles A'AA", B*BB' ' , C'CC" are all right angles,
* t t

S O L U T I O N S

Mo fiiv-ttem l& eve* penmanentVf clo&ed. 7fre edit on mill atum^A te ptea&ed
to cori&l&e* foi p-ii&tlcatLoa new &otutc&ti& 01 new Lfi&Lfht&> an £e&t pto&ttmb .

%
999, [1984: 319] Proposed by Jack Garfunkel, Flushing, N.F.

Let R, r, s be the circumradius, inradiuss and semiperimeter?

respectively, of an acute-angled triangle. Prove or disprove that
s2 > 2R2 + 8Rr + 3r2.
When does equality occur?

Solution by Leon Bankoff, Los Angeles, California.

From (2»56) of Bottema et al, Geometric Inequalities, we read

(!> s i n ^]
where a # , {$' , i* are the angles of a triangle and the sums are over these
di B* 1 * ° ° °
three angles. Replacing «—, £—, j — by 90 -a, 90 -p, 90 -v respectively, we
obtain
(2 cosa)2 < 2 sin2a ,
where af fi, t are the angles of an acute triangle. Using
2 sin2a = (I sina)2 - 21 sina sinp
and the known relations
s
2 sina = ^

„ . . A s2 + 4ftr -i- r 2
2 s i n a sin/3 = —
4/?2
R + r
2 cosa =

we o b t a i n
! 2 s 2 + 4i?r 4- r 3
m Msr-
which is equivalent to the given inequality.

Also solved by O. BOTTEMA, Delft, The Netherlands; D.S. MITRINOVIC and

J.E. PECARIC, University of Belgrade, Yugoslavia; and VEDULA N. MUKTY,
Pennsylvainia State University, Middletown, Pennsylvania. There were three
partial solutions received.
 - 81 -

Most solvers observed that equality holds for the equilateral triangle.
HERTA T. FREIIAG, Roanoke, Virginia, and J.T. GROENMAN, Arnhem, The
Netherlands, proved that the inequality also holds for right triangles, with
equality only for the isosceles right triangle.
liitrinovic and Pecaric point out that the proposed problem has previously
appeared as E2388(ii) in the American Mathematical Monthly 79(1972), p.1135
(solution in AMM 80 (1973), p.1142).
t * *

1001. [1985: 15] Proposed by Mark Kantrowitz, student, liaimonides

School, Brookline, Massachusetts.
In the following exact cryptarithmic division, each X can be any of
the decimal digits except the single digit represented by S. Restore the
digits.
XSXX
XSXv5K3Xm
XSX
XXXX
XXXS
XXXX
XSX
XXXS
XXXS

Solution.
1419
94671342374 (S * 4)
946
3963
3784
1797
946
8514
8514

The solution is unique.

Found by RICHARD I. HESS, Rancho Palos Verdes, California;

ALLAN WM. JOHNSON JR., Washington, B.C.; PATRICIA KUSS, Cleveland, Ohio;
J.A. McCALUUM, Medicine Hat, Alberta; GLEN E. MILLS, Valencia Community
College, Orlando, Florida; KENNETH M. WILKE, Topeka, Kansas; and the proposer.
Kuss and Wilke submitted partial arguments for their solutions; all others
just sent the solution.
* * *
 82

1002. [1985: 15] Proposed by Vedula N. Murty9 Pennsylvania State

Universityf Capitol Campus.
Let m and n be given natural numbersf where m < n. Evaluate the
sum
m

J=l
a..
11

I* Solution by Richard A. Gibbsf Fort L&wi® Col lege f Durango, Colorado.

Since
J]
0 n - fflj

\m\
and

V fa + t ] ("a + £> + 1]
L I a J= I a + 1 J '
f=0
the desired sum is
a? ffl] 12?
1
* J • J
\p - m\

v
J=l SI FT
1
L
\m\ j = o
m

•11%:
m
'-JIBJ
]
W i = l j=i

1 V \n - i + 11
1HT I [n - m + l j
ro i = l
1 f n + 1 1
FT [n - m + 2 J

n + 1
«/ - 1

PI
- [mj j g ( / ? * 1 )
" In - a? + l)(i? - » + Z) *

II, Solution by Frank P. Battles and Laura L, Kelleher, Massachusetts

Maritime Academy, Buzzards Bay, MA.
We consider the more general sum
 - 83 -

m
,'"1
J=l L/J
where m < n and a is a non-negative integer. The above can be rewritten by
some elementary manipulations as

a ml Z (/? -
- J J.a

By repeated use of the familiar reduction formula

ftf - 1"
r ']•£:!]
we o b t a i n t h e i d e n t i t y
m
r /? + a 1= r pi + a - n
[n + a - m + 1J £ [n + a - fflj

which we write in the form

m
1 <T p? + a - J"! 1 f /? + a 1
(*)
£ \n + a - ffij TnT [j? + a - f f i + l j
ffij j = l [fflj
When a = 0 it follows directly from (*) that

^° " n + 1 - m '
We n o t e t h a t t h i s i s problem 258 from t h e Two-Year College Mathematics
Journal, September 1983, p . 3 5 3 .
When a = 1 ( t h e g i v e n problem) i t follows from (*) that
m
1 V (n + 1 - jfl m(n + 1)
FT I [n + 1 - m\ ~ (n + 1 - m) (n + 2 - m) '
fflj 7=1
Since
pi + 1 - j l = n M - j fa - n
[A + 1 - m\ n + 1 - m [n - roj
the above can be rewritten as
{n + 1)ffl
(n + 1)S 0 - S t =
23 + 2 - m

Substituting for S 0 and solving for S x gives

m(n + 1)
Si (n + 1 - m)(n + 2 - m)
When a = 2 it follows from (*) that
 - 84 -

m
1 V (n + 2 - n (n + 2)(n + l)m
TnT Z [n + 2 - raj (n + 1 - m)(n + 2 - m)(n + m)
J=l
Since
>? + 2 ~ J ] _ (i? + 2 - J)(j? + w ~ J
n + 2 - a?J "" (ri 4- 1 - 12?) (12 + " n ~m
the above can be rewritten as
(n + 2)(n + 1)S0 - (2/7 + 3JS, + S 2 = ("+ 2Ji" + 1)m
.
Substituting for S0 and S x and solving for S 2 gives
m(n + l)(w + 73 + 1)
^ 2 " (n + 1 - m)(n + ~2~^ n)(n + 3 - m) '
The above procedure can clearly be continued indefinitely. However,
although the denominator of S remains simple, the numerator becomes quite

unwieldy.

III. Solution by the proposer.

Consider a box containing a = n - m + 1 white balls and b = m black
balls* Balls are drawn from this box at random one by one (without
replacement). Let X denote the number of black balls drawn before the first
white ball is drawn. Then it is known that
b m
E(X) =
FTT h -ffi+ 2 * u;
(If all balls were drawn outf this is the average interval between consecutive
white ballsi counting imaginary initial and terminal white balls. See problem
4 ? page 178 of J.V. Uspensky's Introduction to Mathematical Probability, first
edition.)
Since X can take values 0,1,2,...,Z? with respective probabilities
a b a b b ~ 1 a
9
i + b a + 6'a + b - I ' a + b*a. + b ~ T a T F T f ' *'*
b ^ b - I 1 a
f
a + b'a + b - f TT'i"
we g e t
r b 'b'
a i
E(X) - • i \
FTT L fa + b - r
| 2=1 ^ i

n - m+ 1 !2)
__T~~
2=1 [ij
{1) and (2) e s t a b l i s h t h a t
 - 85 -

m(n + 1)
3 =
W © + l)(n -ffl+ 2) *
Aiso solved by CURTIS COOPER, Central Missouri State University, Warrenshurg,
Missouri; WALTHER JANOUS, Ursul mengymnasium, Innsbruck, Austria;
MARK KANTROWITZ, student, Maimonides School, Brooklme, Massachusetts;
BRUCE WATSON, Memorial University of Newfoundland, St. John's, Newfoundland;
and KENNETH M. WILKE, Topeka, Kansas, Janous points out that the proposed
problem is item 17, page 629, of A.P. Prudnikow, Ju.A. Brychkow, and
0.1. Maiichew, Integrals and Series (Elementary Functions), Nanka, Moscow,
1981 (in Russian)•
% * t

1003. [1985: 15] Proposed by M.S. Klamkin, University of Alberta.

Without using tables or a calculator, show that
,2/5
In 2 >

Solution by Jordan B. Tabov, Sofia, Bulgaria.

From the familiar series
X2 X3 (|x| < 1)
ln(l + x) = x - -J- + 3. ¥
and

ln(l - x) = ~x - x* x" (|x| < 1)

3" T
we obtain
In j - L £ = ln(l + x) - ln(l - x)

Putting x = TJ-, we obtain

In 2 =
277
<2n + 1)3
n=0
Thus we have
2 .2
In 2 > -t +
3
3«3 5-3 7-37
!
9-39
2 2 2 2 2
_ +
7 HT + T2T5- + TS30¥ + 177147
> 0.6666666 + 0.024691 + 0.001646 + 0.0001306 + 0.000011
= 0.6931452
= a .
 - 86 -

Then
az > 0.48045
aA > 0-230832
and so
a5 > (0,230832)(0.693145) = 0.16000004664.
("2"! 2
Hence (In 2 ) s > aB > 0.16 = LJ , which is equivalent to the required

inequality.

Also solved by RICHARD I. HESS, Rancho Palos Verdes, California;

FRIENDff.KIERSTEAD JR., Cuyahoga Falls, Ohio; VEDULA N. MURTY and ELI DAVID,
Pennsylvania State University, Middletown, Pennsylvania; BASIL C. RENNIE,
James Cook University of North Queensland, Australia; and BOB SALER,
Agincourt, Ontario,
% t t

1004, [1985: 15] Proposed by O. Bottema, Delft, The Netherlands.

There exists a right triangle with perimeter p and area F if and
only if the positive numbers p and F satisfy what condition?

I. Solution by Lawrence Somer, George Washington University, Washington,

D»C*
The required condition is that p > (2 + 2JZ)JF* It is known and easily
provable that among all right triangles of a given areaf the one with minimum
perimeter is the isosceles right triangle. Let a be one of the legs of the
isosceles right triangle with area F. Then a 2 = 2F* and so the perimeter is

2a + J2a2 = (2 4- JZ)a = (2 + 2JZ)JF. Thus any right triangle of area F and

perimeter p must satisfy p > (2 4- 2JZ)JF*
Conversely? for a given area F, a right triangle of area F can be found
with perimeter larger than any given number Nl simply choose N for one leg of
the triangle and 2F/N for the other. By continuity, there exists a right
triangle of area F and perimeter equal to any real number greater than or
equal to the perimeter of the isosceles right triangle of area F. Thus the
condition p > (2 + 2J2)JF is also sufficient to ensure a right triangle of
area F and perimeter p exists.

II. Solution by Mark Kantrowitz, student, Maimonides School, Brookline,

Massachusetts.
Given a right triangle with sides a, b, c, we have
 - 87 -

p = a + & + e = a + 5 + Ja2 + b1
and
F = ab/2.
Thus from the A.M.-G.M. inequality,

p a + b , 1
+
[7 7J

= (72" + l)v^
with equality if and only if a = -b. Thus
p 2 > F(2 + 2V2")2 = F(12 + 8V2T).

i l l . Solution and generalization by Hayo Ah 1 bury, Bemdorm, Alicante,

Spam.
Consider a triangle with sides a, 6, c, perimeter p and area F, which
means
„ _ ab sinC
F = 2

and

p~a+b + c - a + b + J a 2 + 5 2 - 2ab cosC •

Then

p2 + £> + Ja2 + b2 - 2ab cosC

= (a + b)2 + a 2 + b2 - 2a£> cosC + 2{a + 6)Ja2 + b2 - 2a6 cosC

= 2(a + & ) 2 - 2a5(l + cosC) + 2(a + b)(p - a - b)
4F
= 2(a + 5)p - - - ^ (1 + cosC)
sine
and so
, , p2sinC + 4F(1 + cosC)
a + b =£ „ -1- L.
ZpsinT
Thus
2F
+ - p'sinC + 4F(1 + cosC)
a sinC "" 2p sinC
which simplifies to
(2p sinC)a2 - (p2sinC + 4F(1 + cosC))a + 4Fp = 0. (1)
Since a is real, this means that
[p2sinC + 4F(1 + cosC)]2 > 16Fp(2p sinC)
or
 - 88 -

p 4 sin 2 C + 8p 2 FsinC(l + cosC) + 16F 2 (1 4- cosC) 2 > 32p 2 FsinC

or
[p 2 sin 2 C - 4F(1 + cosC)] 2 > 16p2FsinC(l - cosC). (2)
Since
c = p - {a + 5)
__ p 2 sinC + 4F(1 * cosC)
p 2p sinC "

(3!
2p sinC ~~f
we have
p a simC - 4F(1 + cosC) > 0,
and so (2) is equivalent to
p 2 sinC - 4F(1 + cosC) > 4 p # V s i n C ( l - cosC)
or
p 2 sinC - 4^FYsinC(l - cosC)p - 4F(1 + cosC) > 0- (4;
Since p > 0, (4) is equivalent to

4JF JsinCjl - coscr + 7T5FsinC(l - cosC) -l- 16FsinC(l * cosC)

p > , . 2" sinC
2JF VsinC r **- ^ , FTl
= —^-^ [VI ~ COSC 4- 7?]
sinC
~"2F~ [1 - cosC n
= 2 X
siinC
I 2 *
or
pj gvg(3inC/2 + 1) ^ (5)

ysin C
Equality holds if and only if (1) is a perfect square. Since 6 satisfies the
same equation as a does, namely (l) f this means that equality holds in (5) if
and only if a = 6.
Condition (5) is not only necessary but also sufficient. Any positive
o
values p, F f and C (< 180 ) satisfying (5) yield, according to equations (1)
and (3), positive values a9 b9 c such that
a + b + c = p
#5 sin C _
= F
2
and

c = Ja 2 + ft2 - 2ab cos C < a + ft ,

and which are therefore sides of a triangle with area F, perimeter p, and
angle C.
 - 89 -

O
To solve the stated problem, put C = 90 ; then (5) becomes
p > (2 + 2JZ)JF
which is the required relation.
In the notation of [1], (5) says that for any triangle,
s i n
> ^( ?/2 + 1) JF .
./sin 7
o
For fixed F, the right side is minimized when 7 = 60 ; hence for any triangle,
« > 3 3 / 4 JF .
This is inequality 4*2 of [1].
Reference:
[1] Bottema et al, Geometric Inequalities.

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

Bay, MA; J.T. GROENMAN, Arnhem, The Netherlands; RICHARD I. HESS, Rancho Palos
Verdes, California; FRIEND H. KIERSTEAD JR., Cuyahoga Falls, Ohio; LEROY F.
MEYERS, The Ohio State University, Columbus, Ohio; VEDULA N. MURTY,
Pennsylvania State University, Middletown, Pennsylvania; MICHAEL M. PARMENTER,
Memorial University of Newfoundland, St. John's, Newfoundland; GEORGE
TSINTSIFAS, Thessaloniki, Greece; and the proposer. Four other solvers
submitted solutions which involved the sides of the triangle as well as p and
F, but I have interpreted the problem to mean that the condition must be only
in terms of p and F* There were also three incorrect solutions submitted.
* * t

1005* [1985: 16] Proposed by Hidetosi Fukagawa, Yokosuka High School,

Tokai-<:ity, Aichi, Japan.
A chord AB divides a circle 7
into two segments. A circle 7 x, of radius
ru is inscribed in one of the segments,
tangent to AB at its midpoint C and to the
arc at D, as shown in the figure* A circle
7 2 , of radius r2, is then inscribed in the
mixtilinear triangle CBD. The common
interior tangent to 7 i and 7 2 meets circle
7 in P and Q. Find the length of PQ in
terms of rx and r 2 .
 - no *

Solution by Dan Sokolowsky, Brooklyn, N.Y•

Let O, Ox, 0 2 denote the respective
centers of 7, ylt y2, and rf rx ? r2 their
respective radii. Let i2 touch 1x at U,
and extend IX? to meet 7 at Q # .
Consider the arbelos w on diameters
ZW' (of 7), DC (of 7 X ) , and CQ\ The circle
J
7 2 is then one of w s "twin circles of
Archimedes", and, as is well-known (e.g.
R.A. Johnson, Advanced Euclidean Geometry, Q' = Q
pp.116-117):
(i) the common internal tangent PQ of yx and iz passes through Q', i.e.
Q = Q*;
r
(ii) r 2 Jr - r
i)

Solving (ii) for r, we obtain

r = ;i)

Draw QWLPQ and let x = 0W. Then

PQ = 2 j r 2 - x 2 . (2)
Since O^/IFQ, JCWG - 4DJIQ. Hence
r
Oji/ " otQ 2F~=~T7 '
and so
rrf (3)
x = 2r
Substituting (1) in (3) we obtain

(4)
rx -1- r 2
Finally, substituting (1) and (4) in (2) and simplifying, we obtain

PQ
4r 1 2 vF7F7
r 2 „ rJ 2
' 1 2

Also solved by LEON BANKOFF, Los Angeles, California; EERFA T. FREITAG,

Roanoke, Virginia; RICHARD I. BESS, Rancho Palos Verdes, California; WALTHER
JANOUS, Ursulinengymnasium, Innsbruck, Austria; GEORGE TSIhTSIFAS,
Thessaloniki, Greece; and the proposer.
This problem is from a lost 1844 sangaku (see the note following the
solution to problem 995 [1986: 60]). It is recorded in Professor Fukagawa's
 - 91 -

book Collection of Sangaku in Aichi Prefecture and also in his book Study of
Sangaku? p.200.
t t t

1006, [1985: 16] Proposed by Hans Havermann, Westonf Ontario.

Given a base-ten positive integer of two or more digits, it is
possible to spawn two smaller base~ten integers by inserting a space somewhere
within the number* We call the left offspring thus created the farmer (F) and
the value of the right one (ignoring leading zeros, if any) the ladder (L). A
number is called modest if it has an F and an L such that the number divided
by L leaves remainder F, (For examplef 39 is modest,)
Considers for n > 1, a block of n consecutive positive integers all of
which are modest* If the smallest and largest of these are a and 6,
respectively, and if a - 1 and b + 1 are not modest, then we say that the
block forms a multiple berth of size n. A multiple berth of size 2 is called
a set of £wz/is, and the smallest twins are {411,412}. A multiple berth of
size 3 is called a set of tripletsf and the smallest triplets are
{4000026,4000027,4000028}.
(a) Find the smallest quadruplets,
(b) Find the smallest quintuplets. (There are none less than 25
million.)

Partial solution by Leroy F. Meyers, The Ohio State University, Columbus,

Ohio.
The positive integer N is modest if and only if there are positive
integers m, F, and L such that
N = 10mF + L, N s F(mod L), and F < L < 10 m .
But
10fflF + L = F(mod L) «=* (10ffi - 1)F = 0(mod L) «=> Lj(10ffi - 1)F.
Hence a way to find modest numbers is to choose positive integers m and F
(with F < 10 - 1) and then find divisors L of (10 - 1)F which lie strictly
between F and 10 . Obviously N < 10 . For m = 1, the divisors of 10 - 1
are 1, 3, and 9, so that the modest numbers obtained in this case are 13, 19,
23, 26, 29, 39, 46, 49, 59, 69, 79 and 89. For m = 2, the divisors of 10ffi - 1
are 1, 3, 9, 11, 33 and 99, and there are 350 modest numbers in this case
(give or take a few). Twins and higher multiplets occur especially when
10 - 1 has many small divisors • The smallest twins occur when m = 2: 411,
412. The smallest triplets seem to occur when m = 6: 4000026, 4000027,
 - 92 -

4000028; the smaJlest quadruplets seem to occur even earlier, when m = 6:

4000011, 4000012, 4000013, 4000014; and quintuplets (possibly the smallest)
occur when m = 18: 20-10 ** + Lf where 35 < L < 39. These quintuplets are a
result of the favourable factorization
10 18 - ] = 34-7-ll-13.19-37.52579-333667.
Note that it is possible forraultipletsto be derived from several values of
m; for example, the twins 2036, 2037 come from m = 2 and m = 3, respectively.

TTie above quadruplets were also found by RICHARD I. BESS, Rancho Palos
Verdes, California; and the proposer. The proposer submitted a computer
printout showing that the above triplets and quadruplets are minimal. For an
earlier problem by the proposer on modest numbers, see #1291, Journal of
Recreational Mathematics 17 (1984) p*140»
* * *

1007. [1985: 16] Proposed by Leroy F. Meyers, The Ohio State University,
It is known that every positive rational number can be written as
the sum of finitely many reciprocals of distinct positive integers (the
Egyptian fraction decomposition). Show that every positive real number can be
written as the sum of infinitely many reciprocals of distinct positive
integers.

Solution by Richard I. Hess, Rancho Palos Verdes9 California.

Consider the harmonic series
1 + r,- + ^ + . . . + - + . . .
Z 3 n
which is known to diverge* For any positive real number x, start in the
harmonic series with the largest term < x and continue to add terms until the
next term would cause the running sum to be > x. The partial sum at this
point is less than x by some amount xx* Now skip to the largest term in the
harmonic series which is < xx and continue as before until the next term would
cause the new running sum to be > x. Define x 2 as x minus the running sum.
Continue these steps forever to produce a running sum which converges to x and
contains an infinite number of distinct reciprocals.
Also solved by WALTHER JANOUS, Ursulinengymnasiumf Innsbruck, Austria;
FRIEND H. KIERSTEAD JR., Cuyahoga Falls, Ohio; EDWIN M. KLEIN, University of
Wisconsin, Whitewater, Wisconsin; DAN SOKOWWSKY, Brooklyn, N.Y>; LAWRENCE
SOMER, George Washington University, Washington, D.C*; STAN WAGON, Smith
College, Northampton, Massachusetts; and the proposer.
t t *
 - 93 -

1008. [1985: 16] Proposed by Jordan B. Tabov, Sofia, Bulgaria.

A circle 7 of diameter AB and two real numbers x and y are given.
A variable point N ranges over 7 * Find the locus of a point M on the line AN
such that AM = \xAN + yBV|. B(032)

Solution by the proposer.

We let A be the origin and B be the
point (0,2), Then we need only find the
locus of M for points N in the first
quadrant, since by reflecting this locus
about the y axis we obtain the entire locus*
Let 9 be the angle between AN and the A(030)
positive x axis, so that 0 < 9 < 90 * Then AN = 2 sine and BN = 2 cose.
We may assume that x > 0 and that x and y are not both 0* Then there is
a unique pair (p,F) such that x = p cosF, y = p sinF, p > 0, and
o o
~90 < ^ < 90 5 (p?^) are just the polar coordinates of the point (x,y).
Now
AM = jxA# •!» yBY|
= Jp cosF«2 sine + p sinF*2 cose |
= 2p|cosF sine + simp cose | ,
that is, M is one of the two points on the line AN at distance
2p|cosF sine + sinF cose |
from the origin. We may therefore assume that M has rectangular coordinates
(2p(cosF sine + sinF cose)cose 9 2p(cosF sine + sinF cose)sine), (1)
as the rest of the locus may be obtained by reflection through the origin.
Let Ox be the point (y,x), that is,
Ox = (p sinF,p cosF). (2)
Then the vector Ujt, from (1) and (2), has components
(2p cosF sine cose 4- 2p sinF cos 2 e - p sinF ,
2p cosF sin2e + 2p sinF sine cose - p cosF)
= (p cosF sin2e + p sinF(2cos2e - 1 ) ,
p sinF sin2e - p cosF(l - 2sin 2 e))
= (p cosF sin2e + p sinF cos2e f p sinF sin2e - p cosF cos2e)
= (p sin(2e 4- T) 9-~p cos(2e + F ) ) ,
and so UlM = p. Hence M lies on the circle with centre u 4 and radius p. Note
that from (2), this circle also passes through the origin.
 - 94 -

From (1) we get that when 9 c 0 ,

M = (2p sinFfO) = (2y,0)
o
and when 0 = 90 ,
$ = (0,2p cosF) = (0,2x).
It follows that the locus of M is a semicircle, oriented as at the left if
y > 0j and as at the right if y < 0:

r^sxr?;
(032x)

v°l
T. I
(2y$0)s
(2y3 0)
y ± o

To complete the locus we make it symmetric with respect to the y axis and
origin, obtaining

y > 0

We note two special cases:

x ± 0S y = 0 x = 03 y i- 0

There was one partial solution submitted.

t t
 - 95 -

1.009. [1985: 17] Proposed by Allan to. Johnson Jr., Washington, B.C.
Prove that every pandiagona] fourth-order magic square can be
written in the form

A + B 4 C D + B - C i) - a 4 e A - B - C

\ D - B - E A - B + E A 4 B - E D H + f

\ A + B - C D 4 B + C B ~ B ~- C A ~ B + e \

D - B 4 E A - B -• £ A + B + E D 4 B - E

Solution by Kenneth M» Wilke, Topeka, Kansas.

Maurice Kraitchik, in Mathematical Recreations (Second Revised Edition,
Dover, 1953) pp.186-187, has shown that all pandiagonal fourth-order magic
squares can be written in the form

a b c S - (a+b+c)

e S - (a+£»4e) a - c 4 e b 4 c - e

S S
a454c - ^ ^ - a ?~5
S S
_ (a-r+e) a+5+e - TJ.
T£ - (fc+c-e) 2~e

The proposer's form can be obtained by taking

a = A 4 £ 4 C
fc = Z) + B - C
c = Z? - B 4 C
e = 0 - B - ff
and
S = 2A 4 20.
For a similar problem, see Crux 605 (solution in 11982: 22]).

Also solved by the proposer.

* * *