Sharpening Everyday Mental

Problem Solving in Mathematics and Beyond
Print ISSN: 2591-7234

Online ISSN: 2591-7242
Series Editor: Dr. Alfred S. Posamentier

Distinguished Lecturer
New York City College of Technology - City University of New York
There are countless applications that would be considered problem solving in

mathematics and beyond. One could even argue that most of mathematics in one
way or another involves solving problems. However, this series is intended to be of
interest to the general audience with the sole purpose of demonstrating the power
and beauty of mathematics through clever problem-solving experiences.
Each of the books will be aimed at the general audience, which implies that the
writing level will be such that it will not engulfed in technical language — rather
the language will be simple everyday language so that the focus can remain on the
content and not be distracted by unnecessarily sophiscated language. Again, the
primary purpose of this series is to approach the topic of mathematics problem-
solving in a most appealing and attractive way in order to win more of the
general public to appreciate his most important subject rather than to fear it. At
the same time we expect that professionals in the scientific community will also
find these books attractive, as they will provide many entertaining surprises for the
unsuspecting reader.
Published
Vol. 30 Sharpening Everyday Mental/Thinking Skills Through Mathematics

Problem Solving and Beyond
by Alfred S Posamentier and Hans Humenberger
Vol. 29 Introduction to Chaos, Fractals and Dynamical Systems

by Phil Laplante and Chris Laplante
Vol. 28 Seduced by Mathematics: The Enduring Fascination of Mathematics

by James D Stein
Vol. 27 Mathematics: Its Historical Aspects, Wonders and Beyond

by Alfred S Posamentier and Arthur D Kramer
For the complete list of volumes in this series, please visit www.worldscientific.com/series/psmb
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
Library of Congress Cataloging-in-Publication Data

Names: Posamentier, Alfred S., author. | Humenberger, Hans, author.
Title: Sharpening everyday mental/thinking skills through mathematics problem
solving and beyond / Alfred S. Posamentier (City University of New York, USA),
Hans Humenberger (University of Vienna, Austria).
Description: New Jersey : World Scientific, [2024] | Series: Problem solving in
mathematics and beyond, 2591-7234 ; volume 30 | Includes index.
Identifiers: LCCN 2023012622 | ISBN 9789811273940 (hardcover) |
ISBN 9789811276392 (paperback) | ISBN 9789811273957 (ebook for institutions) |
ISBN 9789811273964 (ebook for individuals)
Subjects: LCSH: Problem solving. | Problem solving--Methodology. | Mathematics.
Classification: LCC QA63 .P673 2024 | DDC 510--dc23/eng20230911
LC record available at https://lccn.loc.gov/2023012622
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.
Copyright © 2024 by World Scientific Publishing Co. Pte. Ltd.

All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance
Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy
is not required from the publisher.
For any available supplementary material, please visit

https://www.worldscientific.com/worldscibooks/10.1142/13342#t=suppl
Desk Editors: Balasubramanian Shanmugam/Rosie Williamson
Typeset by Stallion Press

Email: [email protected]
Printed in Singapore
About the Authors
Alfred S. Posamentier is currently a Distinguished

Lecturer at the New York City College of Technol-
ogy of the City University of New York. Prior to
that, he was the Executive Director for Internation-
alization and Funded Programs at Long Island Uni-
versity, New York. This was preceded by 5 years
as the Dean of the School of Education and Profes-
sor of Mathematics Education at Mercy University,
New York. Before that, he was at the City College
of the City University of New York for 40 years, at which he is now the
Professor Emeritus of Mathematics Education and the Dean Emeritus of
the School of Education. He is the author and co-author of more than 80
mathematics books for teachers, secondary and elementary school students,
as well as the general readership. Dr. Posamentier is also a frequent com-
mentator in newspapers and journals on topics related to education and
mathematics.
After completing his B.A. degree in mathematics at Hunter College
of the City University of New York, he took a position as a teacher of
mathematics at Theodore Roosevelt High School (Bronx, New York), where
he focused his attention on improving students’ problem-solving skills and
at the same time enriching their instruction far beyond what the traditional
textbooks offered. During his 6-year tenure there, he also developed the
v
vi Sharpening Everyday Mental/Thinking Skills
school’s first mathematics teams (both at the junior and senior levels). He is
still involved in working with mathematics teachers and supervisors, nation-
ally and internationally, to help them maximize their effectiveness.
Immediately upon joining the faculty of the City College of New York
in 1970 (after having received his master’s degree there in 1966), he began
to develop in-service courses for secondary school mathematics teachers,
including such special areas as recreational mathematics and problem solv-
ing in mathematics. As the Dean of the City College School of Education
for 10 years, his scope of interest covered the full gamut of educational
issues. During his tenure, he took the school from the bottom of the New
York State rankings to the top with a perfect NCATE accreditation assess-
ment in 2009. He also raised more than 12 million dollars from the private
sector for innovative education programs. Dr. Posamentier repeated this
successful transition at Mercy University, where he enabled it to become
the only college to have received both NCATE and TEAC accreditation
simultaneously.
In 1973, Dr. Posamentier received his Ph.D. from Fordham University
(New York) in mathematics education and has since extended his reputa-
tion in mathematics education to Europe. He has been a visiting professor
at several European universities in Austria, England, Germany, the Czech
Republic, Turkey, and Poland. In 1990, he was the Fulbright Professor at
the University of Vienna.
In 1989, he was awarded an Honorary Fellow position at South Bank
University (London, England). In recognition of his outstanding teaching,
the City College Alumni Association named him Educator of the Year in
1994 and 2009. New York City had the day, May 1, 1994, named in his
honor by the President of the New York City Council. In 1994, he was
also awarded the Das Grosse Ehrenzeichen für Verdienste um die Republik
Österreich (Grand Medal of Honor from the Republic of Austria), and in
1999, upon approval of Parliament, the President of the Republic of Austria
awarded him the title of University Professor of Austria. In 2003, he was
awarded the title of Ehrenbürgerschaft (Honorary Fellow) of the Vienna
University of Technology and, in 2004, was awarded the Österreichisches
Ehrenkreuz für Wissenschaft & Kunst 1.Klasse (Austrian Cross of Honor
for Arts and Science, First Class) from the President of the Republic of
About the Authors vii
Austria. In 2005, he was inducted into the Hunter College Alumni Hall of
Fame, and in 2006 he was awarded the prestigious Townsend Harris Medal
by the City College Alumni Association. He was inducted into the New
York State Mathematics Educator’s Hall of Fame in 2009, and in 2010 he
was awarded the coveted Christian-Peter-Beuth Prize from the Technische
Fachhochschule, Berlin. In 2017, Dr. Posamentier was awarded Summa
Cum Laude nemmine discrepante by the Fundacion Sebastian, A.C., Mexico
City, Mexico.
He has taken on numerous important leadership positions in mathemat-
ics education locally. He was a member of the New York State Education
Commissioner’s Blue Ribbon Panel on the Math-A Regents Exams, and
the Commissioner’s Mathematics Standards Committee, which redefined
the Mathematics Standards for New York State, and he also served on the
New York City schools’ Chancellor’s Math Advisory Panel.
Dr. Posamentier is still a leading commentator on educational issues
and continues his long-time passion of seeking ways to make mathematics
interesting to teachers, students, and the general public — as can be seen
from some of his more recent books.
For more information and a list of his publications, see https://
en.wikipedia.org/wiki/Alfred_S._Posamentier.
Hans Humenberger is currently a full Professor

of Mathematics with special emphasis on mathe-
matics education at the University of Vienna (Aus-
tria). He began his career studying mathematics
and sports at the University of Vienna. In the 1990s
he was a high school teacher at several Viennese
high schools and also a graduate-assistant at the
University of Natural Resources and Life Sci-
ences, Institute for Mathematics, Vienna. In 1993
he earned his Ph.D. at the University of Vienna, and 1998 he received
his habilitation in the field of mathematics education. Between 2000 and
2005 he was Assistant Professor of mathematics at the University of Dort-
mund (Germany). During this time, he also spent a half year as Interim
Professor at the University of Duisburg-Essen (Germany). He returned to
viii Sharpening Everyday Mental/Thinking Skills
Vienna in 2005 to accept his current professorship for mathematics at the

University of Vienna. Since then, he has been the head of a working group
“Didactics of Mathematics and School Mathematics” and also responsible
for the educational training of preservice mathematics teachers for sec-
ondary school.
Of particular note is the forthcoming German title Anschauliche
Elementargeometrie, written with his former colleague Berthold Schup-
par, which is a geometry book intended largely for future teachers of
mathematics.
He has been interested in mathematics problem solving for many years,
which is reflected in several papers and corresponding seminars for preser-
vice mathematics teachers, which he has held at several universities.
In 2007, he established an opportunity for secondary school students
(grades 5–8) to attend seminars/workshops at the University of Vienna to
explore interesting and challenging problems. This successful program is
funded by the Vienna Department of Education and continues to the present
day. Furthermore, he is particularly interested in making mathematics inter-
esting for the general audience since this is important for the general per-
ception of mathematics in our society.
Since 2007 he has been editor of an Austrian school textbook series in
mathematics for grades 5–8, and in 2022 he was a member of a committee
that established a new syllabus for mathematics at primary and secondary
school levels in Austria.
His main fields of interest are mathematics as a process, applications of
mathematics, problem solving, geometry, and stochastics.
More details and a complete list of his publications can be found at the
following website: https://homepage.univie.ac.at/hans.humenberger/.
Contents
About the Authors v

Introduction xi
Chapter 1. Arithmetic Calculation Surprises 1
Chapter 2. Numerical Novelties 41
Chapter 3. Challenging Problems with Surprising Solutions 113
Chapter 4. Algebra: A Unique Problem-Solving Tool 173
Chapter 5. Geometric Gems 219
Chapter 6. For Experts: Stepwise Solutions to

Challenging Problems 277
Index 319
ix
This page intentionally left blank
Introduction
This book is intended to share the usefulness, power and beauty of mathe-
matics with the general public and beyond! Mathematics is a subject taught
at all grades from kindergarten through high school, and yet it is the one sub-
ject that most adults are almost proud to admit not having been very good
at and, therefore, tend to avoid it where they can. This, unfortunately, is
sometimes passed onto children, which extends this negative legacy. This is
regrettable as it indirectly tends to limit a person’s reasoning ability, which
is something that is trained throughout all of mathematics instruction —
albeit oftentimes indirectly. One wonders why this continues to be the case
decade after decade. At the elementary school level, teachers who reflect
the general public are among those who are not very enamored with math-
ematics. Consequently, they do not exert themselves to make the subject
interesting and exciting and are largely focused on “teaching to the test.”
We believe this is a huge disservice to the students. At the secondary school
level, mathematics teachers have the opportunity to generate a new interest
in mathematics, yet they too are concerned about student test performance
and, therefore, rarely deviate from the standard curriculum. It is exactly that
measured deviation that we believe is essential to demonstrate the beauty of
mathematics, which in turn can generate or rekindle an interest among the
students. There is a multitude of topics and skills in mathematics to which
students are never exposed. In this book, we are attempting to demonstrate
not only to the teachers but, in large measure, to the general public, the
wonders that are hidden in mathematics and that can make it a very popular
subject both in school and beyond.
xi
xii Sharpening Everyday Mental/Thinking Skills
Chapter 1 appropriately presents a wide variety of arithmetic calcula-

tions, which go far beyond the four basic operations: addition, subtraction,
multiplication, and division. Here we investigate unusual and not well-
known calculations and calculating shortcuts that not only give a genuine
insight into arithmetic operations but also improve arithmetic skills both
in writing and in thinking. This leads us nicely into Chapter 2, which
presents a wide spectrum of numerical novelties that provides an extraor-
dinary view of number relationships and numerical properties that can go
a long way to impress the general readership about the hidden beauties in
mathematics.
One of the key factors in mathematics is its ability to enable us to solve
problems. Learning how to solve mathematical problems often has a favor-
able spin-off to solving everyday problems. For example, there are times
when we deal with an everyday situation where we consider “the worst-case
scenario” of the situation, which is analogous to solving a mathematical
problem by considering extremes. Or we might consider the best path to
take from point A to point B, where geometric relationships can be helpful.
These are just some of the everyday problems that a broad background in
mathematics problem solving can be supportive. Therefore, Chapter 3 of
this book focuses on problem-solving strategies in mathematics, which are
intended to be both entertaining and instructional. One of the most basic
tools in mathematics is algebra, which allows us to understand the general
case of certain ideas and concepts, and also to discover answers to com-
plicated problems. Chapter 4 focuses on algebra and its somewhat unusual
applications, which are intended to strengthen algebraic skills while at the
same time demonstrating its power. As a side benefit, it contributes to a
more favorable feeling towards the subject of mathematics in general.
Chapter 5 highlights the one aspect of mathematics that clearly demon-
strates it beauty, which is geometry, perhaps the oldest branch of the field
of mathematics, as it allows us to appreciate physical relationships that
are sometimes counterintuitive and, thereby, can be highly motivating. The
geometry that is offered here is beyond that which is taught at the secondary
school level yet requires nothing more than some basics taught in school.
Finally, in Chapter 6 we provide a somewhat more challenging topic
than offered in the previous chapters, as it explores a problem solving
Introduction xiii
technique, which requires a bit of patience as it deals with problems in

a stepwise procedure. That is, rather than to attack the problem with an
immediate elegant solution, we present problems that can be very neatly
solved by first considering various minor steps that will lead to a proper
solution.
In its totality, this book is intended to demonstrate a variety of
neglected aspects of mathematics that can regenerate interest in the sub-
ject for the general readership as well as for teachers who will, hopefully,
provide a significant timeslot in their teaching each week to motivate stu-
dents with some of the amazing aspects of mathematics we highlight. In
short, we hope that this book will go a long way to popularize the field
of mathematics by demonstrating its power and beauty beyond where
most people believe is possible. So, join us now as we begin our jour-
ney through a myriad of surprising aspects of this most important subject:
mathematics.
Chapter 1
Arithmetic Calculation Surprises
Although in today’s culture arithmetic skills have been relegated to com-

puters and calculators, there is still a great advantage to have insight into
arithmetic calculations. They can not only be helpful and sometimes even
swifter than a calculator, but also as a tool for appreciating mathematics
and its logic. To multiply a two-digit number by 11 can be done more
quickly mentally than with a calculator, if one knows that this product can
be obtained by taking the sum of the digits of the two-digit number and
placing it between the two digits as is the case with 53 × 11 = 583. Nat-
urally, when the sum of the digits exceeds 9, slight adjustments must be
made, such as carrying the 1 to the tens place. There are many astonishing
arithmetic relationships that have been discovered over the past centuries.
In this chapter, we will highlight merely a few with the hope that the reader
will be motivated to seek out others. A note on symbols for multiplication in
this book: There are various ways to indicate multiplication. The first one is
the small ×; a second one is a centered dot; and a third one is simply using
no symbol as with 2y or ab. In this book we will use all three versions as
most convenient.
Clever Addition
One of the most popularly repeated stories from the history of mathematics
is the tale of the famous mathematician Carl Friedrich Gauss who at age 10
was said to have mentally added the numbers from 1 to 100 in response to a
1
2 Sharpening Everyday Mental/Thinking Skills
busy work assignment given by the teacher.1 Although it is a cute story and
generally gets a very favorable reaction, it happens to provide us with a neat
little formula for adding numbers in an arithmetic sequence, which is a list
of numbers that has a common difference between consecutive numbers.
For example, 2, 4, 6, 8, 10, . . . is an arithmetic sequence as is each of the
following:
5, 10, 15, 20, 25, 30, . . .
7, 11, 15, 19, 23, 27, . . .
Perhaps the simplest arithmetic sequence is the natural numbers:
1, 2, 3, 4, 5, . . .. What Gauss did to get the sum of the first 100 natural
numbers without writing a single number was not to add the numbers in the
order in which they appear but rather to add them in the following way:
the first plus the last,
the second plus the next-to-last,
the third plus the third from last,
and so on.
If we do this, we get the following:
1 + 100 = 101
2 + 99 = 101
3 + 98 = 101
4 + 97 = 101
..
.
50 + 51 = 101
Note each pair has the same sum of 101. The sum of these 50 pairs of
numbers is 50 × 101 = 5,050. From this example, we can derive a useful
formula for adding numbers in an arithmetic sequence. We added the first
and the last and multiplied this sum by one-half the number of members of
1 According to E.T. Bell in his book, Men of Mathematics (New York: Simon & Schuster,
1937), the problem given to Gauss was of the sort: 81297 + 81495 + 81693 + · · · + 100899,
where the common difference between consecutive terms was 198 and the number of terms
was 100. Today’s lore uses the numbers to be summed from 1 to 100, which makes the point
just as well but in simpler form.
Arithmetic Calculation Surprises 3
the sequence. We can generalize this and get a formula for an arithmetic
series of n terms (with n an even number), where a is the first term and
l is the last term (using Gauss’ method) as follows: Sum = n2 (a + l). This
formula turns out to be correct also in the case that n is an odd number,
where in this case a + l is an even number. Here, we have an example of
how simple it is to derive a very useful mathematical formula, one based
on a very lovely pattern that wasn’t completely obvious initially.
Surprising Addition
There is not much one can say beyond admiring the following results and
appreciating the surprises that the addition of natural numbers offers:
1+2 = 3
4+5+6=7+8
9 + 10 + 11 + 12 = 13 + 14 + 15
16 + 17 + 18 + 19 + 20 = 21 + 22 + 23 + 24
25 + 26 + 27 + 28 + 29 + 30 = 31 + 32 + 33 + 34 + 35
etc.
Surprising Division
Some numbers lend themselves to a very beautiful symmetric division,

which can be best demonstrated as follows:
22 × 22 484
121 = = = 121
1+2+1 4
333 × 333 110889
12321 = = = 12321
1+2+3+2+1 9
4444 × 4444 19749136
1234321 = = = 1234321
1+2+3+4+3+2+1 16
55555 × 55555 3086358025
123454321 = =
1+2+3+4+5+4+3+2+1 25
= 123454321
This pattern continues for four further steps until the digit 9 in the middle
is reached.
Surprising Multiplication Results
When the number 076923 is multiplied by the numbers 1, 10, 9, 12, 3, and 4,
quite a surprising pattern results; with the numbers staying in the same order,
beginning with the multiplication by 10, the digit of the previous result far
left is moved with each multiplication to the far right. Furthermore, if you
inspect the vertical arrangements of the resulting numbers, you will find the
same pattern.
076923 × 1 = 076923
076923 × 10 = 769230
076923 × 9 = 692307
076923 × 12 = 923076
076923 × 3 = 230769
076923 × 4 = 307692
An analogous pattern can be obtained by multiplying the number 76923

by the numbers 2, 7, 5, 11, 6, and 8 as follows:
76923 × 2 = 153846
76923 × 7 = 538461
76923 × 5 = 384615
76923 × 11 = 846153
76923 × 6 = 461538
76923 × 8 = 615384
Once again you will note the vertical arrangement of the digits of the
numbers on the right side of the equal sign is analogous to the horizontal
arrangements.
Using All Nine Digits to Make 100
The challenge here is to use all 9 digits 1, 2, 3, 4, 5, 6, 7, 8, and 9 to create

an integer plus a fraction whose sum is 100. Here are 11 ways in which this
can be done. An ambitious reader may wish to seek others.
69258 7524 5643 3546 5823

3+ , 81 + , 81 + , 82 + , 91 + ,
714 396 297 197 647
7524 5742 1578 1428 1752
91 + , 91 + , 94 + , 96 + , 96 + ,
836 638 263 357 438
2148
96 +
537
Calculation Surprises
Consider the following addition and multiplication patterns. This is another

example of the hidden beauty of mathematics.

1 1 1
1+ ×3=4 =1+ +3
2 2 2

1 1 1
1+ ×4=5 =1+ +4
3 3 3

1 1 1
1+ ×5=6 =1+ +5
4 4 4

1 1 1
1+ ×6=7 =1+ +6
5 5 5
..
.

1 1 1
1+ × (n + 1) = (n + 2) = 1 + + (n + 1)
n n n
Consider the following multiplication and subtraction pattern:

11 1
1× ==1−
22 2
21 2
2× =1 =2−
33 3
31 3
3× = 2 =3−
44 4
41 4
4× =3 =4−
55 5
51 5
5× =4 =5−
66 6
..
.
n 1 n
n× = (n − 1) + =n−
n+1 n+1 n+1
Consider the following symmetry involving division and addition:

2 1 1 2
1 ÷=2=1 +
3 3 3 3
3 1 1 3
2 ÷=3=2 +
4 4 4 4
4 1 1 4
3 ÷=4=3 +
5 5 5 5
5 1 1 5
4 ÷=5=4 +
6 6 6 6
..
.

1 n+1 1 n+1
n+ ÷ =n+1= n+ +
n+2 n+2 n+2 n+2
Lastly, consider a symmetry between division and subtraction:

1 1 1
4 ÷3=1 =4 −3
2 2 2
1 1 1
5 ÷4=1 =5 −4
3 3 3
1 1 1
6 ÷5=1 =6 −5
4 4 4
1 1 1
7 ÷6=1 =7 −6
5 5 5
..
.

1 1 1
n+ ÷ (n − 1) = 1 + =n+ − (n − 1)
n−2 n−2 n−2
Squaring a Multiple of 5
Consider the challenge of squaring the number 85 mentally. Here is a

way that you can perform this calculation rather easily. All we need to
do, following this curious technique, is to multiply the tens digit, which
is 8, by 8 + 1 = 9, then multiply that by 100 and add 25, as shown
here: 852 = (8 × 9 × 100) + 25 = 7225. With a little practice, this
can go quite quickly and provide another useful arithmetic technique. To
see why this works, let’s consider a general case of a two-digit number
(10a + 5), which is a multiple of 5. When we square this number, we get
(10a + 5)2 = 100a 2 + 100a + 25 = a(a + 1)(100) + 25.
Squaring Numbers from 1 to 125
This technique requires us to partition the numbers from 1 to 125 into three
groups:
(1) The squares of the first 25 numbers 1–25 are most likely known by
many people without calculation.
The squares of the numbers 25–70 are symmetric about the number
50, as 25–50–75, and their squares can be calculated easily with the
following technique:
Add to (subtract from) 25 the distance of the number to 50 and append
on the right its square as a two-digit number (however, if this square
is a three-digit number, then carry its hundreds digit to the last digit of
the first operation).
This can be easily shown with the following examples:

572 : 25 + 7 = 32; 72 = 49 32 |49, which is the number 3,249
362 : 25 − 14 = 11; 142 = 196, then 11 + 1 | 96 or 12 |96, which is
the number 1,296
This always works because (50 ± a)2 = 2500 ± 100a + a 2 =
(25 ± a) · 100 + a 2 .
If one allows a to also be negative, then one does not need to distinguish
between 50 ± a, one can just write (50 + a)2 = 2500 + 100a + a 2 =
(25 + a) · 100 + a 2 .
(2) The squares of the numbers between 75 and 125 (which are symmetric
about 100 as we can see with 75–100–125) can be calculated easily
using the following technique:
Add to (subtract from) the number its distance to 100 and then append
on the right the squared distance as a two-digit number.
Consider the following examples:
1062 : 106 + 6 = 112; 62 = 36 112 | 36, which is the number 11,236
882 : 88 − 12 = 76; 122 = 144 then 76 + 1|44 or 77 | 44, which is
the number 7,744
The reason that this works can be shown as follows: (100 ± a)2 =
1002 ±2(100a)+a 2 = (100±2a)(100)+a 2 = [(100±a)±a](100)+a 2
If one allows a to also be negative, then one does not need to distin-
guish between 100 ± a.
The Unusual Number 9
The first occurrence in Western Europe of the Hindu–Arabic numerals we

use today was in the book Liber abaci written in 1202 by Leonardo of Pisa
(commonly known as Fibonacci).
He begins the book with an introduction that reads:
these are the nine figures of the Indians 9, 8, 7, 6, 5, 4, 3, 2, 1. With

these nine figures, and with the symbol, 0, which in Arabic is called
zephirum, any number can be written, as will be demonstrated . . .
With this book, the use of these numerals was first popularized in Europe.
Before that, Roman numerals were used, which were clearly much more
cumbersome to use for calculation. Fascinated as he was by the arithmetic
calculations used in the Islamic world, Fibonacci, in his book, first intro-
duced the system of “casting out nines”2 as a check for arithmetic. It comes
in handy even today. However, the nice thing about it is that it again demon-
strates a hidden magic in ordinary arithmetic.
Before we discuss this arithmetic-checking procedure, we will consider
how the remainder of a division by 9 compares to removing nines from
the digit sum of the number. When 8,768 is divided by 9, the quotient is
974 with a remainder of 2. This remainder can also be obtained by “casting
out nines” from the digit sum of the number 8,768: We will find the sum
of the digits and, if the sum is more than a single digit, we shall repeat
the procedure. In the case of our given number, 8,768, the digit sum is 29
(8 + 7 + 6 + 8 = 29) so we will repeat the process. Again, the casting-out-
nines procedure is used to get 2 + 9 = 11, and again 1 + 1 = 2, which was,
in fact, the remainder when we divided 8,768 by 9.
We can use casting out nines to check if a multiplication problem is cor-
rect. Suppose we would like to multiply 734 × 879. The answer we get is
645,186. Now casting out nines by taking digit sums, we get the following:
For 734: 7 + 3 + 4 = 14; then 1 + 4 = 5

For 879: 8 + 7 + 9 = 24; then 2 + 4 = 6
For the product of these two
numbers, 645,186: 6 + 4 + 5 + 1 + 8 + 6 = 30
Since 5 × 6 = 30, which yields 3 (casting out nines: 3 + 0 = 3), is
the same as for the sum of the digits of the product, the answer could be
correct. For practice, we will do another casting-out-nines “check” for the
following multiplication: 56,589 × 983,678 = 55,665,354,342.
For 56,589 5 + 6 + 5 + 8 + 9 = 33; 3+3=6

For 983,678 9 + 8 + 3 + 6 + 7 + 8 = 41; 4 + 1 = 5
2 “Casting out nines” refers to an arithmetic check that tells you if your answer is possibly
correct. The process requires taking bundles of nines away from the sum of the digits.
The product of the remainders: 5 × 6 = 30 → 3 + 0 = 3, compares

favorably to the digit-sum remainder as we see here:
55,665,354,342: 5 + 5 + 6 + 6 + 5 + 3 + 5 + 4 + 3 + 4 + 2 = 48;
4 + 8 = 12; 1 + 2 = 3
The same procedure can be used to check the likelihood of a correct

sum, difference, or quotient, simply by taking the sum (difference or quo-
tient) and casting out nines, taking the sum (difference or quotient) of these
“remainders,” and comparing it with the remainder of the sum (difference or
quotient). If these digit-sum remainders are not equal the answer is surely
wrong. If these digit-sum remainders are equal, the answer has a good
chance to be correct.
More Novelties with the Number 9
The number 9 has another unusual feature that enables us to use a surprising
multiplication algorithm. Although it is somewhat complicated, it is never-
theless fascinating to see how it functions and perhaps try to determine why
this actually works. This procedure is intended for multiplying a number
of two digits or more by 9. It is perhaps best to discuss the procedure in
the context of an actual example: Consider multiplying 76,354 × 9. Now
follow the steps shown in Figure 1.1.
Step 1 Subtract the units digit of the 10 – 4 = 6
multiplicand from 10
Step 2 Subtract each of the remaining 9 – 5 = 4, 4+ 4 = 8
digits (beginning with the tens 9 – 3 = 6, 6 + 5 = 11, 1
digit) from 9 and add this 9 – 6 = 3, 3 + 3 = 6, 6 + 1 = 7
result to the previous digit in 9 – 7 = 2, 2+6=8
the multiplicand (for any two
digit sums, carry the tens digit
to the next sum)
Step 3 Subtract 1 from the left-most 7–1=6
digit of the multiplicand
Step 4 List the results in reverse order 687,186
to get the desired product
Figure 1.1
Although it is a bit cumbersome, especially when compared to a calcu-

lator, this algorithm provides some insights into number theory.
Non-Repeating Eight-Digit Numbers and the Number 9
There are many eight-digit numbers with no digits repeated that when mul-
tiplied by 9 yield nine-digit numbers which have some digits repeated, such
as 56,387,412 × 9 = 507,486,708. However, there are some instances in
which a non-repeating-digit eight-digit number multiplied by 9 results in a
nine-digit number with no repeating digits such as
81274365 × 9 = 731469285
76125484 × 9 = 685129347
72645831 × 9 = 653812479
58132764 × 9 = 523194876
Curiously enough, when these four products of nine digits are multiplied
by 2, the result is once again a 10-digit number with no repeating digits as
we can see with the following:
731469285 × 2 = 1462938570
685129347 × 2 = 1370258694
653812479 × 2 = 1307624958
523194876 × 2 = 1046389752
Once again, the number 9 plays an important role in our arithmetic

wonders!
The Magic of the Number 9
The only peculiarity of the number 12,345,679 is that it has the digits in
sequence and is missing the number 8. If we multiply this number by prod-
ucts of 9 with any of the numbers 1, 2, 3, 4, 5, 6, 7, 8, and 9, we get surprising
results:
9 × 1 × 12,345,679 = 111,111,111
9 × 2 × 12,345,679 = 222,222,222
9 × 3 × 12,345,679 = 333,333,333
9 × 4 × 12,345,679 = 444,444,444
9 × 5 × 12,345,679 = 555,555,555
9 × 6 × 12,345,679 = 666,666,666
9 × 7 × 12,345,679 = 777,777,777
9 × 8 × 12,345,679 = 888,888,888
9 × 9 × 12,345,679 = 999,999,999
This unusual feature can be easily explained, since n × 9 =

n(10 − 1). Now multiplying the number n(10 − 1) by 12,345,679 gives us
n(123,456,790 − 12,345,679) = n(111,111,111).
A Secret about the Number 9
There is a curiosity when the number 15 is placed between the digits of the
number 42 = 16.
1156 = 342 .
When we place the number 15 in the middle of the number 1156, we
get 111556 = 3342 .
When we place the number 15 in the middle of the number 111556, we
get 11115556 = 33342 .
When we place the number 15 in the middle of the number 11115556,
we get 1111155556 = 333342 . This pattern continues. The problem is to
show why this continues to work further on.
Solution: We can understand this pattern by considering any one of the num-
bers in our pattern, say, 11115556, which when multiplied by 9 is equal to
100040004 = 10002 2 . This can be shown to hold true by analyzing the
product as follows: 11115556 × (10 − 1) = 111155560 − 11115556 =
100040004. At this point, we should note that numbers of the form
“1000 . . . 4000 . . . 4” are always square numbers. Remember this was gen-
erated by the number 9.
There is one other number, 48, that can analogously generate square
numbers. This time we begin with the number 49 and place the number
48 between the center digits so that the next number will be 4489 = 67 2 .
And then 444889 = 6672 , and 44448889 = 66672 , and so on. Once again,
it is the number 9 that generates this pattern as follows: 444889 × 9 =
4004001 − 20012 .
The Technique for Determining When a Number is

Divisible by 3 or 9
There are moments in everyday life situations where it can be useful to

know if a number can be divisible by 3 or by 9, especially if it can be done
instantly “in your head.” For example, if you are in a restaurant and receive
a check of $71.23, and you want to add a tip but you would want the end
result to be able to be split evenly among three people. Wouldn’t it be nice
if there were some mental arithmetic technique for determining this? Well,
here comes mathematics to the rescue. We are going to provide you with a
simple technique to determine if a number is divisible by 3 and (as an extra
bonus) also divisible by 9.
The rule, simply stated, is as follows:
If, and only if, the sum of the digits of a number is divisible by 3 (or 9),
the original number is divisible by 3 (or 9).
As before, perhaps an example of this technique would best firm up an
understanding of how it works. Consider the number 296,357. Let’s test it
for divisibility by 3 (or 9). The sum of the digits is 2+9+6+3+5+7 = 32,
which is not divisible by 3 or 9. Therefore, the original number is neither
divisible by 3 nor 9.
Let’s now assume a group of three people is given a restaurant check of
$71.23 and would like to give an approximate 20% tip on this check. They
decide to add $14 to the bill, which would make the total $85.23. To check
for divisibility by 3, since they would like to divide the check equally among
the three persons. The three participants would have them add the digits to
get 8 + 5 + 2 + 3 = 18, which is divisible by 3 and, therefore, would allow
them to equally divide the check among the three customers. If the number
reached from the sum of the digits is not easily identifiable as a multiple of
3, then continue to add the digits of that number until you reach a number
that you can easily recognize as a multiple of 3. In this case, it should be
noted that the final result, 18, is also divisible by 9, which implies that the
original number was divisible by 9 as well.
Here is a brief explanation about why this rule works. Consider the
number ab,cde, whose value can be expressed as (where 9M refers to a
multiple of 9)
N = 104 a + 103 b + 102 c + 10d + e

= (9 + 1)4 a + (9 + 1)3 b + (9 + 1)2 c + (9 + 1)d + e
= [9M + (1)4 ]a + [9M + (1)3 ]b + [9M + (1)2 ]c + [9 + (1)]d + e
= 9M[a + b + c + d] + a + b + c + d + e
which implies that divisibility by nine of the number N depends on the

divisibility of a + b + c + d + e, which is the sum of the digits.
A few examples to firm up this technique are as follows:
Is the number 457,875 divisible by 3 or 9? The sum of the digits is
4 + 5 + 7 + 8 + 7 + 5 = 36, which is divisible by 9 (and then, of course,
by 3 as well), so the number 457,875 is divisible by 3 and by 9.
Is the number 27,987 divisible by 3 or 9? The sum of the digits is 2 +
7 + 9 + 8 + 7 = 33, which is divisible by 3 but not by 9. Therefore, the
number 27,987 is divisible by 3 but not by 9.
Now that you are an expert at determining if a number is divisible by
3 or 9, we can go back to the original question about the divisibility of a
restaurant bill of $71.22 + 14.00 = $85.22. Can it be divided into three
equal parts? Because 8 + 5 + 2 + 2 = 17, and 17 is not divisible by three,
therefore $85.22 is not divisible by 3, and they will have to modify the tip
to get everyone to pay the same amount.
Multiplying by 11 Mentally
Multiplying by 11 can be significantly simpler than using a calculator, and

yet, this technique is hardly ever presented in school. What a shame! When
multiplying by 11, the technique for doing it is as fast as you can write the
number you are multiplying by 11. For example, suppose you would like
to multiply 23 × 11. All you need to do is add the digits, 2 + 3 = 5, and

place that digit 5 between the two digits in the number 23 to get 253. What
could be easier! We can state this technique as follows:
To multiply a two-digit number by 11, just add the two digits and place
this sum between the two digits. If the sum of the digits is a two-digit number,
then carry the tens digit of the sum to be added to the former tens digit.
Just for practice, we will apply this technique to multiply 45 by 11.
According to the rule, add 4 and 5 and place it between the 4 and 5 to get
495. This technique can get a bit more difficult when the sum of the two
digits you need to add results in a two-digit number. We no longer have a
single digit to place between the two original digits. So, if the sum of the
two digits is greater than 9, then we would place the units digit of that sum
between the two digits of the number being multiplied by 11 and “carry”
the tens digit to be added to the hundreds digit of the multiplicand. Let’s
try it with 78 × 11. We find that the sum of the two digits is 7 + 8 = 15.
Therefore, we place the 5 between the 7 and 8 and add the 1 to the 7 to get
[7 + 1][5][8] or 858.
You may legitimately ask if the rule also holds when 11 is multiplied by
a number of more than two digits. Consider a larger number such as 12,345
and use this technique to multiply it by 11.
Here, we begin at the right-side digit and add every pair of digits going
from right to left.
1[1 + 2][2 + 3][3 + 4][4 + 5]5 = 135,795.
Suppose we now combine the skill we have garnered through this tech-
nique of multiplying by 11 and apply it to a number, which requires a more
complicated version, where the sum of adjacent digits exceeds 9. Remember,
if the sum of two digits is greater than 9, then use the procedure described
earlier: Place the units digit of this two-digit sum appropriately and carry the
tens digit to the next place. To become an expert at this procedure, we will
do one of these more complicated versions here. Let us consider multiplying
56,789 by 11. This may be a little bit tedious, and perhaps somewhat less
realistic for common use, but we will show it here merely to demonstrate
the extension of this multiplication technique. Follow along as we do this
step by step.
5[5 + 6][6 + 7][7 + 8][8 + 9]9 Add each pair of digits between the
end digits
5[5 + 6][6 + 7][7 + 8][17]9 Add 8 + 9 = 17
5[5 + 6][6 + 7][7 + 8 + 1][7]9 Carry the 1 (from the 17) to the next
sum
5[5 + 6][6 + 7][16][7]9 Add 7 + 8 + 1 = 16
5[5 + 6][6 + 7 + 1][6][7]9 Carry the 1 (from the 16) to the next
sum
5[5 + 6][14][6][7]9 Add 6 + 7 + 1 = 14
5[5 + 6 + 1][4][6][7]9 Carry the 1 (from the 14) to the next
sum
5[12][4][6][7]9 Add 5 + 6 + 1 = 12
5 + 1[2][4][6][7]9 Add 5 + 1 = 6 to get the answer
624,679
This technique for multiplying by 11 could actually be helpful when
least anticipated.
A Technique to Determine When a Number Is Divisible by 11
One can also be fortified with another novelty of the number 11, which
owes its special status to being 1 greater than the base 10 of our number
system. Let’s look at the reverse of the previous arithmetic technique, that
is, division by 11 rather than multiplication by 11. At the oddest times, the
need can come up to determine if a number is divisible by 11. If you have a
calculator at hand, the problem is easily solved. But that is not always the
case. Besides, there is such a clever “rule” for testing for divisibility by 11
that it is worth knowing it even just for its charm.
The rule is quite simple:
If, and only if, the difference of the sums of the alternate digits is divisible
by 11, the original number is also divisible by 11.
This may sound more complicated than it actually is. Let us take this
rule a piece at a time. The sums of the alternate digits mean that you begin
at one end of the number taking the first, third, fifth, etc., digits and adding
them. Then add the remaining (even placed) digits. Subtract the two sums
and inspect for divisibility by 11. If the resulting number is divisible by 11,
then the original number was divisible by 11. And the reverse is also true.
That is, if the number reached by subtracting the two sums is not divisible
by 11, then the original number was also not divisible by 11.
This is probably best demonstrated by an example. Suppose we test
the number 768,614 for divisibility by 11. Sums of the alternate digits are
7 + 8 + 1 = 16 and 6 + 6 + 4 = 16. The difference of these two sums
16 − 16 = 0, which is divisible by 11. (Remember 11 0
= 0.) Therefore, we
can conclude that 768,614 is divisible by 11.
Another example might be helpful to firm up your understanding of this
procedure. To determine if 918,082 is divisible by 11, we need to find the
sums of the alternate digits:
9 + 8 + 8 = 25 and 1 + 0 + 2 = 3
Their difference is 25 − 3 = 22, which is divisible by 11, and so the

number 918,082 is divisible by 11. Practice will make this technique quite
useful.
Here is a brief explanation to justify this rule. Consider the number
ab,cde, whose value can be expressed as follows (here 11M represents a
multiply of 11):
N = 104 a + 103 b + 102 c + 10d + e

= (11 − 1)4 a + (11 − 1)3 b + (11 − 1)2 c + (11 − 1)d + e
= [11M + (−1)4 ]a + [11M + (−1)3 ]b +
[11M + (−1)2 ]c + [11 + (−1)]d + e
= 11M[a + b + c + d] + a − b + c − d + e,
which implies that since the first part of this value of N , namely, 11M[a +
b + c + d], is already a multiple of 11, the divisibility by 11 of the number N
depends on the divisibility of the remaining part, which is a−b+c−d +e =
(a + c + e) − (b + d) and which is actually the difference of the sums of
the alternate digits.
Divisibility by Prime Factors of 10 and Their Powers
Most people can determine if a number is divisible by 2 or by 5, simply by

looking at the last digit (i.e., the units digit) of the number. That is, if, and
only if, by looking the last digit is an even number (such as 2, 4, 6, 8, and
0), then the number will be divisible by 2. Incidentally, you can extend this
simple well-known technique to one where you can determine divisibility
by higher powers of 2, such as the divisibility by 4, 8, 16, etc. Once again,
we look at the end of the number being inspected. If, and only if, the number
formed by the last two digits is divisible by 4, then the entire number itself
is divisible by 4. Also, if, and only if, the number formed by the last three
digits is divisible by 8, then the number itself is divisible by 8. And so,
to test for divisibility by 16, we would focus on the last four digits of the
number, and so on for other higher powers of 2.
In an analogous fashion, we can develop a technique for determining
divisibility by powers of 5. It is common knowledge that if the last digit of
the number being inspected for divisibility is either 0 or 5, then the number
itself will be divisible by 5. A number is divisible by 25 if, and only if, the
last two digits of the number, taken together as a number, is divisible by
25. This is analogous to the technique we used to determine divisibility by
powers of 2. Have you guessed what the relationship here is between powers
of 2 and 5? Yes, they are both factors of 10, the base of our decimal number
system.
Divisibility by Other Prime Numbers
Divisibility by 7
With the proliferation of the calculator, there is no longer a crying need
to be able to detect by which numbers a given number is divisible. Yet,
for a better appreciation of mathematics, divisibility rules provide an inter-
esting “window” into the nature of numbers and their properties. For this
reason (among others), the topic of divisibility still finds a place on the
mathematics-learning spectrum and can provide some further problem-
solving strength. This is especially true for the technique for divisibility
by 7. (By the way, the technique to determine divisibility by 6 is simply to
apply the technique for divisibility by 2 and by 3 — both must hold true for
a number to be divisible by 6.)
We begin by considering the method to determine if a given number is
divisible by 7 and then, as we inspect this technique, we can see how it can
be generalized for other prime numbers.
The technique for divisibility by 7 is as follows:
Delete the last digit from the given number and then subtract twice this
deleted digit from the remaining number. If, and only if, the result is divisible
by 7, the original number is divisible by 7. This process may be repeated
until the result can be determined by simple inspection for divisibility by 7.
For example, suppose we want to test the number 876,547 for divisibility
by 7. Begin with 876,547 and delete its units digit, 7, and subtract its double,
14, from the remaining number: 87,654 − 14 = 87,640. Since we cannot
yet visually inspect the resulting number for divisibility by 7, we continue
the process with the resulting number 87,640 and delete its units digit, 0, and
subtract its double, still 0, from the remaining number; we get 8,764 − 0 =
8,764. This did not bring us any closer to visually being able to check
for divisibility by 7; therefore, we continue the process with the resulting
number 8,764 and delete its units digit, 4, and subtract its double, 8, from the
remaining number; we get 876 − 8 = 868. Since we still cannot visually
inspect the resulting number, 868, for divisibility by 7, we continue the
process with the resulting number 868 and delete its units digit, 8, and
subtract its double, 16, from the remaining number we get 86 − 16 = 70,
which we can easily see is divisible by 7. Therefore, the original number
876,547 is also divisible by 7. Before continuing with our exploration of
divisibility by prime numbers, it would be wise to practice this technique
with a few randomly selected numbers and then check your results with a
calculator.
Why does this rather strange procedure work? Being able to answer
this question is the wonderful thing about mathematics. It doesn’t do things
that, for the most part, we cannot justify. This will all make sense after
you see what is happening with this procedure. To justify the technique of
determining divisibility by 7, consider the various possible terminal digits
(that you are “dropping”) and the corresponding subtraction that is actually
being done by dropping the last digit. In the chart (Figure 1.2), you will
see that by dropping the terminal digit and doubling it, the number being
subtracted gives us in each case a multiple of 7. That is, we have taken

“bundles of 7” away from the original number. Therefore, if, and only if,
the the remaining number is divisible by 7, then so is the original number
divisible by 7 because you have separated the original number into two
parts, each of which is divisible by 7, and, therefore, this must hold true for
the entire number.
Terminal Number subtracted from

digit original
1 20 + 1 = 21 = 3 × 7
2 40 + 2 = 42 = 6 × 7
3 60 + 3 = 63 = 9 × 7
4 80 + 4 = 84 = 12 × 7
5 100 + 5 = 105 = 15 × 7
6 120 + 6 = 126 = 18 × 7
7 140 + 7 = 147 = 21 × 7
8 160 + 8 = 168 = 24 × 7
9 180 + 9 = 189 = 27 × 7
Figure 1.2
Divisibility by 13
The technique for divisibility by 13 is very similar to that for divisibility
by 7 except that the 7 is replaced by 13, and instead of subtracting twice
the deleted digit, we subtract nine times the deleted digit each time. It thus
reads as follows:
Delete the last digit from the given number and then subtract nine times
this deleted digit from the remaining number. If, and only if, the result is
divisible by 13, the original number is divisible by 13.
Let’s try our technique to check for divisibility by 13 for the number
5,616. We begin by deleting its units digit, 6, and this time instead of dou-
bling the number to subtract, we subtract nine times the number, in this
case, 9 × 6 = 54, from the remaining number 561 − 54 = 507. Since we
still cannot visually inspect the resulting number for divisibility by 13, we
continue the process with the resulting number 507 and delete its units digit
and subtract 9 times this digit (9 × 7 = 63) from the remaining number
50−63 = −13, which is divisible by 13, and, therefore, the original number
is also divisible by 13.
Now we might want to see how we determined the “multiplier” 9 in
our technique. We sought the smallest multiple of 13 that ends in 1. That
was 91, where the tens digit is 9 times the units digit. Once again, consider
the various possible terminal digits and the corresponding subtractions in
Figure 1.3.
Terminal digit Number subtracted

from original
1 90 + 1 = 91 = 7 × 13
2 180 + 2 = 182 = 14 × 13
3 270 + 3 = 273 = 21 × 13
4 360 + 4 = 364 = 28 × 13
5 450 + 5 = 455 = 35 × 13
6 540 + 6 = 546 = 42 × 13
7 630 + 7 = 637 = 49 × 13
8 720 + 8 = 728 = 56 × 13
9 810 + 9 = 819 = 63 × 13
Figure 1.3
In each case, a multiple of 13 is being subtracted one or more times

from the original number. Hence, if, and only if, the remaining number is
divisible by 13, then the original number is divisible by 13.
Divisibility by 17
The technique for divisibility by 17 is as follows:
Delete the units digit and subtract five times the deleted digit each time
from the remaining number until you reach a number small enough to visu-
ally determine its divisibility by 17.
We justify the technique for divisibility by 17 as we did the techniques
for 7 and 13. Each step of the process requires us to subtract a “bunch of
17s” from the original number until we reduce the number to a manageable
size by which we can make a visual inspection of divisibility by 17. This
time we can see that the multiplier is 5, since we will be deducting multiples
of 17, such as 51, 102, 153, and so on, from the original number.
Divisibility by larger prime numbers

The patterns developed in the preceding three divisibility techniques (for 7,
13, and 17) should enable you to produce analogous techniques for testing
divisibility by larger primes. Figure 1.4 presents the “multipliers” of the
deleted digits for various primes.
To test
7 11 13 17 19 23 29 31 37 41 43 47
divisibility by
Multiplier 2 1 9 5 17 16 26 3 11 4 30 14
Figure 1.4
You may want to extend this chart as it will increase your percep-
tion of mathematics, while at the same time extending your toolkit of
problem-solving techniques. You may also want to extend your knowledge
of divisibility rules to include composite (i.e., non-prime) numbers. Why
the following rule refers to relatively prime factors and not just any factors
is something that will sharpen your understanding of number properties.
Perhaps the easiest response to this question is that relatively prime factors
have independent divisibility rules, whereas other factors may not.
The following technique is for divisibility by composite numbers:
A given number is divisible by a composite number, if, and only if, it is
divisible by each of its relatively prime factors. (Two or more numbers are
relatively prime if they have no common factors other than 1.)
Figure 1.5 offers illustrations of this rule.
To be divisible by 6 10 12 15 18 21 24 26 28
The number must
2, 3 2, 5 3, 4 3, 5 2, 9 3, 7 3, 8 2, 13 4, 7
be divisible by
Figure 1.5
To this point, you have had an opportunity to broaden your arithmetic

problem-solving skills that give you a deeper insight into mathematics as
well as an appreciation for its wonders. There are times when a calculator
is not at hand, and we need to do mental arithmetic. Thus, we continue on
with arithmetic problem-solving skills.
Amazing Sum of Squares
Consider any eight consecutive numbers, such as 6, 7, 8, 9, 10, 11, 12, and
13. Quite unexpectedly, the squares of these eight numbers can be organized
into equal parts such as 62 + 92 + 112 + 122 = 382 = 72 + 82 + 102 + 132 .
Analogously, we can show that for the eight consecutive numbers 7, 8, 9, 10,
11, 12, 13, and 14 we have the equal sums as follows: 7 2 +102 +122 +132 =
462 = 82 + 92 + 112 + 142 . Why does this happen?
Solution: We can best explain this amazing phenomenon using simple alge-
bra. Based on the pattern above, where we let n represent the first number,
(n+1) the second number, . . . (n+7) the eighth number, we have established
the following pattern:
n 2 + (n + 3)2 + (n + 5)2 + (n + 6)2 = 4n 2 + 28n + 70
and the remaining numbers used:
(n + 1)2 + (n + 2)2 + (n + 4)2 + (n + 7)2 = 4n 2 + 28n + 70.
This proves that for any eight consecutive numbers, this pattern will hold
true.
A Squaring Curiosity
Consider the amazing symmetry by squaring the sum of two numbers where
the result is a number formed by the cumulation of the digits. For example,
(5288 + 1984)2 = 52,881,984. Are there other numbers that share this
curious property?
Solution: The answer to the question is clearly yes. One such example would
be (6048 + 1729)2 = 60,481,729. What we seek are numbers where the
following would be true: (a + b)2 = ab, or (ab + cd)2 = abcd, where ab
is a two-digit number and abcd is a four-digit number. In the first case, we
can have (0 + 1)2 = 01 or (8 + 1)2 = 81. For pairs of two-digit numbers,
some examples are as follows:
(20 + 25)2 = 2,025

(30 + 25)2 = 3,025
(98 + 01)2 = 9,801
This last number when divided by 9 yields the reversal 1,089.

In the meantime, we can also consider pairs of three-digit numbers that

share this unusual characteristic.
(088 + 209)2 = 088,209

(494 + 209)2 = 494,209
(998 + 001)2 = 998,001
There are even four-digit numbers that exhibit this amazing symmetry
such as
(9998 + 0001)2 = 99,980,001
The Russian Peasant’s Method of Multiplication
It is said that the Russian peasants used a rather strange, perhaps even
primitive, method to multiply two numbers. It is actually quite simple, yet
somewhat cumbersome. Consider the problem of finding the product of
43 × 92. We begin by setting up a chart of two columns with the two
members of the product in the first row, as shown in Figure 1.6. One column
will be formed by doubling each number to get the next, while the other
column will take half the number and drop the remainder. We select the
left-side column to be the doubling column, and the right-side column to be
the halving column. Note that by halving the odd number such as 23 (the
third number in the second column), we get 11 with a remainder of 1 and
we simply drop the 1. The rest of this halving process should be clear.
43 92
86 46
172 23
344 11
688 5
1376 2
2752 1
Figure 1.6
Find the odd numbers in the halving column (here the right column) and
then get the sum of the partner numbers in the doubling column (in this case
the left column). These are highlighted in bold type. This sum gives you the
originally required product of 43 and 92. In other words, with the Russian
peasant’s method, we get 43 × 92 = 172 + 344 + 688 + 2752 = 3956.
In the example above, we chose to have the left-side column, the doubling
column, and the right-side column the halving column. We could also have
done this Russian peasant’s method by halving the numbers in the first
column and doubling those in the second. See Figure 1.7.
43 92
21 184
10 368
5 736
2 1472
1 2944
Figure 1.7
To complete the multiplication, we find the odd numbers in the halving

column (in bold type) and then get the sum of their partner numbers in
the second column (now the doubling column). This gives us 43 × 92 =
92 + 184 + 736 + 2944 = 3956.
You are not expected to do your multiplication in this high-tech era
by employing the Russian peasant’s method. However, it should be fun to
observe how this primitive system of arithmetic actually does work. Explo-
rations of this kind are not only instructive but should also be entertaining.
Here you see what was done in the above multiplication algorithm:
∗
43 × 92 = (21 × 2 + 1)(92) = 21 × 184 + 92 = 3956
∗
21 × 184 = (10 × 2 + 1)(184) = 10 × 368 + 184 = 3864
10 × 368 = (5 × 2 + 0)(368) = 5 × 736 + 0 = 3680
∗
5 × 736 = (2 × 2 + 1)(736) = 2 × 1472 + 736 = 3680
2 × 1472 = (1 × 2 + 0)(1472) = 1 × 2944 + 0 = 2944
∗
1 × 2944 = (0 × 2 + 1)(2944) = 0 + 2944 = 2944
3956
Here in the next-to-last column, the numbers are added to get 3,956.
For those familiar with the binary system (i.e., base 2), one can also
explain this Russian peasant’s method with the following representation:
(43)(92) = [(1 × 25 ) + (0 × 24 ) + (1 × 23 ) + (0 × 22 )
+ (1 × 21 ) + (1 × 20 )](92)
= (20 × 92) + (21 × 92) + (23 × 92) + (25 × 92)
= 92 + 184 + 736 + 2944
= 3956
Whether or not you have a full understanding of the discussion of the

Russian peasant’s method of multiplication, you should at least, now, have a
deeper appreciation for the multiplication algorithm you learned in school,
even though most people today multiply with a calculator. There are many
other multiplication algorithms, yet the one shown here is perhaps one of
the strangest and it is through this strangeness that we can appreciate the
powerful consistency of mathematics that allows us to conjure up such an
algorithm.
Another Unusual Method to Multiply Two Numbers
Let’s delve right into this unusual method for multiplying a pair of two-digit
numbers with some examples, and then see why this technique works. Take,
for example, the multiplication 95 × 97. The following steps can be done
mentally (with some practice, naturally!):
Step 1: 95 + 97 = 192
Step 2: Delete the hundreds digit = 92
Step 3: Tag two zeros onto the number = 9,200
Step 4: (100 – 95) · (100 – 97) = 5 × 3 = 15
Step 5: Add the last two numbers = 9215, which is the
product being sought!
Here is another multiplication calculation using this technique to mul-
tiply a pair of two-digit numbers: 93 × 96 =?
93 + 96 = 189
189 (Delete the hundreds digit.)
Tag on two zeros = 8,900
Then add (100 − 93) × (100 − 96) = 7 × 4 = 28 to the previously
obtained number to get 8,928, which is what we sought 93 × 96.
This technique also works when seeking the product of a pair of two-digit
numbers that are further apart, such as 89 × 73 =?
89 + 73 = 189
162 (Delete the hundreds digit.)
Tag on two zeros = 6,200
Then add (100 − 89) × (100 − 73) = 11 × 27 = 297, since this is
a three-digit result, we add the hundreds digit, 2, to 62 and then follow
the previous procedure to get 6,497, which again provides the product of
89 × 73.
For those who might be curious to know why this unusual technique
works, we can use simple algebra to justify it. We begin with a pair of
two-digit numbers:
(100 − a) and (100 − b) (where 0 < a and b < 100).
Step 1: (100 − a) + (100 − b) = 200 − a − b

Step 2: Delete the hundreds digit — which means subtracting 100
from the number: (200 − a − b) − 100 = 100 − a − b
Step 3: Tag on two zeros, which means multiply by 100:
(100 − a − b) × 100 = 10,000 − 100a − 100b
Step 4: a×b
Step 5: Add the last two results from Steps 3 and 4, and then do
some factoring:
10,000 − 100a − 100b + ab

= 100(100 − a) − (100b − ab)
= 100(100 − a) − b(100 − a)
= (100 − a)(100 − b), which is what we set out to show.
Now you just need to practice this technique to master it!

A Quick Multiplication Technique
There are people who are very adept at rapid multiplication. Here is one
of the procedures that is used to do this quick multiplication of a pair of
two-digit numbers. It is done as follows:
• Begin by multiplying the units digits of the two numbers. If a two-digit
number results, then write the units digit and carry the tens digit to the
next step.
• Next, multiply the units digit of one number by the tens digit of the other
number, and then multiply the tens digit of the first number by units digit
of the second number and add the two products. Afterwards, add any
number carried over from the previous step. Place the units digit of the
result to the left of the previously obtained single digit and carry the
tens digit to be added to the next step’s calculation.
• Lastly, multiply the two tens-digits and add the carried-over number
from the previous step. Place this result to the left of the two previously
obtained numbers, and you will have your final result of the multipli-
cation.
At first sight, this appears to be a rather complicated procedure, but after
we do an example, the procedure will become rather simple and clear and
relatively fast to do. As our example, let us use this procedure to multiply
59 × 38.
• First, we will multiply the units digits of the two numbers to be multi-
plied: 9 × 8 = 72. This will have given us the units digit of our ultimate
answer, namely, the 2, and we will carry the 7 to the next step.
• We will now multiply the units and tens digits of the two numbers:
5×8 = 40 and 9×3 = 27. We now add these two numbers 40+27 = 67
and then add the 7 carried from the previous step to get 67+7 = 74. We
place the 4 to the left of the previously obtained 2 and carry the 7 to the
next step. So far, we have the tens and units digits of our sought-after
product, namely, 42.
• We now multiply the two tens digits to get 3 × 5 = 15 and add the
carried 7 to get 15 + 7 = 22, which we now place to the left of the
two previously obtained final digits to get 2,242, which is the product
of 59 × 38.
A quick review of this procedure is simply to multiply the pair of units

digits, and the cross product of tens and units digits, and then the pair of tens
digits — each time carrying the tens digit as appropriate. To be a successful
user of this technique, you will need to practice this procedure with a number
of examples so that your calculating speed will be impressive. An ambitious
reader may want to extend this multiplication technique to multiplying two-
digit numbers by three-digit numbers, and three-digit numbers by other
three-digit numbers.
A Multiplication Technique for Special Numbers
Here is a technique for multiplying by two-digit numbers whose units digit

is 1, such as 21, 31, and 41. Follow along step by step and eventually the
technique will become easier.
To multiply by 21: Double the number, then multiply by 10, and add the
original number.
For example: To multiply 37 × 21, double 37 yields 74, multiply by 10
to get 740, and then add the original number 37 to get 777.
Here’s another example:
To multiply by 31: Triple the number, then multiply by 10, and add the
original number.
For example: To multiply 43 × 31, triple 43 yields 129, multiply by 10
to get 1,290, and then add the original number 43 to get 1,333.
For some more practice, we offer the following:
To multiply by 41: Quadruple the number, then multiply by 10, and add
the original number.
For example: To multiply 47 × 41, quadruple 47 yields 188, multiply
by 10 to get 1,880, and then add the original number 47 to get 1,927.
By now you should be able to recognize the pattern and apply the rule
to other such two-digit numbers.
This technique also works for multiplying by two-digit numbers whose
units digit is a 9, such as 19, 29, and 39.
To multiply by 19: Double the number, then multiply by 10, and subtract
the original number.
For example: To multiply 37 × 19, double 37 yields 74, multiply by 10

to get 740, and then subtract the original number 37 to get 703.
Although experiencing the solution to arithmetic problems can be useful
when a computer or calculator is not at hand, considering the previously
presented techniques provides a keen insight into mathematics, which could
open the door for further arithmetic investigations.
Shopping with Mathematical Support
Most supermarkets today provide the unit cost of an item. This is very
helpful as it allows the consumer to decide whether it makes sense to buy
two 12 oz. jars of mayonnaise costing $1.35 per jar or one 30 oz. jar of the
same brand of mayonnaise costing $3.49. We have been trained to think
that the larger quantity is generally the better price value. However, there is
a neat little trick to determining which is the better price per ounce when it
isn’t provided by the market. First, we need to establish the price per ounce
for each of the two jars:
For the 12-ounce jars, the price per ounce is $1.35 12
For the 30-ounce jar, the price per ounce is $3.49 30
To compare the two fractions 1.35 12
, 3.49
30
in order to see which is larger,
there is a neat little algorithm to accomplish this task. We will cross multiply,
writing the products under the fraction whose numerator was used (see
Figure 1.8).
Figure 1.8
The larger product, in this case 41.88, determines that the fraction 3.49
30
is a larger fraction, and therefore, the more expensive cost of mayonnaise,
in this case the larger jar was more expensive per ounce than the smaller jar.
Although this is not typically expected, it does occur, and for that reason, a
good consumer will make these comparisons.
Successive Percentages
We encounter challenges in our everyday lives and don’t even realize that
they can be properly understood with just a little mathematics. We often
visit stores that are running a sale and then on a special day will add a
percentage on top of the one having been advertised previously. The typi-
cal response is to add to percentages and conclude that the total savings
for the day would be the sum of the two percentages. This is clearly
a wrong calculation. Most folks defer thinking about percentage prob-
lems as they see them as nothing but a nemesis. Problems get particularly
unpleasant when multiple percentages need to be processed in the same
problem. However, we shall see how such successive percentages lend
themselves very nicely to a delightfully simple arithmetic algorithm that
leads us to lots of useful applications and provides new insight into succes-
sive percentage problems. This not-very-well-known procedure should be
enriching.
Let’s begin by considering the following problem:
Wanting to buy a coat, Barbara is faced with a dilemma. Two com-

peting stores next to each other carry the same brand coat with the
same list price but with two different discount offers. Store A offers
a 10% discount year-round on all its goods but on this particular
day, offers an additional 20% on top of its already discounted price.
Store B simply offers a discount of 30% on that day in order to stay
competitive.
Are the two end prices the same? If not, which gives Barbara the
better price?
At first glance, you may assume there is no difference in price, since

10 + 20 = 30, which would appear to be yielding the same discount in both
cases. Yet with a little more thought you may realize that this is not correct,
since in store A only 10% is calculated on the original list price, while the
20% discount is calculated on the lower price (namely, the 10% discounted
price), while at store B, the entire 30% is calculated on the original price.
Now, the question to be answered is as follows: What percentage difference
is there between the discount in store A and store B?
To determine the difference in the prices, one procedure might be to

assume the cost of the coat to be $100 and then calculate the 10% discount
yielding a $90 price, and an additional 20% of the $90 price (or $18) will
bring the price down to $72. In store B, the 30% discount on $100 would
bring the price down to $70, giving a discount difference of $2 between the
two stores, which in this case would be a 2% difference. This procedure,
although correct and not too difficult, is a bit cumbersome and does not
always allow a full insight into the situation as you will soon see.
We shall provide an interesting and quite unusual procedure for a deeper
look at this situation, as well as for entertainment. We will consider a some-
what mechanical method for obtaining a single percentage discount (or
increase) equivalent to two (or more) successive discounts (or increases).
Follow this four-step procedure:
(1) Change each of the percents involved into decimal form:

.20 and .10
(2) Subtract each of these decimals from 1.00:
.80 and .90 (for a percent increase, add to 1.00)
(3) Multiply these decimals:
.80 × .90 = .72
(4) Subtract this number from 1.00:
1.00 − 0.72 = 0.28, which, written as a percentage is 28%, represents
the combined discount.
If the result of Step 3 is greater than 1.00, subtract 1.00 from it to obtain
the percent of increase.
Therefore, we can conclude that the combined percentage of 28% differs

from the single discount of 30% by 2%.
Following the same procedure, you can also combine more than two
successive discounts. In addition, successive increases, combined or not
combined with a discount, can also be accommodated in this procedure by
adding the decimal equivalent of the increase to 1.00, while the discount was
subtracted from 1.00 and then continuing the procedure in the same way.
If the end result comes out greater than 1.00, then this will have resulted in
an overall increase rather than the discount as found in the above problem.
A conundrum often facing consumers is that of determining if a discount

and increase of the same percentage leave the original price unchanged. For
example, suppose a store just increased all of its prices by 10% and then
notes that its business has declined substantially, whereupon they then resort
to discounting all of these recently increased prices by the same percentage
of 10%. Have they then restored the prices to their original level? Using this
technique, we find ourselves multiplying 1.1 times 0.90 to get 0.99, which
would indicate that the original price had dropped by 1%. For many people,
this is a counterintuitive result.
This procedure not only streamlines a typically cumbersome situation
but also provides some insight into the overall picture. For example, con-
sider the following question: Is it advantageous to the buyer in the above
problem to receive a 20% discount and then a 10% discount, or the reverse,
10% discount and then a 20% discount? The answer to this question is
not immediately intuitively obvious. Yet, since the procedure just presented
shows that the calculation is merely multiplication, a commutative opera-
tion, we can immediately conclude that there is no difference between the
two.
So here you have a delightful algorithm for combining successive dis-
counts or increases or combinations of these to calculate the combined
result. Not only is it useful, but also it gives you some newfound power in
dealing with percentages when a calculator might not be available.
Another shopping situation where mathematics can be helpful is when
there are discounts of different types. Suppose you have two sales-promotion
coupons for the same store: One that says “20% off” (independent of the
purchase) and one that only applies above a specified amount, for example,
“$15 off for purchases exceeding $49.99.” Assuming that the two coupons
cannot be combined, then which one would be more advantageous, if the
item we want to buy costs, say, $80? The 20% off coupon would yield a
price of $64, while the $15 reduction would yield a price of $65. It might
be nice to know at what price the 20% coupon will be more advantageous.
To approach this problem, we may consider two extreme cases. Let’s
consider an item costing $50, since the minimal purchase that qualifies for
the second coupon is $50, which when reduced by $15 would be $35. On the
other hand, the fixed-percentage coupon would yield a reduction of 20%,
or one-fifth, of $50, which is $10. That means we would have to pay $40, if
we used the fixed-percentage coupon. Thus, the $15 coupon would be the
better choice for a $50 purchase.
The other extreme case could be infinity, but since we can only spend
a finite amount of money, let’s assume our budget is very high, say it is
$150. One-fifth of $150 is $30. Therefore, the 20% coupon yields a reduced
price of $120, while the other coupon (a reduction of $15) would result
in a price of $135. Obviously, the better choice of coupon depends on the
total sum of the purchase. For the fixed-percentage coupon, the amount of
money saved increases with the price, while we cannot save more than $15
with the other coupon. The extreme cases we have considered show that
there must be some break-even point X between $50 and $150, at which
price, both coupons yield the same discount. For purchases at $50, where
the 20% reduction will be $10 so that the $15 is preferred, yet, for purchases
above $50 and below X,the $15 coupon is still the better choice. To find the
break-even price X for a 20% coupon and a $15 coupon, we just have to
compute X − .20X = X − $15 and then X = $75. Therefore, if we want
to buy an item for more than $75, we should use the 20% coupon.
Occasionally, you may also encounter different-type coupons that are
combinable, although this is a rare phenomenon since most stores are
usually not that generous to their customers. Let’s take a look at such a
situation, since it provides an example for the mathematical notion of non-
commutativity. Suppose we were allowed to use both coupons for the same
purchase, that is, coupon 1 with a 20% discount as well as coupon 2 with a
$15 reduction. Now the question arises whether the order matters or not. If
it were to matter, which of the two coupons should be used first? Denoting
the price without any discount by P (which we assume to be at least $50
for the sake of simplicity), we obtain the following:
• A reduced price by using coupon 1 first as p1,2 = P · 0.8 − $15 and;

• A reduced price by using coupon 2 first as p2,1 = (P − $15) · 0.8.
Since p2,1 = (P − $15) · 0.8 = P · 0.8 − $12, which is more than p1,2 .
Therefore, we should apply coupon 1 first, unless P · 0.8 is less than $50.
In this case, we would have to use coupon 2 first.
Two operations that will in general lead to different results, if their

order is reversed, are called “non-commutative” in mathematics. As our
analysis showed, these different types of discounts are an example of non-
commutative operations, meaning that the order does matter! You should
think about that, if you are offered combinable discounts of different types.
However, requesting your preferred order might be a bit of a challenge.
Raising Interest
We are often confronted with advertisements by savings institutions offering

attractive interest rates and frequent compounding of interest on deposits.
Since most banks have a variety of programs, it is valuable for potential
depositors to understand how interest is calculated under each of the avail-
able options. In our discussion of interest rates and practices, we will use
the formula for compound interest to calculate the return on investments at
any rate of interest, for any period of time, and for any commonly used fre-
quency of compounding, including instantaneous (continuous) compound-
ing. They will also determine which of the two or more alternatives gives
the best return over the same time period.
Let’s consider the following interesting problem:
In the year 1626, Peter Minuit bought Manhattan Island for the
Dutch West India Company from the Lenape Native Americans
for trinkets costing 60 Dutch guilders or about $24. Suppose the
Lenapes had been able to invest this $24 at that time at an annual
interest rate of 6%, and suppose further that this same interest rate
had continued in effect all these years. How much money could the
present-day descendants of these original Lenapes collect if (1) only
simple interest was calculated and (2) interest was compounded (a)
annually, (b) quarterly, and (c) continuously?
The answers to these questions could be surprising.
You might recall that simple annual interest is calculated by taking the
product of the principal P, the annual interest rate r, and the time in years
t. Accordingly, you have the formula I = Prt, and in the above problem,
I = (24)(.06)(397) = $571.68 as simple interest. Add this to the principal
of $24.00 to obtain the amount A of $595.68 available in 2023. You have

just used the formula for “amount,” A = P + Prt.
With this relatively small sum in mind (for a return after 397 years!),
let’s investigate the extent to which this return would have been improved
if interest had been compounded annually instead of being calculated on
only a simple basis. With a principal P, an annual rate of interest r, and a
time t = 1, the amount A at the end of the first year is given by the formula
A1 = P + Pr = P(1 + r). (The subscript indicates the year at the end of
which interest is calculated.) Now A1 = P(1 + r) becomes the principal
at the beginning of the second year, upon which interest will be credited
during the second year.
Therefore, A2 = P(1 +r) + P(1 +r)r = P(1 +r)(1 +r) = P(1 +r)2 .
Since the last expression represents the principal at the beginning of the third
year, you have
A3 = P(1 + r)2 + P(1 + r)2 r = P(1 + r)2 (1 + r) = P(1 + r)3 .
By now, you will see the emerging pattern and should be able to suggest
the generalization for the amount after t years, At = P(1 + r)t .
Now try this formula on the $24 investment made in 1626. Assuming
annual compounding at 6% per annum, you have A397 = 24(1 + .06)397 =
267,079,006,366. This means that the original $24 is now worth more than
$200 billion! One should be surprised by the huge difference between this
figure and the $595.68 obtained by computing simple interest.
Most banks now compound not annually but quarterly, monthly, daily,
or continuously, so we shall next generalize the formula A = P(1 + r) t
to take into account compounding at more frequent intervals. Bear in mind
that if interest is compounded semi-annually, the periodic rate would be
only one-half the annual rate, but the number of periods would be twice the
2t
number of years so that A = P 1 + r2 . Likewise, if the interest is com-
4t
pounded quarterly, we have A = P 1 + r4 . In general, if the interest is
nt
compounded n times a year, the formula would be A = P 1 + nr , which
can be used for any finite value of n. Letting n = 4 in our original problem
4(397)
yields A = 24 1 + .06 4
= 24 (1.015)1588 = 444,924,512,737. Our
$24 has now risen to about $445 billion. Note that changing the compound-
ing from annually to quarterly increased the yield by about $178 billion.
You may now wonder whether the yield can be increased indefinitely by
simply increasing the frequency of compounding. A complete treatment of

this question requires a thorough development of the concept of limits, but
an informal, intuitive approach will suffice here. We shall first explore the
simpler problem of an investment of $1 at a nominal annual interest rate
n
of 100% for a period of one year. This will lead us to A = 1 1 + 100 n
.
You should now prepare a table of values for A for various common val-
ues of n, such as n = 1 (annual compounding), n = 2 (semiannual),
n = 4 (quarterly), and n = 12 (monthly). Note that the amount A does
not rise astronomically as n increases but rather rises slowly from $2.00
(n = 1) to about $2.60 (n = 12). The amount A would approach, but
not quite reach, the value $2.72. To explain this limiting value takes us a
bit beyond the realm of this book. Yet for the more advanced reader, it is
n
lim 1 + n1 = e = 2.71828.... This number e is called “Euler number”.
n→∞
Since investments generally do not earn 100% interest, we must convert
to a general interest rate, say r. By letting nr = k1 , we have n = kr, and A =
nt krt k rt
P 1 + nr , which then becomes A = P 1 + 1k = P 1 + 1k .
Clearly, as n approaches infinity so does k, since r is finite, so the expression
in brackets approaches the value e as a limit. You then have the formula
A = Pert for instantaneous compounding, where r is the nominal annual
rate of interest and t is the time in years. You might be interested in knowing
that this formula is a special representation of the general “Law of Growth,”
which is usually written in the N = N0 ert form, where N represents the final
amount of a material whose initial amount was N0 . This law has applications
in many other areas such as population growth (of people, bacteria in a
culture, etc.) and the radioactive decay of elements, in which case it becomes
the “law of decay,” as N = N0 e−rt .
Completing the investment problem, using 2.72 as an approximation to
e, you have A = 24(2.72).06(397) = 539,066,738,490. So, you can see that
the “ultimate” return on a $24 investment (at a nominal annual interest rate
of 6% for 397 years) is over $500 billion.
Banks currently offer much lower interest rates and compounding is
commonly done quarterly, monthly, daily, or continuously. You can work
problems with varying principals, periodic rates, frequencies of compound-
ing, and time periods and compare yields. You may be surprised at the
outcomes.
The Rule of 72
We often want to know how a certain interest rate in a bank will affect our
total holdings. Naturally, there are traditional ways of calculating interest,
which we have just experienced in the previous unit. However, there is an
unusual quirk of our number system that allows us to calculate how long
it will take to double our money in a bank with a yearly compounding
procedure at any given annual percentage rate. The procedure is as fast as
you can divide 72 by another number. This is clearly good to know, but it is
the unusualness of this rule that allows us to exhibit it here. So, enjoy it. It
is called the “Rule of 72” as it is based on this number as you will soon see.
The “Rule of 72” states that, roughly speaking:
Money will double in 72 r
years when it is invested at an annual com-
pounded interest rate of r%.
So, for example, if we invest money at an 8% compounded annual inter-
est rate, it will double its value in 72
8
= 9 years. Similarly, if we leave our
money in the bank at a compounded rate of 2%, it would take 36 years
for this sum to double its value. The interested reader might want to bet-
ter understand why this is so, and how accurate it really is. The following
discussion will explain that.
To investigate why or if this really works, we consider the compound

r n
interest formula: A = P 1 + 100 , where A is the resulting amount of
money and P is the principal invested for n interest periods at r% annually.
We need to investigate what happens when A = 2P.

r n
The above equation then becomes 2 = 1 + 100 (1)
log 2
It then follows that n = log(1+ 100
(2)
r
)
A table of values (Figure 1.9) from the above equation established with
the help of a calculator is shown.
If we take the arithmetic mean (the usual average) of the nr values, we
get 72.04092314, which is quite close to 72, and so our “Rule of 72” seems
to be a very close estimate for doubling money at an annual interest rate of
r% for n interest periods.
If r is restricted to more realistic values between, say, .5% and 5%, then
one could use also the “Rule of 70.”
r n nr
1 69.66071689 69.66071689
3 23.44977225 70.34931675
5 14.20669908 71.03349541
7 10.24476835 71.71337846
9 8.043231727 72.38908554
11 6.641884618 73.0607308
13 5.671417169 73.72842319
15 4.959484455 74.39226682
Figure 1.9
An ambitious reader, or one with a very strong mathematics background,

might try to determine a “rule” for tripling and quadrupling money, similar
to the way we dealt with the doubling of money. The above equation (2)
for k-tupling would be n = log log k
, which for r = 8, gives the value for
(1+ 100
r
)
n = 29.91884022(log k). Thus nr = 239.3507218(log k), which for k = 3
(the tripling effect) gives us nr = 114.1993167. We could then say that for
tripling money, we would have a “Rule of 114.” For k = 4, we even do
not need any further calculations. Since two times doubling is equivalent to
quadrupling, we know that here we would have a “Rule of 144” (or “Rule of
140” if we used the “Rule of 70” for doubling). Although we use computer
calculation, for the most part, to determine compound interest, here we see
a simple trick that allows us to skip sizable portion of calculation and still
come up with a useful answer. Once again, we see how mathematics helps
us with everyday finances.
Chapter 2
Numerical Novelties
Mathematics harbors a plethora of numerical curiosities. There are times

when one simply stumbles on some of these and there are other times when
one seeks a creative challenge to find such curiosities. In this chapter, we
present some of these unusual numerical relationships with the hope that
they will not only bring a new life into one’s view of mathematics, but also
perhaps motivate readers to seek other analogous relationships. We begin
with a few simple ones and then progress further to some more challenging,
yet amazing, numerical novelties.
The Nine-Digit Number Yields an Unexpected Digit Sum
This sum needs no further explanation. Just admire it!

987,654,321 → digit sum 45
−123,456,789 → digit sum 45
864,197,532 → digit sum 45
Curiously, 16 Fours Can Equal 1,000
Here, the number 4 is used 16 times to create the number 1,000.

444 + 444 + 44 + 44 + 4 + 4 + 4 + 4 + 4 + 4 = 1,000
41
Unexpectedly, 8 Eights Can Equal 1,000
Here, the digit 8 is used 8 times to create the number 1,000.
888 + 88 + 8 + 8 + 8 = 1,000
Using Nine Digits to Create the Number One-Half
Note how all 9 digits are used exactly once to create the number 12 .
9,327 1
=
18,654 2
Using Nine Digits to Create 99,999
Here is an addition calculation that uses all 9 digits exactly once with a sum
of 99,999:
98,765
+1,234
99,999
An Unexpected Pattern
Patterns in mathematics tend to crop up when we would least expect them.

We offer the following example of how a rather innocuous series of numbers,
when squared, results in a totally unexpected pattern.
42 = 16
342 = 1156
3342 = 111,556
33342 = 11,115,556
333342 = 1,111,155,556
3333342 = 111,111,555,556
Numerical Novelties 43
Although the ambitious reader will probably be able to continue this pattern,
it would be pleasantly challenging to seek other patterns of a similar kind.
Such unusual occurrences in mathematics are always entertaining as well
as motivating.
An Unexpected Prime Relationship
Here, we encounter an interesting phenomenon in arithmetic, where we

begin by choosing any sequence of natural numbers, beginning with 1 and
ending before it reaches the next prime number. Then, multiply those num-
bers and add 1 to this product, and you will find that this result is divisible
by the prime number at which you stopped the sequence.
To show how this works, let’s consider the first few such sequences that
you might select, such as the sequence that ends before the prime number
3, and perform the requested operation:
1 × 2 + 1 = 3, which we find is divisible by 3
Now, consider the sequence that ends before the next prime number,
namely, 5:
1 × 2 × 3 × 4 + 1 = 25, which is divisible by 5
Considering the sequence that ends before the next prime number, namely, 7:
1 × 2 × 3 × 4 × 5 × 6 + 1 = 271, which is divisible by 7, since 721

7
= 103
Taking this one step further to the next prime number, namely, 11:
1 × 2 × 3 × 4 × 5 × 6 × 7 × 8 × 9 × 10 + 1 = 3,628,801,
3,628,801
which is divisible by 11, since = 329,891
11
This then continues on and once again demonstrates amazing numerical

relationships.
Favorable Numerical Arrangements of Digits 1–9
The challenge here is to arrange the numbers 1, 2, 3, 4, 5, 6, 7, 8, 9 in

sequence using only addition and subtraction to reach the number 100.
Here is one possible solution:
123 − 45 − 67 + 89 = 100
We are offering some other solutions that you might get from your
audience just so you will be properly prepared:
123 + 4 − 5 + 67 − 89 = 100
123 + 45 − 67 + 8 − 9 = 100
123 − 4 − 5 − 6 − 7 + 8 − 9 = 100
12 − 3 − 4 + 5 − 6 + 7 + 89 = 100
12 + 3 + 4 + 5 − 6 − 7 + 89 = 100
1 + 23 − 4 + 5 + 6 + 78 − 9 = 100
1 + 2 + 34 − 5 + 67 − 8 + 9 = 100
12 + 3 − 4 + 5 + 67 + 8 + 9 = 100
1 + 23 − 4 + 56 + 7 + 8 + 9 = 100
1 + 2 + 3 − 4 + 5 + 6 + 78 + 9 = 100
−1 + 2 − 3 + 4 + 5 + 6 + 78 + 9 = 100
To take this one step further, we can also try to do this in reverse, such as
9 + 8 + 76 + 5 − 4 + 3 + 2 + 1 = 100
Here are several more examples of how to create sums of 100 using the
numbers 1–9 in reverse order:
98 − 76 + 54 + 3 + 21 = 100
9 − 8 + 76 + 54 − 32 + 1 = 100
98 − 7 − 6 − 5 − 4 + 3 + 21 = 100
9 − 8 + 7 + 65 − 4 + 32 − 1 = 100
9 − 8 + 76 − 5 + 4 + 3 + 21 = 100
98 − 7 + 6 + 5 + 4 − 3 − 2 − 1 = 100
98 + 7 − 6 + 5 − 4 + 3 − 2 − 1 = 100
98 + 7 + 6 − 5 − 4 − 3 + 2 − 1 = 100
98 + 7 − 6 + 5 − 4 − 3 + 2 + 1 = 100
98 − 7 + 6 + 5 − 4 + 3 − 2 + 1 = 100
98 − 7 + 6 − 5 + 4 + 3 + 2 − 1 = 100
98 + 7 − 6 − 5 + 4 + 3 − 2 + 1 = 100
98 − 7 − 6 + 5 + 4 + 3 + 2 + 1 = 100
9 + 8 + 76 + 5 + 4 − 3 + 2 − 1 = 100
−9 + 8 + 76 + 5 − 4 + 3 + 21 = 100
−9 + 8 + 7 + 65 − 4 + 32 + 1 = 100
−9 − 8 + 76 − 5 + 43 + 2 + 1 = 100
You might also extend your newly developed talent by creating the
number 100 using all 10 digits with only the operations of addition
and multiplication. One such possibility using all digits from 0 to 9 is
(9 × 8) + 7 + 6 + 5 + 4 + 3 + 2 + 1 + 0.
Using All Nine Digits 1–9 as Mixed-Number Fractions
Here, we try to find a way of representing the number 100 using all nine
digits in the form of a mixed-number fraction. We offer 11 ways to do
that and they are by no means simple, but perhaps after seeing a few of
them, you may discover other ways to accomplish this feat. Here are the 11
possibilities:
69258 5643 7524 3546 5742 5823

3 , 81 , 81 , 82 , 91 , 91 ,
714 297 396 197 638 647
7524 1578 1428 1752 2148
91 , 94 , 96 , 96 , 96
836 263 357 438 537
More Fun with All 10 Digits 0–9
Here, we seek a way to arrange the digits 0–9 in fraction form to reach the
number 1. One possible solution is 35
70
+ 148
296
= 12 + 21 = 1.
Another challenge could be to create the number 10 using all 10 digits.
One possible way to create the number 10 by using all digits from 0 to 9
exactly once is 1 35
70
+ 8 46
92
= 10.
An Amazing Result by Multiplying 12,345,679 by a Multiple of 3
Here, we can have some fun by multiplying the number 12,345,679 (note
that 8 is missing) by various multiples of 3 to get some surprising results.
Keeping in mind that the numbers 45, 48, and 51 are all multiples of 3 and
so we offer the following examples:
12345679 × 45 = 555,555,555
12345679 × 48 = 592,592,592
12345679 × 51 = 629,629,629
Here are some more examples using other multiples of 3, such as 63,
54, and 72:
12345679 × 63 = 777,777,777
12345679 × 54 = 666,666,666
12345679 × 72 = 888,888,888
You might wonder, which multiples of 3 resulted in numbers which had

a repetition of 9 digits. There are many other surprising results through
these multiple-of-3 multiplications. Each will probably bring surprise and
pleasure.
A Surprising Division
Today, it is quite likely that you have a calculator on hand which will allow
you to experience a rather pretty result, namely, where our 10 digits keep
repeating as shown in the following (only the first three repetitions; there-
fore, we use the symbol ≈ instead of =):
137,174,210
≈ 0.123456789012345678901234567890
1,111,111,111
This beautiful result requires no further explanation.
Returning to a Starting Number
Select any three-digit number and write it twice to form a six-digit number.
For example, if you choose the number 357, then write the six-digit number
357,357. We now — perhaps using a calculator — divide this number by
7, then divide the resulting quotient by 11, and lastly, divide that quotient
by 13, as we have done as follows:
357,357
= 51,051
7
51,051
= 4,641
11
4,641
= 357
13
You will recognize 357 as the first number we started with! You may wish
to try this with other three-digit numbers to convince yourself that this
technique always works.
Explanation: The reason that this works is that to form the original six-digit
number, you actually multiplied the original three-digit number by 1,001.
That is, 357×1,001 = 357,357. However, 1,001 = 7×11×13. Therefore,
by dividing successively by 7, 11, and 13, we have undone the original
multiplication by 1,001, leaving the original number. To those unaware of
this relationship, the result will surely be surprising.
While we are on the subject of the number 1,001, we can see how this
number can also help us multiply other number combinations:
221 × 77 = (17 × 13) × (11 × 7) = 1,001 × 17

= 1,000 × 17 + 1 × 17 = 17,017
264 × 91 = (24 × 11) × (13 × 7) = 1,001 × 24
= 1,000 × 24 + 1 × 24 = 24,024
407 × 273 = (37 × 11) × (3 × 7 × 13) = 1,001 × 111
= 1,000 × 111 + 1 × 111 = 111,111
Arithmetic can indeed expose some hidden numerical treasures.
Products of 91 and the Numbers 1–9
To appreciate the next suggested multiplications with the number 91, we

need to do all the multiplications from 1 through 9 and then admire the
results vertically. Note the list of units digits, tens digits, and hundreds
digits:
91 × 1 = 091
91 × 2 = 182
91 × 3 = 273
91 × 4 = 364
91 × 5 = 455
91 × 6 = 546
91 × 7 = 637
91 × 8 = 728
91 × 9 = 819
A Strange Coincidence
Can you imagine that the first six prime numbers could be divisors of six
consecutive numbers? Well, there is such a case, where the consecutive
numbers, 788, 789, 790, 791, 792, and 793, are divisible by the first six
prime numbers: 2, 3, 5, 7, 11, and 13, respectively. That is,
788 789 790 791

= 394, = 263, = 158, = 113,
2 3 5 7
792 793
= 72, = 61
11 13
How to Establish a Prime Number
There are times when you will appreciate a clever technique for doing
something that could have been taught in high school but clearly was not.
One such case is how to establish whether a number is a prime number or
not. Before we set up a technique for establishing prime numbers, we need to
review the meaning of a factorial, such as n! = 1×2×3×4×5×6×· · · ×n.
The rule for establishing a prime is that if n! + 1 is divisible by n+ 1, then
n +1 is a prime number. Suppose we would like to test this to see if the
number 11 is a prime number. Therefore, we say that 11 = n+1, whereupon
n = 10. We now seek 10!, which is equal to 3,628,800. Thus, 3,628,800 +
1 = 3,628,801 = 11 × 329,891. Therefore, we can conclude that 11 is a
prime number. Although this may appear to be somewhat complicated, it is
much simpler than taking a large number and trying to divide it by so many
various smaller numbers until you can conclude that it has no factors other
than itself and 1.
Explanation: Why does this work? This is part of the so-called Wilson’s
theorem. One direction of this famous number theorem (the one we used
above) is easy to prove without a deep knowledge of number theory: The
product 1 ×2 ×3×4×5×6×· · · ×n + 1 cannot be divisible by 2, 3, . . . , n
(because the first summand 1 × 2 × 3 × 4 × 5 × 6 × · · · × n is divisible by
these numbers. In this case, 1 would have to be divisible by these numbers,
which is obviously not the case). Thus, n + 1 is the smallest number greater
than 1 which can possibly be a divisor of n + 1, and the smallest divisor

greater than 1 of a number must obviously be a prime number.
Twin Primes
Prime numbers can also be seen by their place on the list of primes. When
two prime numbers have a difference of 2, they are considered twin primes.
It is suspected that there are infinitely many twin primes, but this has never
been proved or disproved. The first few twin primes are (3, 5), (5, 7), (11,
13), (17, 19), (29, 31), (41, 43), (59, 61), (71, 73), (101, 103), (107, 109),
(137, 139)…. We note that 5 is the only number that will appear twice in the
list of twin primes. You may be curious to know what the largest twin prime
pair that has been discovered to date. As of June 2023, the largest twin prime
pair is (2996863034895 × 21290000 − 1, 2996863034895 × 21290000 + 1),
with 388,342 decimal digits, which was discovered in September 2016.
You may ask if there is a general format for expressing twin prime pairs.
The answer is that — with the exception of the first twin prime pair, namely,
(3, 5) — they can be expressed in the form of (6n − 1), (6n + 1), where
n is a natural number. A clever observer will also note that every number
between a pair of twin primes will be a multiple of 6, which can be easily
substantiated by examining the first several twin primes.
The Sum of a Two-Digit Number and its Reversal
The sum of a two-digit number and its reversal is always divisible by 11. As
an example, consider the number 89 + 98 = 187, which is divisible by 11.
Explanation: Now, consider the number sum ab + ba, which, when written
out properly, is (10a + b) + (10b + a) = 11a + 11b = 11(a + b), making
11 a factor of the number we started with.
The Numbers 9 and 11
The numbers 9 and 11 have very peculiar properties in the base 10 system
since they are on either side of the base number 10. For example, in the
“cubic world,” 113 = 1331, a palindromic number, and the number 9 also
produces a very unusual pattern since 93 = 729 = 13 + 63 + 83 = 36 . An

ambitious reader may wish to search for other analogous patterns.
Further Peculiarities with the Numbers 9 and 11
The two numbers, 9 and 11, can also be tied together in other ways. Such as
1
9
= 0.111111111111111111111 . . . and 11 1
= 0.0909090909090909 . . ..
We know that the product of 9 and 11 is 99 and that also provides a rather
1
unusual unit fraction: 99 = 0.0101010101010101 . . .. While on the topic
of 99, here is another curiosity that can be entertaining: 992 = 9801, and if
we split and add, 98 + 01 = 99.
Taking the number 9 a step further, we can consider the num-
ber 999, which is the product of 27 × 37. Now, taking the recipro-
cals of these two numbers, another nice pattern and relationship evolves
1
27
= 0.037037037037037037 . . . and when we compare that to 37 1
=
0.027027027027027027 . . ., we note a fascinating relationship between
these two numbers that came from the factors of 999.
Not wanting to ignore 9’s partner, the number 11 also provides some
curious patterns, such as
11 = 62 − 52
1111 = 562 − 652
111,111 = 5562 − 4452
11,111,111 = 55562 − 44452
Multiplying Our Numbers by Multiples of 9
The number 9 and its multiples can produce a rather unexpected pattern
of results when multiplied by a number formed with our 9 numerals in
descending order. Once again, the calculator may be required to make this
calculation run smoothly. It requires quite a bit of multiplication before the
beauty of the pattern emerges.
Multiplying 987654321 by multiples of 9,

987654321 × 9 = 8888888889
987654321 × 18 = 17777777778
987654321 × 27 = 26666666667
987654321 × 36 = 35555555556
987654321 × 45 = 44444444445
987654321 × 54 = 53333333334
987654321 × 63 = 62222222223
987654321 × 72 = 71111111112
987654321 × 81 = 80000000001
987654321 × 90 = 88888888890
987654321 × 99 = 97777777779
987654321 × 108 = 106666666668
987654321 × 117 = 115555555557
987654321 × 126 = 124444444446
987654321 × 135 = 133333333335
987654321 × 144 = 142222222224
987654321 × 153 = 151111111113
987654321 × 162 = 160000000002
987654321 × 171 = 168888888891
987654321 × 180 = 177777777780
987654321 × 189 = 186666666669
987654321 × 198 = 195555555558
987654321 × 207 = 204444444447
987654321 × 216 = 213333333336
987654321 × 225 = 222222222225

987654321 × 234 = 231111111114
987654321 × 243 = 240000000003
987654321 × 252 = 248888888892
987654321 × 261 = 257777777781
987654321 × 270 = 266666666670
987654321 × 279 = 275555555559
987654321 × 288 = 284444444448
987654321 × 297 = 293333333337
987654321 × 306 = 302222222226
An ambitious reader may wish to continue to watch this pattern grow

impressively. Patterns such as this, which are completely unexpected, serve
well to exhibit the hidden beauty of mathematics.
Yet, we can take this a step further by reversing the order of the consec-
utive digits — without the number 8 — and multiplying by multiples of 9
and discover how another surprising pattern of numbers emerges:
12345679 × 9 = 111,111,111
12345679 × 18 = 222,222,222
12345679 × 27 = 333,333,333
12345679 × 36 = 444,444,444
12345679 × 45 = 555,555,555
12345679 × 54 = 666,666,666
12345679 × 63 = 777,777,777
12345679 × 72 = 888,888,888
12345679 × 81 = 999,999,999
More Strange Multiples of 9
Multiplication with large numbers has become trivial using the calcula-
tor. However, some numbers have unusual properties and remain so after
multiplication, especially when multiplied by the number 9, as we have seen

in the previous example. It might be challenging to find numbers that are
composed of eight different numerals that when multiplied by 9, result in
numbers consisting of nine different numerals.
This can be an interesting challenge. We offer a few examples here:
58132764 × 9 = 523194876
76125483 × 9 = 685129347
72645831 × 9 = 653812479
81274365 × 9 = 731469285
What is particularly notable here is that the numeral 9 is missing from

each number being multiplied by 9 and yet appears in each of the end
products. Had we multiplied these original numbers by 18, we would be
able to get numbers with 10 different digits; however, this time, the 0 would
be included, which was excluded from the above products. Here are the
products, each consisting of 10 different digits:
58132764 × 18 = 1046389752
76125483 × 18 = 1370258694
72645831 × 18 = 1307624958
81274365 × 18 = 1462938570
As neat as this appears, it still would be a challenge to find other such

eight-digit numbers that when multiplied by 9, result in products that have 9
different digits, and when multiplied by 18, have products with 10 different
digits.
More Fun with 9s
There are times when you can be amused with some unusual patterns in
mathematics. The 9s lend themselves nicely to such a situation as you can
see here.
9 × 9 = 81
99 × 99 = 9,801
999 × 999 = 998,001
9,999 × 9,999 = 99,980,001
99,999 × 99,999 = 9,999,800,001
999,999 × 999,999 = 999,998,000,001
9,999,999 × 9,999,999 = 99,999,980,000,001
We also get an unexpected pattern of numbers when reverse consecutive-

digit numbers are multiplied by 9 and are sequentially added to reverse
consecutive numbers as shown here:
0 × 9 + 8 = 8
9 × 9 + 7 = 88
98 × 9 + 6 = 888
987 × 9 + 5 = 8,888
9,876 × 9 + 4 = 88,888
98,765 × 9 + 3 = 888,888
987,654 × 9 + 2 = 8,888,888
9,876,543 × 9 + 1 = 88,888,888
98,765,432 × 9 + 0 = 888,888,888
This time, we take multiples of 9 and multiply them by the number

37,037, as shown in the following, to once again get some surprising results:
37,037 × 3 = 111,111
37,037 × 6 = 222,222
37,037 × 9 = 333,333
37,037 × 12 = 444,444
37,037 × 15 = 555,555
37,037 × 18 = 666,666
A Program for Determining the Number of Digits Used
A computer program prints in sequence the natural numbers 1, 2, 3, 4, 5, 6,

7, 8, 9, 10, 11, 12, 13, 14, 15, …. When we reach the number 15, there were
21 digits that have been printed. What is the largest natural number that
can be entirely printed by this computer program generating consecutive

numbers, if it can print at most 1,000,000,000 digits?
Explanation: There are nine numbers with 1 digit, 90 with 2 digits, 900
with 3 digits, etc.
It will not be possible to print all numbers through 999,999,999, however,
let us try the number with one fewer digit 9. Consider the following: What is
the number of all printed digits of the numbers from 1 through 99,999,999?
It is (1 × 9) + (2 × 90) + (3 × 900) + (4 × 9,000) + (5 × 90,000) + (6 ×
900,000) + (7 × 9,000,000) + (8 × 90,000,000) = 788,888,889.
Thus, 1,000,000,000 − 788,888,889 = 211,111,111 digits remaining.
All the further numbers from 100,000,000 onwards have nine digits. There-
fore, another
211,111,111
= 23,456,790.111111111111111111111111 ≈ 23,456,790
9
numbers can be printed, where the last completely printed number is
99,999,999 + 23,456,790 = 100,000,000 + 23,456,789 = 123,456,789.
A Numerical Peculiarity
Look at the following pattern and marvel at the relationships:
19 × 1 = 19 and 9 + 1 = 10, and 1 + 0 = 1

19 × 2 = 38 and 3 + 8 = 11, and 1 + 1 = 2
19 × 3 = 57 and 5 + 7 = 12, and 1 + 2 = 3
19 × 4 = 76 and 7 + 6 = 13, and 1 + 3 = 4
19 × 5 = 95 and 9 + 5 = 14, and 1 + 4 = 5
19 × 6 = 114 and 11 + 4 = 15, and 1 + 5 = 6
19 × 7 = 133 and 13 + 3 = 16, and 1 + 6 = 7
19 × 8 = 152 and 15 + 2 = 17, and 1 + 7 = 8
19 × 9 = 171 and 17 + 1 = 18, and 1 + 8 = 9
19 × 10 = 190 and 19 + 0 = 19, and 1 + 9 = 10
Note the end results!

For further amazement, consider the following:
9109 × 1 = 09,109 and 0 + 9 + 1 + 0 + 9 = 19

9109 × 2 = 18,218 and 1 + 8 + 2 + 1 + 8 = 20
9109 × 3 = 27,327 and 2 + 7 + 3 + 2 + 7 = 21
9109 × 4 = 36,436 and 3 + 6 + 4 + 3 + 6 = 22
9109 × 5 = 45,545 and 4 + 5 + 5 + 4 + 5 = 23
9109 × 6 = 54,654 and 5 + 4 + 6 + 5 + 4 = 24
9109 × 7 = 63,763 and 6 + 3 + 7 + 6 + 3 = 25
9109 × 8 = 72,872 and 7 + 2 + 8 + 7 + 2 = 26
9109 × 9 = 81,981 and 8 + 1 + 9 + 8 + 1 = 27
Note the sequence of the resulting numbers!
The Sum of Squares Equals the Sum of More Squares
Here is an arithmetic surprise that sounds very simple and yet requires
a little bit of thought. Begin by taking the sum of any three squares and
multiplying the sum by 3. We now need to find four squares that will have
the same sum. For example, 3(22 +32 +42 ) = 87 = 92 + 22 + 12 + 12 . Or
perhaps as another example: 3(22 + 32 + 32 ) = 66 = 52 + 42 + 42 + 32 . This
can be sometimes rather frustrating and, yet, also delightfully challenging,
which is then considered entertaining.
Explanation: For those who might want to see a justification so that they
don’t feel that we left them with an unsolvable situation, we provide a simple
algebraic proof:
3(a 2 + b2 + c2 ) = (a + b + c)2 + (b2 − 2bc + c2 )

+(c2 − 2ca + a 2 ) + (a 2 − 2ab + b2 )
= (a + b + c)2 + (b − c)2 + (c − a)2 + (a − b)2
Using this relationship will also provide other possible solutions.

Sums of Powers
For those who saw the 2015 movie “The Man Who Knew Infinity,” they
will recall the last scene where the famous Indian mathematician Srinivasa
Ramanujan (1887–1920) instantly cites from his hospital bed that the num-
ber 1729 is the smallest number that can be expressed as a sum of two
cubes in two different ways. That is, 1729 = 123 + 13 = 103 + 93 . By the
way, 1729 is a number which is divisible by the sum of its digits. That is,
1729
1+7+2+9
= 1729
19
= 91. From this, we also have another curiosity: 1729 =
19 × 91.
Building on the previous curiosity, the number 6578 is the smallest
number that can be expressed as a sum of 3 fourth powers in two different
ways. That is, 6578 = 14 + 24 + 94 = 34 + 74 + 84 . As a follow-up, one
may search for two-digit numbers that can be expressed as a sum of squares
in two different ways. One such number is 65 = 82 + 12 = 72 + 42 . By
the way, the number 65 can also be expressed as the sum of two cubes:
65 = 43 + 13 . By now, you should be able to see how we can constantly
seek such relationships among our numbers. Just searching for such number
patterns can be quite amusing and, upon success, rewarding at the same time.
While we are fixed on the sum of squares, we can find a curious arrange-
ment as follows. If we multiply the sum of two different squares by the sum
of two other different squares, the result would be the sum of two squares
in different ways. We can show this symbolically in the following fashion:
2
a + b2 · c2 + d 2 = (ac + bd)2 + (ad − bc)2 or
2
a + b2 · c2 + d 2 = (ac − bd)2 + (ad + bc)2
Let’s see how that works with the numbers a = 2, b = 5, c = 3, and d = 6
so that
2
2 + 52 · 32 + 62 = 29 · 45 = 1,305

We then can set up the following: 22 + 52 · 32 + 62 = (2 · 3 + 5·6)2 +
(2 · 6 − 3 · 5)2 = 362 + (−3)2 = 1296 +9 = 1,305.

And alternatively: 22 + 52 · 32 +62 = (2 · 3 − 5 · 6)2 +
(2 · 6 + 3 · 5) = 24 + 27 = 576 + 729 = 1,305.
2 2 2
You might like to try to see if you can come up with another set of two
sums of squares.
Guessing the Birthday Date
Alfred says:
“A trickster wanting to impress her audience tells a person

that she can guess his birthdate if he does the following
arithmetic: Double the number of the days of the month to
reach your birthday and then add 5. Multiply this result by
50 and then add the number of the months to reach your
birthday month. Then the participant is to tell the trickster
the result of the calculations, and she will determine the
birthday date.”
How and why does that work?
Explanation: Let d be the number of days in the month to reach your

birthday (1 ≤ d ≤ 31) and let m be the number of months in the year to
reach the birthday month (1 ≤ m ≤ 12).
This can be seen algebraically as (2d + 5) 50 + m = 100d + m + 250.
That indicates that Alfred must simply subtract 250 from the result. The
first two digits denote the number of days from the start of the month to
reach the birthdate, and the last two digits denote the number of months to
reach the birthdate. For example, suppose the calculations is as follows: 18
days to the birthday, doubled and adding 5 yields 2 × 18 + 5 = 41, then
41 × 50 = 2,050, then adding the 10th month (October) leads to a result of
2,060, Alfred then subtracts 250, which yields 1,810, which we then note
as 18–10, and that indicates the 18th of October, which was the sought-after
birthday.
Natural Numbers and Their Digits
Consider this challenge: How many natural numbers less than 100 million
have the property that the first digit (from the left) equals the number of
digits?
Explanation: Once again, we will seek a pattern in order to resolve the
question. Considering one-digit numbers, this holds only for the number 1.
Considering two-digit numbers, this holds for all numbers 2*, where * can
be all 10 digits. Now, considering three-digit numbers, we have the numbers

of the form 3** of which there are 100 possibilities. The pattern continues
so that we have 1 + 10 + 100 + · · · + 100,000,000 = 111,111,111 natural
numbers.
Determining Remainders of Special Divisions
Begin by considering the following simple question: What is the smallest

two-digit number that always yields the remainder 1 when divided by 3, 4,
and 5? Now, consider larger numbers, where respective remainders from
division by 2, 3, 4, 5, 6, 7, 8, 9, and 10 are 1, 2, 3, 4, 5, 6, 7, 8, and 9, where
the remainder is always 1 less than the divisor.
Explanation: We know that the smallest number divisible by 3, 4, and 5 is

merely the product of these three numbers. Therefore, by adding 1 to the
product, we find the number which divided by these three numbers has a
remainder of 1. That is, 3·4·5+1 = 61. Because the remainders are always
1 less than the divisor, we have to subtract 1 from the respective multiples,
and this has to work for all powers of primes which divide the dividend,
and this yields
2 · 3· 5 · 7 −1 = 2,520 − 1 = 2,519

3 2
divisible by
2,3,4,5,6,7,8,9,10
Inverting the Order of the Digits
Let us denote a four-digit number by the digits “abcd”. Are there four-
digit numbers “abcd” so that the multiplication by 9 inverts the order of
digits: “abcd” · 9 =“dcba”? If yes, what are these numbers? If no, why
not? A further question would be as follows: Are there five-digit num-
bers “abcde” so that the multiplication by 9 inverts the order of dig-
its: “abcde”· 9 = “edcba”? If yes, what are these numbers? If no, why
not?
Explanation: The answer to the first question is yes, as it holds for the
number 1089 · 9 = 9801. We can analyze this as follows:
“abcd” · 9 = “dcba”; this implies that a = 1 and d = 9 so that “1bc9”×9 =
“9cb1”; this yields
9000 + 900b + 90c + 81 = 9000 + 100c + 10b + 1, or
90b + 9c + 8 = 10c + b
89b + 8 = c
and from that, we can conclude b = 0 and c = 8.
Alternative explanation: It is clear that “abcd” < 1111. Consider
“1bc9” · 9 = “9cb1”, where for the value of b, there are only two pos-
sibilities b = 0, b = 1: In case of b = 1, we have “11c9” · 9 = “9c11”,
but that is impossible because on the one hand, this would yield c = 9
(100 · 9 = 900), and on the other hand, the number would be greater than
1111. Hence, b = 0 so that “10c9” · 9 = “9c01”. That means c · 9 (tenths
place) must end with 2 because at the ones place, we have 9 · 9 = 81, carry
over 8, and at the tenths place, we must have 0; thus, only c = 8 is possible.
This can be extended to a five-digit number, where 10989 · 9 = 98901,
where symbolically we have “abcde” · 9 = “edcba”, which yields a = 1
and e = 9 so that “1bcd9” · 9 = “9dcb1”. This leads us to the following:
90000 + 9000b + 900c + 90d + 81 = 90000 + 1000d + 100c + 10b + 1
⇒ 899b + 80c + 8 =
91d
≤819
Therefore, b = 0 and 80c + 8 = 91d. Hence, 91d must end with

8 ⇒ d = 8, c = 9.
Another alternative explanation: It is clear that “abcde” < 11111, hence
in “1bcd9” · 9 = “9dcb1”, where for the value for b there are only two possi-
bilities b = 0, b = 1. In case of b = 1, we had “11cd9” · 9 = “9dc11”, but
that is impossible because then we would have d = 9 (1000 · 9 = 9000):
“11c99” · 9 = “99c11” but that contradicts “abcde” < 11111. Hence,
b = 0: “10cd9” · 9 = “9dc01”, which means d · 9 (tenths place) must end
with 2 because in the ones place, we have 9 · 9 = 81, carry over 8, and
at the tenths place, we finally must have 0, thus, only d = 8 is possible:
“10c89” · 9 = “98c01” ⇒ “c89” · 9 = “8c01” ⇒ c = 9.
Palindromic Numbers
There are certain categories of numbers that have particularly strange char-
acteristics. Here, we consider numbers that read the same in both directions:
left to right, and right to left. These are called palindromic numbers. First,
note that a palindrome can also be a word, phrase, or sentence that reads
the same in both directions. Figure 2.1 shows a few amusing palindromes.
A
EVE
CIVIC
LEVEL
RADAR
REVIVER
RACECAR
ROTATOR
DON’T NOD
LEPERS REPEL
MADAM I’M ADAM
STEP NOT ON PETS
NEVER OD OR EVEN
PULL UP IF I PULL UP
NO LEMONS, NO MELON
DENNIS AND EDNA SINNED
ABLE WAS I ERE I SAW ELBA
WAS IT A CAR OR A CAT I SAW
A MAN, A PLAN, A CANAL, PANAMA
A SANTA LIVED AS A DEVIL AT NASA
SUMS ARE NOT SET AS A TEST ON ERASMUS
ANNE, I VOTE MORE CARS RACE ROME TO VIENNA
ON A CLOVER, IF ALIVE, ERUPTS A VAST, PURE EVIL; A FIRE VOLCANO
DOC, NOTE I DISSENT:A FAST NEVER PREVENTS A FATNESS. I DIET ON COD
Figure 2.1
A palindrome in mathematics would be a number such as 666 or 123321

that reads the same in either direction. For example, the first five powers of
11 are palindromic numbers:
110 =1
111 = 11
112 = 121
113 = 1331
114 = 14641
Once again, using a calculator, there are some unusual aspects, which result
from squaring numbers consisting of all 1s, which are often called repunits.
Unexpectedly, these powers of repunits result in palindromic numbers, as
we can see here:
112 = 121
1112 = 1331
11112 = 1234321
111112 = 123454321
1111112 = 12345654321
11111112 = 1234567654321
111111112 = 123456787654321
1111111112 = 12345678987654321
A small curiosity is that the second smallest number with an even num-
ber of digits whose square is a palindrome with an even number of digits
is the number 798,644 since 798,6442 = 637,832,238,736, which is a
palindrome.
Now for the entertaining aspect of palindromic numbers. Here, we have
a procedure to see how a palindromic number can be generated from a
given number. All you need to do is to continually add a given number to its
reversal (that is, the number written in the reverse order of digits) until you
arrive at a palindrome. For example, a palindrome can be reached with a
single addition such as with the starting number 23: the sum 23 + 32 = 55,
a palindrome.
Or it might take two steps, such as with the starting number 75: The two
successive sums are 75 + 57 = 132 and 132 + 231 = 363, which lead us to
a palindrome. Or it might take three steps, such as with the starting number
86: 86 + 68 = 154, 154 + 451 = 605, and 605 + 506 = 1111.
The starting number 97 will require 6 steps to reach a palindrome; the
starting number 98 will require 24 steps to reach a palindrome. It is important
to be cautioned about using the starting number 196 as this one has not yet
been shown to produce a palindromic number, even when tried with over
three million reversal additions. We still do not know if this number will
ever reach a palindrome.
There are a few quirky results in this procedure. If you were to try to
apply this procedure to 196, you would eventually — at the 16th addition —
reach the number 227574622. Yet, amazingly, you would also reach that
same sum at the 15th step of the attempt to get a palindrome from the
starting number 788. This would then tell you that applying the procedure
to the number 788 has also not yet been shown to reach a palindrome. As
a matter of fact, among the first 100,000 natural numbers, there are 5,996
numbers for which we have not yet been able to show that the procedure of
reversal additions will lead to a palindrome. Some of the starting numbers
of these non-palindrome results are : 196, 691, 788, 887, 1675, 5761, 6347,
and 7436.
You might want to take this to another level by showing some unusual
aspects in this process. For example, using this procedure of reversal and
addition, we find that some numbers yield the same palindrome in the same
number of steps, such as 554, 752, and 653, which all produce the palin-
drome 11011 in three steps. In general, all integers, in which the corre-
sponding digit pairs symmetric to the middle 5 have the same sum, will
produce the same palindrome in the same number of steps. The three sam-
ple numbers, 554, 752, and 653, have this characteristic since the pair of
digits on either side of the middle 5 have the same sum, namely, 9.
There are other integers that produce the same palindrome, yet with a
different number of steps, such as the number 198, which, with repeated
reversals and additions, will reach the palindrome 79497 in five steps, while
the number 7299 will reach this same number in two steps, that is, 7299 +
9927 = 79497.
We can determine the number of additions that we will have to do to
reach a palindrome using this procedure. For a two-digit number ab with
digits a = b, the sum a + b of its digits determines the number of steps
needed to produce a palindrome. Clearly, if the sum of the digits is less than
10, then only one step will be required to reach a palindrome, for example,
25 + 52 = 77. If the sum of the digits is 10, then ab + ba = 110 and
110 + 011 = 121, so two steps will be required to reach the palindrome.
The number of steps required for each of the two-digit sums 11, 12, 13, 14,
15, 16, and 17 to reach a palindromic number is 1, 2, 2, 3, 4, 6, and 24,
respectively.
Now, we can take this yet to another level to appreciate some unusual
aspects of palindromic numbers. We can arrive at some lovely patterns
when dealing with palindromic numbers. For example, some palindromic
numbers when squared also yield a palindrome. For example, 222 = 484
and 2122 = 44944. On the other hand, there are also some palindromic
numbers that, when squared, do not yield a palindromic number, such as
5452 = 297,025. Of course, there are also non-palindromic numbers that,
when squared, yield a palindromic number, such as 262 = 676 and 8362 =
698,896. These are just some of the entertainments that numbers provide.
A motivated reader may want to search for other such curiosities.
Taking Palindromic Numbers Further
There are also some palindromic numbers that, when cubed, again yield
palindromic numbers.
To this group belong all numbers of the form n = 10k + 1, for k =
1, 2, 3, . . .. When n is cubed, it yields a palindromic number, which has
k − 1 zeros between each consecutive pair of 1,3,3,1, as we can see with
the following examples:
k = 1, n = 11 113 = 1331
k = 2, n = 101 1013 = 1030301
k = 3, n = 1001 10013 = 1003003001
k = 7, n = 10000001 100000013 = 1000000300000030000001
Using algebra, one can show that this pattern continues:

k 3
10 + 1 = 103k + 3·102k + 3·10k + 1 and written in another way,
3k

. . 01 = 1 0
1 0 . 3
. . . . . .
.
03 0 03 0 . . . 01
k k k k

2k
Especially in this representation, one can see the pattern immediately: As
the value of k increases by 1, the number of zeros in each block increases
by 1!
We can continue to generalize to reach some interesting patterns, such

as when n consists of three 1s and any even number of zeros symmetrically
placed between the end 1s when cubed will give us a palindrome, such as,
1113 = 1367631
101013 = 1030607060301
10010013 = 1003006007006003001
1000100013 = 1000300060007000600030001
Taking this even another step further, we find that when n consists of four
1s and 0s in a palindromic arrangement, where the places between the 1s
do not have the same number of 0s, then n 3 will also be a palindrome, as
we can see with the following examples:
110113 = 1334996994331
101001013 = 1030331909339091330301
100100010013 = 1,003,003,301,900,930,390,091,033,003,001
However, when the same number of zeros appears between the 1s, then
the cube of the number will not result in a palindrome, as in the follow-
ing example: 10101013 = 1030610121210060301. As a matter of fact,
the number 2201 is the only non-palindromic number, which is less than
280,000,000,000,000, and, when cubed, yields a palindrome 22013 =
10662526601.
Just for further amusement, consider the following pattern with palin-
dromic numbers:
333 · 333
12321 =
1+2+3+2+1
4444 · 4444
1234321 =
1+2+3+4+3+2+1
55555 · 55555
123454321 =
1+2+3+4+5+4+3+2+1
666666 · 666666
12345654321 =
1+2+3+4+5+6+5+4+3+2+1
and so on until the digit 9 is in the middle.
An ambitious reader may search for other patterns involving palindromic
numbers.
The Common Divisor of Palindromic Numbers
Now that we are thoroughly familiar with palindromic numbers, we can

consider the following question: What is the greatest common divisor of all
four-digit palindrome numbers? The answer is the number 11. Now, let’s
see why that is the case.
Explanation: A four-digit palindrome is of the form “anna” = a1001 +

n110 = (a × 7 × 11 × 13) + (n × 2 × 5 × 11) with the possibilities of
a ∈ {1, . . . , 9} and n ∈ {0, . . . , 9}, where also a = n is possible.
A Surprising Pattern of Odd Numbers
Odd numbers can be presented in such a way that they generate cubic
numbers. This can be quite surprising as well as enchanting.
1= 1 = 13
3+5= 8 = 23
7 + 9 + 11 = 27 = 33
13 + 15 + 17 + 19 = 64 = 43
21 + 23 + 25 + 27 + 29 = 125 = 53
31 + 33 + 35 + 37 + 39 + 41 = 216 = 63
Consider what the sum of the tenth line will be. The tenth line would have
a sum equal to 103 = 10,000. Amazement is typically generated by how
such a simple relationship can generate cubes.
The Digits Remain in Use
Simple multiplication (of course, having a calculator handy makes the expe-
rience much more less cumbersome) can generate some unexpected results.
Here are several multiplication examples that continue to use only the dig-
its of the two given numbers in the product; as you will note, in the first
product, only the digits 0, 1, 3, and 5 are used.
30 × 51 = 1,530
21 × 87 = 1,827
80 × 86 = 6,880
60 × 21 = 1,260
93 × 15 = 1,395
41 × 35 = 1,435
Naturally, this can be taken for larger numbers as well. For example, take
a pair of larger numbers and consider a case in which each of the digits in
the original multiplication is used in the product as many times as it appears
in the original modification.
9,162,361,086 × 1,234,554,321 = 11,311,432,469,283,552,606
Some More Cute Number Patterns
It is not very difficult to note that one can express a number as the sum of
three other numbers. However, with the number 118, we can also express
it as the sum of four arrangements of three numbers, and the amazing thing
is that the product of each of these sets of three numbers is the same for all
four groups of three, namely, 37,800. Take a look here:
15 + 40 + 63 = 118 and 15 × 40 × 63 = 37,800

14 + 50 + 54 = 118 and 14 × 50 × 54 = 37,800
21 + 25 + 72 = 118 and 21 × 25 × 72 = 37,800
18 + 30 + 70 + 118 and 18 × 30 × 70 = 37,800
More amazingly, this number 118 is the smallest number for which this
can be done. You might want to challenge yourself to come up with some
other such arrangements for other numbers.
Here is another nice number relationship to enjoy! Admire the

symmetry:
133 − 37 = 2197 − 2187 = 13 − 3
53 − 27 = 125 − 128 = −(5 − 2)
Do not try to find another such a pair of numbers because no other such pair
has yet been found!
There are many other curious number relationships. We will present a
few of them here. Note that in each case, the exponents are consecutive:
43 = 42 + 33
63 = 62 + 33
135 = 11 + 32 + 53
175 = 11 + 72 + 53
518 = 51 + 12 + 83
598 = 51 + 92 + 83
1306 = 11 + 32 + 03 + 64
1676 = 11 + 62 + 73 + 64
2427 = 21 + 42 + 23 + 74
And here is one where the exponents match the base:
3435 = 33 + 44 + 33 + 55
You might want to see if there are other such relationships or even some
such as
244 = 13 + 33 + 63 , and 136 = 23 + 43 + 43
Note the amazing relationship here!
If you want to keep this to one exponential number, you can do this with
the following numbers:
153 = 13 + 53 + 33
370 = 33 + 73 + 03
371 = 33 + 73 + 13
407 = 43 + 03 + 73
We can take this even a step further by considering the four-digit numbers,
such as the following:
1634 = 14 + 64 + 34 + 44
8208 = 84 + 24 + 04 + 84
9474 = 94 + 44 + 74 + 44
A Surprising Pattern
Begin by considering the following problem, which on the surface seems

rather harmless but could get a bit cumbersome. What is the sum of
13 + 23 + 33 + 43 + · · · + 93 + 103 ?
If carefully done with the aid of a calculator, this should yield the correct
answer. However, if we do not have a calculator at hand, the multiplication
and addition could prove quite complicated and messy! Let’s see how we
might solve the problem by searching for a pattern. Let’s organize our data
as shown below:
13 = (1) =1 = 12
13 + 23 = (1 + 8) =9 = 32
13 + 23 + 33 = (1 + 8 + 27) = 36 = 62
13 + 23 + 33 + 43 = (1 + 8 + 27 + 64) = 100 = 102
Note that the number bases in the final column (namely, 1, 3, 6, 10, …) are
triangular numbers, which are numbers that represent points that can be
arranged to form equilateral triangles; they are 1, 3, 6, 10, 15, 21, …. The
nth triangular number is formed by taking the sum of the first n integers.
That is, the first triangular number is 1. The second triangular number is
3 = (1 + 2). The third triangular number is 6 = (1 + 2 + 3). The fourth
triangular number is 10 = (1 + 2 + 3 + 4), and so on.
Thus, we can rewrite our problem as follows:
13 = (1)2 = 12 = 1
13 + 23 = (1 + 2)2 = 32 = 9
13 + 23 + 33 = (1 + 2 + 3)2 = 62 = 36
13 + 23 + 33 + · · · + 93 + 103 = (1 + 2 + 3 + · · · + 9 + 10)2 = 552 = 3025
By this point, you should have gotten a “feel” for the advantage of looking
for a pattern in solving a problem. It may take some effort to find a pattern,
but when one is discovered, it not only simplifies the problem greatly but
also once again demonstrates the beauty of mathematics.
Equal Number of Numbers with Equal Sums
The challenge is to determine for which numbers n can the set {1, 2, . . . , n}
be partitioned into two subsets with an equal number of numbers and equal
sums.
For example, it is impossible with n = 3, having the set {1, 2, 3}, yet
it is possible for n = 4, where we have {1, 2, 3, 4}, which can be par-
titioned as {1, 4} and {2, 3}, where in both cases 2 numbers, with equal
sums of 5.
How might the challenge be modified if we are not limited to an equal
product of numbers and only have equal sums? In that case, we would
have n = 3 as a solution since {1, 2, 3} can be partitioned in {1, 2} and
{3}, where in both cases the sum equals 3. Before reading the expla-
nation, a enthusiastic reader might seek other such examples as shown
above.
Explanation: We can show that it is possible for n = 4, 8, 12, . . . , so that

n = 4k. For example, when n = 8, we have two possibilities {3, 4, 5, 6}
and {1, 2, 7, 8} with sums of 18, as well as {2, 3, 6, 7} and {1, 4, 5, 8} with
sums of 18, as we shown in Figure 2.2.
When n is a multiple of 4 (n = 4k), it works with pairs of equal sums
from the “boundaries” to the “center.” One of the possibilities is shown in
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Figure 2.2
Figure 2.2. The odd numbers such as (4k + 1 and 4k + 3) for n cannot work
because one cannot divide {1, 2, . . . , n} into two subsets with equal number
of numbers. Furthermore, when n = 4k + 2, it is also not possible because
there the sum 1 + 2 + · · · + n would be odd, and one could not get two
equal sums. Although it is good to know it, one does not necessarily need
the formula for the sum of this arithmetic series: 1 + 2 + · · · + n = n(n+1)
2
.
The sum 1 + 2 + 3 + 4 = 10 is even; with the series up to 5, the sum is
odd, and with the series up to 6, the sum is again odd. Yet, when the series
extends to 7, the sum is even, and when the series extends to 8, the sum
is also even; this “pattern,” of course, continues beyond 8, hence, one can
conclude that the sum from 1 to 6 is odd (21), as the sums are also odd for
series up to 10, 14, 18, …, 4k + 2, ….
We seek to have the sum 1 + 2 + · · · + n to be even. Such a sum is
even for the first time with n = 3, where 1 + 2 + 3 = 6, and again with
n = 4, where 1 + 2 + 3 + 4 = 10; yet with n = 5, the sum is odd,
and it is again odd with n = 6; then again, the sum is even with n = 7
and n = 8; then again, twice odd, etc. That means all the multiples of
4 are a possibility (as shown above), but now their predecessors are also
possible, namely, n = 3, 7, 11, . . . n = 4k − 1. If one has a partition in two
subsets with equal sums for n, one has such a partition also for n + 4 (see
Figure 2.3).
1 2 n–1 n n+1 n+2 n+3 n+4
Figure 2.3
And since we have a concrete partition for n = 3, we know that alto-

gether there are such partitions for the numbers, n = 4k and n = 4k − 1
(k = 1, 2, 3, . . .).
Splitting Numbers
We can even split numbers other than into individual digits and still end up
with some spectacular results such as the following:
1,233 = 122 + 332

8,833 = 882 + 332
5,882,353 = 5882 + 23532
94,122,353 = 94122 + 23532
1,765,038,125 = 176502 + 381252
2,584,043,776 = 258402 + 437762
There are many more such amazements as we focus on a “reverse” situation,

where we could take the difference of the squares of a split number rather
than the sum of squares. For instance, we reverse the two parts so that from
48, we would consider the reverse and split the digits 8 and 4 and subtract
the squares: 48 = 82 − 42 . Here are several more such examples:
3,468 to be split as 34 and 68 so that 682 − 342 = 3,468
2,661,653 to be split as 266 and 1653 so that 16532 − 2662 = 2,661,653
59,809,776 to be split as 5980 and 9776 so that 97762 −59802 = 59,809,776
There are many more such examples that you might feel compelled to
search for.
We can then take this a step further in our effort to further highlight
unusual relationships. Consider displaying portions of numbers as the sum
of cubes, as we show with a few examples here: 41,833, which we can split
up as follows to get the sum of cubes: 43 + 183 + 333 = 41,833.
221,859 = 223 + 183 + 593

444,664 = 443 + 463 + 643
487,215 = 483 + 723 = 153
336,701 = 333 + 673 + 013
982,827 = 983 = 283 = 273
983,221 = 983 = 323 + 213
166,500,333 = 1663 + 5003 + 3333
Once again, this is not an exhaustive list and there are more such numbers
that lend themselves to this unusual splitting arrangement.
We can always look for nice relationships between numbers. With some
creativity, we can establish another form of “friendliness” between numbers.
Some of them can be truly mind-boggling! Take, for example, the pair of
numbers 6,205 and 3,869.
At first glance, there seems to be no apparent relationship. But with some
luck and imagination, we can get some fantastic results. 6,205 = 382 + 692
and 3,869 = 622 + 052 .
We can even find another pair of numbers with a similar relation-
ship. Consider these: 5,965 = 772 + 062 and 7,706 = 592 + 652 . Quite
spectacular!
A Random Division
David writes the 16 digits 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9 in an arbi-

trary order and randomly inserts the division symbol “÷” between two digits
indicating a division. Is it possible that the result of this division is exactly
2? Hint: Use the divisibility rule for 3, where the sum of the digits must be
divisible by 3, which was presented on page 13.
Explanation: Let G be the written number 2,233,445,566,778,899 without
including the “÷” sign in any place. Then, by using the well-known rule
for divisibility by 3, namely, that a number is divisible by 3 if, and only
if, the sum of its digits is divisible by 3, we know that G is not divisible
by 3 since the sum of the digits of the number is 88, which is not divisible
by 3. Therefore, no divisor of G can be a multiple of 3 (later, this fact will
serve as a contradiction, see the following). Now, let us denote the left part
of G (i.e., to the left of the “÷” sign) with L, and denote the right part
of G with R (i.e., to the right of the “÷” sign). For an indirect proof, let
us assume that the division L ÷ R = 2 was without a remainder, then it
would follow directly from the algorithm of division that the result of G ÷ R
G

is G ÷ R = ∗ ·
· · ∗ ∗
·
· · ∗ ÷R = 2 0
·
· · 01, where the number n of
L R n digits
m digits n digits
digits after 2 in the result (quotient) equals the number of digits in R.
We can also explain this phenomenon in another way; we will keep
the explanation simple and not use all 16 digits as originally proposed.
Consider, for example, L = 482 and R = 241, then L ÷ R = 482 241
= 2.
Now, consider L ÷ R = 241 = 2001. This could also be written as
482241
3 +241
G
R
= 2·241·10
241
= 2 · 1000 + 1 = 2001. Again, we see that the three digits
after the digit 2 in 2001 come from the three digits in 241.
Now, we have the desired contradiction: From the above, we know that
G and no divisor of G can be divisible by 3. But the assumption L ÷ R = 2
led us to G ÷ R = 2 0 .
. . 01 which is a divisor of Gdivisible by 3, which
n digits
is a contradiction.
The Surprising Sum and Product of Special Fractions
Given three positive (randomly selected) integers, such as 4, 5, and 14, we

shall create three fractions using these numbers in the following fashion:
The numerator of each fraction will be the sum of two of these numbers and
the denominator will be the third number so that we get 4+5 14
, 4+14
5
, 5+14
4
=
9
14
, 5 , 4 . Next, we will subtract the sum of these three fractions 14 +
18 19 9
18
5
+ 19
4
= 90+504+665
140
= 1259
140
= 2518
280
from their product 149
· 18
5
· 19
4
= 3078
280
to get 3078
280
− 2518
280
= 560
280
= 2. You may want to take three other starting
numbers and follow this procedure. Amazingly, you will find that the final
result will always be 2.
Explanation: To justify that the final result will always be 2, we would use
simple algebra. Let a, b, c be our starting integers:

a+b b+c a+c a+b b+c a+c
· · − + +
c a b c a b

ab + ac + b + bc · (a + c)
2
= −
abc
ab · (a + b) + bc · (b + c) + ac · (a + c)
abc
a 2 b a 2 c + ab2 + abc + abc + ac2 + b2 c + bc2
= −
abc
a 2 b + ab2 + b2 c + bc2 + a 2 c + ac2
abc
2abc
= =2
abc
Friendly Numbers
Within the realm of numerical novelties, there are numbers that are con-
sidered to be “friendly numbers.” Mathematicians have decided that two
numbers are to be considered friendly (or as sometimes used in the more
sophisticated literature, “amicable”) if the sum of the proper divisors1 (or
factors) of one number equals the second number and the sum of the proper
divisors of the second number equals the first number as well. Just take a
look at the smallest pair of friendly numbers: 220 and 284. The divisors (or
factors) of 220 are 1, 2, 4, 5, 10, 11, 20, 22, 44, 55, and 110. Their sum is
1 + 2 + 4 + 5 + 10 + 11 + 20 + 22 + 44 + 55 + 110 = 284.
The divisors of 284 are 1, 2, 4, 71, and 142, and their sum is 1 + 2 +
4 + 71 + 142 = 220.
This shows that these two numbers can be considered friendly numbers.
1 Proper divisors are all the divisors, or factors, of the number except the number itself. For
example, the proper divisors of 6 are 1, 2, and 3 but not 6.
A second pair of friendly numbers, which were discovered by the famous

French mathematician Pierre Fermat (1601–1665), is 17,296 and 18,416.
In order for us to establish their friendliness relationship, we need to find
all of their prime factors, which are 17,296 = 24 × 23 × 47 and 18,416 =
24 × 1151. Then, we need to create all the numbers from these prime factors
as follows:
The sum of the factors of 17,296 is 1+2+4+8+16+23+46+47+92+
94+ 184+ 188+368+ 376+ 752+ 1081 +2162 +4324+ 8648 = 18,416.
The sum of the factors of 18,416 is 1 + 2 + 4 + 8 + 16 + 1151 +
2302 + 4604 + 9208 = 17,296.
Once again, we note that the sum of the factors of 17,296 is equal to
18,416, and conversely, the sum of the factors of 18,416 is equal to 17,296.
This qualifies them to be considered a pair of friendly numbers.
There are many more such pairs. Here are a few more pairs of friendly
numbers:
1,184 and 1,210
2,620 and 2,924
5,020 and 5,564
6,232 and 6,368
10,744 and 10,856
9,363,584 and 9,437,056
111,448,537,712 and 118,853,793,424
An ambitious reader might want to verify the above pairs’ “friendliness!”
Explanation: For the experts, the following is one method for finding pairs
of friendly numbers: Let a = 3·2n −1, b = 3·2n−1 −1, and c = 32 ·22n−1 −1,
where n is an integer greater than or equal to 2, and a, b, and c are all prime
numbers. It then follows that 2n ab and 2n c are friendly numbers. We should
note that for n = 2, 4, and 7, we have a, b, and c that are all prime for n
less than or equal to 200. Another form of friendliness can be seen with the
following examples:
3869 = 622 + 052 = 6205 = 382 + 682
5965 = 772 + 062 = 7706 = 592 + 652
Are there other numbers that exhibit such friendliness?
We can even set up an analogous cycle using cubes: Starting with 55:
53 + 53 = 250, then 250: 23 + 53 + 03 = 133, and then 133: 13 + 33 +
33 = 55, which is the number we started with. This can be done with other
sequences of numbers, such as
136, 244, 136
919, 1459, 919
160, 217, 352, 160
The Magic of Square Numbers
Let’s consider a certain “magic” of square numbers. But first, let’s take a
slight detour to marvel at another curiosity. Sometimes, peculiarities are so
simple and yet can be interesting. Take, for example, the fact that there are
only two numbers, 2 and 11, where their squares increased by 4 will yield
a cube.
22 = 4, then by adding 4, we get 4 + 4 = 8 = 23
112 = 121, then by adding 4, we get 121 + 4 = 125 = 5 3
Now, let’s take a look at a list of square natural numbers and see if there
is any pattern to be recognized. Patterns always seem to provide enrichment
or enlightenment.
12 22 32 42 52 62 72 82 92 102 112
1 4 9 16 25 36 49 64 81 100 121
122 132 142 152 162 172 182 192 202 212
144 169 196 225 256 289 324 361 400 441
One thing that may be quickly noted among the square numbers listed is
that the units digits, which we have bold underlined above, follow a specific
pattern, namely, 1, 4, 9, 6, 5, 6, 9, 4, 1, 0, 1, 4, 9, 6, 5, 6, 9, 4, 1, 0, 1,….
This pattern will continue without end. Seeing this, one would be able to
surmise that there are certain digits, which can never appear in the units-
digit position of a square number since they are missing from the repetitions
list. That is, the digits 2, 3, 7, and 8 will never be the units digit of a square
number. Furthermore, these numbers separated by the zero is a palindromic
arrangement which can easily be spotted in the sequence
1, 4, 9, 6, 5, 6, 9, 4, 1, 0, 1, 4, 9, 6, 5, 6, 9, 4, 1, 0, 1…
and will continue ongoing.
There is probably no limit to the number of curiosities we can offer
about square numbers. For example, the numbers 13 and 31, which are
reversals of one another, have, respectively, squares that are also reversals
of one another, that is, 169 in 961. Furthermore, if we take the product of
these two numbers, we get 169 × 961 = 162,409 = 4032 ; yet another square
appears. If we want to take this a step further, the sum of the digits of 169
is 1 + 6 + 9 = 16 = 42 , and the sum of the digits of the square root of 169,
which is 13, is 1 + 3 = 4 = 22 , are in both cases square numbers. To add to
this surprisingly beautiful relationship, there is another pair of numbers that
has the same characteristic: These numbers are 12 and 21. If we follow the
same pattern as we did with the numbers 13 and 31, we will get 122 = 144
and 212 = 441. The product of these two numbers is 144 × 441 = 63,504
= 2522 . In addition, 1 + 4 + 4 = 9 = 32 and 1 + 2 = 3, which is analogous to
what we have done with the numbers 13 and 31.
While we are admiring square numbers, there are numbers, called auto-
morphic numbers, whose squares end in the same digits, such as
52 = 25
62 = 36
762 = 5,776
3762 = 141,376
6252 = 390,625
90,6252 = 8,212,890,625
890,6252 = 793,212,890,625
1,787,109,3762 = 3,193,759,921,787,109,376
8,212,890,9252 = 67,451,572,418,212,890,625
After observing this pattern, the question is: How can we create other
automorphic numbers?
Explanation: Suppose we take the next-to-last automorphic number above

and chop off several of its left-side digits so that we consider the num-
ber 921,787,109,376, which when we square it, we get the number
849,691,475,011,761,787,109,376. You will note that the last 10 digits are
the same. This can be done with any of the above automorphic numbers,
as we can also see with the previously calculated numbers: 90,625 2 =
8,212,890,625 and 890,6252 = 793,212,890,625.
At this point, you may like to experiment by tagging on to the front of
some of these numbers a few random digits keeping the terminal digits as
shown above and finding that in each case, automorphic numbers will be
created. Keep in mind that at most, there are two suffix groups of a specific
number of digits that can be used to create automorphic numbers. For exam-
ple, the numbers 625 and 376 are the only numbers of three digits that can
be used to make automorphic numbers, such as the number 1,234,625 2 =
1,524,298,890,625.
We should note that the number 90,625 is the only five-digit automorphic
number. We can see a few uses of that five-digit automorphic number above.
Here are the automorphic numbers up to 1015 :
1, 5, 6, 25, 76, 376, 625, 9376, 90625, 109376, 890625,
2890625, 7109376, 12890625, 87109376, 212890625, 787109376,
1787109376, 8212890625, 18212890625, 81787109376, 918212890625,
9918212890625, 40081787109376, 59918212890625, 259918212890625,
740081787109376
At this point, you have lots to experiment with to try to create numbers
whose square ends up with the same end digits as the original number. Lots
of fun lurks in the future!
What Are Lucky Days?
A day should be called lucky day if, when written in the format
MM/DD/YYYY, it uses eight different digits (days and months with num-
bers less than 10 are preceded by a 0). For example, the date 04/23/1965
was a lucky day. We are now in the 21st century. When is the next lucky
day after 2023?
Explanation: The next lucky day is 06/17/2345. In all coming years of

the form 20**, there cannot be a lucky day because one cannot find a
corresponding month since we cannot use the 0 which covers the first 10
months, and we cannot use November or December as well. In the years
21**, the month must contain 0, but then it is impossible to find a corre-
sponding day. Next, we are looking at the years of the form 23**. The month
of a lucky day then must contain 0 and hence the day must start with 1. The
next possible year is thus 2345 with the first possible month 06 and the first
possible day 17. Therefore, the next lucky day is June 17, 2345.
Typographical Errors That Are Correct
Just for entertainment, here are some typographical errors that turn out to be
correct. Consider the following, where the × was missing and misplaced,
such as with
73 × 9 × 42 = 7 × 3942
73 × 9 × 420 = 7 × 39420
Then there are some where the × and the exponents dropped out, such as
with
25 25
25 × = 25
31 31
25 × 92 = 2592
34 × 425 = 34425
34 × 4250 = 344250
1 1
112 × 9 = 1129
3 3
9 9
212 × 4 = 2124
11 11
6 6
132 × 7 = 1327
7 7
There are also some very complicated versions of these curious typograph-
ical errors, such as
6 6
132 × 7857142 = 1327857142 .
7 7
There are many more such strange coincidences, where symbols are deleted
or misread and still the correct answer results. We provide this merely as
an entertaining activity!
Wrong Arithmetic
Imagine an elementary school student who is learning multiplication of

fractions and finds that the following appears to be correct: 14 × 85 = 18
45
= 25 .
In other words, the student feels that to do the multiplication, you need
merely to combine the digits in the numerator and the denominator to get
the right answer since clearly, 14 × 85 = 20 8
= 25 . Unconvinced with your
reasoning that this doesn’t work, the student shows you another example,
where this method does work, such as 26 × 65 = 26 65
= 25 . Does this imply
that the student has come up with a new method of multiplying fractions?
This will certainly give you something to think about. The student may very
well say we can flip the two previous fractions and it still works, as we see
with 62 × 56 = 6526
= 52 , which is a correct result with a wrong procedure.
There are 14 such examples, where it works so that you can experience the
limits of this weird and incorrect multiplication. They are as follows:
1 8 18 1 5 15 1 4 14 1 6 16
× = , × = , × = , × = ,
4 5 45 2 4 24 6 3 63 6 4 64
1 9 19 4 9 49 2 6 26
× = , × = , × =
9 5 95 9 8 98 6 5 65
Each of these can be flipped to get another seven such examples. Using
simple algebra, we can show that these are the only seven examples, where
the digits are not the same. Let a, b, c, and d be the digits from 1 to 9 so that
a
b
× dc = 10b+d
10a+c
, which then can be reduced to the equation ac(10b + d) =
bd(10a + c), which then leads you to the seven examples above and their
flips.
Howlers! Reducing Fractions Incorrectly
In our early years of schooling, we learned to reduce fractions. For this,

there were specific ways to do it correctly. Some wise guy seems to have
come up with a shorter way to reduce some fractions. Is he right? He was
asked to reduce a fraction and did it in the following way:
26 2
=
65 5
That is, he just canceled out the 6s to get the right answer. Is this procedure
correct? Can it be extended to other fractions? If so, then we were surely
treated unfairly by our elementary school teachers who made us do much
more work. Let’s look at what was done here and if it can be generalized.
Explanation: In his book, Fallacies in Mathematics, E.A. Maxwell refers
to the following cancelations as “howlers”:
16 1 26 2
= =
64 4 65 5
Perhaps when someone did the fraction reductions this way and still got the
right answer, it could just make you howl.
Begin by reducing the following fractions to the lowest terms:
16 19 26 49
, , ,
64 95 65 98
After you have reduced to the lowest terms each of the fractions in the usual
manner, one may ask why it couldn’t have been done in the following way:
16 1
=
64 4
19 1
=
95 5
26 2
=
65 5
49 4 1
= =
98 8 2
At this point, you may be somewhat amazed. Your first reaction is probably
to ask if this can be done to any fraction composed of two-digit numbers of
this sort. Can you find another fraction (comprised of two-digit numbers)
where this type of cancelation will work? You might cite 55 55
= 55 = 1 as an
illustration of this type of cancelation. This will hold true for all two-digit
multiples of 11.
10x + a
Further explanation: Consider the fraction .
10a + y
The above four cancelations were such that when canceling the as, the
fraction was equal to xy . Therefore, 10x+a
10a+y
= xy . This yields y(10x + a) =
x(10a + y) and then 10x y +ay = 10ax +x y, and it follows that 9x y +ay =
10ax. Thus, y = 9x+a10ax
.
At this point, we shall inspect this equation. It is necessary that x, y, and
a are integers since they were digits in the numerator and denominator of a
fraction. It is now our task to find the values of a and x for which y will also
be integral. To avoid a lot of algebraic manipulation, you will want to set
up a chart which will generate values of y from y = 9x+a 10ax
. Remember that
x, y, and a must be single digit integers. In Figure 2.4, we have a portion
of the table we constructed. Note that the cases where x = a are excluded
since ax = 1.
x\a 1 2 3 4 5 6 9
1 20 30 40 50 60 90
=4 =5
11 12 13 14 15 18
2 20 60 80 100 120
=5
19 21 22 23 24
3 30 60 120 150 180
28 29 31 32 33
4 360
=8
45
Figure 2.4
The portion of the chart pictured above already generated two of the
four integral values of y, that is, when x = 1, a = 6, then y = 4, and
when x = 2, a = 6, then y = 5. These values yield the fractions 16 64
and
26
65
, respectively. The remaining two integral values of y will be obtained
when x = 1 and a = 9, yielding y = 5, and when x = 4 and a = 9,
yielding y = 8. These yield the fractions 19
95
and 49
98
, respectively. This should
convince you that there are only four such fractions composed of two-digit
numbers.
You may now wonder if there are fractions composed of numerators and
denominators of more than two digits, where this strange type of cancelation
holds true. Try this type of cancelation with 49 9 . You should find that 499 =
9 98
998
4
8
= 2.
1
A pattern is now emerging, and you may realize that:
49 499 4999 49999 499999 4 1

= = = = = ··· = =
98 998 9998 99998 999998 8 2
16 166 1666 16666 166666 1
= = = = = ··· =
64 664 6664 66664 666664 4
19 199 1999 19999 199999 1
= = = = = ··· =
95 995 9995 99995 999995 5
26 266 2666 26666 266666 2
= = = = = ··· =
65 665 6665 66665 666665 5
Enthusiastic readers may wish to justify these extensions of the original
howlers. Readers who, at this point, have a further desire to seek out addi-
tional fractions, which permit this strange cancelation should consider the
following fractions. They should verify the legitimacy of this strange can-
celation and then set out to discover more such fractions.
332 32 2
= =
830 80 5
385 35 7
= =
880 80 16
138 18 2
= =
345 45 5
27 5 25 5
= =
770 70 14
163 1
=
326 2
Aside from providing an algebraic application, which can be used to intro-
duce a number of important topics in a motivational way, this topic can also
provide some recreational activities. Here are some more of these “howlers”:
484 4 545 5 424 4 249 24 1

= = = = =
847 7 654 6 742 7 996 96 4
48484 4 54545 5 42424 4
= = =
84847 7 65454 6 74242 7
3243 3 6486 6 3
= = =
4324 4 8648 8 4
147 14 14 7 878048 8
= = =
71468 68 34 987804 9
1428571 1 2857142 2 1 3461538 3
= = = =
4285113 3 8571426 6 3 4615384 4
767123287 7 3243243243 3
= = ,
876712328 8 4324324324 4
1025641 1 3243243 3 4571428 4
= = =
4102564 4 4324324 4 5714285 5
4848484 4 5952380 5 4285714 4 2
= = = =
8484847 7 9523808 8 6428571 6 3
5454545 5 6923076 6 3 4242424 4
= = = =
6545454 6 9230768 8 4 74 24242 7
5384615 5 2051282 2 1 3116883 3
= = = =
7538461 7 8205128 8 4 8311688 8
6486486 6 3 484848484 4
= = =
8648648 8 4 848484847 7
This problem shows how elementary algebra can be used to investigate

a number theory situation, one that is also quite amusing. Mathematics
continues to hold some hidden treasures.
Curious Ways to Represent the Number 1,000,000
The number 1,000,000 can be written in many ways as a sum of four even
numbers. The number 1,000,000 can also be written in many ways as a
sum of four odd numbers. The question being posed here is as follows: Are
there more ways to represent the number 1,000,000 with four even numbers
or with four odd numbers (disregarding the order of the summands, which
should all be positive)? Or are there equally many ways?
Explanation: The sum of the even numbers is 2a + 2b + 2c + 2d =
1,000,000, which gives us a + b + c + d = 500,000, and implies that one
has to count all the possibilities for representing 500,000 as a sum of four
natural numbers a, b, c, d ≥ 1.
The sum of odd numbers can be written as (2a − 1) + (2b − 1) +
(2c − 1) + (2d − 1) = 1,000,000, which gives us a + b + c + d =
500,002, that is, one has to count all the possibilities for representing
500,002 as a sum of four natural numbers a, b, c, d ≥ 1, and these are
more than in the case of even numbers because 500,002 > 500,000.
We can simplify this by considering the number 10 instead of 1,000,000
where this effect is rather clearer. There is only one possibility regarding 10
as the sum of four even numbers (that is, 2 + 2 + 2 + 4); however, there are
three possibilities for the sum of four odd numbers, namely 1 + 1 + 1 + 7,
1 + 1 + 3 + 5, and 1 + 3 + 3 + 3.
Counting Paths
Counting paths can be a challenge unless a pattern evolves. Consider the

grid shown in Figure 2.5 and then determine the number of permitted ways
one can go from A to B by moving from square to square either to the right
or upwards.
Explanation: Start at A and fill in every square with the number of possibil-
ities to get there. Then, one quickly recognizes that every such number is the
sum of the number below and the number to the left. This then allows one
to fill the squares with larger numbers without testing the number of ways
being indicated. Hence, there are 486 ways to reach square B, as shown in
Figure 2.6.
Figure 2.5
Figure 2.6
A Most Unusual Magic Square
Typically, magic squares are square arrangements of numbers, where the

sum of each row, column, and diagonal is the same. However, there are
times in most unusual cases where this can be extended from a normal
magic square to one that has a surprising result. Consider the 3 × 3 magic
square shown in Figure 2.7.
2 7 6
9 5 1
4 3 8
Figure 2.7
In each case, the sums of the rows, columns, and diagonals are 15, which
is what is expected in a proper magic square. However, this particular magic
square can be modified when its elements are exponents of the number
2 because a new magic square evolves where the products of the rows,
columns, and diagonals are the same. We show this in Figures 2.8 and
2.9, where the products of the rows, columns, and diagonals are all 32,768.
Thus, we have created an unusual magic square based on a common product
emanating from a common sum.
22 27 26
29 25 21
24 23 28
Figure 2.8
4 128 64
512 32 2
16 8 256
Figure 2.9
An ambitious reader may choose to try to create a product magic square

using a base other than 2.
A Peculiar Large Number
A huge number n consists of 300 units digits and an unknown number of

zeros. Can this number be a square number? If yes, in which cases, if no,
why not?
Explanation: Recall that a number is divisible by 3, if, and only if, the sum
of its digits is divisible by 3, and a number is divisible by 9, if, and only
if, the sum of its digits is divisible by 9. The sum of the digits of n is 300
since there are 300 units digits, therefore, the number n is divisible by 3. In
order for n to be a square number, the sum of the digits would have to be
divisible by 9 and not by 3 alone, among others. Unfortunately, the sum of
the digits is 300, which is not divisible by 9, and therefore, n itself is not
divisible by 9 and, thus, cannot be a square number.
The Sum of Consecutive Odd Numbers (Starting at 1) is Always

a Perfect Square
1 + 3 + 5 + · · · + (2n − 1) = n 2
There are numerous proofs for this phenomenon, one of the easiest needs
just a picture (Figure 2.10), not a formal proof by mathematical induction.
Figure 2.10
The following is clear:
(1) Helpful squares appear (starting from a square 1 × 1 the side length
(both width and height) increases in every step by 1, which means that
the shape remains a square).
(2) The number of circles increases by 2 from every angle hook to its
successor (by 1 on each leg), which means the number of circles —
starting from 1 — remains odd.
Hence, this pattern surely continues.
A Remarkable Relationship
The following three equations demonstrate some very beautiful symmetry

and equality. Does this pattern continue indefinitely? Or does it end with
these three equations?
1+2 = 3
4+5+6=7+8
9 + 10 + 11 + 12 = 13 + 14 + 15
.. ..
. .
Explanation: If you continue this process, we find that the pattern

continues:
16 + 17 + 18 + 19 + 20 = 21 + 22 + 23 + 24
25 + 26 + 27 + 28 + 29 + 30 = 31 + 32 + 33 + 34 + 35
Proof: Using 1 + 3 + 5 + · · · + (2k − 1) = k 2 (see above), we can be sure

that every line, say line number k, begins with k 2 , and in this line, k + 1
summands are on the left-hand side of the equation and k summands on the
right-hand side of the equation (in both cases consecutive natural numbers).
So, we have to prove

k2 + k2 + 1 + · · · + k2 + k = k2 + k + 1 + · · · + k2 + k + k

k·k
k summands k summands
But this is easy to prove: Take the k many ks of k 2 = k · k (first summand

above) and put in each of the following brackets on the left-hand side one
summand k, then we have exactly the right-hand side of the equation.
Another Remarkable Relationship
Here is another surprising relationship to further enhance your appreciation

for mathematical wonders:
1 1+3 1+3+5 1+3+5+7 1
= = = = ··· =
3 5+7 7 + 9 + 11 9 + 11 + 13 + 15 3
The numerator is the sum of the first k odd numbers, and the denominator
is the sum of the next k odd numbers. How might you justify this result
without using the formula for the sum of an arithmetic series?
Explanation: Consider the following “proof-without-words” for the case
k =4, where
1+3+5+7 1
=
9 + 11 + 13 + 15 3
The odd numbers in Figure 2.11, which are 1, 3, 5, 7, 9, 11, 13, 15, are
represented by “right-angle hooks,” and there are equally many such angle
hooks of light dots and darker dots (k = 4 in both cases). And thus, we
can see that the light dots are 13 of the darker dots because they are 14 of the
whole picture.
Of course, the same works with other values of k! Obviously, such
pictures can prove to be significant explanations (“proofs without words”)
and in many cases, better than using elaborate formulas.
The Paved Paths
Here, we set out to create paved paths consisting of rectangular tiles, shown
in Figure 2.12, each of which has a length of 2 ft and a width of 1 ft. The
challenges before us are as follows:
• How many possibilities are there to pave a path of length 10 ft and width
2 ft with 10 such rectangular tiles?
• How many possibilities are there to pave a path of length 15 ft and width
2 ft with 15 such rectangular tiles?
• Describe generally for the number of possibilities with n such tiles and
a path of length n.
Figure 2.11
Figure 2.12
Explanation: One can easily find the number of possibilities for smaller
numbers of n:
n (number of tiles, length of the path) 1 2 3 4 5

s(n) (number of solutions) 1 2 3 5 8
n (number of Possible paths Number

tiles, length of of paths
the path)
1 1
2 2
3 3
4 5
5 We shall leave these for the reader 8

6 We shall leave these for the reader 13
Figure 2.13
If we take the list (Figure 2.13) a bit further, we can see that with n = 6,
there are 13 possible ways to pave the path. This pattern may remind you
of the Fibonacci numbers, and, indeed, this is the underlying principle. The
explanation is easy in retrospect but not so easy to find when dealing with
the problem for the first time. For every paved path of length n, there are
two possibilities for the beginning; either it begins with two horizontal tiles
or with a vertical tile, as shown in Figure 2.14.
Figure 2.14
In the first case, there are s(n−2) possibilities to pave the remaining path,
and in the second case, there are s(n −1) possibilities for the remaining path.
Altogether, we can see that the number s(n) can be split up into s(n − 2)
cases on the one hand and s(n − 1) cases on the other hand. Therefore,
s(n) = s(n − 1) + s(n − 2).
Diagonal Properties of a Rectangle
On a graph paper consisting of squares, a rectangle is drawn with its vertices

at square intersection points, as shown in Figure 2.15. Consider a rectangle
with side lengths m (horizontally) and n (vertically) and then draw one
diagonal. The challenge is to determine how many small squares are hit by
this diagonal. We then need to establish a general formula. A small square
is said to be “hit” by a diagonal if the diagonal passes through its interior;
if the diagonal meets only one of its vertices, it is not considered to have
been hit by the diagonal.
Figure 2.15
Explanation: Let us in the first case assume that m and n are relatively
prime. Then (as in Figure 2.15), the diagonal will not pass through vertices
of the small squares somewhere in the interior of the rectangle. How many
small squares are hit in this case? Every time the diagonal passes one of the
interior m − 1 vertical lines, a small square will be hit. Analogously, every
time the diagonal passes one of the interior n − 1 horizontal lines, a small
square is hit. Together with the first situation, we have 1+(m−1)+(n−1) =
m + n − 1 hit small squares.
Next, we shall consider the general case, where m and n are not neces-
sarily relatively prime numbers, with gcd(m, n) = d ≥ 1. Then, we divide
m and n by d yielding m = md and n = dn , where m and n are again rela-
tively prime. In such a smaller rectangle with dimension m × n , we have
m + n − 1 hit small squares, as shown in Figure 2.15. Since the rectangle
with dimension m × n has d such “sections” (horizontally or vertically) in
which we have smaller rectangles of dimension m × n , we have in sum

d m + n − 1 = d md + n
d
− 1 = m + n − d hit small squares in the
general case.
Strategy Game on the Whiteboard to the Year 2017

On a whiteboard are the numbers 1, 2, 3, . . . , 2017. David and Lisa alter-
nately delete one number each, until only two numbers remain. David
starts. If the sum of the last two numbers is divisible by 8, then David
wins, otherwise Lisa wins. Is there a strategy for winning for David or
Lisa? If yes, for whom, and how does it work? In which year could this
game be played again with the same strategy or perhaps with a similar
one?
Explanation: David has a strategy for winning. He thinks of building pairs

with a sum 2016: (1, 2015); (2, 2014); . . . ; (1007, 1009), that is, if Lisa
selects one of these numbers in the pair, David will select the other paired
number. Following this procedure, there will remain the pair (1008, 2016)
and the number 2017. David’s strategy is that if he initially deletes 2017,
then all the other pairs remain (with a sum divisible by 8). After each deletion
by Lisa, he can then respond by deleting the corresponding 2016-sum-pair
number. For example, when Lisa deletes 3, he deletes 2013, and when Lisa
deletes 1008, he deletes 2016. Thus, it is guaranteed that after his deletion,
the sum of the remaining numbers is always divisible by 8, and he will win!
That works exactly in the same way with all numbers of the form 16k +1;
the next year that this scheme will work after 2017 is in the year 2033
(16 years later). There, David deletes the last number 16k + 1 and then
thinks of pairs with sum 16k such as the following: (1, 16k − 1), (2, 16k −
2), . . . , (8k − 1, 8k + 1) and the pair (8k, 16k) — this is the only one not
having sum 16k. And with the analogous strategy, he can guarantee that
after his move, the sum of the remaining numbers will be divisible by 8.
With a slight change, it also works for all numbers of the form 8k − 1,
such as in the year 2023. Building pairs with sum 2024 (in general 8k)
yields (1, 2023); (2, 2022); … ; (1011, 1013) and the number 1012 (in
general 4k), which means that David should initially delete 1012 (4k), and
then he can answer each of Lisa’s moves advantageously (so that the sum
of the remaining numbers will always be divisible by 8). For example, if
Lisa deletes 1011, he should delete the corresponding “partner” 1013 since
1011 + 1013 = 2024.
An Arithmetic Phenomenon
Here is a wonderful opportunity to show the usefulness of algebra, for it

will be through algebra that the curiosity will be quenched. Consider the
following:
Select any three-digit number with all digits different from one
another. Write all possible two-digit numbers that can be formed
from the three digits selected. Then, divide their sum by the sum
of the digits in the original three-digit number. Everyone in the
audience should get the same answer, 22.
Let’s consider the three-digit number 365. Take the sum of all the pos-
sible two-digit numbers that can be formed from these three digits: 36 +
35 + 63 + 53 + 65 + 56 = 308. Then, we get the sum of the digits of the
original number, which is 3 + 6 + 5 = 14. When we are to divide 308 by 14,
we get 22, which everyone should have gotten regardless of which original
three-digit number was selected.
Explanation: Let’s analyze this unusual result, where everyone arrives at
the number 22 regardless of which three-digit number was initially selected.
We will begin with a general representation of the selected number: 100x +
10y + z. We now take the sum of all the two-digit numbers taken from the
original three digits:
(10x + y) + (10y + x) + (10x + z) + (10z + x) + (10y + z) + (10z + y)

= 10(2x + 2y + 2z) + (2x + 2y + 2z)
= 11(2x + 2y + 2z)
= 22(x + y + z)
When this value 22(x + y + z) is divided by the sum of the digits (x + y + z),
the result is 22.
With this algebraic explanation, we ought to get a genuine appreciation
as to how nicely algebra allows us to understand arithmetic curiosities.
Understanding a Very Special Six-Digit Number 142857
It has many striking properties:
(a) If the last digit is removed and put as the first digit, the result is five
times the number:
142,857 → 714,285 = 5 × 142,857
(b) If the first digit is removed and put as the last digit, the result is three
times the number:
142,857 → 428,571 = 3 × 142,857

(c) If the first two digits are removed and put as the last digits, the result is
two times the number:
142,857 → 285,714 = 2 × 142,857
(d) If the last two digits are removed and put as the first digits, the result is
four times the number:
142,857 → 571,428 = 4 × 142,857

(e) If the first (last) three digits are removed and put as the last (first) digits,
the result is six times the number:
142,857 → 857,142 = 6 × 142,857
Are there other six-digit numbers with the properties shown for the number?
The challenge is to see if there are other such numbers, or, if not, we need
to prove there are no other such numbers.
We will now consider the properties mentioned above and noted as (a)
to (e) to check their veracity:
(a) Let us denote the six-digit number z by z = 10a + b, where a is
a five-digit number and b a single digit: z = a
b =
5-digit number
10a + b. Then, the number z̄ with b placed as the first digit is
z̄ = b a
= 100,000b + a and the condition 5z = z̄
5-digit number
can be written as
50a +5b = 100,000b+a ⇔ 49a = 99,995b ⇔ 7a = 14,285b (*)

Since 14,285 is not divisible by 7 and we see from (*) that 14,285b must
be divisible by 7, we can conclude b = 7 and a = 14,285, altogether
z = 142,857, which is the only such six-digit number.
(b) Now, the roles of z and z̄ are interchanged and the factor 5 is replaced
by 3: 3z̄ = z.
10a + b = 300,000b + 3a ⇔ 7a = 299,999b ⇔ a = 42,857b
Thus, 42,857b must have five digits since a has five digits, hence, the
only two possible values of b are b = 1 and b = 2 (with b ≥ 3, the
number 42,857b would have more than five digits); we then have two
solutions:
b = 1: a = 42,857, z̄ = 142,857, z = 3z̄ = 428,571

b = 2: a = 85,714, z̄ = 285,714, z = 3z̄ = 857,142
(c) Now, a is a four-digit number and b is the moved two-digit number:

z =
b a = 10,000b + a.
2-digit 4-digit
number number
Then, the number z̄ with b put as the last two digits is
z̄ = b = 100a+b and the condition z̄ = 2z can be written
a
4-digit 2-digit
number number
as 100a+b = 20,000b+2a ⇔ 98a = 19,999b ⇔ 14a = 2,857b
(**)
Since 2,857 is a prime number, it follows from (**) that a is a multiple
of 2857 and b is the same multiple of 14. As for corresponding ⎧ fac-
⎨1
tor k (“the same multiple”), there are only three possibilities k = 2
⎩
3
⎧ ⎧
⎨ 14 ⎨ 2857
because a = k·2857 must stay at four digits: b = 28 a = 5714
⎩ ⎩
42 8571
⎧
⎨ 142,857
yielding three solutions z = 285,714 .
⎩
428,571
(d) In relation to (c), again the roles of z and z̄ are interchanged and the
factor 4 is replaced by 4: we have 4z̄ = z.
400a + 4b = 10,000b + a ⇔ 399a = 9,996b ⇔ 19a = 476b
Since 19 is a prime number, it follows that b is a multiple of 19 and
a is the same multiple of 476. As corresponding ⎧ factor k (“the same
⎨3
multiple”), there are only three possibilities k = 4 because b = k ·19
⎩
5
must⎧have two digits ⎧ and a = k · 476 must have four digits:
⎨ 1428 ⎨ 57
a = 1904 b = 76 yielding three solutions
⎩ ⎩
2380 95
⎧ ⎧
⎨ 142,857 ⎨ 571,428
z̄ = 190,476 , z = 4z̄ = 761,904 .
⎩ ⎩
238,095 952,380
(e) Now, a and b both are three-digit numbers:

z = b = 1000a + b. Then, the number z̄ with a and b
a
3-digit 3-digit
number number
interchanged is z̄ = a = 1000b + a and the condition 6z = z̄
b
3-digit 3-digit
number number
can be written as 6000a + 6b = 1000b + a ⇔ 5999a = 994b ⇔
857a = 142b.
Since 857 is a prime number, it follows that b = 857 and a = 142,
thus, the only solution is z = 142,857.
Remark: The background from number theory (not needed in the formu-
lation of the problem) is the notation of 7i (i = 1, . . . , 6) as periodical
decimal fractions:
1 2 3 4
= 0.142857, = 0.285714, = 0.428571, = 0.571428,
7 7 7 7
5 6
= 0.714285, = 0.857142
7 7
The Ever-Present Number 6174
There are some numbers in our decimal system that have unique character-
istics. One such number is 6147. To exhibit this strange characteristic, we
begin by selecting any four-digit number, where the digits are not all the
same. Following the procedure that we provide in the following, using any
such four-digit number, will end up with a number 6174.
(1) Begin by selecting any four-digit number — except one that has all
digits the same.
(2) Rearrange the digits of the number so that they form the largest number
possible. (In other words, write the number with the digits in descending
order.)
(3) Then, rearrange the digits of the number so that they form the smallest
number possible. (That is, write the number with the digits in ascending
order. Zeros can take the first few places.)
(4) Subtract these two numbers (obviously, the smaller from the larger).
(5) Take this difference and continue the process, over and over and over,
until you note something disturbing happening. Don’t give up before
something unusual happens.
You will note that this entertaining exercise will eventually arrive at the
number 6,174 — perhaps after one subtraction or after several subtractions.
Once you arrive at this number 6,174, you will find yourself in an endless
loop, which means that by continuing the process with the number 6,174,
you will continue to end up with 6,174. Remember that you will have begun
with an arbitrarily selected number.
Here is an example of how this works with our arbitrarily selected start-
ing number 3,927:
• The largest number formed with these digits is 9,732;
• the smallest number formed with these digits is 2,379;
• the difference is 7,353.
Now, using this number, 7,353, we continue the process:
• The largest number formed with these digits is 7,533;
• the smallest number formed with these digits is 3,357;
• the difference is 4,176.
Again, we repeat the process.
• The largest number formed with these digits is 7,641

• The smallest number formed with these digits is 1,467
• The difference is 6,174
When one arrives at 6,174, the number continuously reappears. Note

that the largest number that can be formed with these digits is 7,641, the
smallest number is 1,467 and the difference as we have seen above is 6,174.
Remember, all this began with an arbitrarily selected four-digit number
and will always end up with the number 6,174, which then gets you into
an endless loop by continuously getting back to 6,174. This nifty loop
was first discovered by an Indian mathematician, Dattathreya Ramachandra
Kaprekar (1905–1986), in 1946.2 We often refer to the number 6,174 as the
Kaprekar constant.
By the way, just as an aside, the number 6,174 is also divisible by the
sum of its digits. That is,
6174 6174
= = 343
6+1+7+4 18
Some Variations of the Karprekar Constants
• If you choose a two-digit number (not one with two same digits), then the
Kaprekar constant would be 81 and you would end up in a loop of length
5: [81, 63, 27, 45, 09 (, 81)]. There is no loop of length 1 for two-digit
numbers.
• If you choose a three-digit number (not one with all same digits), then
the Kaprekar constant would be 495 and you would end up in a loop of
length 1: [495 (, 495)].
2 Dattatreya Ramchandra Kaprekar announced it at the Madras Mathematical Conference

in 1949. He published the result in the paper “Problems involving reversal of digits” in
Scripta Mathematica in 1953; see also Kaprekar, D. R., “An interesting property of the
number 6174.” Scripta Mathematica 15(1955), 244–245.
• If you choose a four-digit number (not one with all same digits), then the
Kaprekar constant would be 6,174 — as we have seen before — and you
end up with a loop of length l: [6174 (, 6174)].
• If you choose a five-digit number (not one of all same digits), then there
are three Kaprekar constants: 53,955, 61,974, and 62,964.
One of length 2: [53,955, 59,994 (, 53,955)]
and two of length 4: [61,974, 82,962, 75,933, 63,954 (, 61,974)]
[62,964, 71,973, 83,952, 74,943 (, 62,964)]
You can follow this scheme with six-digit numbers, and you will also
find yourself getting into a loop. One number you may find leading you into
the loop is 840,852, but do not let this stop you from further investigating
this mathematical curiosity.3 For example, consider the digit sum of each
difference. Since the sums of the digits of the subtrahend and the minuend 4
are the same, the difference will have a digit sum that is a multiple of 9. For
three- and four-digit numbers, the digit sum is 18. In the case of five- and
six-digit numbers, the digit sum appears as 27. It follows that for seven-
and eight-digit numbers, the digit sum is 36. Yes, you will find that the digit
sum, when this technique is used on nine- and ten-digit number, is 45. You
will be pleasantly surprised when you check to see what the digit sum is for
even larger numbers.
Another Amazing Curiosity Generated by the Number 1089
This is about a number that has some truly exceptional properties. We begin
by showing how the number 1089 just happens to “pop up” when least
3 If you choose a six-digit number (not one with all of the same digits), then there are also
three Kaprekar constants: 549,945, 631,764, and 420,876.
Two of length 1: [549,945 (, 549,945)], [631,764 (, 631,764)] and one of length 7:
[420,876, 851,742, 750,843, 840,852, 860,832, 862,632, 642,654 (, 420,876)].
If you choose a seven-digit number (not one with all of the same digits), then there is only
one Kaprekar constant: 7,509,843.
There is a loop of length 8: [7,509,843, 9,529,641, 8,719,722, 8,649,432, 7,519,743,
8,429,652, 7,619,733, 8,439,552 (, 7,509,843)].
4 In a subtraction, the number in the subtrahend is subtracted from the number in the
minuend to get the result, referred to as the difference.
expected and then we’ll take another look at this number. We shall begin
by having you select any three-digit number, where the units and hundreds
digits are not the same and follow the following instructions:
Follow these instructions step by step, while we do it along in the boxes
below each instruction.
Choose any three-digit number (where the units and hundreds digits
are not the same).
We will do it with you here by arbitrarily selecting

825
Reverse the digits of the number you have selected.
We will continue here by reversing the digits of 825 to get

528
Subtract the two numbers (naturally, the larger minus the smaller).
Our calculated difference is 825 − 528 = 297
Once again, reverse the digits of this difference.
Reversing the digits of 297, we get the number

792
Now, add your last two numbers.
We then add the last two numbers to get 297 + 792 = 1089
Your result should be the same as ours even though your starting num-
ber was different from ours.
You will probably be astonished that regardless of which number you
selected at the beginning, you got the same result as we did, 1089. How does
this happen? Is this a “freak property” of this number? Did we do something
devious in our calculations?
Unlike the previous unit, which depended on a peculiarity of the decimal

system, this illustration of a mathematical oddity depends on the operations.
Before we explore (for the more motivated reader) why this happens, let
yourself be impressed with a further property of this lovely number 1089.
Let’s look at the first nine multiples of 1089:
1089 × 1 = 1089
1089 × 2 = 2178
1089 × 3 = 3267
1089 × 4 = 4356
1089 × 5 = 5445
1089 × 6 = 6534
1089 × 7 = 7623
1089 × 8 = 8712
1089 × 9 = 9801
Do you note a pattern among the products? Look at the first and ninth
products (i.e., 1089 and 9801). They are the reverse of one another. The
second and the eighth products (i.e., 2178 and 8712) are also reverse of one
another. And so, the pattern continues, until the 5th product, 5445, which
is a palindromic number (see page 62 for a quick refresher).
Note, in particular, that 1089 × 9 = 9801, which is the reversal of the
original number. The same property holds for 10989 × 9 = 98901, and
similarly, 109989 × 9 = 989901. You should recognize that we altered
the original number, 1089, by inserting a 9 in the middle of the number to
get 10989, and extended that by inserting 99 in the middle of the number
1089 to get 109989. It would be nice to conclude from this that each of the
following numbers has the same property: 1099989, 10999989, 109999989,
1099999989, 10999999989, and so on.
As a matter of fact, there is only one other number with four or fewer
digits where a multiple of itself is equal to its reversal, and that is the
number 2178 (which just happens to be 2 × 1089), since 2178 × 4 = 8712.

Wouldn’t it be nice if we could extend this, as we did with the above example,
by inserting 9s into the middle of the number to generate other numbers that
have the same property? Yes, it is true that
21978 × 4 = 87912
219978 × 4 = 879912
2199978 × 4 = 8799912
21999978 × 4 = 87999912
219999978 × 4 = 879999912
2199999978 × 4 = 8799999912
and so on.
As if the number 1089 didn’t already have enough cute properties, here is
another one that (sort of) extends the 1089: We will consider the number
1089 in two parts, the numbers 1 and 89.
Let’s see what happens when you take any number and get the sum of
the squares of its digits. Then, continue this process of find the sum of the
squares of the digits. Each time, curiously enough, you will eventually reach
1 or 89. Take a look at some examples that follow.
We will begin with the number 30. So, we can say that n = 30, and we
will find the sum of the digits of this number:
32 + 02 = 9, 92 = 81, 82 + 12 = 65, 62 + 52 = 61, 62 + 12 = 37,

32 + 72 = 58, 52 + 82 = 89, 82 + 92 = 145, 12 + 42 + 52 = 42,
42 + 22 = 20, 22 + 02 = 4, 42 = 16, 12 + 62 = 37, 32 + 72 = 58,
52 + 82 = 89, . . .
Once we reached 89, we got into what we call a loop, since we always
seem to get back to the number 89, when we repeat the process. Let’s try
this with the number 31.
So, we will let n = 31: 32 + 12 = 10, 12 + 02 = 1, 12 = 1
Again, for the number 1, a loop is formed, getting us back to 1 over and
over.
We shall now try 32, and we let n = 32: 32 + 22 = 13, 12 + 32 = 10,

1 + 02 = 1, 12 = 1
2
n = 33: 32 + 32 = 18, 12 + 82 = 65, 62 + 52 = 61, 62 + 12 = 37,

32 + 72 = 58, 52 + 82 = 89, 82 + 92 = 145, 12 + 42 + 52 = 42,
42 + 22 = 20, 22 + 02 = 4, 42 = 16, 12 + 62 = 37, 32 + 72 = 58,
52 + 82 = 89, . . .
n = 80: 82 + 02 = 64, 62 + 42 = 52, 52 + 22 = 29, 22 + 92 = 85,
82 + 52 = 89, 82 + 92 = 145, 12 + 42 + 52 = 42, 42 + 22 = 20,
22 + 02 = 4, 42 = 16, 12 + 62 = 37, 32 + 72 = 58,
52 + 82 = 89, . . .
n = 81: 82 + 12 = 65, 62 + 52 = 61, 62 + 12 = 37, 32 + 72 = 58,
52 + 82 = 89, 82 + 92 = 145, 12 + 42 + 52 = 42, 42 + 22 = 20,
22 + 02 = 4, 42 = 16, 12 + 62 = 37, 32 + 72 = 58,
52 + 82 = 89, . . .
n = 82: 782 + 22 = 68, 62 + 82 = 100, 12 + 02 + 02 = 1, 12 = 1
n = 85: 82 + 52 = 89, 82 + 92 = 145, 12 + 42 + 52 = 42,
42 + 22 = 20, 22 + 02 = 4, 42 = 16, 12 + 62 = 37, 32 + 72 = 58,
52 + 82 = 89, . . .
Now, let’s go back to the original oddity of 1089, where we used digit
reversals in order to generate 1089 from a selected three-digit number. We
assumed that any number we chose would lead us to 1089. How can we
be sure? Well, we could try all possible three-digit numbers to see if it
works. That would be tedious and not particularly elegant. An investigation
of this oddity requires nothing more than some knowledge of elementary
algebra. For the reader who might be curious about this phenomenon, we
will provide an algebraic explanation as to why it “works.”
We shall represent the arbitrarily selected three-digit number, htu, as

100h + 10t + u, where h represents the hundreds digit, t represents the tens
digit, and u represents the units digit.
Let h > u,5 which would be the case either in the number you selected
or the reverse of it.
In the subtraction, u − h < 0; therefore, take 1 from the tens place (of
the minuend) making the units place 10 + u.
Since the tens digits of the two numbers to be subtracted are equal, and 1
was taken from the tens digit of the minuend, then the value of this digit is
10(t − 1). The hundreds digit of the minuend is h − 1 because 1 was taken
away to enable subtraction in the tens place, making the value of the tens
digit 10(t − 1) + 100 = 10(t + 9).
We can now do the first subtraction
100(h − 1) + 10(t + 9) + (u + 10)
−(100u + 10t + h)
100(h − u − 1) + 10(9) + u − h + 10
Reversing the digits of this difference gives us
100(u − h + 10) + 10(9) + (h − u − 1)
Now, adding these last two expressions gives us
100(9) + 10(18) + (10 − 1) = 1089
It is important to stress that algebra enables us to inspect the arithmetic

process, regardless of the number.
Before we leave the number 1089, we should point out to the reader,
who is now so motivated so as to inspect this curious number further, that
there is still another oddity, namely, 332 = 1089 = 652 − 562 , which is
unique among two-digit numbers.
By this time, you must agree that there is a particular beauty in the
number 1089.
The number 1089 also lends itself to another interesting numerical pat-
tern. When we multiplied 1089 by each of the digits from 1 through 9, the
5 A reminder: The symbol > means “greater than” and the symbol < means “less than”.
products have a very curious property: The units digits decreased by 1 each
time, beginning with the number 9; the tens digits decreased by 1 each time
beginning with the number 8; the hundreds digits increase each time start-
ing with 0; and the thousandths digits increased each time again with the
number 1.
Also of note is that the last entry in Figure 2.16 shows that 9801 is
a multiple of its reversal 1089. There is only one number of five distinct
digits, whose multiple is a reversal of the original number. That number is
21,978, since 4 × 21,978 = 87,912, its reverse number. The number 1089
gives rise to similarly structured numbers that provides even more fodder
for entertaining an audience.
1 × 1089 1089
2 × 1089 2178
3 × 1089 3267
4 × 1089 4356
5 × 1089 5445
6 × 1089 6534
7 × 1089 7623
8 × 1089 8712
9 × 1089 9801
Figure 2.16
Remember that 1089 × 9 = 9801, which is the reversal of the original

number. The same property holds for 10989 × 9 = 98901, and similarly,
109989 × 9 = 989901. You should recognize that we altered the original
number, 1089, by inserting a 9 in the middle of the number to get 10989
and extended that by inserting 99 in the middle of the number 1089 to
get 109989. It would be nice to conclude from this that each of the fol-
lowing numbers has the same property: 1099989, 10999989, 109999989,
1099999989, 10999999989, and so on.
An Amazing Phenomenon
As we close this chapter, we present a truly amazing numerical relationship.

This one will clearly baffle you and impress you with a true wonder about
the field of arithmetic. We begin with an equation of numbers equal to 0.
1237892 + 5619452 + 6428642 − 2428682 − 7619432 − 3237872 = 0
Looking at this relationship, there is nothing particularly strange other

than the fact that the numbers are rather large. However, when we delete
the 100-thousands place (the left-most digit) from each number, we get the
following relationship which remains equal to 0.
237892 + 619452 + 428642 − 428682 − 619432 − 237872 = 0
When we repeat this process by deleting the left-most digit of each

number, we are left with another relationship which is again equal to 0.
37892 + 19452 + 28642 − 28682 − 19432 − 37872 = 0
When we continue this process of deleting the leftmost digit in each

case, the result remains 0.
7892 + 9452 + 8642 − 8682 − 9432 − 7872 = 0

892 + 452 + 642 − 682 − 432 − 872 = 0
92 + 52 + 42 − 82 − 32 − 72 = 0
At this point, you may think that the problem was rigged, as well it might
have been. However, we can take the same sequence and repeat the process.
But this time deleting the units digit sequentially from each of the numbers,
and once again noting that the equation is equal to 0.
1237892 + 5619452 + 6428642 − 2428682 − 7619432 − 3237872 = 0

123782 + 561942 + 642862 − 242862 − 761942 − 323782 = 0
12372 + 56192 + 64282 − 24282 − 76192 − 32372 = 0
1232 + 5612 + 6422 − 2422 − 7612 − 3232 = 0
122 + 562 + 642 − 242 − 762 − 322 = 0
12 + 52 + 62 − 22 − 72 − 32 = 0
If you are not yet impressed enough, then you have an opportunity to
consider a bit more wonders by combining the two types of deletions that we
have done above, simultaneously! That is, that we will remove the rightmost
and leftmost digits at the same time, and once again, amazingly, we achieve
a sum of zero with each pair of deletions.
1237892 + 5619452 + 6428642 − 2428682 − 7619432 − 3237872 = 0
23782 + 61942 + 42862 − 42862 − 61942 − 23782 = 0
372 + 192 + 282 − 282 − 192 − 372 = 0
With this wild challenge, we figured to have impressed you sufficiently
and can move on beyond arithmetic to the other areas of mathematics, which
should also provide a vast variety of entertainment.
Chapter 3
Challenging Problems with Surprising Solutions
To solve problems in everyday life, as in mathematics, there are specific

strategies that we tend to use, and we are often not even aware that we
are using these specific strategies. Yet, mathematics problems dramatically
exhibit some powerful problem-solving techniques. One example of a sim-
ple mathematics problem in everyday is a simple decision-making situa-
tion; we would say “in the worst-case scenario such and such would be
the case,” which is, in effect, saying that we are looking for an extreme
situation to help solve a problem or concern. The technique of considering
extremes is also very enlightening in solving some mathematics problems.
Another example is that when traveling from point A to point B, we typ-
ically plan a route by determining the roads emanating from the endpoint
B and working our path backwards to starting point A. The technique of
working backwards to solve problems in mathematics can have some favor-
able and dramatic effects. Essentially, as we just said, mathematics helps
demonstrate the power of specific problem-solving techniques. This chapter
presents selected problems that demonstrate how solving problems can be
made much simpler by using some of these techniques.
113
Logical Thinking
Problem: How can you write the numbers 1–100 in a sequence, where no
two consecutive numbers have a sum of less than 50?
Solution: Starting with 50 or 49 rather than 51 is also possi-

ble. Some logical thinking will lead to the following sequence:
51, 1, 52, 2, 53, 3, 54, 4, 55, 5, . . . , 47, 98, 48, 99, 49, 100, 50.
Logical Thinking
Problem: Show how you can remove only four matchsticks in the square
arrangement in Figure 3.1 so that the remaining figure shows only five
identically sized squares.
Figure 3.1
Solution: In Figure 3.2, we show the four matchsticks removed leaving

exactly five identically sized squares.
Challenging Problems with Surprising Solutions 115
Figure 3.2
Thinking Out of the Box
Problem: Suppose you are presented with a collection of toothpicks

arranged, as shown in Figure 3.3, where each of the two outside rows and
two outside columns contains 11 toothpicks.
Figure 3.3
The problem posed is to remove one toothpick from each outside row
and from each outside column and still end up with 11 toothpicks in each of
these rows and columns. This seems to be impossible, since we are actually
removing toothpicks, and yet we are asked to keep the same number of
toothpicks in each row and column, as before.
Solution: In Figure 3.4, we see that we have taken a toothpick from the
center portion of each of the rows and each of the columns and then placed
another of these center toothpicks in the corner position so that they could
be counted twice.
Figure 3.4
Thus, we have achieved our goal of having 11 toothpicks in each of the

two rows and each of the two columns. This is a skill that merits attention
so that one can go through life by analyzing situations in a more critical
fashion.
Problem: Consider a regular hexagon formed by 12 matchsticks as shown

in Figure 3.5. Show how four matchsticks can be removed and placed else-
where so that the result will be three equilateral triangles instead of the six
equilateral triangles shown in Figure 3.5.
Figure 3.5
Solution: In Figure 3.6, we show with dashed lines the matchsticks that
are to be removed and placed accordingly so that the result will be three
equilateral triangles.
Figure 3.6
Problem: Here, you are given a lattice of nine dots as shown in Figure 3.7
and are asked to draw four straight connected lines to touch all the nine
dots, without lifting the pencil from the paper.
Figure 3.7
Solution: The typical attempts are to connect the dots along the sides and
then note that the four lines have not included the center dot. The trick
here is to realize that you have to “think out of the box.” In other words,
you do not have to be restricted to stay on or within the square formed by
the nine dots. In Figure 3.8, we show one possible solution to this trick
question.
Figure 3.8
Further Thinking Out of the Box
Problem: In Figure 3.9, there are six dots. The challenge is to move one
dot to another position so that there will be four rows of dots with three dots
in each row.
6 2
5
4
Figure 3.9
Solution: When dot number 6 is moved to the position shown in Figure 3.10,
the result is that there are four rows of dots with three dots in each row.
1
2
6
5
4
Figure 3.10
Logical and Organized Thinking
Problem: You are given ten coins and asked to place them in such a way
that they form five straight lines, each of which contains four coins. There
is more than one solution.
Solution: In Figure 3.11, we offer two possible solutions. Note in the dia-
grams that each of the five lines contains exactly four dots representing the
coins. An ambitious reader might seek other solutions.
Figure 3.11
Adopting a Different Point of View
Problem: How many numbers in the set {100, 101, 102, 103, . . . , 999} do
not include the digits 3, 5, 7, and 8?
Solution: Rather than to list all the numbers and then identify those that do
not have these four digits included, it would be wise to look at the problem
from a different point of view. The numbers that we are considering all have
three digits. The first digit cannot include zero and so there are five ways
to fill the first digit. There are six ways to fill the second and third digits.
Therefore, the number of three-digit numbers that fit our requirement is
5 × 6 × 6 = 180.
Finding a Pattern
Problem: What is the units digit of the number 3999 ?

Solution: To find the answer using a calculator would probably exceed
the capability of most common calculators. Therefore, one clever approach
would be to seek a pattern. To do this we will inspect the first 12 powers of
3 as follows:
Values of n Values of 3n
1 3
2 9
3 27
4 81
5 243
6 729
7 2,187
8 6,561
9 19,683
10 59,049
11 177,147
12 531,441
13 1,594,323
14 4,782,969
15 14,348,907
16 43,046,721
As we inspect these powers of 3, starting with 31 , we find that there

are repetitions of the units digits every 4 powers (3 → 9 → 7 → 1 →
3 → . . .). This is clear because the units digit of 3x depends only on the
units digit of x. So that as we consider the number 3999 , we realize that the
value of 3996 has a units digit of 1, since 996 is divisible by 4, and, therefore,
the value of 3999 , which is 3 powers further along, must have a units digit
of 7.
Logical Reasoning
Problem: On a grid of square spaces, such as that shown in Figure 3.12, an

equilateral triangle is to be placed containing dots on its sides so that it has
maximum area while at no time containing any of the dots in its interior.
How can this equilateral triangle be placed?
Figure 3.12
Solution: The largest equilateral triangle that can be drawn is one which
would include the most dots on its sides, with no dots in its interior, as
shown in Figure 3.13. That means that the two sides AB and AC will
contain two dots, points D and E, thus, forming equilateral triangle ADE,
the sides of which when extended will allow the third side of the triangle
to contain points F and G. This produces the largest triangle ABC, which
does not have any of the dots in its interior. The logic used here was to
maximize the area by maximizing the number of dots on the sides of the
triangle.
D E
B F G C
Figure 3.13
Logical Reasoning
Problem: Consider the sequence of numbers 6, 12, 24, 48, 96, 192,
384, . . . , where the difference between consecutive numbers doubles each
time. Will there ever be a perfect square among these numbers?
Solution: With logical reasoning, we will realize that a perfect square must
have factors in pairs, such as the number 36, whose factors are 2 ×2 ×3×3.
Each of the numbers in our given sequence has only one factor of 3 (and
can be written as 3 × 2n ), which will never have a partner number 3, and
therefore, there will never be a perfect square in this sequence.
Logical Reasoning
Problem: One of three sources should empty a swimming pool. Which of

these three options for emptying the pool is most effective?
Option 1: Three pipes each with a diameter of 20 inches
Option 2: Two pipes each with a diameter of 30 inches
Option 3: One pipe with a diameter of 60 inches
Solution: At first sight, it would seem that there is no difference because

the diameter sums in each case are 60 inches. However, logical reasoning
tells us that the diameter is not the significant factor here, it is the area of the
circular cross-section of the pipe that is important. For Option 1, each of the
three pipes has a radius of 10 inches; therefore, each pipe has a cross-section
area of 100π for a total of the cross-sections of the three pipes being 300π .
For Option 2, each of the pipes has a radius of 15 inches and, therefore, a
cross-section area of 225π for a total cross-section area of 450π . However,
Option 3, which has a radius of 30 inches, therefore, has a cross-section area
of 900π . Hence, the largest single pipe is better than the smaller collection
of pipes.
Making a Visual Representation
Problem: Two proofreaders are reading a manuscript in which one proof-

reader found 94 errors and the second proofreader found 75 errors. It turns
out that 50 errors were found by both proofreaders. How many errors did
the manuscript contain?
Solution: There are number of ways to approach this problem, however, a
clever technique would be to use a Venn diagram, as we show in Figure 3.14.
Since there were 50 errors found by both proofreaders, that would be placed
in the intersection of the two circles. One circle, therefore, has 94 errors
and the other circle has 75 errors. When we add the three regions, we have
44 + 50 + 25 = 119 errors.
44
25
50
Figure 3.14
Organizing Data
Problem: There are two identical bags shown in Figure 3.15. Bag A has 2
white balls and 3 black balls, and Bag B has 3 white balls and 4 black balls.
One bag is chosen at random, and a single ball is removed from it. It was a
white ball. What is the probability that the bag chosen was Bag A?
Solution: To organize the data we need to note, as shown in Figure 3.15,
that there are 5 balls in Bag A and 7 balls in Bag B. Since the product of
the number of balls is 5 × 7 = 35, we will imagine having selected 35 balls
from each of the two bags. From Bag A, a white ball will be selected 2 out
of 5 times or 14 out of 35 tries. From Bag B, a white ball will be chosen 3
out of 7 times, or 15 out of 35 tries. In total, a white ball will be selected
14 + 15 = 29 times.
14
Therefore, the probability that it was selected from Bag A is 29
.
Bag-B
Bag-A
2 white balls 3 white balls
3 black balls 4 black balls
Figure 3.15
Organizing Data
Problem: How many squares are contained in Figure 3.16?
Figure 3.16
Solution: The big mistake would be to just search through the figure and
identify as many squares as one can without repeating. However, by organiz-
ing data and selecting squares of different sizes and orientations, the process
becomes quite simple. We can begin with the obvious complete square and
then count the small squares in the interior according to organization and
orientation (see Figure 3.17).
Number Location Shape
12
Figure 3.17
Therefore, there are 1 + 9 + 4 + 5 + 12 = 31 squares in Figure 3.17.

Logical Thinking
Problem: Consider 10 stacks of coins where 9 of the stacks consist of coins

of 1-ounce weight and one stack has coins of 2-ounce weight. With one
weighing on a typical homestyle scale, how can you determine which stack
of coins has the 2-ounce coins?
Solution: The clever way to approach this problem is to find a way to
distinguish the stacks. This can be done by taking 1 coin from the first
stack, 2 coins from the second stack, 3 coins from the third stack, and so on
until you get to the tenth stack where you will have selected 10 coins. If all
the coins were the same weight, which they are not, then the total weight
would be 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10 = 55. However, when you
find that the weight is 59, then it shows that 4 extra ounces were added by
one of the stacks, which, in this case, had to be the fourth stack since that
would have added 4 extra ounces.
Pattern Recognition
Problem: Consider the following sequence, which is a geometric sequence

of multiples of 3:
1, 3, 9, 27, 81, 243, 729, 2187, 6561, 19683, 59049, 177147, 531441
The challenge here is to determine how to get the sum of the sequence
without the normal addition process.
Solution: If we search for a pattern, we will begin with the smallest numbers
and note that each number is three times as large as its predecessor. However,
in the absence of the formula presented in high school for a geometric series,
we can search for a pattern. Suppose we take the sum of the first three
numbers 1 + 3 + 9 = 13, which we can also reach by 9 × 3 − 1 = 26, and
then 26 ÷ 2 = 13. To see if this pattern works, we can try it by taking the
sum of the first four numbers, 1 + 3 + 9 + 27 = 40, which could also be
reached by multiplying 27×3− 1 = 80, then 80 ÷ 2 = 40. Now that we see
a pattern, we can apply this procedure, which appears to determine the sum
of a geometric series by applying it to the original challenge. Therefore, we
begin by taking 531,441×3− 1 = 1,594,322, which we then divide by 2 to
get 797,161, which is the sum we originally sought. We see that discovering
a pattern can be a very useful technique, although this approach is not a proof
that the recognized pattern is valid in the general case.
Pattern Recognition
Problem: The challenge here is to determine the sum of the digits of all the
numbers from 1 to 1,000,000.
Solution: One way to answer this question is to begin by summing the digits
of consecutive numbers starting with 1 (see Figure 3.18).
The number The Digit Sum

1 1
2 2
3 3
... ...
35 3+5=8
36 3+6=9
37 3 + 7 = 10
38 3 + 8 = 11
etc.
Figure 3.18
This does not seem to be a very efficient, or elegant, way to proceed.

To determine this sum, rather than simply trying to add the digits of all the
numbers, we will use a better arrangement of the numbers and see if we can
obtain the sum without actually doing all this tedious addition.
Let’s consider the list of the numbers from 0 to 999,999 in two directions:
ascending order and descending order (see Figure 3.19).
We will leave the number 1,000,000 till a bit later. Since every pair
of numbers has a digit sum of 54, then the million pairs have a digit sum of
54 × 1,000,000. The two columns have the same numbers, so the sum of
the digits in one column is 27 × 1,000,000. We must now add the digit sum
of the last number, 1,000,000, which is 1.
Ascending order Descending order Digit Sum of Each Pair of Numbers

0 999 999 54
1 999 998 54
2 999 997 54
3 999 996 54
... ... ...
127 999 872 54
... ... ...
257 894 742 105 54
... ... ...
999 997 2 54
999 998 1 54
999 999 0 54
Figure 3.19
Therefore, the digit sum of all numbers from 1 to 1,000,000 is

1,000,000 × 27 + 1 = 27,000,001.
From this example, you can see that arithmetic is more than just doing
the basic operations of addition, subtraction, multiplication, and division —
it requires a bit of thinking as well.
Alternate solution
Let us determine how often the digit 1 is written in the numbers from
0 to 999,999. In every block of ten numbers, namely, 0−9, 10−19, . . .
999,990−999,999, the digit 1 is written once as a units digit. Since there
are 100,000 such blocks of ten numbers, the digit 1 appears 100,000 times as
units digit. It appears as a tens-digit also 10 times in every block of hundreds,
such as 0−99, 100−199, . . . , 999,900−999,999, which yields the digit 1
another 100,000 times. Furthermore, the digit 1 appears another 100 times
as a hundreds digit in every block of thousands. This pattern continues for
each other power of 10 block, until we reach 100,000. Therefore, since we
have 6 such groups of powers of 10, we have the digit 1 appearing 600,000
times. The same holds true for the other digits 2 through 9. So the digit sum
is (1 + 2 + · · · + 9) × 600,000 = 27,000,000. Finally, the last digit 1 from
1,000,000 yields 27,000,001.
Problem: Of the many strategies that are available to us to solve mathe-

matical problems, the one that allows us to avoid “running into the wall,”
namely, avoiding frustration, is that of approaching the problem from a dif-

ferent point of view. We present here a problem that clearly illustrates how
the most common method leads to a correct answer, but it is cumbersome,
and more prone to a possible arithmetics error. However, there is a dramati-
cally different, but far simpler, solution method. The problem is as follows:
At a school with 25 classes, each of these classes has a basketball team to
compete in a schoolwide tournament. In this tournament, a team that loses
one game is immediately eliminated. The school only has one gymnasium,
and the principal of the school would like to know how many games will
be played in this gymnasium in order to get a winner.
The typical solution to this problem could be to simulate the actual
tournament by beginning with 12 randomly selected teams playing against
a second group of 12 teams, with one team drawing a bye — that is, passing
up a game. This would then continue with the winning teams playing against
each other as shown here:
Any 12 teams vs. any other 12 teams, which leaves 12 winning teams
in the tournament.
6 winners vs. 6 other winners, which leaves 6 winning teams in the
tournament.
3 winners vs. 3 other winners, which leaves 3 winning teams in the
tournament.
3 winners + 1 team (which drew a bye) = 4 teams.
2 remaining teams vs. 2 remaining teams, which leaves 2 winning
teams in the tournament.
1 team vs. 1 team to get a champion!
Now counting (summing up) the number of games that have been played
(Figure 3.20), we get that the total number of games played is 12 + 6 + 3 +
2 + 1 = 24.
Teams playing Games played Winners
24 12 12
12 6 6
6 3 3
3+ 1 bye = 4 2 2
2 1 1
Figure 3.20
This seems like a perfectly reasonable method of solution and is certainly

a correct one. Approaching this problem from a different point of view
would be vastly easier by considering the losers, rather than winners, which
is what we did in the previous solution. In that case, we ask ourselves, how
many losers must there have been in this competition in order to get one
champion? Clearly, there had to be 24 losers. To get 24 losers, there needed
to be 24 games played. And with that the problem is solved. Looking at
the problem from an alternative point of view is a curious and often clever
approach that can be useful in a variety of contexts.
Another alternative point of view would be to consider these 25 teams
with one of them — only for our purposes — being considered a professional
basketball team that would be guaranteed to win the tournament. Each of
the remaining 24 teams would be playing the professional team only to
lose. Once again, we see that 24 games are required to get a champion.
This should demonstrate the power of this problem-solving technique of
considering the problem from another point of view.
Solving a Simpler Analogous Problem
Problem: Our challenge here is to determine the relationship between the

lengths of the four segments, a, b, c, and d, joining the randomly selected
point E in rectangle ABCD, which is shown in Figure 3.21.
A D
a c
E
d b
B C
Figure 3.21
Solution: At first view, one might be a bit confused as how to approach

this problem. If the point E were in the center of the rectangle, then the
problem would be rather simple since the four line segments would be of
equal length. Furthermore, if the point E were at the center of the rectangle,
the Pythagorean theorem would also be helpful. However, we can invoke

the Pythagorean theorem by drawing a horizontal and vertical line through
E to form four rectangles, as shown in Figure 3.22. Each of these rectangles
has a diagonal which is the hypotenuse of right triangles.
A u F v D
a c x
x
E
G K
d b y
y
B u v C
H
Figure 3.22
Therefore, we will apply the Pythagorean theorem for each of our four
initial segments: a 2 = x 2 +u 2 , b2 = y 2 +v 2 , c2 = x 2 + v 2 , and d 2 = y 2 +u 2 .

A quick overview tells us that since a 2 + b2 = x 2 + u 2 + y 2 + v 2 and

c2 + d 2 = x 2 + v 2 + y 2 + u 2 , we can clearly see that a 2 + b2 = c2 + d 2 ,
which shows how adopting a different point of view leads us to a successful
response.
Solving a Similar Analogous Problem
Problem: Here we are asked to find the center of gravity of the shape
shown in Figure 3.23. In other words, we need to find the point in this shape
consisting of rectangles where it could balance on a pin.
12
8
18
Figure 3.23
Solution: The balancing point of various geometric figures should be

known. Perhaps it is intuitive that a rectangle would be balanced at the point
of intersection of the diagonals. This figure is comprised of two equal-area
rectangles as we can see in Figure 3.24. We draw the diagonals of each of
the two rectangles, and since they are equal in area (48 square units), we join
their two centers of gravity and at the midpoint, M, of the line segment AB,
and we have located the required point of balance for the original figure.
12 8
B
4 M
A
18
Figure 3.24
An extension: This technique (just taking the midpoint of A and B) would

not be possible if the rectangles in question were not of equal area. One
would have to find a point dividing the line segment AB in the ratio ba,
if the areas of the left and right rectangles were a and b. But in such a
case, there is still an easier solution: We locate the centroids1 of two pairs
of rectangles and note that the centroid G must lie on both line segments
connecting the partial centroids. Therefore, as we show in Figure 3.25, the
pairs of centroids of the two pairs of rectangles are G 1 , G 2 and G 3 , G 4 and
their intersection point is G. It is possible that G could also lie in the exterior
of the L-shaped region, but this would obviate the pin balancing.
1 The centroid is the center of gravity or balancing point of a plane figure.

G4
G2
G1
G3
Figure 3.25
Solving a Simpler Analogous Problem
Problem: In Figure 3.26, the ladder is shown with equally spaced steps.
The top step has length 20 and the bottom step has length 90. We are asked
to find the length of the step marked EF.
A 20 D
E
? F
90
B C
Figure 3.26
Solution: At first glance, we note that quadrilateral ABCD is a trapezoid.

A simpler analogous problem is to find the length of the median, which in
this case is EF and is equal to one-half the sum of the bases. Therefore,
EF = 12 (20 + 90) = 55. We should not let ourselves be distracted by
the other steps, rather use them to determine that EF is a median of the
trapezoid.
When an equilateral triangle is drawn on each side of square ABCD, as

shown in Figure 3.27, and the remote vertices, E, F, G, and H , of the
equilateral triangles are joined, a quadrilateral EFGH is formed. If the sides
of square ABCD are each 4 units long, our challenge is to find the area of
quadrilateral EFGH.
A D
F H
B C
Figure 3.27
Solution: Typically, one is tempted to find the sum of the areas of the
equilateral triangles and of the square and the four congruent isosceles tri-
angles, AEF, BFG, CGH, and HDE. However, adopting a different point
of view would have us consider drawing FH, intersecting AB and CD at
points K and L, respectively. Draw EM, where M is the midpoint of KL,
and EM intersects AD at point N . Using the Pythagorean theorem (or
recognizing that there are 30◦ , 60◦ , and 90◦ triangles), we get segments
of lengths noted in Figure 3.28. It is now rather trivial to find the area
ofthe quadrilateral EFGH by finding the area of triangle EFH, which is
√ √ √ √
1
2
2 3 + 4 + 2 3 2 + 2 3 = 8 3+16. Thus, the area of the quadri-
√
lateral EFGH is double the area of triangle EFH or 16 3 + 32.
2 3
A N D
4 2 2 2 4
2 3 2 3
F
K H
M L
4
B C
Figure 3.28
Working Backwards
Problem: Charlie has an 11-liter can and a 5-liter can. Charlie’s problem is
to determine how he can measure exactly 7 liters of water using these cans.
A Common Solution
Most people will simply guess at the answer and keep “pouring” back and
forth in an attempt to arrive at the correct answer, a sort of “unintelligent”
guessing and testing.
A cleverer solution
The problem can be solved in a more organized manner by using the strategy
of working backwards. Charlie needs to end up with 7 liters in the 11-liter
can, leaving a total of 4 empty liters in the can. But how can Charlie capture
4 empty liters? (see Figure 3.29).
Figure 3.29
To obtain 4 liters, he must leave 1 liter in the 5-liter can. Now, how can he
obtain 1 liter in the 5-liter can? He fills the 11-liter can and pours from it
twice into the 5-liter can, then discarding the water. This leaves 1 liter in
the 11-liter can (see Figure 3.30).
Figure 3.30
He then pours the 1 liter into the 5-liter can. Then he fills the 11-liter can and
pours off the 4 liters needed to fill the 5-liter can. This leaves the required
7 liters in the 11-liter can, which is what he wanted to originally have (see
Figure 3.31).
Figure 3.31
Note that problems of this sort do not always have a solution. That is, if you
wish to construct additional problems of this sort, you must bear in mind
that a solution only exists when the difference of multiples of the capacities
of the two given cans can be made equal to the desired quantity. In this
problem, 2(11) − 3(5) = 7.
This concept can lead to a discussion of parity. We know that the sum of
two like parities will always be even (i.e., even + even = even and odd +
odd = even), whereas the sum of two unlike parities will always be odd
(odd + even = odd). Thus, if two even quantities are given, they can never
yield an odd quantity. Further discussion can be particularly fruitful as it
provides much-needed insight into some valuable number properties and
concepts.
Working Backwards
Problem: Here we have a seemingly simple challenge: If x = a 2 + b2 , can

2x be the sum of two squares as well?
Solution: We would like to know if 2x = 2a 2 + 2b2 can also be true from
the given value x. In other words, can 2x also be a sum of squares? Working
backwards, we can create the sum of two squares (a − b)2 + (a + b)2 =
2
a − 2ab + b2 + a 2 + 2ab + b2 = 2a 2 + 2b2 . Thus, we have shown
how 2x can be the sum of two squares.
Working Backwards
Problem: Our challenge is to find out if it is possible to find integer values

for a, b, and c, where
√ none of these is a zero or a perfect square so that the
√ √
equation a + b = c can be true.
Solution: Working backwards, we realize that we can find a solution if the
radicals can be converted to be the same under the radical symbol: a “kernel.”
Therefore, we must find values of a, b, and c that
√ will√ generate√“equal radical
√ 8 + √18 =√ 50. All
kernals.” One such is 8, 18, and 50, so that √ these
radicals can be√expressed
√ as multiples
√ of 2 so that 8 + 18 = 50 can
be written as 2 2+3 2 = 5 2. With this technique, we can find infinitely
many such solutions. Let n be not a perfect square (as the common “kernel”
of all the radicals), then take a = a12 n, b = b12 n, and then consequently
c = (a1 + b1 )2 n with arbitrary natural
√ numbers a1 , b1 . This surely yields a
√ √ √ √
solution since then we have a + b = a1 n + b1 n = (a1 + b1 ) n =
√
c.
Working Backwards
Problem: In Figure 3.32, we are asked why the sum of the angles is as
follows:
A + B + C + D = AE D + B FC
Figure 3.32
Solution: By working backwards, we note that we would need to get a

polygon with which we are familiar. That could be a quadrilateral. If we draw
line EF, we will have created two quadrilaterals, namely, ABFE and CDEF.
To make matters simpler, we have noted in Figure 3.33 the four angles at
points E and F. In quadrilateral ABFE, the sum of the angles A + B +
n + m = 360◦ . Yet, along line EF, we get x + m + y + n = 360◦
or m + n = 360◦ − ( x + y). Therefore, by substitution A + B+ =

360◦ − ( n + m) = 360◦ − (360 − ( x + y)) = x + y.
In a similar fashion, working backwards with quadrilateral CDEF, we
get C + D = w + z. Thus, by addition we have A+ B + C + D =
x + y + w + z = AED + BFC.
n y
m z
x Fs
w kE
Figure 3.33
Working Backwards
Problem: Here is a very simple-sounding problem, which typically could

be solved with a good bit of algebra. However, you will be entertained by
a classic example of how working backwards can make the solution trivial.
You are given that the sum of two numbers is 2, and the product of these
same two numbers is 5. The problem is to find the sum of the reciprocals
of these two numbers.
A common approach
The problem immediately suggests forming two equations in two variables:
x+y=2
xy = 5
These two equations can be√solved simultaneously by using the quadratic

formula, which is x = −b± 2ab −4ac , for ax 2 + bx + c = 0. However, the
2
method yields complex values for both x and y, namely, 1 + 2i and 1 − 2i.
Following the requirements of the original problem, we now need to take
the sum of the reciprocals of these two roots.
1 1 (1 − 2i ) + (1 + 2i ) 2
+ = =
1 + 2i 1 − 2i (1 + 2i )(1 − 2i ) 5
We should emphasize here that there is nothing wrong with this method,
it is just not the most elegant way to solve this problem.
A cleverer solution
Before embarking on a problem, it usually makes sense to step back from it
and see what is being required. Curiously, this problem is not asking for the
values of x and y but rather the sum of the reciprocals of these two numbers.
That is, we seek to find 1x + 1y . Using a strategy of working backwards, we
could ask ourselves from what might this sum have come. Adding these
two fractions could give us the answer. Therefore, 1x + 1y = x+y xy
. At this
point, the required answer is immediately available to us since we know the
sum of the numbers is 2, and the product of the numbers is 5, we merely
substitute these values in the last fraction to get 1x + 1y = x+y
xy
= 25 , and our
problem is solved.
Working Backwards
Problem: Another challenge of similar nature is to find the sum of the

squares of two numbers, where the sum of the two numbers is 10 and the
product of these same two numbers is 20.
Solution: As with the previous challenge, the common way to approach this
problem is to find the two numbers and then find the sum of the squares.
However, with the numbers offered here, this would be a rather tedious task.
Yet, this problem can be approached by working backwards. Since we need
not determine the two numbers but rather the sum of their squares, we can
begin by referring to the two numbers as a and b. Working backwards, we
can find the sum of the squares of these two numbers, namely, a 2 + b2 .
By taking the sum of the two numbers and squaring, we get (a + b)2 =
a 2 + b2 + 2ab. This can be written as (a + b)2 − 2ab = a 2 + b2 . Thus, by
working backwards from our desired conclusion, we never needed to find
the actual numbers as we have the sum of squares in terms of the sum of
the two numbers and the product of the two numbers, that is, a 2 + b2 =
(10)2 − 2(20) = 100 − 40 = 60. Working backwards has been a great
timesaver!
Problem: Here is a challenge which directs us to approach the problem

from an unusual point of view. We are given that x 2 + y 2 = 1 and x + y = 1
and we are asked to find the value of x 7 + y 7 .
Solution: Typically, an approach would be to try to multiply these two
equations and hopefully come up with something that is workable. However,
using a different point of view, we can inspect the value of x + y. For
starters, x 2 + y 2 = x 2 + 2x y + y 2 , therefore, 2xy = 0. This implies that

1 =(x+y)2 =1
either x or y must equal 0. Thus, either x 2 or y 2 must equal 1. Consequently,
x 7 + y 7 = 1.
Considering Extreme Cases
Problem: When working with infinity unusual aspects arise, which we will
appreciate in the
.
challenge to find the positive value of x that satisfies the
..
xx
xx
equation: x = 2.
Solution: At first glance, most people would be overwhelmed by the concept
of infinity and not know how to approach the problem. We could look at
this as being somewhat of an extreme situation by noting that there is an
infinite number of x’s in this series or tower of powers. Eliminating one of
the x’s would not have any effect on the end result because of the nature of
infinity. Therefore, by removing the first x, we find that all those remaining
in the tower of x’s must also equal 2. This then permits us to rewrite this
√
equation as x 2 = 2. It then follows that x = ± √ 2. If we remain in the set
of positive real numbers, then the answer is x = 2.
In the following, you can see how the successive increases get ever closer
to 2:
√
2 = 1.414213562 . . .
√ √2
2 = 1.632526919 . . .
√ √2
√ 2
2 = 1.760839555 . . .
√
√ 2
√
√ 2
2
2 = 1.840910869 . . .
√
√
√ 2 2
√
√ 2
2
2 = 1.892712696 . . .
√
√ 2
√
√ 2 2
√
√ 2
2
2 = 1.926999701 . . .
...
And so, we have a surprisingly simple solution to a very complicated-

looking problem. Making sure of the existence of a solution is often crucial.
Here, we were lucky because the assumed. solution exists. But if we wanted
..
yy
to solve with the “same technique” y y = 4, we would get y 4 = 4 ⇒
√ ...
√ √ √ √2 2
y = 4 4 = 2; but we know from above that 2 = 2 (and not 4).
The mistake here is to assume that there is a solution and write y 4 = 4
instead of the original equation. We must be careful in posing problems that
cannot be generalized.
Problem: Maximizing the area of the triangle by drawing a line through

given lines can be an interesting challenge. In Figure 3.34, point P is situated
between lines AB and AC. When a line is drawn through P intersecting AB
and AC, a triangle AEF is created. How might this line be drawn through
point P in order to yield the smallest area for triangle AEF?
C
E
A
P
Figure 3.34
Solution: Considering extreme cases, the line EF could be placed in such

a way that it intersects AC at a very distant position. In Figure 3.34, the
point E could be as far as desired as long as it intersects AB and point P.
This could lead to an extremely large triangle. Therefore, in order to avoid
this extreme situation, we can conjecture that the smallest triangle could be
formed when point P is the midpoint of the line segment XY, as we show
in Figure 3.35. We claim that under these circumstances, triangle AXY has
the smallest area. There is an easy way to prove this. Let RS be another
line through P and XQ RY, then XPQ ∼ = RYP and the area of triangle
ARS using the line PR is larger than the area of triangle AXY (the difference
being the area of the triangle XQS). Therefore, using the line XY produces
the smallest triangle area.
Figure 3.35
Problem: Now having been challenged to find the smallest possible triangle
under given circumstances, our next challenge is to maximize the area of
triangle ABC as we are given three lines AP, BP, and CP, as shown in
Figure 3.36, with lengths AP = 3, BP = 5, and CP = 7. How should these
three line segments be placed so that a triangle ABC has maximum area?
B A
Figure 3.36
Solution: To maximize the area, we need to consider various extreme cases.

Suppose we begin by leaving points A, B, and P in their current posi-
tion for the moment and look to maximize the area of triangle ABC by
moving point C to a maximum position. That would be when CP would
be perpendicular to AB, as shown in Figure 3.37. Any other position for
point C would produce a shorter altitude to line AB and, thereby, a smaller
area.
B A
Figure 3.37
Repeating this procedure, by extending each of BP and AP to be perpen-

dicular to AC and BC, respectively, as shown in Figure 3.38, will produce
the maximum area triangle, as we maximized the length of each of the
altitudes.
B A
Figure 3.38
Problem: In Figure 3.39, we show a square of side length 9 and an isosceles

right triangle whose equal sides have length 12 and is so situated that the
vertex of the right angle is at the center of the square. The problem here is
to find the area of the shaded quadrilateral region.
6 3
12
12 2
9 9
12
Figure 3.39
Solution: The first attempt might be to try to draw various line segments
and then try to find the areas of the square and of the isosceles right triangle.
This is a mere distraction. The technique here is to take an extreme situation
by rotating the triangle about the right-angle vertex until the sides of the
triangle contain the two vertices of the square, as shown in Figure 3.40. It
can be easily shown that the two shaded triangles are congruent, and so the
area of the original quadrilateral is, therefore, equal to the area in the square
formed by the two equal sides of the isosceles right triangle and the side of
the square. That area is exactly 14 of the area of the square. Therefore, the
area of that original quadrilateral is then 14 of the area of the square which
is 14 · 81 = 20 14 .
6 3
12
12 2
9 9
12
Figure 3.40
Problem: Beginning with a randomly drawn triangle ABC, a square is con-

structed on each of its sides. Referring to Figure 3.41, we need to compare
the area of each of the three shaded triangles, ADJ,BGH, and CEF,
to the area of ABC.
E
J
H
B
C
G F
Figure 3.41
Solution: Using one of the area formulas to find the areas of each of these
four triangles would be quite challenging, so we will adopt a different point
of view. Suppose we focus on comparing the areas ABC and BHG. Since
there are two right angles at point B, we know that HBG and ABC are
supplementary. Therefore, if we rotate BHG 90◦ about point B so that BG
coincides with BC as we show in Figure 3.42 solution, we find that ABH
is a straight line, where AB = BH . Thus, areaABC = areaBH C, since
they have equal bases and the same altitude from C to ABH . Therefore, as
we have shown that triangle BHG is equal in area to triangle ABC, we can
use the same procedure to show that each of the shaded triangles has an
area equal to that of triangle ABC.
E
J
H
B G'
C
H'
G F
Figure 3.42
Problem: Comparing triangle sizes often presents challenging problems.

Here we have right triangle AFB placed on square ABCD so that the
hypotenuse is coincident with the side of the square, as shown in Figure 3.43.
From the center of the square, point P, a line is drawn to the right-angle
vertex F. What is the relationship between AFP and BFP?
A D
B C
Figure 3.43
Solution: Since point P is the point of intersection of the diagonals, we

draw them and find that APB is also a right angle. Looking at this problem
from another point of view, we focus on quadrilateral APBF, which has
supplementary opposite angles at vertices P and F. This makes it a cyclic
quadrilateral and allows us to circumscribe a circle about it, as we show in
Figure 3.44. Since AP = BP, we also have arc AP = arc BP, and therefore,
AFP = BFP, which answers our question.
A D
B C
Figure 3.44
Discovering a Pattern
Problem: Here we are in search of a point P inside triangle ABC, shown in

Figure 3.45, that will allow the triangles APB, BPC, and CPA to have
equal areas.
C
B
Figure 3.45
Solution: One needs to determine how a triangle can be divided into

two equal triangles. This can be done by drawing the median, where in
Figure 3.46 triangles ABM and ACM have equal bases, BM = CM, and the
same altitude AD so that their areas are equal.
B C
D M
Figure 3.46
Having established that the median of a triangle divides the triangle

into two equal areas, we can apply this as follows (see Figure 3.47):
since area (ABM) = area(ACM), and in triangle BPC, area (BPM) =
area(CPM), which by subtraction indicates that area (APB) =
area(APC). This can be repeated at any other portion of triangle ABC
so that we will find area (APB) = area(APC) = area(BPC). Thus,
the point P that partitions the original triangle into three equal-area triangles
is the point of intersection of the medians and is known as the centroid of the
triangle.
N R
B C
M
Figure 3.47
Working Backwards
Problem: A circle O is drawn so that it is tangent to sides BC and ED of a

regular pentagon ABCDE, as shown in Figure 3.48. The challenge is to find
the measure of arc BE of the circle with center O.
B E
C D
Figure 3.48
Solution: Working backwards, it would be good to be able to determine the

measure of angle BOE, shown in Figure 3.49, as that would determine the
measure of arc BE. At this point, the only angle measure we have available
is the measure of any of the interior pentagon angles which are each equal
to 108◦ . Since the radii to the point of tangency are perpendicular to the
tangents, we have OE⊥DE.
Line AO extended is perpendicular to CD at point F and also bisects
angle ABE so that EAO = 12 108◦ = 54◦ . Furthermore, AEO =
108◦ −90◦ = 18◦ . Therefore, the exterior angle of AOE, namely, EOF =
54◦ + 18◦ = 72◦ and arc BME = 2 × 72◦ = 144◦ .
54°
O
72°
18°
B E
C F D
Figure 3.49
Perhaps, an easier way to get the angle measure of EOF is to note that
quadrilateral EOFD, whose opposite angles at vertices E and F are right
angles, is a cyclic quadrilateral and, therefore, since EDF = 108◦ , then
its supplement EOF = 72◦ . And thus, EOB = 144◦ , which is then the
measure of arc BME.
Intelligent Guessing and Testing
Problem: We are given the triangle shown in Figure 3.50 where the sides
are marked with three colors, which in clockwise order are red, blue, and
yellow. Here we are asked to determine if it is possible to use this triangle

in whatever position to form the pyramid also shown in Figure 3.50 so that
the colors remain in their relative positions.
Yellow Red
Blue
Figure 3.50
Solution: We show in Figure 3.51 the solution. It requires intelligent testing,

specifically regarding the face in the rear, which has to be viewed from the
back view in order for it to be correct and the triangle at the base needs to
be seen in the reverse as well.
Yellow Blue
Red
Red
Yellow
Blue
Figure 3.51
Problem: Here we are faced with the problem of dividing a circle into four
equal parts. To do this symmetrically provides lots of opportunities such as
those shown in Figure 3.52. However, to find a way to partition a circle into
four equal parts which are not symmetric is the challenge here.
Figure 3.52
Solution: Draw semicircles on the diameter of the circle with diameters

, , and 34 of the large circle’s diameter so that we get a partitioning, as is
1 1
4 2
shown in Figure 3.53, where each region is one-quarter of the area of the
original circle. Unlike the partitioning in Figure 3.52, these are not entirely
symmetric. They are partly symmetric, namely, the outer two regions are
symmetric to each other, and the two inner regions are also symmetric.
Figure 3.53
Suppose that the diameter of the big circle is 8. Then the area of one of

the two symmetric inner regions is π2 22 − 11 + 32 − 22 = 4π , and the

area of one of the two symmetric outer regions is π2 12 + 42 − 32 = 4π ,
thus, all four regions have the same area.
Organizing Data
Problem: Suppose Lisa and David have two pots of coins. Lisa says to
David: “If you give me one coin, I will have twice as many coins as you
have.” David says to Lisa: “If you give me one coin, I will have as many
coins as you have.” How many coins does each of them have?
Solution: This appears as an elementary problem, especially when com-

pared to others offered here. However, this problem is being included to
demonstrate how using very elementary algebra can make a possibly con-
fusing problem trivial. To simplify matters, we will let x equal the number
of coins that David has, and let y equal the number of coins that Lisa has.
Therefore, in the first case, we have x − 1 = y + 1. And in the second case,
we have x + 1 = 2(y − 1). When we solve these equations simultaneously,
we get y = 5 and x = 7. We can check this by noting that when David
gives one coin to Lisa, they will both have 6 coins and when Lisa gives one
coin to David, he will have twice as many coins as Lisa.
Problem: Here is a challenge from an early nineteenth century British

geometry book that can be solved very easily with the proper technique. We
are given parallelograms ABCD and APQR, with point P on side BC and
point D on side RQ, as shown in Figure 3.54. If the area of parallelogram
ABCD is 18, what is the area of parallelogram APQR?
B P C
A
D
Figure 3.54
A common solution
This is by no means an easy problem to solve. First attempts to solve the
problem would be to look for congruent relationships that would lead to
equal areas. This method will lead nowhere. A clever method, although
rather “off the beaten path,” is to draw the line segment PD, as shown in
Figure 3.55.
B P C
A
D
Figure 3.55
Then note that triangle APD can be shown to be one-half the area of
each of the two triangles, since in each case it shares a base with each of
the parallelograms as well as the related height. Although this is a rather
clever approach to a challenging problem, there is yet an even shorter way
to approach this problem, but this approach does not explain/prove that the
position of P is indeed not relevant; it simply takes this for granted.
A cleverer solution
When the problem was posed, we were merely told that point P was on
side BC but not where along the side it was to be placed. We can consider
an extreme case. Therefore, we could have placed P to overlap point B.
Similarly, point D, which was to be placed on side RQ, could just as easily
have been placed to overlap point R. Under these circumstances, this would
certainly fit the original problem’s statement; the two parallelograms would
overlap and consequently would have the same area. Therefore, the area of
parallelogram APQR is 18.
Problem: In Figure 3.56, the dimensions of rectangle AEFK are given as

width AK = 8, while the length AE is divided into four segments such that
AB = 1, BC = 6, CD = 4, and DE = 2. This problem will allow us to
appreciate the cleverness of an alternate way to determine what the area of
the four shaded triangles is.
K J I H G F
A 1 B 6 C 4 D 2 E
Figure 3.56
A common approach
The obvious approach is to find the area of each of the four triangles and
get their sum. In all four cases, the altitude of the triangle equals the length
of AK = 8. Therefore, the areas of the four triangles are as follows:
1
ABJ =×1×8 =4
2
1
BCI = × 6 × 8 = 24
2
1
CDH = × 4 × 8 = 16
2
1
DEG = × 2 × 8 = 8
2
The sum of these areas is 4 + 24 + 16 + 8 = 52 square units.
A cleverer solution
We can make use of our strategy of adopting a different point of view to
solve this problem. The triangles each have the same altitude, namely, 8.
The sum of the bases of the four triangles equals the length of the longer
side of the rectangle, which is 13. Thus, the area of the four shaded triangles
is half the area of the rectangle or 12 × 13 × 8 = 52.
Working Backwards
Problem: A game can also demonstrate the power of mathematics. Max,

Sam, and Jack are playing an unusual card game. In this game, when a player
loses, he gives each of the other players as much money as they have. Max
loses the first game and gives Sam and Jack as much money as they each
have. Sam loses the second game and gives Max and Jack as much money
as they each have. Jack loses the third game and gives Max and Sam as
much money as they each have. They then decide to stop playing and each
has exactly $8.00. How much money did each of them start with?
A common approach
The problem suggests we set up a series of equations designed to represent
each game. We shall begin by representing the starting money for each as
follows:
Max starts with x, Sam starts with y, and Jack starts with z.
Game # Max Sam Jack

1 x−y−z 2y 2z
2 2x − 2y − 2z 3y − x − z 4z
3 4x − 4y − 4z 6y − 2x − 2z 7z − x − y
From the last transaction, we find that each of these values is 8. This
gives us the following three equations in three variables:
4x − 4y − 4z = 8 or x − y − z = 2
−2x + 6y − 2z = 8 or − x + 3y − z = 4
−x − y + 7z = 8 or − x − y + 7z = 8
When we solve these three equations simultaneously, we obtain x = 13,

y = 7, and z = 4.
A cleverer solution
Note that the problem gave the end situation and asked for the starting
situation. This might give us a clue of a problem that will usually benefit
from a strategy of working backwards. Note, too, that the statement of the
situation shows that the same amount of money (namely, 3 × $8 = $24) is
always “in play.” Working backwards should provide an elegant solution.
Max Sam Jack Total

Game # 3 8 8 8 24
Game # 2 4 4 16 24
Game # 1 2 14 8 24
Start 13 7 4 24
Max started with $13, Sam started with $7, and Jack started with $4, the
same answers as before but found in a more elegant fashion.
Problem: Here is a popular mathematics question: A car is driving along

a highway at a constant speed of 55 miles per hour. The driver notices a
second car, exactly 12 mile behind him. The second car passes the first car
exactly 1 minute later. How fast was the second car traveling, assuming
their speeds are constant?
A common approach
The traditional solution is to set up a series of “Rate × Time = Distance
boxes,” which has been the popular method to approach this sort of problem.
This would be done as follows:
Rate × Time = Distance

1 55
55 60 60
1 x
x 60 60
55 1 x
+ = , and x = 85
60 2 60
The second car was traveling at a speed of 85 miles per hour.
A cleverer solution
An alternate approach would be using the strategy of considering extremes.
We assume that the first car is going extremely slowly, that is, at 0 miles per
hour. Under these conditions, the second car travels 12 mile in one minute
to catch the first car. Thus, the second car would have to travel 30 miles per
hour. When the first car is moving at 0 miles per hour, then the second car is
traveling 30 mph faster than the first car. If, on the other hand, the first car
is traveling at 55 miles per hour, then the second car must be traveling at 85
miles per hour (within the legal limit, of course!). Also, without thinking
of extreme cases, one could argue as follows: With its additional velocity,
the faster car needs 1 minute for the additional distance of 12 mile, thus, the
additional velocity must be 30 mph.
Problem: We have two one-liter bottles. One contains a half liter of red wine
and the other contains a half liter of white wine. We take a tablespoonful
of the red wine and pour it into the white wine bottle and thoroughly mix
the two-colored wines. Then we take a tablespoon of this new mixture (red
wine and white wine) and pour it into the red wine bottle.
Is there more red wine in the white wine bottle, or more white wine in
the red wine bottle?
A common solution
There are several common approaches, where problem solvers attempt to
solve the problem using the given information, namely, the tablespoon. With
some luck and cleverness, a correct solution may evolve, but it will not be
easy and often not convincing.
A cleverer solution
We can see that the size of the spoon does not really matter, since there are
large and small tablespoons: Suppose we use a very large tablespoon, one
that is enormously large and actually can hold half liter of liquid — this
would be an extreme consideration. When we pour the half liter of the red
wine into the white wine bottle, the mixture is then 50% red wine and 50%
white wine. After mixing these two together, we take our half-liter spoon
and take one half quantity of this red wine–white wine mixture and pour it
back into the red wine bottle. The mixture is now the same in both bottles;
so, to answer our question, we can conclude that there is as much red wine
in the white wine bottle as there is white wine in the red wine bottle.
Another way of approaching this problem is as follows: The missing red
wine in the red wine bottle must be in the white wine bottle and vice versa,
and since the volume of the liquid in both bottles is equal (half liter), these
two parts (red wine in the white wine bottle and white wine in the red wine
bottle) must be equal. This is further evidence why the size of the spoon is
not important.
Making a Visual Representation
Problem: Time for a cute problem. If, on the average, a hen and a half can
lay an egg and a half in a day and a half, how many eggs should six hens
lay in eight days?
A common solution
This is an old problem that has survived the test of time. Traditionally, the
problem is solved as follows: Since 32 hens work for 32 days, we may speak

of the job of laying an egg and a half 32 eggs as taking 32 32 or 94
“hen-days.” Similarly, the second job takes 6 × 8 or 48 “hen-days”. Thus,

we form the following proportion:
Let x = the number of eggs laid by 6 hens in 8 days.
9
hen-days 3
eggs
Then 484 hen-days = x2 eggs .

Multiplying the product of the means and extremes, we get 94 (x) =
3
48 2 and x = 32.
A cleverer solution
However, as an alternate solution, we may set up the following visual rep-
resentation (here in the form of a tabular layout) of the situation: 32 hens lay
3
2
eggs in 32 days.
3
Double the number of hens: 3 hens lay 3 eggs in 2
days
Double the number of days: 3 hens lay 6 eggs in 3 days
One-third of the number of days: 3 hens lay 2 eggs in 1 day
Double the number of hens: 6 hens lay 4 eggs in 1 day
Eight times the number of days: 6 hens lay 32 eggs in 8 days
Therefore, 6 hens should lay 32 eggs in 8 days.
Solving a Problem from a Different Point of View
Problem: Consider two trains serving the Pittsburgh to New York route, a
distance of 400 miles, starting toward each other at the same time (along
the same tracks). One train is traveling uniformly at 60 mph and the other
at 40 mph. At the same time, a bee begins to fly from the front of one of
the trains, at a speed of 80 mph, toward the oncoming train. After touching
the front of this second train, the bee immediately — without losing any
time — reverses direction and flies toward the first train (still at the same
speed of 80 mph). The bee continues this back-and-forth flying until the two
trains collide, crushing the bee. The challenge is to determine how many
miles the bee had flown.
Solution: One is naturally drawn to find the individual distances that the
bee traveled. An immediate reaction is to set up an equation based on the
relationship speed ×time = distance. However, this back-and-forth path is

velocity
rather difficult to determine because it requires considerable calculation.
Even then, it is very difficult to solve the problem in this way.
A cleverer solution
A much more elegant approach would be to solve the problem from a
different point of view. We seek the distance the bee traveled. If we knew
the time the bee traveled, we could determine the bee’s distance because
we already know the bee’s speed.
The length of time the bee traveled can be easily calculated because
it traveled the entire time the two trains traveled (until they collided). To
determine the time, t, the trains traveled, we set up an equation as follows:
The distance of the first train is 60t and the distance of the second train
is 40t. The total distance the two trains traveled is 400 miles. Therefore,
60t + 40t = 400 and t = 4, which is also the time the bee traveled. We can
now find the distance the bee traveled, which is (4)(80) = 320 miles.
Problem: Consider the globe of the earth with a rope wrapped tightly around
the equator. The rope will be about 24,900 miles long. We now lengthen the
rope by exactly 1 yard. We position this (now loose) rope around the equator
so that it is uniformly spaced off the globe. The challenge is to determine
if a mouse can fit under the rope (Figure 3.57).
Figure 3.57
Solution: The traditional way to determine the distance between the cir-
cumferences is to find the difference between the radii. Let R be the length
of the radius of the circle formed by the rope (circumference C + 1) and
r the length of the radius of the circle formed by the earth (circumference
C), as shown in Figure 3.58.
O
Earth
Rope
Figure 3.58
The familiar circumference formulas give us the following: C = 2πr,

or r = 2π
C
, and we have C + 1 = 2π R, or R = C+1 2π
. We need to find the
difference of the radii, which is R − r = 2π − 2π = 2π
C+1 C 1
≈ 0.159 yards ≈
1
5.7 inches. That is, there is a space of over 5 2 inches for a mouse to crawl
under. Imagine, by lengthening the 24,900-mile rope by 1 yard, it lifted off
the equator by about 5 12 inches!
A cleverer solution
Now for an even more elegant solution, which will consider the problem by
using an extreme case — without loss of generality.
Keep in mind that the solution to the problem is independent of the

circumference of the earth, since the end result did not include the circum-
1
ference in the calculation. It only required calculating 2π .
Therefore, using an extreme case, we may suppose the inner circle in
Figure 3.58 is very small, so small that it has a zero-length radius, which
essentially says that the circle has transformed into just a point. This is,
obviously, the extreme case. We seek to find the difference between the
radii, R − r = R − 0 = R. Therefore, all we need to find is the length
of the radius of the larger circle and our problem will be solved. With the
circumference of the smaller circle now equal to 0, we apply the formula for
the circumference of the larger circle: C + 1 = 0 + 1 = 2π R, then R = 2π 1
,
1
which is approximately equal to 5 2 inches. Thus, using the extreme case
we solved the problem immediately.
Problem: In Figure 3.59, chord AB of the larger of the two concentric circles
is tangent to the smaller circle at point T . As the length of chord AB = 8,
We are challenged to find the area of the shaded region between the two
circles.
A T B
Figure 3.59
Solution: We know that a radius is perpendicular to a tangent at the point

of contact (T ). Furthermore, a radius perpendicular to a chord divides the
chord into two equal segments. Thus, in Figure 3.60, AT = BT = 4.

We know that the area of the region between the two circles (the doughnut
shape) can be found by obtaining the difference between the areas of the
two circles.
C
R-r
A 4 T 4 B
Figure 3.60
Thus, the area of this region between the two circles equals
π R 2 − πr 2 = π(R 2 − r 2 ). Now, OC = R and OT = r, C T = (R + r) and
DT = (R − r), which can be seen in Figure 3.60. Recall that the product
of the segments of two intersecting chords of a circle is equal. Therefore,
we obtain (R − r)(R + r) = 4 × 4, and then R 2 − r 2 = 16. Thus, the area
of the region between the two circles equals 16π square units.
We could have also solved this problem by drawing line segment OA.
We then created a right triangle ATO, in which (OA)2 = (OT )2 + 42 or
R 2 − r 2 = 16, which again gives us the area as 16π square units.
A cleverer solution
We can also look at this problem by considering an extreme case. Let’s
assume that the smaller circle gets smaller and smaller, until it becomes a
point that coincides with point O. Then, AB becomes a diameter of the
larger circle, and the area of the region between the two circles becomes
the area of the larger circle, which equals π R 2 = 16π.
Chapter 4
Algebra: A Unique Problem-Solving Tool
When thinking of the basics of mathematics, the topic of algebra follows

immediately after arithmetic. Although algebra is often considered a pow-
erful tool to solve problems, there are many problems where the use of
algebra can lead to unnecessarily complicated solutions. Before we exhibit
numerous examples where algebra shows its power, we will consider a few
problems whose solutions are complicated by using algebra techniques. In
this chapter you will experience some of the mistakes people can make with
algebra, which then, of course, lead to absurd results. Your journey through
this chapter should give you a good impression of the power and usefulness
of algebra as well as the issues one should be aware of.
A Problem with Marbles
Sandy has a bag containing marbles. She gives half of them to Bernie and
then one-third of the remaining marbles to Peter. Finally, she has 6 marbles
left. How many marbles were in Sandy’s bag initially?
If one denotes the unknown number of marbles with x and establishes

an equation following the text, the result would be x − x2 − 13 x − x2 = 6.
One could argue that solving this equation leaves one open to making trivial
mistakes. Therefore, by working backwards and making a sketch, such as
a circle diagram as shown in Figure 4.1, it should be not difficult to see that
173
the right half of the circle must have 9 marbles, and thus, the whole circle
(which corresponds to Sandy’s initial bag) must have contained 18 marbles.
Figure 4.1
In the following problem, we have an example where “algebra” is only

one powerful tool among others (here “using illustrative figures”).
Sum of Three or More Consecutive Natural Numbers
Here we can experience algebra as a short and instructive tool for visualizing
the sum of three consecutive natural numbers. We can write this sum as
(n − 1) + n + (n + 1); when simplified, we get 3n, which indicates that such
a sum is always divisible by 3. The same is possible for 5, 7, 9, . . . , 2k + 1,
in fact for every odd number of consecutive natural numbers. Consider the
situation for 7 consecutive numbers:
(n − 3) + (n − 2) + (n − 1) + n + (n + 1) + (n + 2) + (n + 3) = 7n.
Also, the sum of an even number of consecutive natural numbers can be
well understood by using algebra. What can be said about the divisibility
of the sum of an even number of consecutive natural numbers, such as 4 or
6 consecutive numbers?
Sum of 4 consecutive numbers is as follows:
(n − 1) + n + (n + 1) + (n + 2) = 4n + 2 = 2(2n + 1)
Algebra: A Unique Problem-Solving Tool 175
Sum of 6 consecutive numbers is as follows:
(n − 2) + (n − 1) + n + (n + 1) + (n + 2) + (n + 3) = 6n + 3 = 3(2n + 1)
Sum of 2k consecutive numbers is as follows:
(n − (k − 1)) + (n − (k − 2)) + · · · + n + · · · + (n + (k − 1)) + (n + k)

= (2k)n + k = k(2n + 1)
One can see, with the help of algebra, that it is never possible that a sum
of an even number (2k) of consecutive natural numbers is divisible by that
even number 2k, but this sum is always divisible by k. This is a famous,
and rather elementary, example of how algebra can help explain numerical
patterns with minimal effort.
An impressive alternative to explaining these concepts is done by using
illustrative figures.
In case of an odd number 2k + 1 of consecutive natural numbers, there
is always a mean value, say m. Then one can imagine to cut off the part of
numbers greater than m and move this part to the left yielding a rectangle
with a width 2k + 1 and a height m, and it can be seen visibly that such a
sum is divisible by 2k + 1 (see Figure 4.2).
Figure 4.2
In case of an even number 2k of consecutive natural numbers, one can

take the right part and put it onto the left part yielding a rectangle with width
k, which demonstrates visibly that such a sum is always divisible by k (see
Figure 4.3).
Figure 4.3
Now that we’ve experienced a comparison of algebraic solutions to logic

techniques and visual methods, we shall consider problems that are best
approached algebraically.
An Unusual Equation in Search for Specific Natural Numbers
The challenge is to find all pairs (n, m) of natural numbers (including 0)

where n + m = n · m. Furthermore, we need to prove that we have found
all such possible solutions.
Solution: Two solutions seem to arise quickly. Namely, n = 0 = m and

n = 2 = m. But why are there no others? Here is where we can use algebra
to demonstrate two possibilities to prove this result.
(1) The given equation n +m = n ·m is equivalent to 1 = n ·m −n −m +1,

and further this is equivalent to 1 = (n − 1)(m − 1). And therefore,
the only two ways that 1 is a product of two integers are 1 = 1 · 1 and
1 = (−1) · (−1). In the first case, we would have n = 2 = m and in
the second case, we would have n = 0 = m.
(2) For n = 0, we have m = 0, and it is clear that n = 1 can never be
a solution since that would mean 1 + m = m. So, we can assume for
what follows n ≥ 2:
n + m = n · m ⇔ n = m · (n − 1) ⇔ n−1 n
= m ⇔ 1 + n−1 1
= m,
the only way that (for n ≥ 2) n−1 is an integer is n = 2 which yields
1
m = 2.
Sum of Squares Equals Sum of More Squares
Here is one problem that sounds very simple and yet requires a little bit of
thought. Begin by taking the sum of any three squares and multiplying it
by 3. We now need to find four squares that will have the same sum. For
example, 3(22 + 32 + 42 ) = 87 = 92 + 22 + 12 + 12 . Or perhaps as another
example: 3(22 + 32 + 32 ) = 66 = 52 + 42 + 42 + 32 . This can be sometimes
rather frustrating and yet also delightfully challenging when successful. At
this point, we might want to see a justification so that we are not left with
an unsolvable situation. We provide a simple algebraic proof:
3(a 2 + b2 + c2 ) = (a + b + c)2 + (b2 − 2bc + c2 )

+ (c2 − 2ca + a 2 ) + (a 2 − 2ab + b2 )
= (a + b + c)2 + (b − c)2 + (c − a)2 + (a − b)2
Using this relationship will provide other possible solutions.
Arithmagons with Four Vertices
Let us first define an arithmagon, which is a polygon-shaped figure where

there is an encircled number at each vertex and a boxed number on each side
so that the boxed number is the sum of the two circled numbers between
which it is located.
Now it is very easy to find the numbers on the sides (in squares) given
the numbers in the circles (at vertices), but the other way round, it could
be more interesting. That is, given the numbers in the squares on the edges
(a, b, c, d), shown in Figure 4.4, we find possible numbers for the circles
at the vertices (all these numbers should be natural numbers).
(a) Are there solutions in the following four-sided arithmagons shown in

Figure 4.4? If yes, how many? If no, then why?
(b) Give necessary and sufficient conditions for a, b, c, d ∈ N so that there
are solutions! How many such conditions do exist, if any exist at all?
25 30
c
18 d b 20 10 38
a
13 22
Figure 4.4
Solution: For the first arithmagon, we can use trial and error to easily find
a solution. Let us consider the bottom-left vertex (circle) with the value 7
(Figure 4.5). Then at the bottom-right circle, there must be 6 so that the
sum of 13 is on the bottom side. On the top-right vertex, there must be 14
in order to justify the sum of 20 on the right-side square. Finally, on the
top-left vertex, we have 11 so that we can have the sum of 25 on the top
side, and this fits as well to the sum 18 on the left side.
Figure 4.5
Had we started at the bottom-left vertex with the number other than 7,
that is, using any other natural number of {1, 2, . . . , 12}, we also would
have got a solution. Suppose we had taken 3, then we would have arrived
at another solution as shown in Figure 4.6.
Figure 4.6
There must be a solution also with any number from the set {1, 2, . . . , 12}
at the bottom left because as the value of the number at the bottom-left
vertex gets smaller (or larger), the number at the top-left vertex gets bigger
(or smaller) by the same amount. Thus, at the end, it must fit with the same
sum of 18. So, altogether we have 12 solutions, and if we count 0 among the
natural numbers, the number on the bottom-left vertex could be one of these
numbers {0, 1, . . . , 12, 13}, thereby, resulting in 14 solutions. In general,
the arithmagon will have solutions if we let a be the minimum (smallest
value) of {a, b, c, d} then the number of solutions is a−1 or a+1, depending
on whether or not one counts 0 among the natural numbers.
When we try the same technique for the second arithmagon, we fail by
trying 7 again at the bottom-left vertex, as shown in Figure 4.7.
Figure 4.7
At the top-left vertex, on the one hand, we should write 3 to get a sum
of 10 on the left edge, and on the other hand, we should write 7 so as to get
a sum of 30 on the top edge. A conflict arises. But also any other number
7 + x at the bottom-left vertex instead of 7 does not work (see Figure 4.8):
At the top-left vertex, on the one hand, we should write 3 − x to get a sum
of 10 on the left edge, and on the other hand, we should write 7 − x so as
to get a sum of 30 on the top edge, again a contradiction!
Figure 4.8
Our task would be to determine if there is a solution by merely inspecting

a, b, c, d. Since the sum of two natural numbers is at least 2 (not counting 0
as part of the natural numbers), we have each of the numbers a, b, c, d ≥ 2
must hold (a necessary condition) for a solvable four-sided arithmagon. Fur-
thermore, considering Figure 4.9, with a and c, two vertices are summed up.
The sum of the bottom two vertices should be a, and the sum of the top two
vertices should be c. Therefore, for a + c, we have the sum of the four ver-
tices, and the same holds for b + d. Thus, a + c = b + d is another necessary
condition for a four-sided arithmagon to have a solution. And these two con-
ditions together are also sufficient, because in this case one can always find
appropriate numbers, starting with, say, x, at the bottom left vertex, and at the
end at the top left vertex one gets d − x on the one hand, and a +c −b − x on
the other hand, which is the same because of a + c = b + d, as we can see in
Figure 4.9.
Figure 4.9
If one considers 0 to be included in the set of natural numbers, N, then the

condition a, b, c, d ≥ 2 can be omitted for a solvable arithmagon because
in this case, the minimum sum of two natural numbers is 0 and this does
not yield a new condition.
Arithmagons with Three Vertices
There is a substantial difference between four-sided arithmagons and

three-sided arithmagons, which have either no solution or exactly one solu-
tion. In case of a solvable four-sided arithmagon, it is rather easy to find
a solution, but three-sided arithmagons are not so easily solved. We use
algebra for two purposes: First, to see how to arrive at a solution, if one
exists, although, this could also be accomplished without algebra as well;
second, to see under which conditions for a, b, c ∈ N there is a solution
with natural numbers — here algebra is a powerful tool. We begin with the
diagram shown in Figure 4.10.
Figure 4.10
In Figure 4.11, we will begin with x at the bottom-left vertex, then at

the bottom-right vertex there must be c − x and the value on the top must
be on the one hand b − x, and on the other hand a − (c − x) = a − c + x;
these two numbers have to be equal. Therefore, b − x = a − c + x, which
is equivalent to x = b+c−a
2
.
Figure 4.11
Therefore, we have x = b+c−a 2

, which determines x and subsequently
determines the two numbers at the other vertices. Furthermore, we can see
that b + c − a must be an even natural number so that the result will not be a
fraction. That means b + c > a, and analogously a + c > b and a + b > c.
Either all three of a, b, c ∈ N are even or two of them are odd and one is
even. If one includes 0 among the natural numbers, the conditions b+c > a,
a + c > b, and a + b > c change to b + c ≥ a, a + c ≥ b, and a + b ≥ c.
For instance, the left three-sided arithmagon in Figure 4.12 is solvable
in N, as one can find the missing numbers also without algebra. The right
arithmagon is not solvable in N (three odd numbers).
Figure 4.12
Interchanging × and + Without Changing the Value (I)
It is easy to see 6 × 65 = 6 + 65 , since 6×6

5
= 36
5
and 30
5
+ 65 = 36
5
.
Are there other such examples? If so, then we need to find all of them
and prove we have them all.
Solution: Here is an example where algebra becomes essential. The first

guess could be that it works with numbers of the form n × n−1 n
= n + n−1n
with n > 1. Actually, this is true and can be seen by applying algebra to
this equation by multiplying both sides by (n − 1), to get n 2 = n 2 − n + n,
which implies that the equality above is true for every such n. However, the
question still remains: Are there other such examples?
Let us use three different variables. Once again, we begin with
a × bc = a + bc , where from the beginning we have the condition c = 0.
In case of a = 0 = b, the value of c = 0 could be chosen arbitrarily.
The above equation can be written equivalently as (a − 1) × bc = a. Obvi-
ously, for a = 1, this is not possible, since 0 is not equal to 1. For a = 1,
we get bc = a−1a
. Here we see, the reduced version of bc must be a−1 a
, so
essentially there are no other such examples as we had above, namely,
n × n−1
n
= n + n−1 n
.
Interchanging × and + Without Changing the Value (II)
It is easy to see 6 × 1.2 = 6 + 1.2, since both sides are equal to 7.2.
Are there other such examples? If yes, we need to find all of them and
then prove all have been identified.
Solution: This is very similar to the previous problem; the difference is

that in the previous problem, we had (non-negative) fractions, and here we
have real numbers that can be positive or negative. As we embark on the
challenge presented above, we have to solve the equation x · y = x + y and
not be limited to integers. First, we see that x = 1 is impossible because
then y = 1 + y is a contradiction. And for x = 1, we can say x · y =
x=1
x + y ⇔ y · (x − 1) = x ⇔ y = x−1 x
= x−11
+ 1 and, thus, we see that
all points (x, y) lie on the graph of the function y = x−1
1
+ 1 which is a
hyperbola (see Figure 4.13).
Figure 4.13
This is a generalization of the above problem (Interchanging × and +

without changing the value (I)) where we had the restrictions that x was a
non-negative integer, moreover, the variables were n or a instead of x, and
n

we had y as a positive rational number. Note: Above we had n−1 or bc .
A Curiosity with Mixed Numbers

It is easy to see 5 24 = 5 24
5 5
(with a mixed number under the first square
root). This can be equated as follows:

5 24 · 5 + 5 5 · (24 + 1) 125
5 = = = ,
24 24 24 24
√
5
and the right side is 5 24 = 25 · 24 5
= 25·524
= 125
24
.
Is this the only such example of this relationship or are there others? If
there are others, then our challenge is to find all of them for non-negative
numbers.

Solution: Another example to consider would be 6 35 6
= 6 35 6
. Are all

possible examples built like n n2n−1 = n n2n−1 ? Algebra can help us to
have an overview over all possible cases.
Algebraically, we need to solve the equation x yz = x yz , where z is
apositive integer and x, y are non-negative integers. For x = 0, we have
y
z
= 0 with the solution y = 0 and z > 0 an arbitrary positive integer. For

x > 0, when we square both sides of x yz = x yz , we get z = y(x+1)(x−1)
x
,
and x = 1 or else z = 0, which is not acceptable here. Considering the
greatest common divisor, we see gcd(x, x + 1) = 1 = gcd(x, x − 1).
Therefore, the only chance for z ∈ N is for x to be a divisor of y so that
y = kx, (k ∈ N). We, thus, have the following solutions:
(1) x = 0 = y, z > 0 arbitrary

0; x = 1 arbitrary, y = k · x(k ∈ N), z = k · (x 2 − 1)
(2) x =
So essentially, there are no other interesting solutions than what we had

above.
Just extending
the fraction with k is also not a new solution:
n (n2 −1)·k = n (n2 −1)·k . Maybe one may have had the right conjecture
n·k n·k
earlier, but now, with the use of algebra, one knows what is happening.
A Surprise in a Sequence
We begin by taking two arbitrary initial positive numbers. Then with the
following procedure, we create the next number: Add 1 to the second number
and divide the result by the first number and then add 1 to that result and
divide it by the second number. Let us do that with an example by taking
initially the numbers 34 and 2. Then the first step would yield 2+1
3 = 33 = 4,
4 4
the second step would be 4+1 2
= 52 , and if one continues this process, the
sequence will look like this: 34 , 2, 4, 52 , 78 , 34 , 2, . . .

arbitrarily
chosen
Here we can see that the sixth number in the sequence equals the
first number (where we began) and the seventh number equals the sec-
ond number. Take other initial numbers and this will hold true. Once
again, it will be through algebra that we will explain this pheno-
menon.
Solution: One can take any numbers a1 = 0 and a2 = 0 as starting numbers;

we need to show that they will always yield a6 = a1 and a7 = a2 . With
algebra, the proof involves just a few calculations with fractions and is very
a2 +1
a1 +1 a1 +a2 +1
expedient. The calculation for a4 goes like this: a4 = a2
= a1 a2
. The
a1 +a2 +1
a1 a2 +1 a1 +a2 +1+a1 a2 (a1 +1)(a2 +1) a1 +1
one for a5 like this: a5 = a2 +1 = a2 (a2 +1)
= a2 (a2 +1)
= a2
;
a1
a1 +1
a2 +1 a1 +a2 +1 a1 +1
then we get a6 = a1 +a2 +1 = a1 +a2 +1 = a1 and finally a7 = a1 +1 = a2 .
a1 a2 a1 a2
Altogether we can summarize:
a +1 a1 +a2 +1 a1 +1
a1 ,a2 , 2 , a1 a2 , a2 ,
a1 , a2 · · ·
a1

arbitrarily a6 a7
a3 a4 a5
chosen
There is another interesting question for the motivated reader. Is it possible

that the equality an = a1 , an+1 = a2 can hold true for some value of n < 6?
We know from the above that it holds true at the latest for n = 6. Let
us try n = 2: This would mean a2 = a1 , a2a+1 = a2 , and this equation
1
a3
√
system has the only positive solution a1 = a2 = 1+2 5 = ϕ ≈ 1.618, which
happens to be the golden ratio. Explanation: The equation system yields
a1 +1
a1
= a1 , which is equivalent with a12 − a1 − 1 = 0 and solving this with
the well-known
formula for quadratic equations yields the two solutions
a1 = 12 ± 14 + 1; one of them is negative and the other one can be written
as above and is the golden ratio. So, if we start with a1 = a2 = ϕ, we have
a constant sequence.
What about n = 3? This would mean
a

3
a2 +1
a1
= a1
a1 +a2 +1
= a2
a
1 a2
a4
and√ this equation system has, again, the only positive solution a1 = a2 =
1+ 5
2
= ϕ ≈ 1.618, and the same holds for n = 4 and for n = 5. Therefore,
the only way that the equality an = a1 , an+1 = a2 happens
√ for some n < 6
is the constant sequence in the case of a1 = a2 = 1+2 5 = ϕ ≈ 1.618.
Sum of Unit Fractions
In ancient Egypt, aside from the fraction 23 , all measurements were made by
using unit fractions. Therefore, it was important to determine how fractional
measurements can be expressed as the sum of various unit fractions. Our
challenge here is to show that all unit fractions between 0 and 1 of the
form n1 (n ≥ 2) and fractions of the form n2 (n ≥ 3) can be represented as
sums of different unit fractions. (Note: The representation n2 = n1 + n1 is not
acceptable because the unit fractions are not different, as required.)
Solution: One can easily check some small numbers for n: 21 = 13 + 16 ,

1
3
= 14 + 12 , 4 = 16 + 12
1 1 1
= 15 + 20 1
. . .. Sometimes, one will find more
than one such representation as a sum of different unit fractions as we
have shown above with the fraction 14 . Yet, we are challenged to prove
that there is always such a representation available. Suppose we start in
the case of n1 with the largest possible unit fraction as the first summand:
1
n
= n+1 1
+ x. And we are lucky, as x will turn out as a different unit fraction
x = n − n+1
1 1
= n+1−n
n(n+1)
= n(n+1)
1
. (Note: For n ≥ 2, we have n(n+1)
1
< n+1
1
.)
Thus, we have demonstrated what was requested.
In case of n2 (n ≥ 3), we can distinguish two cases. If, on the one hand,
n is an even number, say, n = 2k, then n2 = 1k , and from this we know that
there is such a representation. If on the other hand, n is an odd number, say,

n = 2k − 1, then n2 = 2k−12
.
We use the same technique as above, that is, to try the biggest possible
unit fraction. Here we take 1k , as the first summand, and we are successful
as the remaining summand must then be a unit fraction as well: 2k−1 2
− k1 =
1
k(2k−1)
, and thus, 2k−1 = k + k(2k−1) .
2 1 1
A Number Pyramid with Fractions
In an arbitrarily big number pyramid, as shown in Figure 4.14, the left-

most “bricks” are filled with unit fractions, beginning at the top with 11 . The
interior bricks are filled following the well-known rule:
Each brick is the sum of the two bricks below it.
1
1
1
2
1
3
1
4
1
5
and so on
Figure 4.14
Then one can see two surprising phenomena:
(a) The whole number pyramid contains only unit fractions (after reducing
the fractions).
(b) The number pyramid is symmetric.
Our challenge here is to justify why this is true.
Solution: If one fills the empty bricks above, one gets Figure 4.15.
1
1
1 1
2 2
1 1 1
3 6 3
1 1 1 1
4 12 12 4
1 1 1 1 1
5 20 30 20 5
and so on
Figure 4.15
But such a filled-in number pyramid, even if it were 1,000 times larger,
is not a proof. And for that purpose, we introduce variables, and once again
an algebraic procedure is welcomed.
Let us number the oblique columns running from top left to bottom right
with 1, . . . , n, . . . and the oblique columns running from top right to bottom
left with 1, . . . , m, . . . and denote the numbers in each brick with am,n as
shown in Figure 4.16.
Figure 4.16
Then we know that a1,n = n1 (first given oblique column at the left).
Then we can easily determine a2,n = a1,n − a1,n+1 = n1 − n+1 1
= n(n+1)
1
,
again a unit fraction. Now we come to a3,n = a2,n − a2,n+1 = n(n+1) 1
−
1
(n+1)(n+2)
= n(n+1)(n+2)
n+2−n
= n(n+1)(n+2)
2
, and since the denominator is a prod-
uct of three consecutive natural numbers, it is surely divisible by 2, and
we know that also a3,n is a unit fraction (after reducing it). Then we go to
a4,n = a3,n − a3,n+1 = n(n+1)(n+2)
2
− (n+1)(n+2)(n+3)
2
= n(n+1)(n+2)(n+3)
2(n+3)−2n
=
1×2×3
n(n+1)(n+2)(n+3)
.
With an analogous discussion to that above, we can be sure that the
denominator is divisible by 2 and 3. In the general case, we get with the
same technique:
(m − 1)! (m − 1)!(n − 1)!
am,n = =
(n+m−1)! (n + m − 1)!
(n−1)!
Here we used for abbreviation the factorial notation,

which is n! = n × (n − 1) × (n − 2) × · · · × 2 × 1.
One could produce a proof using mathematical induction for m. This
term is on the one hand symmetric in m and n,which proves the sym-
metry of the number pyramid, and on the other hand it can be written as
am,n = (n+m−1)!
1
= ⎛ 1 ⎞ , which proves that it is a unit fraction. Here
(m−1)!(n−1)! n+m−1 ⎠
n⎝
m−1
we used the well-known notation of binomial coefficients which always
represent natural numbers since they
⎛ ⎞
indicate a number of possibilities. In
n
general, the binomial coefficient ⎝ ⎠ describes the number of possibilities
k
to select k different elements out of n ≥ k different
⎛ ⎞ elements. It is standard
n n!
knowledge of elementary combinatorics that ⎝ ⎠ = k !(n−k) ! , which was
k
⎛ ⎞
10
used above. As an example, there are ⎝ ⎠ = 4!10×6! ! = 10×9×8×7
4×3×2×1
= 210
4
possibilities to choose a team of 4 persons from a group of 10 persons.
Which Is the Larger Fraction?
There are countless curiosities in mathematics that at first sight generate

amazement. Through the use of algebra, we can make some sense out of
these surprising and counterintuitive phenomena. Consider the two frac-

tions:
√ √ √ √
59 59 + 3 3 59 59 − 3 3
(a) √ √ (b) √ √
59 3 + 3 59 59 3 − 3 59
The challenge is to determine which of these two fractions (a) or (b) is

the greater. Furthermore, we need to justify our response, which is where
algebra can be extremely useful.
Solution: The first issue to consider is if the numbers 59 and 3 are critical to
this question. For the first approach,
√ one√ will take a calculator or √
a computer
√
and find the following values: 59 3+3 59 ≈ 3.660205 and 59 3−3 59 ≈
59√59+3√ 3 59√59−3√ 3
5.660205.
Unexpectedly, we find that the second fraction is larger, despite the fact
that it has the same numbers and involves minus signs.
However, to see if there is any pattern independent of the numbers, we
shall use another pair of numbers, 45 and 5, with the same pattern to see if
anything noteworthy results.
√ √ √ √
43√43+5√ 5
43 5+5 43
≈ 2.273573 and 43 √43−5√ 5 ≈ 4.273573, which suggests the
43 5−5 43
conjecture that (b) is always 2 greater than (a). But even if we had 10 or 100
more examples, we would not have a proof. Therefore, √
we
√
have√to introduce
√
variables and use algebra. We want to prove that a b−b√a − aa √a+b
a a−b b
√
b
√ = 2.
b+b a
First, let us briefly consider necessary conditions for a and b. Negative
numbers for a and b are forbidden from the very beginning because we
have to calculate square roots, but also a = 0 or b = 0 cannot be used
because then the denominators would be zero, and, finally, a = b cannot be
used because the first denominator would vanish. Altogether we have the
conditions a, b > 0 and a = b. First, we will combine the fractions with a
common denominator:
√ √ √ √ √ √ √ √
(a a − b b)(a b + b a) − (a a + b b)(a b − b a)
√ √ √ √
(a b − b a)(a b + b a)
and then expanding the products appropriately yields

√ √ √ √
a 2
b2
ab + a 2 b − ab2 − ab − a 2ab + a 2 b − ab2 + b2 ab
a 2 b − b2 a
2a 2 b − 2ab2 2ab(a − b)
= = =2
ab(a − b) ab(a − b)
Thus, we have shown that the difference of these two fractions is always
equal to 2.
Proving an Inequality
√ √
In the above problem, one can note that all the fractions of the form aa √a−b √
b−b a
b
seem to be greater than 3. However, is this always true? It is our challenge

to prove that this fraction is always greater than 3.
√ √
a a−b b
Solution: We have to prove that √ √
a b−b√ a
> 3 for all a, b > 0 and
√ √ √
a = b. Since we showed above that aa √a−b b a a+b b
√ − √ √ = 2, our conjecture
√ √ √ √ b−b a a b+b a
a a−b b a a+b b
√ √ > 3 is equivalent to √ √ > 1, which, in turn, is equivalent to
√
a b−b a
√ √ √ a b+b a
a a + b b > a b + b a. √
Squaring on both
√ sides of this inequality yields a 3
+ b 3
+ 2ab ab >
a b + b a + 2ab ab, which can then be written as (a − b)(a − b ) > 0.
2 2 2 2
For a > b, both parenthetical expressions are positive, and for a < b, both
are negative; in both cases the product is positive, and thus, we have proved
our conjecture.
Which Is the Larger Number?

√ √
Let a > 1 be a real number. What is the bigger number a− a − 1 or
√ √
a + 1 − a?
Solution: There is a rather hidden relationship among consecutive square
+3 +5 +7
numbers as can be seen here: 1 −→ 4 −→ 9 −→ 16 . . . . This pat-
tern of differences seems to continue. Might this be the same with square
√ √
roots? If so, then we might expect that a + 1 − a should increase
as the numbers increase. Let us take an example, say, for a = 9:
√ √ √ √ √ √
a− a − 1 = 3− 8 ≈ 0.1716 and a + 1− a = 10−3 ≈ 0.1623.
It seems to be the reverse! This can be proved in general by using algebra.
√ √ √ √
We have to prove a − a − 1 > a + 1 − a, which is equivalent to
√ √ √
2 a > a + 1 + a − 1, and by squaring both sides, and then dividing
√
both sides by 2, we get 2a > a + (a + 1)(a − 1), and this, in turn, is
√
equivalent to a > (a + 1)(a − 1) or a 2 > (a + 1)(a − 1) = a 2 − 1 which
proves our conjecture.
To make our understanding of algebra more solid, we should embark
on a pursuit of some important algebraic mistakes. Possibly one of the
most important rules in mathematics is that one is not allowed to divide
by zero. Some people even refer to this as the “eleventh commandment.”
There are times when division by zero is so well camouflaged that one
violates this commandment without knowing it. It is interesting to see what
happens when it is breached. Hopefully we will learn from each of these
transgressions. What is interesting (or entertaining) is to discover when this
rule — dividing by zero — has been violated, thus allowing us to arrive at
ridiculous results. Let’s now consider a few of these mistakes.
Does 1 = 2? A Mistake Based on Division by Zero
If we square both sides of the equation a = a, we get a 2 = a 2 . Then

subtracting a 2 from both sides of the equation, we have a 2 − a 2 = a 2 − a 2 .
We will factor the common term a on the left side and factor the difference
of two squares on the right side to get a(a − a) = (a + a)(a − a). As
a + a = 2a, this can be rewritten as a(a − a) = 2a(a − a). When we now
divide both sides of this equation by a(a − a), we get 1 = 2. Where did
we make the mistake? We have a − a = 0. Therefore, the above divisor
a(a − a) is zero, and we have violated the important rule of not dividing by
zero resulting in an absurd statement, namely, 1 = 2.
Here is another simple example of this sort of mistake — of dividing by
zero — that leads to an absurdity.
Is 1 = 2? A Mistake Based on Division by Zero
This time we will begin with the statement that a = b.

We then multiply both sides by b to get:
a · b = b · b or ab = b2
Then subtract a 2 from both sides of the equation so that:
ab − a 2 = b2 − a 2
Factoring the common factor on the left and the difference of two squares
on the right:
a(b − a) = (b + a)(b − a)
Dividing both sides by (b − a) give us:
a =b+a
However, since a = b (which was given), b = b + b or b = 2b, which

when we divide both sides by b has us resulting with 1 = 2.
In the following example, division by zero is more camouflaged, thereby
making it a bit more difficult to detect.
If a > b , Then a = b : A Mistake Resulting from Division by Zero
We begin with a > b, which can be restated as a = b + c, where a, b, and

c are positive numbers. We shall now multiply both sides of this equation
by a − b to get a 2 − ab = ab + ac − b2 − bc. We then subtract ac from
both sides of the equation giving us a 2 − ab − ac = ab − b2 − bc.
Then, factoring both sides of the equations, we get a(a − b − c) =
b(a − b − c), and then dividing both sides by (a − b − c) leaves us with
a = b.
How can it be that a = b, when we were told at the beginning that
a > b? Here we see that, once again, (a − b − c) is equal to 0 because we
began with a = b + c. We have, thus, violated the rule that forbids dividing
by zero.
All Integers are Equal: An Error resulting from Dividing by

Zero
Once again, we will use the mistake of dividing by zero — in a somewhat

hidden fashion — to show a silly result. We begin by accepting the correct
quotient that is as follows: x−1
x−1
= 1. We now multiply both sides of the
x 2 −1
equation by x + 1 to get x−1
= x + 1, since (x + 1)(x − 1) = x 2 − 1.
x 3 −1
Since (x 2 + x + 1)(x − 1) = x 3 − 1, we get x−1
= x 2 + x + 1.
4 −1
Since (x 3 + x 2 + x + 1)(x − 1) = x 4 − 1, we get xx−1 = x 3 + x 2 + x + 1.
And since (x n−1 + x n−2 + · · · + x + x + x + 1)(x − 1) = x − 1, we
3 2 n
n −1
get xx−1 = x n−1 + x n−2 + · · · + x 3 + x 2 + x + 1.
Now suppose we let x = 1. The absolute values of the right sides of the
above equations equal 1, 2, 3, 4, . . . , n.
The left side of the above equations will all be the same, as they are in
n −1
the form 11−1 , and, therefore, all the right-side numbers must be equal, or
1 = 2 = 3 = 4 = · · · = n. Surely, by now you will have realized that the
denominators are all 1 − 1 = 0. This cannot be permitted to exist, since if
it did then absurd conclusions would follow, such as that all the numbers
1, 2, 3, . . . , n are equal. We can see here that 00 cannot be a number, or all
of these weird results would follow.
The Hidden Nemesis of Dividing by Zero and Not Noticing It

Immediately
There are times when the division by zero is well camouflaged. Take, for
example, the equation 3x−3011−x
= x+2
x−7
− 4, which allows the right side to
be combined as 11−x =
3x−30 x+2−4(x−7)
x−7
. This can be then simplified to be
3x−30
11−x
= 3x−30
7−x
. Since the numerators are equal, the denominators must also
be equal, and, therefore, 11 − x = 7 − x or 11 = 7. Quite an absurdity!
It doesn’t appear that we divided by zero this time, and yet we ended up
with an absurd result.
Had we solved the equation 3x−3011−x
= 3x−30
7−x
in the traditional way, we
would find that x = 10, which would make the two numerators equal to
zero. Still, that doesn’t show that we divided by zero.
So, we should consider the following: If ab = ac , then we multiply both

sides by bc to get ac = ab. Dividing both sides by a gives us b = c, which
we expected. However, if a = 0, then this would not be valid, since we
would have divided by zero.
Let us now return to the equation 3x−30
11−x
= 3x−30
7−x
, which led us to an
absurd result. We found that x = 10. With that value of x, the numerators
3x −30 = 0, and, therefore, in this case we cannot equate the denominators.
Note how slyly division by zero hid from us to deliver a ridiculous result.
More of the Hidden Nemesis: Division by Zero
In a similar vein — but equally well hidden — we can show that +1 = −1.
We begin with the equation
x +1 x −1
=
p+q +1 p+q −1
By subtracting 1 from each side of this equation, we get
x+1
p+q+1
− p+q+1
p+q+1
= p+q−1
x−1
− p+q−1
p+q−1
, which can be simplified to get
x+1−( p+q+1)
p+q+1
= x−1−( p+q−1)
p+q−1
, or x− p−q
p+q+1
= x− p−q
p+q−1
.
Since the numerators are equal, the denominators must also be equal so
that p + q + 1 = p + q − 1 or +1 = −1, which is an absurdity! Why did
this happen? Might the previous example give a clue?
x+1
If you solve the original equation p+q+1 = p+q−1
x−1
for x, we find that
x = p + q.
Therefore, we have the same situation as above, where the numerators
of the two equal fractions (x − p − q) were zero.
x+1
The initial equation p+q+1 = p+q−1
x−1
is not as general as we would at first
imagine. It is relevant only for the case where x = p + q and p + q = ±1.
To better understand this result, we can look at a simpler version: From
a
b
= ab , we cannot simply conclude that a+c
b+c
= a−c
b−c
, since that is only true
if:
(1) a = b and (b + c)(b − c) = 0 or
(2) c = 0 and b = 0.
In other words, we need to make sure that the denominator is not zero.
Finding the Division by Zero Before It Misleads Us
There are many examples of division by zero mistakes that follow a similar
pattern. However, division by zero is usually camouflaged and sometimes
difficult to find. There are terms that hid the zero so well that it can be
easily overlooked, especially when you have no reason to suspect it being
there. Let’s considerthe following
√ √example: Suppose a term T1 is divided by
another term T2 = 4 − 2 3 − 3 + 1; we would not be at all suspicious.
However, as you will see in a moment, the term T2 is of a nature that will
violate our now-familiar “eleventh commandment” (thou shall not divide
by zero), if we use it as a divisor. In fact, the term T2 is equal to zero! Follow
along the algebra and you will see that it equals √zero.
√ √ √
4−2 3 = 3−2 3+1 = ( 3)2 − 2 · 1 · 3 + 12 =
√ √
( 3 − 1)2 = 3 − 1, then it follows that

√ √
T2 = 4 − 2 3 − 3 + 1 = 0
The zero divisor can be even more hidden as shown in the following:

3 √ 3 √ √
T3 = 5+2+ 5−2− 5
One may now wonder how we can get to show that T3 = 0. Here is a hint
that should help you show that T3 = 0.

3 √
√
3 √
√
Note that 5 + 2 = 5+1 and 5 − 2 = 5−1
, and
√ √ √ 2 2 √
5+1
2
+ 2 = 5. So, now in the value of T3 when we subtract 5, we
5−1
get 0.
An Absurd Result Stemming from the Well-Known Mistake

Based on Division by Zero
Before we begin this example, let’s look at a basic principle from algebra.
Consider the proportion ab = dc . From this we can conclude that b−d
a−c
= dc ,
only if b = d and d = 0. To show that this rule actually is true, we
begin by recognizing that from the original proportion ad = bc (by cross
multiplication). The cross multiplication for the second proportion (above)
gives us the following: (a − c)d = (b − d)c or ad − cd = bc − dc, which,

when adding cd to both sides, gives us ad = bc. This is the same as we had
gotten for the first proportion. Now having established the rule above, we
shall apply it to the following situation: Given x, y, z and the proportion
3y−4z
3y−8z
= 3x−5z
3x−z
, we shall now apply the rule we established above to get
3y−4z−(3x−z)
3y−8z−(3x−5z)
= 3x−z
3x−5z
, and then 3y−4z−3x+z
3y−8z−3x+5z
= 3x−z
3x−5z
, which simplifies to
3y−3z−3x
3y−3z−3x
=1 = 3x−z
3x−5z
. It then follows that 3x − 5z = 3x − z, which gives
us 5 = 1.
There must be something wrong, since we ended up with an absurd
result.
The mistake here is a bit more difficult to find. The equation 3y−4z
3y−8z
=
3x−z
3x−5z
is satisfied when x − y + z = 0. Now substituting x = y − z, we get
3x−z
3x−5z
= 3(y−z)−5z
3(y−z)−z
= 3y−4z
3y−8z
.
Using a similar substitution, we have 3y−3z−3x = 3y−3z−3(y−z) =
3y − 3z − 3y − +3z = 0 and 3y − 3z − 3x = 3y − 8z + 5z − 3x =
3y − 8z − (3x − 5z) = 0. Both are equal to zero. Thus, the denominator
3y−4z−(3x−z)
of the fraction 3y−8z−(3x−5z) is equal to zero. This is an example of where
division by zero is well camouflaged and exemplifies the kind of subtle
mistakes that mathematicians have to be cautious about throughout their
investigations.
Absurd Results Based on a Mistake of Interpretation
We are asked to solve the system of equations:
(1) a + b = 1
(2) a + b = 2.
Our initial reaction is to note that since the left sides of these equations
are equal, then so must the right sides be equal. Thus, we find that 1 = 2.
“Proved!” Or is it?
We could also have embarked on this system of equations by subtracting
the two equations — knowing that the difference of equals is also equal.
When we subtract the first equation from the second equation, on the left
side we would get 0 and on the right side we would get 1. Thus, 0 = 1.
Once again, an absurd result.
To take this one step further, with this set of equations we could also
show that 1 = −1. When we subtracted the first equation from the second
equation, we got 0 = 1. If we, now, subtract the second equation from the
first equation, we get 0 = −1. We can take this absurdity even one foolish
step further.
Since 0 = 1 and 0 = −1, one could then conclude that 1 = −1, as both
are equal to 0.
Our series of absurd conclusions above result from the fact that these
two equations have no common solution. Were we to graph them, they
would appear as two parallel lines — thus having no intersection or point
in common. The mistake here was to embark on the two equations, seeking
a solution and not recognizing immediately that there cannot be a solution
when the two lines representing these equations are parallel and, therefore,
have no common point of intersection.
Some mistakes in algebra can be seen better graphically, as we can see
in Figure 4.17. Consider the two equations 5x + y = 15 and x = 4 − 5y . If
we substitute the value of x from the second equation into the first equation,

we will get 5 4 − 5y + y = 15. This simplifies to 20 − y + y = 15 or
20 = 15. Now, there must be something clearly wrong here. Where was the
mistake? If we multiply both sides of the second of the two given equations
by 5, we get 5x + y = 20. Were we to graph these two equations, we would
find them to be parallel and would, therefore, have no point of intersection,
or to put it another way, no common solution (see Figure 4.17). Therefore,
it makes no sense to try to solve these two equations simultaneously as we
did above — thus, clearly, leading to an absurd result!
This time the parallelism of the two equations was not as obvious as in
the first case above. Yet, to avoid such absurd results, we have to be cautious
not to make some of the mistakes of interpretation shown here.
With this reasoning we can also prove that 5 = 16. To do this, we begin
with the equation
(x + 1)2 − (x + 2)(x + 3) = (x + 4)(x + 5) − (x + 6)2
Then doing the indicated multiplications, we get
x 2 + 2x + 1 − (x 2 + 5x + 6) = x 2 + 9x + 20 − (x 2 + 12x + 36)
By combining like terms, we then get −3x − 5 = −3x − 16.
Adding 3x to both sides gives us −5 = −16.
y
5
–10 –5 5 10 x
–5
–10
5x + y = 20 –15
5x + y = 15
–20
Figure 4.17
Lastly, multiplying by (−1), we get the absurd result that 5 = 16.

From a given equation the mathematical mistake is less obvious, but as
you approach the ridiculous result, you will note the similarity to the pre-
vious oversights. You then ask yourself, where is the mistake? We mul-
tiplied and added correctly. The mistake lies in the original equation,
which only becomes clear at the end of this discussion. In other words,
assuming the existence of a solution (but there is none) of this last equa-
tion, we find the contradiction 5 = 16. From a contradiction, we con-
clude that one of our assumptions or some reasoning must be wrong. But
here all steps are correct except our assumption that the equation has a
solution. Therefore, we can conclude that our original equation has no

solution.
Simultaneous Equations Leading to a Strange Result through an

Algebraic Mistake
We begin with the following two equations to be solved as a pair of simul-

taneous equations:
x y
+ =2 (1)
y x
x−y=4 (2)
Then by multiplying equation (1) by xy, we arrive at x 2 + y 2 = 2xy, which

can then be reworked to give us the following: x 2 − 2xy + y 2 = 0.
This can be written as (x − y)2 = 0, whereupon x − y = 0, or put
another way, x = y.
If we substitute this value of y in equation (2), we end up with the
ridiculous result x − x = 0 = 4, which then results in 0 = 4. So where is
the mistake?
From the given we would assume that x and y are not zero. Actually,
this system of equations has no solution. If we look at this graphically, we
will note that we actually have two parallel lines. With no intersection of
the two lines there can be no common solution as can be seen graphically
in Figure 4.18.
–5 5
x y x–y=4
+ = 2
y x
–2
Figure 4.18
A Mistake Based on Faulty Square Root Extraction Leading to

an Absurd Conclusion
In this case just follow the steps shown in the following:
2=2
3−1=6−4
1−3=4−6
9 9
1−3+ =4−6+
4 4
3 9 3 9
1−2· + =4−4· +
2 4 2 4
2
3 3 2
1− = 2−
2 2
3 3
1− =2−
2 2
1=2
So where is the error?

It is hidden in the step where we took the square root of both sides of the
2 2
following 1 − 32 = 2 − 32 , and arrived at this expression 1− 32 = 2− 32 ,
which ignored the negatives that need to be considered.
We should have
gotten the absolute 3 3
as follows: 1 − 2 = 2 − 2 , which would have
values

led us to − 12 = 12 , which results in something that is certainly reasonable
(and correct) 12 = 12 . Ignoring the proper square root extraction can lead to
countless mistaken and silly results.
For example, suppose we begin with −20 = −20, and write it as 16 −
36 = 25−45. If we add 81 to both sides, we get 16−36+ 81 = 25−45+ 81 ,
4 9 2 2
4 4
which is equivalent to 4 − 2 = 5 − 2 . Now taking the square root of
9
both sides (although incorrectly, as stated above) we get the following:

4 − 92 = 5 − 92 , which then results in the ridiculous result 4 = 5.
However, had we done the work correctly by considering that the square
root shouldresult in an absolute-value
statement, we would have gotten the
9 9
following: 4 − 2 = 5 − 2 , which leads to a sensible result 12 = 21 .
A Subtle Mistake in Solving an Equation Will Cause an Error

√
We are asked to solve the equation 3x − 2x − 4 = 4x − 6 (where x ∈
Real Numbers, x ≥ 2). The usual way to solve this equation is to isolate
the radical term and then square both sides as follows:
√
2x − 4 = −(x − 6)
2x − 4 = (−(x − 6))2 = (x − 6)2
Simplifying this we get x 2 − 14x + 40 = 0. This then produces two roots

x1 = 10 and x2 = 4. By substituting these two values into the original
equation, we find that only x2 is a solution and not x1 . Why is x1 not a valid
solution? The mistake we made was to consider taking the square root of
both sides, which is not an equivalent change. In other words, from T1 = T2
it follows that T12 = T22 . However, the reverse is not true — as we have just
experienced. In Figure 4.19 we graph each side of the original equation and
can see where the two equations intersect.
50
g(x) = 4x 6
40
f (x) = 3x 2x – 4
30
20
10
20
10
Figure 4.19
To “Prove” That 0 = 100: Stemming from a Mistake with

Square-Root Extraction
We begin by letting y = 100 and z = 0. Suppose we have x = y+z 2

. Then
2x = y + z.
By multiplying both sides by y − z, we get 2x(y − z) = (y + z)(y − z).
Now multiply as indicated: 2x y − 2x z = y 2 − z 2 . Then rearrange terms:
z 2 − 2x z = y 2 − 2x y. Then adding x 2 to both sides of the equation:
z 2 − 2x z + x 2 = y 2 − 2x y + x 2 . Factor each of the sides of the equation:
(z − x)2 = (y − x)2 . Taking the square root of each side gives us: z − x =
y − x. This leaves us with z = y, or by substituting the original values for
y and z, we have 0 = 100. Once again, the mistake lies in concluding that
a 2 = b2 implies that a = b. But the conclusion can only be a = ±b. So
that in the next-to-last step we could have gotten z − x = −(y − x), which
then gets us back to where we began with x = y+z 2
.
To Show That If a = b, Then a = b: Another Mistake Stemming

from Square-Root Extraction
There are times when our mistake could be of the same nature as that
above, but well hidden, and therefore, easily overlooked as in the following
example.
We begin with a = b and assume (without loss of generality) a < b.
Further, let be c = a+b
2
. That means that a + b = 2c. By multiplying both
sides by a − b, we get a 2 − b2 = 2ac − 2bc.
Then adding b2 − 2ac + c2 to both sides of the previous equation, we get
a − 2ac + c2 = b2 − 2bc + c2 . Each of the two sides of this equation is a
2
perfect square, and, thus, the equation can (a − c)2 = (b − c)2 .

be written as
Taking the square root of both sides (a − c)2 = (b − c)2 gives us
a − c = b − c or a = b. Recall that we began by stating that a = b. There
must be a mistake somewhere in our work. We seem to have done every
step of this algebraic process correctly. However, in the last step, where we
took the square root of both sides, we neglected to consider the negative
values. Had we taken the result of extracting the square root of both sides

of (a − c)2 = (b − c)2 to get a − c = −(b − c), we would have gotten
a − c = −b + c, which is our original equation, a + b = 2c.
Caution Must be Taken in Solving Equations or Else a Mistake

May be Encountered
Another error arising from a surprise mistake is one very subtly hidden in
√ √
the procedure of solving the equation 1 + x + 2 = 1 − 12 − x.
We begin our solution to this equation by adding −1 to both sides and
then squaring the two sides. This yields x + 2 = 12 − x, which results
in x = 5. If we substitute this value of x into the original equation, we
√ √
get 1 + 5 + 2 = 1 − 12 − 5, and then adding −1 to both sides and
√ √
then squaring both sides of 5 + 2 = − 12 − 5, we get 7 = 7. This
would have us think that the value of x is the correct value. It is not!√If we
substitute
√ 5 in place of x in the original equation, we will have 1 + 7 =
1 − 7, which is not correct. There is no answer to this equation. This can
be seen from the very beginning, since the initial equation is equivalent with
√ √
x + 2 = − 12 − x. The only way a square root is a negative square root
is that the radicand is 0. But this would imply on the left side x = −2 and
on the right side x = 12.
Where, then, has the mistake been made? When taking the square root,
we must take the positive and the negative into account. We violated that
rule in this process!
It is good to remember that squaring both sides of an equation is not an
equivalence transformation. It yields a new equation with possibly more
solutions than the given equation. Therefore, not every solution of the
“squared equation” is a solution of the original equation. This is an important
rule very often not explicitly expressed. This is when mistakes appear.
√
Consider the equation x + 5 − x + 5 = 6. This can be written as
√
x−1 = x + 5. Squaring both sides and solving, we get x 2 −2x+1 = x+5,
or simplified as x 2 − 3x − 4 = 0. Then x = 4 or x = −1. However, whereas
x = 4 is a solution, x = −1 is not a solution. This is a typical mistake made
in algebra classes. Again, the square root process did not take the negative
into account.
This absurdity can be taken further, for if we want to prove that 5 = 1,
we then subtract 3 from both sides to get 2 = −2, and then squaring both
sides, we get 4 = 4. Therefore, 5 must have equaled 1!
A Mistake with Powers Can Lead Awry

x 3
Imagine someone solving the equation 23 = 32 as follows: Begin with
x 3
the given 23 = 32 and then apply the powers to the fractions to get
2x 3
3x
= 323 . Multiplying by the common denominator, 23 · 3x leaves us with
23 · 2x = 33 · 3x .
Following the rules of exponents, we have 23+x = 33+x .
If two equal powers have equal exponents, we might conclude that the
bases must be equal as well, therefore, 2 = 3. Something is not correct here.
Where might the mistake be? The last step is wrong! The correct solution
is x = −3, which makes each of the two sides equal to 1.
You may want to see what this looks like graphically; we provide the two
functions f (x) = 23+x and g(x) = 33+x and show the graph in Figure 4.20.
Figure 4.20
A Subtle Oversight — or Mistake — with the Binomial Theorem
We know that (a + b)2 = a 2 + 2ab + b2 . This formula is an application of

the binomial theorem, which provides a formula for obtaining the value of
a binomial raised to any positive integer power n. The formula is
n(n − 1) n−2 2
(a + b)n = a n + na n−1 b + a b +
2
n(n − 1) 2 n−2
··· + a b + nabn−1 + bn
2
If n = 2, we get (a + b)2 = a 2 + 2ab + b2 . If n = 1, we get
(a + b)1 = a 1 + b1 = a + b. Using the formula of the binomial theo-
rem, we can also be led by mistake to a weird conclusion. When n = 0, we
get 1 = 1 + 0 + 0 + 0 + · · · + 0 + 0 + 0 + 1, or 1 = 2. (Because when
n = 0, then (a + b)0 = 1, a 0 = 1, b0 = 1.) That we reached an absurd
result would have us believe that a mistake was made, and indeed one was.
But where is the error? This seems to be alright: (a + b)n = (a + b)0 = 1.
Now look at the right side more precisely, for therein lies the reason for
our mistake. On the right side of the equation, there is for n = 0 only one
term (not two or more terms), which we produced by substituting into the
binomial formula. That term is 1a 0−0 b0 = 1 · a 0 b0 = 1 · 1 · 1 = 1. We
can also see that on the Pascal triangle, which can be used to determine the
coefficients of the terms of the binomial expansion as follows and shown in
Figure 4.21.
(a + b)0 = 1
(a + b)1 = 1a + 1b
(a + b)2 = 1a 2 + 2ab + 1b2
(a + b)3 = 1a 3 + 3a 2 b + 3ab2 + 1b3
···
n n(n − 1) n−2 2 n(n − 1) 2 n−2 n n−1
(a + b)n = 1a n + a n−1 b + a b +···+ a b + ab + 1bn
1 2 2 1
1
1 2 1
1 3 3 1
1 4 6 4 1
Figure 4.21
But be careful: If a = b = 0, you could have a problem! What is

the value of 00 ? This is an expression that we leave undefined, and many
calculators do not react to this input, yet some deliver the value 1, since
“anything” taken to the zero power seems to be 1.
As we said at the outset, the binomial theorem only holds for positive
values of n. Note the mistake that using zero has caused. That is why we
specify the value of n to avoid such ridiculous results!
To Show That If p Is Positive, It Is Actually Negative: A Mistake

with Inequalities
We begin by assuming that both p and q are positive, and we are going
to show that p is negative. Clearly, the inequality 2q − 1 < 2q is a true
statement. Suppose we multiply both sides by − p to get −2pq+ p < −2pq.
If we then add 2pq to both sides of this inequality, we end up with p < 0,
which is to say p is negative. How can this be when we started off with a
positive p? Where was the mistake?
We violated a rule for inequalities as follows:
When multiplying (or dividing) both sides of an inequality by a negative
number, the inequality symbol must be reversed.
Look at a simple example: 2 < 3. But, when we multiply both sides
of the inequality by −1, we have the following: 2 · (−1) = −2, which is
greater than 3 · (−1) = −3, or simply written −2 > −3.
We can see this mistake played out in a less obvious way in the following
example.
To Show That Any Positive Number Is Greater Than Itself
We shall begin with the two positive numbers p and q, where p > q. We
will now multiply both sides of this inequality by q to get pq > q 2 . Now
subtracting p2 from both sides of the inequality gives us pq − p2 > q 2 − p2 .
By factoring both sides, we get p(q − p) > (q + p)(q − p). Dividing both
sides by (q − p) leaves us with p > q + p, which says that p is greater
than itself. That’s absurd! So where was the mistake made? Since p > q, it
must follow that (q − p) is negative. We made the mistake of not reversing
the inequality symbol when dividing both sides of the inequality by the
negative term q − p.
The structure of this example follows the “proof” of “1 = 2: A mistake
based on division by zero” — instead of division by zero in an equation we
had a division by a negative term for an inequality. This absurd result can
be taken a step further. With the given p > q and our newly found result
p > q + p, we can add the two inequalities to get 2 p > 2q + p. Now
subtracting p from both sides, we get p > 2q. So if having p > q, and
p > 2q, we can conclude that p > 2q, similar reasoning would allow us
to then conclude that p > 4q. This can continue in the same way to further
absurdities.
Mistaken Application of Complex Numbers to Prove That

−1 = +1
√
We begin with the product of two imaginary numbers −1 and apply
√ √ √
the rules we know about real numbers: −1 · −1 = (−1) · (−1) =
√ √ √
+1 = 1. This time we will evaluate the product as follows: −1· −1 =
√
( −1)2 = −1. Therefore, since the given product provides us with two val-
ues, we would have to conclude that −1 = +1. Something must be wrong,
since clearly −1 is not equal to +1. This mistake is one that depends on a
definition in mathematics — one that we may√even think arose to avoid this
√ √
dilemma. That is, that the product a· b = a · b does not hold true when
√ √ √
a and b are negative. Therefore, in a sense −1 · −1 √ = (−1) · (−1)
√
is wrong! However, it is true that (−1) · (−1) = 1. Another way to
avoid (or explain) the above dilemma is the following: In the real numbers,
taking a square root is a process with a unique
√ result; solving an equation
may have more solutions, for example, 4 = 2 but the equation x 2 = 4
has the two solutions ±2. In the world of complex numbers, this is not true
any longer, square roots always have two possible values, cube roots always
have three possible values, and so on. In terms of complex numbers solving
the equation x 2 = 4 is the same as taking a square root, which is a big
difference between real and complex numbers! So, there is no unique value
√ √
of −1, it is not only i,1 rather we have −1 = ±i. And seeing it this
way, the above equation does not produce a contradiction, it merely says
(±i) · (±i) = 1 which is correct, because the left side is nothing other than
±i 2 = ±(−1). Analogously, a similar dilemma arises when√ we take for

granted the rule for real numbers in the quotient operation as √ab = ab and
carelessly extend it to negative numbers.
The following is clearly true since both sides of the equation are equal to
√
−1. Now observe
what
happens when we accept the above generalization:
√ √
Starting with −1 1
= −1 1
would then lead to √−11
= √−11
.
Now clearing fractions (perhaps by either multiplying
√ 2 by √ the common
denominator or simply cross-multiplying), we get ( 1) = ( −1)2 . This
essentially tells us that 1 = −1. Again, the definition was abused leading to
a mistaken result. To debunk this “proof,” one need not know much about
complex numbers, just a familiarity with the characteristics of the familiar
operations. We note that there are times when our time-honored operations
take on other characteristics.
A Subtle Mistake Leads to an Absurdity
Follow along as we solve this equation for x (a real number): 6

x−3
− 9
x−2
=
1
x−4
− x−1
4
.
1 Where i is the famous imaginary number with i 2 = −1.

We begin by multiplying the fractions on each side by their respective

lowest common denominator:
6(x−2)
(x−2)(x−3)
− 9(x−3)
(x−2)(x−3)
= x−1
(x−1)(x−4)
− 4(x−4)
(x−1)(x−4)
.
We then clear parentheses and add the fractions on each side of the equa-
−3x+15
tion: 6x−12−9x+27
x 2 −3x−2x+6
= xx−1−4x+16
2 −4x−x+4 . By combining like terms, we get x 2 −5x+6 =
−3x+15
x 2 −5x+4
. Now divide both sides by (−3x + 15) to get x 2 −5x+6
1
= x 2 −5x+4
1
.
We then equate denominators, since the numerators and the fractions are
equal: x 2 − 5x + 6 = x 2 − 5x + 4. By subtracting x 2 − 5x from both sides
of the equation, we end up with 6 = 4.
With this absurd result, you would think that the original equation has
no solution. This is wrong! The solution of this equation is x = 5, as you
can see from the following, where we show that when x = 5, each side of
the original equation has the same value, namely, 0:
6 9 6 9
− = − = 3 − 3 = 0 and
5−3 5−2 2 3
1 4 1 4
− = − =1−1=0
5−4 5−1 1 4
We note that x cannot take on the values of 1, 2, 3, and 4, since that would
produce a zero denominator in one of the fractions of the original equation.
So then, where might the error lie? When we divided by −3x + 15, we had
to eliminate the possibility that −3x + 15 = 0. However, this case is the
one that provides us with the correct answer since 3x = 15, and, therefore,
x = 5. Therefore, we once again — surprisingly — divided by zero. Our
old nemesis!
A Confusing Equation: Ripe for a Mistake (or an Omission)
We seek to solve the following equation for the real number x:

3 − 1+x
2
= 3x+1
2−x
.
Add the terms on the left side of the equation to get 3(1+x)
1+x
− 1+x
2
= 3x+1
1+x
.
This gives us 1+x = 2−x . Since the numerators are equal, the denominators
3x+1 3x+1
must be equal as well so that 1 + x = 2 − x. Solving for x, we get 2x = 1

and x = 12 . This would appear to be a solution to the equation. Let’s check to
see if this checks out properly. Substituting the 12 for x we get the following:
The left side of the equation: 3 − 1+2 1 = 3 − 23 = 3 − 43 = 53 . The right
2 2
3· 1 +1 5
1 = 3 = 3 . All appears to be fine. Unfortunately,
2 2 5
side of the equation: 2−
2 2
that is not the only solution to this equation. Let’s begin another method for
solving this equation: 3 − 1+x 2
= 3x+1
2−x
. Multiply both sides of the equation
by (1+x)(2−x) to get 3·(1+x)(2−x)−2·(2−x) = (3x +1)(1+x). Then
clearing parentheses, we get −3x 2 + 3x + 6 − 4 + 2x = 3x 2 + 3x + x + 1.
Then simplifying: −3x 2 + 5x + 2 = 3x 2 + 4x + 1. We will now add
3x 2 − 5x − 2 to both sides of the equation to get 0 = 6x 2 − x − 1. By
dividing both sides of the equation by 6, we get x 2 − 16 x − 16 = 0.
Using the well-known quadratic formula to solve this quadratic
equation,
we get the two values for x: x1,2 = 1
12
± 1212 + 16 = 12 1
± 144
1
+ 144
24
=
1
12
± 12
5
. Or, written separately: x1 = 1
12
+ 12
5
= 12 and x2 = 12
1
−− 12
5
= − 13 .
Then x2 = − 13 is then a second solution to this equation (along with the
previously found x1 = 12 ).
We ought to check the second solution to see if it is, in fact, a correct
solution. Substituting the second solution, x2 = − 13 , on the left side of the
original equation: 3 − 1−2 1 = 3 − 22 = 3 − 3 = 0. Then substituting the
3 3
value x2 = − 13 on the right side of the equation, we have ( 2+3 )1 = 07 = 0.

3 − 1 +1
3 3
So clearly the second solution is also correct. Our mistake here was to solve
an equation and to be perhaps satisfied with only one of two possible correct
solutions.
An Equation That Leads to a Mistaken Solution
Suppose we search for the solution (x, y) of the equation x 2 + 9y 2 = 0,

where x and y are the real numbers. Follow the following solution:
x 2 + 9y 2 = 0. Subtract x 2 from both sides of the equation: 9y 2 = −x 2 .
2
Divide both sides equation by 9: y 2 = − x9 . Take the square root of both
sides of the equation: y = ± x3 . This is a mistake and not the correct
solution.
Since x 2 ≥ 0 and 9y 2 ≥ 0, we know that x 2 + 9y 2 ≥ 0. However,

we are given that x 2 + 9y 2 = 0. Therefore, it must follow that x 2 =
0 and 9y 2 = 0, which implies that x = 0, as well as y = 0, and not
y = ± x3 .
(The original equation has its only solution x = 0 = y in the real
numbers yet in the complex numbers, the following is a solution: y = 3i x
or y = − 3i x.)
Caution Must Be Taken When Solving Inequalities — Otherwise

We Will Be Making a Mistake
We begin with the real numbers a and b. For which values of a and b is
the following inequality satisfied? ab + ab > 2. We realize that the values
of a and b cannot be zero, or else the division indicated would be invalid.
We begin by multiplying both sides by ab to get: a 2 + b2 > 2ab. Then add
to both sides the following: −ab − b2 to get: a 2 − ab > ab − b2 .
Factoring the common factor on both sides of the equation gives the
following: a(a − b) > b(a − b). Our last step is to divide both sides by
(a − b), which results in a > b. The solution to this inequality appears to
be a > b. Is this answer correct? Clearly, with the division by (a − b), we
realized that a = b. If a = b then following would be true: aa + aa = 1+ 1 =
2, which contradicts the given inequality.
So now let’s look at an alternate solution to this inequality: ab + ab > 2.
We begin by multiplying both sides by ab. If, in the first case, ab > 0
(that means a and b have the same sign; and notice that a = b) then we
get a 2 + b2 = 2ab which is equivalent to (a − b)2 > 0, which is true
for every such a, and b. In the second case, ab < 0 (that means a and
b have different signs), we would end up with (a − b)2 < 0 which is
never true. The correct answer is the following: (a) a = b and a, b > 0,
and (b) a = b and a, b < 0. In other words, when a and b have differ-
ent signs, the inequality is not satisfied. This can be seen graphically in
Figure 4.22.
b b=a
4
a b
3 + >2
b a
2 a b
+ >2
1 b a
–6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 a
–1
a b
+ >2 –2
b a
a b –3
+ >2
b a
–4
b=a
Figure 4.22
More Mistakes That Lead to Correct Results
As we have seen from the previous example, mistakes don’t always lead to
an absurd result. We could also have mistakes that lead to a correct answer.
These are not to be condoned but just provide us with some amusement.
We begin with the equation x − 2 = 3, which is the same as x = 5.
Now we will make a deliberate mistake and add 12 to only the left side
of the original equation to get x + 10 = 3. Then we will multiply both
sides of the equation by x − 5 to get (x + 10)(x − 5) = 3(x − 5). We
now subtract 3(x − 5) from both sides of the equation, which gives us
x 2 + 5x − 50 − (3x − 15) = 0 or in its simplified form, x 2 + 2x − 35 = 0.
By factoring, we get (x + 7)(x − 5) = 0. Dividing both sides by x + 7 gives
us x − 5 = 0 or x = 5, which is what we had as our initial value of x. So,
despite our earlier mistake of adding 12 to only one side of the equation,
we still got the right result.
Had we not added 12 to only one side of the equation, but to both
sides, as we should have, we would have subtracted 15(x − 5) instead of
3(x − 5). This would have given us (x − 5)2 = 0, implying that x = 5. The
“wrong” solution x = −7 will have disappeared through the division by

x + 7.
Another comical mistake that leads to a correct answer is to incorrectly
add the two binomials instead of the indicated multiplication.
We begin with the equation that we are asked to solve for x to get
(5 − 3x)(7 − 2x) = (11 − 6x)(3 − x). Now adding instead of multiply-
ing, as is indicated in the given equation, we get (5 − 3x) + (7 − 2x) =
(11−6x)+(3− x). This can (correctly) be converted to 12−5x = 14−7x.
This yields 2x = 2 or x = 1, which, surprisingly, is correct!
Compare this to solving the equation (5−3x)(7−2x) = (11−6x)(3−x)
correctly, which leads to 6x 2 − 31x + 35 = 6x 2 − 29x + 33, which has the
sole solution x = 1.
Hereis a rather silly series of two mistakes that leads to the correct
√
answer: 2.8 70
= 0.04 = 0.2. In other words, the second error corrects the
first error.
A Correct Start Followed by a Series of Silly Mistakes Leads to

Correct Solution
We are given the following equation and asked to solve for x:

x−7
x+7
+ x+10
x+3
= 2. But multiplying both sides of the equation by (x +7)(x +3),
as we would normally begin in solving this equation, we get the following.
(x − 7) · (x + 3) + (x + 10) · (x + 7) = 2(x + 7) · (x + 3)
x −7+x +7=2
2x = 2
x =1
This is then followed by the mistaken cancellations shown in the following,

which is then followed by some further sloppiness that brings us to a correct
answer. This mistake may appear comical but has been already committed
by some.
Mistake for Which We Can Blame the Calculator
Some mistakes in mathematics may not be our fault. Rather, they may be
the fault of the calculator on which we seem to have unquestioned reliance.
1 √
Suppose we take an algebraic fraction such as √a+b− a
and correctly find
√ √
a+b+ a
its equivalent by first multiplying by 1 in the form of √ √ and then
a+b+ a
doing the algebra as shown here:
√ √ √ √
1 1 a+b+ a a+b+ a
√ √ =√ √ ·√ √ = √ √
a+b− a a+b− a a+b+ a ( a + b)2 − ( a)2
√ √ √ √
a+b+ a a+b+ a
= =
(a + b) − a b
Let us now compare the way the calculator evaluates the two equal algebraic
expressions:
√ √
1 a+b+ a
√ √ and
a+b− a b
Assigned Calculator √ √
√ 1 √ a+b+ a
values result to n b
a+b− a
places
a = 1,000 8 63,291.139 63,245.569
b = 0.001 20 63,245.569014751992618 63,245.569014751934636
a = 100 8 2000.4001 2000.05
b = 0.01 20 2000.0499987500624968 2000.0499987500624960
Note the differences — or we should say, note the mistakes — that the
calculator has made. These are not exactly mathematical mistakes. They
are rounding-off mistakes that result from the fact that the first denominator
is a difference of approximately equal numbers. And such differences in a
denominator are, numerically, seen to be very dangerous! The more accurate

result we get from the second fraction.
Mistaken Relationships
It is an unfortunate mistake when one misunderstands what a proportion is.

The common mistake is one that can be seen when a woman is asked her
age and reasons as follows: I was 20 years old when I got married to my
husband, who at that time, was 30 years old. Since he is 60 years old today,
which is twice 30, I suspect that I am 40 years old, which is twice 20.
x
In other words, the woman figured as follows: 20 = 60
30
, x = 20×2 = 40.
Unfortunately, this is a mistake. This cannot be handled as a propor-
tion, since it merely requires a constant difference. In this case, there is a
difference of ten years; therefore, her correct age is 60 − 10 = 50.
Another Absurdity
In a proportion, if the first term is greater than the second term, then the
third term must be greater than the fourth term. Therefore, if ad = bc, then
a
b
= dc . Suppose a > b, then c > d.
Now if we let a = d = 1, and b = c = −1, we have satisfied the
equation ad = bc, where a > b. It then should follow that c > d, which in
this case would indicate that −1 > 1. This is clearly a mistake, but where
is the error?
The mistake here is that we said earlier that ab = dc , and suppose
a > b, then c > d. This only holds true for positive numbers but
not in general. For example, if ab = dc , and a > b, we could have
5
4
= −10−8
and 5 > 4, but −10 is not greater than −8. Therefore, here
c < d.
Chapter 5
Geometric Gems
The origin of geometry lies, on the one hand, in problems of field measure-
ment, and, on the other hand, in astronomy. As the field of mathematics
evolved, one of the early formalities was Euclid’s Elements (300 BCE),
where we find the first attempt to introduce a system of axioms and postu-
lates to establishing theorems. In a sense, geometry was the first exact and
accurate science in which logic (formal) reasoning was required! Further-
more, geometry is a very concrete and demonstrative discipline; one can see
and examine figures, shapes, solids, and patterns. Many students learn to
really appreciate geometry — especially when it is well taught — and end
up preferring it over arithmetic and algebra, largely because of its visual
aspects.
Some Important Aspects of Geometry
Forms, figures, shapes, and patterns

The very graphic, concrete, clear, and vivid aspect of geometry presented
in school offers many people something visually fascinating. Later in life,
geometry helps define patterns and symmetry.
Geometry serves as a model for perceiving and experiencing

space and objects around us
This includes everyday objects, such as mirrors, balls, playing fields, shapes
of houses, crystals, and honeycombs.
219
Building a logic system using rigor with axioms, postulates, and

proved theorems
Our intuition should be controlled by more exact considerations because
otherwise we are tempted to draw wrong conclusions, such as that for a
rectangle, where a false conclusion such as that the length of the diagonal
equals the sum of the rectangle’s length and width. This faulty statement
evolves from the following “reasoning.” In Figure 5.1, the sum of the steps
along diagonal d is equal to a + b, since the sum of the steps is actually
the sum of the horizontal parts and vertical parts. The same is true for
Figure 5.2. On the one hand, the sum of the lengths of all stairs from A
to C stays constant a + b (even in case of arbitrarily small stairs). On the
other hand, the whole stairway comes arbitrarily near to the diagonal when
the number of stairs increases and the stairs become arbitrarily small, and
we are led to the result that d = a + b, which we know is incorrect as it is
contrary to the famous Pythagorean theorem which states that a 2 +b2 = d 2 .
We cannot clarify here the deeper details where infinity plays a curious role.
Figure 5.1
Figure 5.2
Geometric Gems 221
To truly appreciate the field of geometry, we now offer a collection of

geometry problems that show the power and beauty of geometry with its
unusual aspects.
Intersection Angle on a Clock
Problem: On the circular dial of a clock, the points at 2 and 8 are joined by
a straight line, and the points at 6 and 11 are also joined by a straight line.
These 2 lines form the angle ϕ. The challenge here is to determine the size
of this angle ϕ (see Figure 5.3).
Figure 5.3
Solution: There are several ways to approach this challenge. We will con-
sider three ways here.
(1) In Figure 5.4, we draw a parallel line to the line segment joining 6 and
11 and containing point M, which is the center of the circular clock.
With points P and Q as the midpoints between with 5–6 and 11–12,
respectively, we construct PQ containing the center of the clock at point
M and it is parallel to line 11S6 as shown in Figure 5.4. Since the angle
measured between numbers on the face of the clock is 30◦ , we then
have the central angle 8M P = 75◦ . However, since we have alternate
interior angles of parallel lines, we can conclude that ϕ = 8M P =

75◦ .
Figure 5.4
(2) We know that 11M 8 = 90◦ and arc 118 = 90◦ and, therefore, the
inscribed angle 1168 = 45◦ = S 68; since 86 M is equilateral we
know 68 M = 60◦ = 68 S; thus, in the triangle S 86 we know
ϕ = 8S 6 = 180◦ − (60◦ + 45◦ ) = 75◦ .
(3) Another method begins with 6M 11 = 90◦ + 60◦ = 150◦ , that means
M 11 S = 15◦ (note that the triangle M 116 is isosceles) and, there-
fore, in S M 11 we have ϕ = M S 11 = 180◦ − (90◦ + 15◦ ) = 75◦ .
Angle Sum in Star Polygons
Problem: The challenge is to determine if there is a constant sum of

the angle measures at the vertices of a pentagram (a star with five ver-
tices) or of a heptagram (a star with seven vertices), which are shown in
Figure 5.5.
Geometric Gems 223
(a) (b)
Figure 5.5
Is the angle sum the same for all such figures? If yes, what is the respec-
tive angle sum? If not, give examples with different angle sums.
Pentagram solution
As shown in Figure 5.6, we let α, β, and γ be the interior angles of the
AGF, whereby α + β + γ = 180◦ . For the corresponding exterior angles,
we have CGE = 180◦ − α and BFD = 180◦ − β.
Figure 5.6
In the triangle GCE, we have GEC + GCE = α. Analogously, in

FBD, we have FDB+ FBD = β. Thus, the angle sum of the pentagram
is α + β + γ = 180◦ , as we have noted above.
Heptagram solution
Again, we have an angle sum of 180◦ . The reasoning is essentially the
same as above in the case of a pentagram. One can see that in Figure 5.7
the exterior angle of triangle QDG is DQC = θ + δ and for triangle
JEA the exterior DJE = α + ε. Then for triangle PHJ, the exterior angle
QPC = α + ε + ϕ + β, and for triangle PQC, the sum of the angles
180◦ = PCQ + DQC + QPC = γ + (θ + δ) + (α + ε + ϕ + β).

E
ε
F D
φ δ
H P
J
Q γ
θ C
M
G
α
β
A
B
Figure 5.7
What is the Constant Angle?
Problem: In Figure 5.8, we show two congruent equilateral triangles that

share a common vertex at point C. Their other vertices are joined by two
Geometric Gems 225
lines forming angle BPE. Our challenge here is to find the measure of angle
BPE.
B E
D
A
Figure 5.8
Solution: In Figure 5.9, we show two isosceles triangles, namely, ACE

and DCB, which generate equal base angles, EAC = AEC and BDC =
DBC. However, triangle BCE is also isosceles. It is not obvious that angle
BPE does not depend on how far apart the two “upper” vertices B and E
are. We have two congruent isosceles triangles ACE and DCB, as their
legs are sides of the congruent equilateral triangles, and the angles between
them are equal: AEC = DBC. The angle whose measure we seek is
BPE, which is an exterior angle of APB and, therefore, is the sum of the
not adjacent interior angles:
BPE = BAP + ABP = BAP + (60◦ + DBC)

= BAP + (60◦ + CAE) = 60◦ + BAP + CAE
= 60◦ + 60◦ = 120◦
B E
A D
Figure 5.9
Strange Similar Triangles
Problem: In Figure 5.10, the point D on side BC of ABC divides BC in

the ratio B D : DC = 1:2. Furthermore, we are given the measures of two
angles: CBA = 45◦ and CDA = 60◦ . Our challenge is to prove that the
triangles ABC and ADC are similar.
Hint: Draw the perpendicular from C to AD, which will facilitate a proof.
Figure 5.10
Geometric Gems 227
Solution: As suggested, we will draw the perpendicular from C to AD with

the intersection point E, as shown in Figure 5.11. Then EDC is a 30◦ –
60◦ –90◦ triangle, and, therefore, we have CD = 2ED so that, consequently,
we have ED = BD. From this it follows that EBD is isosceles with base
angles of 30◦ . Hence, also the triangle EBC is isosceles with base angles
of 30◦ , and therefore, triangle ABE is isosceles since it has equal base
angles of 15◦ . Thus, we can conclude that AE = EC. The isosceles right
triangle AEC has base angles of 45◦ . Thus, we have two similar triangles
ABC and ADC as they both have angles of 45◦ , 60◦ , and 75◦ .
Figure 5.11
The Semicircle in a Quarter Circle
Problem: A semicircle is inscribed in a quarter circle as shown in

Figure 5.12. We need to determine what portion of the quarter circle is
the area of the semicircle.
Figure 5.12
Solution: Let r denote the radius of the semicircle shown √ in Figure 5.13.
Then we can draw a square with √ side r and diagonal r 2. The right tri-
angle AFG with legs r and r √ 2 has hypotenuse AG = R, which by the
Pythagorean theorem is R = r 3. Therefore, the area of the semicircle is
√ 2
π r , and the area of the quarter circle is π R = π ( ) = π 3r . Thus, we
2 2 r 3 2
2 4 4 4
have the fraction representing the semicircle part of the quarter circle as
2
π r2 1
2
3r 2
= 3 = .
2
π 4 4
3
C G
R
r F
D
r
r r 2
r
A r E B
Figure 5.13
Geometric Gems 229
Determine an Unusual Angle Relationship
Here we are faced with a curious conundrum. In Figure 5.14, a semicircle

with center M and diameter AB has a chord CD drawn parallel to AB. When
we know that BAC = α = CMD, what is the measure of α?
Figure 5.14
Solution: Since angle CAB is inscribed in arc CB, we have arc BC = 2α,
and because C D||AB, we also know that arc AD = 2α. Therefore, CD
= 180◦ − 4α. Thus, for the central DMC we know α = 180◦ − 4α and
◦
α = 180
5
= 36◦ .
Discovering the Missing Area
A rectangle ABCD is partitioned into four triangles by an interior point P.

Three areas are given as shown in Figure 5.15. Our challenge here is to find
the area of the fourth APD.
Figure 5.15
Solution: One method for determining the area of triangle APD is to draw
the parallel line EF to AB through P, as shown in Figure 5.16. Since triangle
APB and rectangle APFE have the same base and altitude, we know that
the area of rectangle ABFE is twice the area of triangle APB or 30. Analo-
gously, the area of rectangle CDEF is twice the area of triangle CDP or 50.
Thus, the area of the rectangle ABCD is 50 + 30 = 80, and by subtracting
the sum of the three given triangle areas, 25 + 28 + 15 = 68 from the area
of the rectangle ABCD we find that the missing area, that of triangle APD,
is 12.
Figure 5.16
It should be noted that the above problem could also have been presented
as a parallelogram ABCD rather than a rectangle ABCD, and the method of
solution would have been analogous to the one presented here.
Perpendicular Bisectors of Extended Sides of a

Cyclic Quadrilateral
Assume that in a cyclic quadrilateral ABCD the opposite sides are not par-
allel, as shown in Figure 5.17. If these sides are extended, one gets the
intersection points P and Q, respectively. Unexpectedly, we need to justify
that the angle bisectors of the angles APD and DQC are perpendicular.
Geometric Gems 231
Figure 5.17
Solution: As we can see in Figure 5.18, the interior angles DAB and ABC
of the cyclic quadrilateral are denoted as α and β, respectively. Since the
opposite angles of a cyclic quadrilateral are supplementary, we get ADC =
180◦ − β and CDQ = β = ADP. From that we can conclude that
in triangle ADP, we have APD = 180◦ − (180◦ − α + β) = α − β,
whereupon, because of the angle bisector, we have EPD = α−β 2
. We also
180◦ −(α+β) ◦ α+β
see in ABQ we have AQE =

2
= 90 − 2 . And because the
◦
angle sum of a quadrilateral (also in concave ones such as
Ä PEQD) is 360
ä , we
α−β α+β
can determine that PEQ = 360 −(180 +β)− 2 − 90 − 2 = 90◦ ,
◦ ◦ ◦
which is what we had to show.

Figure 5.18
A Curious Square Consisting of Eight Rectangles

with Equal Areas
Square ABCD consists of eight rectangles of equal area, as is shown in

Figure 5.19. If the width of the shaded rectangle is 35, we need to find the
area of the square ABCD.
Figure 5.19
Geometric Gems 233
Solution: Here algebra is a helpful tool to find the sought-after area. As

shown in Figure 5.20, we need to get the values of a, f , and x, which will
then provide the length of a side of the square and subsequently the area of
the square.
Figure 5.20
In Figure 5.20, the various lengths are labeled a, b, c, d, e, f, x. Now,

we will generate the steps to determine the values of the crucial values
a, f , and x, which will determine the length of the side of the square. We
begin by recognizing that the area AHOI = area INRJ, which gives us
ab = 2ac, and, thereby, b = 2c. Then since area EFPR = area INRJ,
we can conclude that 35 × (b + c) = 2ac, which then yields 2a = 105.

3c
Now consider that the area INRJ = area JPMK, which yields 105c = 140d
and d = 34 c. Next, we will consider that area FGQM = area JPMK and

then get f 3 + 34 c = 140 34 c , which gives us f = 28. We then

15 d
4 Ä ä
use area KQLD = area FGQM to get 168e = 15 4
c 28, which results
inÄe = 8äc. Finally,Äweä use area GBCL = area KQLD and conclude
5
x 154
+ 58 c = 168 58 c , thus, x = 24. We then have the bottom side
length of the square as a + a + 35 + f + x = 105 + 35 + 28 + 24 = 192.
Therefore, the area of the square is 1922 = 36,864. We could have found
the area of the square by obtaining the values b, c, d, e, however, this
would have been more difficult since these would not have been integer
values.
Comparison of Circular Areas
In Figure 5.21, we show a circle and semicircle with the same center
point at O. The point B of the semicircle is the intersection point of the
tangents to the circle at A and C, respectively. Our challenge here is to
determine how the area of the circle compares to the area of the (bold)
semicircle.
Figure 5.21
Geometric Gems 235
Solution: Let us begin by assuming that the radius of the circle is 1. With
the perpendiculars at points A, O, and C, we have quadrilateral ABCO as
a rectangle. However, since tangents AB and CB are equal, we know that
ABCO is a square. Then in Figure 5.22,√by the Pythagorean theorem, since
AO = OC = AB = 1, we get OB = 2, which is the radius of the bold
semicircle. Thus, the area of the√circle is π(AO)2 = π , and the area of the
(bold) semicircle is π(O B) = ( ) = π . Quite unexpectedly, we find that
2
2 π 2
2 2
the two areas are equal.
Figure 5.22
Comparison of a Circular Area with a Square
In Figure 5.23, the square and the circle share a common center point, P.
The line PQSR = PA and is parallel to the side of the square. Also point
Q is on a side of the square, while point S is on the circle and SR = 2SQ.
We now need to find how the area of the circle compares to the area of the
square.
Figure 5.23
Solution: If the side length of the square is 1, then by the Pythagorean

theorem
√ √ √
2 2 1 1 2 1
AP = = PR, and RQ = − , whereupon, SQ = − .
2 2 2 3 2 2
Then
√ √
1 1 2 1 2+ 2
PS = PQ + PS = + − = .
2 3 2 2 6
√
Thus, the area of the circle is π · ( 36 ) ≈ 1.017. The area of the square is
2
2+ 2
simply 1, therefore, the circle is about 1.7% larger than the square.
What Is the Area of the Circle?
In Figure 5.24, we have two squares ABCD and DEFG with areas 18 and
4, respectively. The two squares have a common vertex at point D, and
between them CDG = 45◦ . We need to find the area of the circumscribed
circle.
Geometric Gems 237
B
G
O
18
F
4
45°
Figure 5.24
Solution: In Figure 5.25, we find that FDG = 45◦ = CDG, which

means that together they form a right-angle FDC. This allows us to conclude
that ADF is a straight line and, therefore, we find that BAF is a right
triangle. Thus, BOF is the diameter of the circle. Now, since the area of
square DEFG is 4, we have GF = 2 and applying
√ the Pythagorean theorem
to triangle DEF, we find that DF = 2 2. Since AB and √AD are sides √
of square ABCD whose area is 18, we get AB = AD = 18 = 3 2.
Applying» the Pytharorean theorem to triangle ABF, we find that the diameter
√ √ √ √ √ √
BF = (3 2)2 + (3 2 √ + 2 2)2 = 18 + 50 = 68 = 2 17. Thus,
the√radius of the circle is 17, and the area of the circumscribed circle is
π( 17)2 = 17π ≈ 53.4.
Figure 5.25
Comparing Related Circle Areas
An equilateral triangle and a regular hexagon have the same perimeter, as

shown in Figure 5.26. We need to determine the ratio of the areas of their
inscribed circles.
Figure 5.26
Solution: One can answer this question rather quickly without any signif-
icant calculations. We must realize that the two polygons, the triangle and
Geometric Gems 239
the hexagon, have the same height as shown in Figure 5.26. Therefore, half
the height, h, would be the radius of the circle inscribed in the equilateral
Ä ä2
hexagon, yielding an area of π h2 = πh4 . The center of the inscribed cir-
2
cle in the equilateral triangle is at the point of intersection of the medians,

which is a trisection point for each of the medians, as in this case the medi-
ans are also the angle bisectors, which determine the center of the inscribed
circle. This then implies that the radius of the circle inscribed in the equilat-
eral triangle is h3 . Thus, the area of the triangle’s inscribed circle is equal to
Ä ä2
π h3 = πh9 . Thus, the ratio of the corresponding areas is 4:9. We could
2
have come to this response more quickly by considering that since the ratio
of the radii of the two circles is 2:3, then the ratio of the areas is 4:9.
Another Comparison of Areas
In Figure 5.27, a semicircle is placed inside a square so that the diameter of

the semicircle forms 45◦ angles with two sides of the square and the other
two sides of the square are tangent to the semicircle. The challenge here is
to determine how the area of the semicircle (the shaded region) compares
to the non-shaded regions in the square.
Figure 5.27
Solution: Let us denote the radius of the semicircle by r, so its area is π2 r 2 ≈

1.57r 2 . The side length of a square with diagonal r is √r2 . Therefore, the
area of the unshaded isosceles right triangle EBH, as shown in Figure 5.28,
2
is 2 · √r2 = r 2 . The upper left unshaded “corner” (KDJ) of the square
can be found by taking one-quarter of the circle from which the semicircle
came and subtracting it from the square KMJD. That gives us r 2 − π4 r 2 .
Each of the two other unshaded regions near the√corners A and C, which are
congruent, has an area of r √r2 − r4 − π8 r 2 = 4 2−2−π
2
8
r 2 , which is obtained
by MG = √2 = MF. The sum of the non-shaded regions, therefore, is
r
√ √
2 − π4 + 2 4 2−2−π
8
r 2 = 3+2 2 2−π r 2 ≈ 1.34r 2 . Thus, the ratio of the
areas of the shaded to the non-shaded region is approximately 1.57:1.34 ≈
1.17:1.
Figure 5.28
Finding the Area of an Awkwardly Placed Square
In Figure 5.29, the square ABCD is placed on a circle so that vertices C and
D lie on the circle which is tangent to side AB of the square. Line segment
EF, whose length is 1, is the perpendicular bisector of DC at point E and
Geometric Gems 241
intersects the circle at point F. The challenge here is to find the area of
square ABCD.
A G D
E F
B H C
Figure 5.29
Solution: Let a represent the side length of the square and let r represent
the radius of the circle. If we imagine that the square had slid a distance
of EF away from the right side of the circle (where, before moving, due to
symmetry reasons it had been tangent to the circle at F and the left vertices
had been at G and H ) so that it is tangent to the left side of the circle, we can
conclude that BH must indicate the distance moved, so that BH = EF = 1.
We then have GD = AD − AG = a − 1. In Figure 5.30, we see that
angle GDC is a right angle, therefore, GC must be the diameter of the
circle. Then, by the Pythagorean theorem applied to triangle GDC, we have
(2r)2 = (a − 1)2 + a 2 ; if we insert 2r = a + 1, we get a = 4, which yields
the area of the square a 2 = 16.
A G D
a-1
r
a a
O 1
E F
r
B H C
Figure 5.30
Comparison of Perimeters
In Figure 5.31, we show a square with two of its vertices on the circle, and
the opposite side tangent to the circle. Our task here is to determine how
the two perimeters compare. Furthermore, we also would like to know the
fraction of the vertical edges of the square which lie outside the circle.
A E D
B C
Figure 5.31
Solution: We denote half of the side length of the square with a, and
the circle’s radius with r, as we are shown in Figure 5.32. Applying √ the
Pythagorean theorem to triangle BOF, we find that √ OF = r − a 2 .2
However, we can also represent OF as 2a−r. Therefore, r 2 − a 2 = 2a−r,

and by squaring both sides of the equation, we get r 2 −a 2 = 4a 2 −4ar +r 2 ,
which then gives us r = 54 a. Thus, the perimeter of the square is 8a, and the
circumference of the circle is 2πr = 2π 54 a = 5π 2
a ≈ 7.854a. So, we can
see that the perimeter of the square is slightly longer than the circumference
of the circle. √ »
The vertical overlap is 2a − 2 r 2 − a 2 = 2a − 2 25 16
a 2 − a 2 = a2 ,
thus, the overlap is exactly one-fourth of the square’s side length.
Geometric Gems 243
A a E a D
2a
r r 2 – a2
2a – r
B a F a C
Figure 5.32
The Unexpected Region Evolving from Two Overlapping Quarter

Circles
In Figure 5.33, there are two-quarter circles in a square. The challenge here
is to determine the fraction of the square that is indicated by the shaded
region determined by points ADF.
Figure 5.33
Solution: Let the side length of the square be 1. In Figure 5.34, we see
that triangle FED is an isosceles right triangle and FDE = 45◦ . There-
fore, DF extended to point B becomes the diagonal of square ABCD. Also,
AD = FD = 1, which is the hypotenuse of triangle FED. Therefore, apply-
ing the Pythagorean theorem to isosceles right triangle DEF, the small
quarter circle has radius FE = √12 . The area by the shaded region bounded
by ADF can be determined by
â ì
1 ·π √1 · √1
π 2 2 2 π π 1 1
− − = − − = .
8
4
2 8 8 4 4
half of the big small isosceles
quarter circle quartercircle right triangle FED
Figure 5.34
Comparing the Area of Seemingly Unrelated Triangles
The semicircle on AB is intersected by a circle at points P and R, which

is shown in Figure 5.35. Line AP extended intersects the circle at point D,
and the line BR extended intersects the circle at point C. Finally, AC and
Geometric Gems 245
BD intersect at E. The challenge here is to find the relationship between the

areas of triangles ADE and BCE.
D C
E
P
A B
Figure 5.35
Solution: In Figure 5.36, when we draw PR and CD, we note that there are
two cyclic quadrilaterals: ABRP and PRCD. This allows us to generate the
ABRP PRCD
following: 180◦ − BAP = PRB = PDC = ADC and therefore,
is cyclic is cyclic
supplementary angles BAD and ADC enable us to conclude that AB||CD.
From this, it follows that the triangles ACD and BCD have equal areas
since they share the same base CD and have equal altitudes. If we now
subtract the common area of ECD from these two equal-area triangles,
we are left with the area of triangle ADE equal to the area of triangle BCE.
D C
P E
A B
Figure 5.36
Four Equilateral Triangles and Two Given Areas
In Figure 5.37, we have four equilateral triangles: ABC, CDE, HEF,

and AFG. Triangle HEF is placed to that its sides are perpendicular to the
sides of triangle ABC. We know that know that the area of triangle CDE is
5, and the area of triangle AFG is 20. Here the challenge is to find the area
of triangle HEF.
G F
E D
A H C
Figure 5.37
Solution: Since the area of triangle AFG is four times the area of trian-
gle CDE, the sides are in a ratio of 1:2 so that we can let EC = a and
AF = 2a, and their areas are in a ratio of a 2 :4a 2 . We can also show
that AHF ∼ = BFE ∼ = CEH because each one has a 60◦ angle, a 90◦
angle and an equal hypotenuse. We can apply the Pythagorean √ theorem to
one of these congruent triangles: a + x = (2a) and x = a 3, which
2 2 2
is a side of the equilateral triangle HEF. Therefore, the area of triangle

HEF is in the ratio with the areas of the other two√triangles as follows:
areaCDE : areaAFG : areaHEF = a 2 :(2a)2 :(a 3)2 = a 2 :4a 2 :3a 2 .
Thus, the area of triangle HEF is 3 × 5 = 15. As a bonus, we can find
that the area of the equilateral triangle ABC would then be 9 × 5 = 45.
Geometric Gems 247
Five Congruent Triangles in a Rectangle
In Figure 5.38, we show five congruent right triangles fitted into a rectangle.
The challenge here is to find what fraction of the rectangle’s area is covered
by the shaded triangles.
Figure 5.38
Solution: As we show in Figure 5.39, we can represent the sides of one

of the congruent right triangles as a ≤ b ≤ c, so that by the Pythagorean
theorem we get a 2 + b2 = c2 . Furthermore, we see that the length of the
rectangle can be represented in two ways, which gives us the following
equation: 2c = 2a + b then b = 2(c − a). If we insert b = 2(c − a) into
the Pythagoras equation and solve for c (c ≥ a), we getÄc = 53 a, ä and if
we then substitute that into the value of b, we get b = 2 3 a − a = 43 a.
5
This indicates the ratio ab = 43 , which leads us to the notion that we have
a 3-4-5 right triangle. The area of one of these congruent right triangles is
1
2
ab = ( 12 )(3)(4) = 6, and with h as the altitude to the hypotenuse of one of
the five congruent triangles, this also can be expressed by 12 ch = 12 (5h) = 6
so that the altitude h = 2.4. Thus, the ratio of the sum of the areas of the
5× 3×4
five right triangles to the area of the rectangle is 10×6.4
2
= 30
64
= 15
32
.
G B H
a b a b
h
h
A D
c C c
F
c c b
b
c a
J a b E a I
Figure 5.39
Two Overlapping Congruent Right Triangles
In Figure 5.40, we show two congruent overlapping right triangles. Our

challenge is to find the fraction of the entire shaded figure that is covered
by the overlapping triangles.
Figure 5.40
Geometric Gems 249
Solution: In Figure 5.41, triangles ABC and DEB are congruent with an
area of X . Since ACB = DBE we know that the darker-shaded triangle
BFC is isosceles so that FC = FB. Because FBA = 90◦ − DBE and
BAF = 90◦ − ACB, we have FBA = BAF, and, thus, ABF is isosceles.
It follows that FA = FB and then FC = FA. Therefore, we have triangle

AFB equal in area to triangle CFB, since they both have the same altitude
from B to CFA and have equal bases. This then tells us that the area of the
triangle BFC is X2 . The complete shaded area is 2X − X2 = 3X 2
, and thus,
X
the sought-after fraction is 2
3X = 13 .
2
E
F
B C D
Figure 5.41
Discovering a Square’s Area
Square ABCD has its opposite vertices A and C connected with three
lines with perpendicularity shown as in Figure 5.42 and lengths CF = 3,
EF = 1, and AE = 4. Our challenge here is to find the area of square
ABCD.
B C
3
F
1
s
E
A D
Figure 5.42
Solution: As we show in Figure 5.43, we begin by extending AE to intersect

at point G the perpendicular from point C to AE extended. From rectangle
CGEF, we find that CG = 1√and AG = 7. Following the Pythagorean
theorem, we find that AC = 50. This gives us one aspect of the size of
the square, which will then allow us to find its area. The side of the square
ABCD is then easily found by the Pythagorean
√ theorem applied to triangle
ABC, where 2s 2 = AC 2 , then 2s 2 = ( 50)2 , and s = 5, and this yields an
area of 25.
B C
1
3
G
F
3
1
s
E
A D
Figure 5.43
Geometric Gems 251
Squares on the Sides of a Parallelogram
On the sides of the parallelogram ABCD, squares are erected, as shown in

Figure 5.44. The centers of the four squares are E, F, G, and H . Curiously,
regardless of the shape of the parallelogram, these four points E, F, G, and
H turn out to be the vertices of a square. We need to show how this can be
proved.
C
F
H A
Figure 5.44
Solution: Consider the four triangles HAE, HDG, EBF, and FCG
shown in Figure 5.45. We can show that these four triangles are all congruent
by simply choosing two of these triangles to prove their congruence and then
realizing that the procedure can be followed for the other two as well.
C
F
H
A
Figure 5.45
Consider the two triangles FBE and HDG. We have many 45◦ angles
in the diagram such as GDC = HDA = FBC = EBC = 45◦ , and
we also know that ADC = ABC. Therefore, by addition, we find that
GDH = EBF. Furthermore, DH = BF and GD = BE. Thus, FBE ∼ =
HDG and GH = FE. Using this procedure, we can show that the four
triangles are all congruent with one another. This makes quadrilateral GHEF
a rhombus. However, we can also show that each of its angles is a right angle,
since DHG = AHE and because DHG + AHG = 90◦ , it follows that
AHE + AHG = 90◦ , which is one of the angles of the rhombus, thus,
making it a square.
Geometric Gems 253
Unexpected Similar Triangles
On the sides of a triangle XYZ, similar triangles (with equal angles α, β, γ )

are drawn outwardly so that each of the angles is at the outer vertex exactly
once, as shown in Figure 5.46. The points P, Q, and R are the centers of
the circumscribed circles of the external triangles. We need to show that the
triangle PQR is actually similar to the three similar triangles.
Figure 5.46
Solution: As shown in Figure 5.47, let A be the intersection point of

the circumcircles p and q, then we have the supplementary opposite
angles of cyclic quadrilaterals ZAXE and ZAYG, ZAX = 180◦ − β and
ZAY = 180◦ − α, respectively. Thus XAY = 360◦ − ( ZAX + ZAY) =
360◦ − (180◦ − β + 180◦ − α) = α + β = 180◦ − γ . And this, in turn,

means that point A lies on the circumcircle r, and we have proved that the
three circumcircles intersect each other at point A. Furthermore, the sides of

triangle PQR, which are PQ, QR, and RP, are perpendicular to AZ, AX, and
AY, respectively. As we had earlier, ZAY = 180◦ − α and ZAY = DAC,
thus, we have DAC = 180◦ − α. Therefore, since quadrilateral DPAC is
cyclic, because it has two opposite right angles, we then have P = α.
This same procedure can be used for the other two angles of triangle PQR.
Hence, for the interior angles of the triangle PQR, we get α, β, γ and
thus, PQR is similar to the three externally constructed triangles.
Figure 5.47
Geometric Gems 255
Tangential Quadrilateral
As we can see in Figure 5.48, the circle with center I is inscribed in quadri-
lateral ABCD. Perpendiculars HE, EF, GF, and HG are drawn to AI, BI, CI,
and DI, respectively. These perpendiculars determine quadrilateral EFGH.
Unexpectedly, we find that the diagonals EG and FH intersect at point I .
Figure 5.48
Solution: Since one pair of opposite angles of each of the quadrilaterals

AIBE, DICG, BICF, and HDIA are right angles, each of these quadrilaterals
is cyclic. Let the interior angles of the circumscribed quadrilateral ABCD be
2α, 2β, 2γ , 2δ, as shown in Figure 5.49. Since inscribed angles measured
by the same arc are equal, we can see how the angles at E, F, G, H are
determined by α, β, γ , δ. As an example, consider the angles at point
H : Since IAD = α and in the circumcircle of the cyclic quadrilateral
AIDH, the angle IHD is another inscribed angle measured by ID, ı we can
conclude that IHD = α. And using the arc IA ı in the same circle, we
have ADI = δ = AHI. Since the sum of the interior angles of EFGH
is 360◦ , we know that 2α + 2β + 2γ + 2δ = 360◦ , which implies that
α + β + γ + δ = 180◦ , and this is the sum of the angles in triangle HEF.
Hence, EIG is a straight line passing through point I , essentially indicating
that the diagonal EG contains point I .
Figure 5.49
The Moving Semicircle
In Figure 5.50, a semicircle is placed with its diameter’s endpoints on two

perpendicular lines. A point P is placed somewhere on the arc of the semi-
circle. As the endpoints of the diameter move along the perpendicular axes,
we need to determine what the path of the point P on the semicircle’s arc is.
Geometric Gems 257
O A
Figure 5.50
Solution: First, one must realize that when the semicircle is completed,
it must pass through the perpendicular intersection, since the diameter
of the circle generates a right angle, which happens to be point O. This
can be seen in Figure 5.51. As the points B and A slide along the two
axes, one constant will be POB = PAB since they are both mea-
sured by arc PB. Therefore, the point P will always travel along the
line PO.
O A
Figure 5.51
Shaded Area in a Rectangle
In Figure 5.52, we show a rectangle ABCD, with two congruent right trian-
gles, AFB and DEC, whose right angles are at points E and F, which are
on sides AB and DC, respectively. Our task at hand is to determine what
fraction of the rectangle is shaded.
Figure 5.52
Solution: As we show in Figure 5.53, we draw the line segment EF and

two rectangles ADFE and BCFE arise. We can easily see that one-fourth
Geometric Gems 259
of each of the two rectangles’ respective areas is shaded. This results in

one-fourth of the total rectangle ABCD being shaded.
D F C
G
H
A E B
Figure 5.53
A Surprise Chord Length
On diameter AB of a semicircle with radius r, an arbitrary point X is cho-

sen, and two lines are drawn from point X to the semicircle’s arc, which
makes three 60◦ angles with AB, as shown in Figure 5.54, and intersects the
semicircle at points C and D. The challenge here is to determine the length
s of chord CD in relation to the radius of the circle.
Figure 5.54
Solution: As we show in Figure 5.55, we draw a perpendicular to AB from

point D which intersects the circle at point F. This produces two right
triangles XED and XEF with angles of 60◦ and 30◦ . Because of the
three 60◦ angles at point X , one can see that F lies also on the extension
of CX, and that yields the 30◦ inscribed angle CFD with arc CD. Thus,
the corresponding central angle COD is 60◦ . We then have an isosceles
triangle CDO with the vertex angle of 60◦ , thereby, producing equilateral
triangle CDO, where s = r.
D
60°
s
C r
60° 60°
60° 60°
A 60° B
X O E
30°
Figure 5.55
Area of a Curious Circle
Figure 5.56 shows four squares sharing a common vertex. The areas of three
of these squares are shown to be 16, 25, and 100. A circle passes through
one vertex of each of the squares. The challenge here is to find the area of
the circle.
16
25
100
Figure 5.56
Geometric Gems 261
Solution: The side lengths of the three squares with given areas are 4, 5,
and 10. By joining the points where the circle intersects the vertices of the
squares, as we do in Figure 5.57, we find that, because the products of the
segments of intersecting chords are equal, we have a × c = d × b and
5a = 4 × 10 = 40 so that a = 8. Since DB and A C are perpendicular to
AC, we have DB parallel √ to A C. Therefore, the arc A B = arc CD. Thus,
we get A B = DC = c2 + d 2 from the Pythagorean theorem applied to
triangle DEC. We then
√ apply the Pythagorean theorem again, first to triangle
AEB to get AB = a 2 + b2 and then to triangle A A B to find √ that A A =
√ a 2 +b2 +c2 +d 2
a 2 + b2 + c2 + d 2 . Thus, the radius of the circle is AM = .
Å√ ã2 2
= (
a +b +c +d
2 2 2 2 π a +b +c +d )
2 2 2 2
The area of the circle is, therefore, π 2 4
,
and for the given lengths, the area of the circle is 64+100+25+16
4
π = 205
4
π ≈
161.
Figure 5.57
A Square on Two Parallel Lines
A square ABCD with side length a is placed onto two parallel lines whose
distance apart is a, such that the square has one vertex on the bottom parallel,
one vertex above the top parallel line, and two vertices between the parallel
lines. We show two such cases in Figure 5.58. We need to show that in
this situation the “overhanging” triangle AEF has a constant perimeter
regardless of the position of the square. Furthermore, we need to find the
value of this perimeter.
Figure 5.58
Solution: We can begin by conjecturing that the perimeter of triangle AEF

is always 2a, which is twice the side length of the square. This value can
be conjectured by considering an extreme position, where A gets arbitrarily
close to the top parallel (or where we would find A = E or A = F). Then
the triangle AEF degenerates to the double line EF, where EF obviously
moves to overlap the side length a of the square. But now let us prove this
conjecture by showing that this holds for all positions of A above the top
parallel.
Suppose the square ABCD is rotated around point C. We have to show
that the perimeter of AEF does not depend on the position of point A, as
Geometric Gems 263
shown in Figure 5.59. Draw a line perpendicular to the bottom parallel at

point C and it intersects the top parallel line at point J .
Figure 5.59
Now two quadrilaterals appear: CDFJ and CJEB. Each has two congru-
ent right triangles with equal legs and a common hypotenuse, so that trian-
gles CBE and CJE are congruent, as are triangles CDF and CJF. Therefore,
EJ = EB and JF = FD, and one can see that the perimeter of triangle EFA
is AE + EJ + AF + FJ = AE + BE + AF + FJ = 2a, which is twice the
side length of the square.
Unexpected Equal Line Segments
Let ABCDE be a pentagon with equal sides and two right angles at con-
secutive vertices C and D. Furthermore, we let P be the intersection
point of the diagonals AC and BD. Our challenge here is to prove that
PA = PD.
Solution: We intentionally did not provide a diagram following the state-

ment of the problem, since making such a sketch is closely tied to the
solution.
Figure 5.60
As shown in Figure 5.60, the pentagon has five equal sides and two
right angles at consecutive vertices C and D; we can then conclude that
EBCD is a square and ABE an equilateral triangle. Therefore, the angle
at B, namely, ABC = 150◦ , which establishes the base angles of isosceles
triangle ABC to be 15◦ . Therefore, EAC = 60◦ − 15◦ = 45◦ . As the
diagonal of the square bisects the angles, we have EDB = 45◦ . Moreover,
AED is isosceles and has equal base angles, namely, EDA = EAD.
Thus, by subtraction, PDA = PAD. Hence, we have isosceles APD
with PA = PD.
A Surprising Angle Sum
In Figure 5.61, we show two equal-area isosceles triangles ABC and

ABD, where AB = CB and AB = BD. Our challenge here is to find the
sum of ACB + ADB.
Geometric Gems 265
C
D
A B
Figure 5.61
Solution: Since the area of triangle ABD is equal to the area of triangle
ABC, and they share the same base, therefore, their altitudes must be equal,
thus, making CD parallel to AB. Let α = ACB and β = ADB. Because of
ABC is isosceles it has equal base angles α, and then ABC = 180◦ − 2α.
As we can see in Figure 5.62, the alternate-interior angles of the parallel
lines, ADC = BAD = β. Therefore, the base angles of the isosceles
triangle CBD are 2β. Once again, the alternate-interior angles of the parallel
lines gives us ABC = BCD, which is that 180◦ −2α = 2β, or 2α +2β =
180◦ , which then provides us with our sought-after conclusion α +β = 90◦ .
Figure 5.62
Determine the Angle
In Figure 5.63, we show two isosceles triangles, ABC and ABD, that have
equal areas. We also show that AB = BD, AC = BC, and ACB = 90◦ . Our
challenge is to find the measure of angle ADB.
C D
A B
Figure 5.63
Solution 1: First, we know that the altitudes of the two isosceles triangles
are equal since they share the same base, AB, and have equal areas. This
implies that in Figure 5.64 C D||AB, and therefore, BAD = ADC, as
they are alternate-interior angles of the parallel lines, thus ADC = α. If x
denotes the length of √ each of the two legs in the right triangle ABC, we
have BD = √AB = x 2. With the help of the law of sines applied to BCD,
2
◦
x 2
one gets sin(45 ◦ ) = sin(2α) , and this implies sin(2α) =
x √ ) = 1 , and since
sin(45
2 2
sin 30◦ = 21 , α = 15◦ .
C D
45°
x x
x 2
A x 2 B
Figure 5.64
Solution 2: In Figure 5.65, we note that the altitude (y) in ABC is half
the length of its base AB. Then because AB = BD, we have the right
triangle BED, with the hypotenuse BD twice the length of DE, which is a
2 This law states that in a triangle ABC with sides a, b, c, the following equation holds:
a b c
sin A = sin B = sin C .
Geometric Gems 267
property of a 30◦ –60◦ –90◦ triangle, where BDE = 60◦ , so that BDC =
DBE = 30◦ . Because of the equality of alternate-interior angles, we have
ADC = DAB = ADB. Therefore, ADB = 1 (30◦ ) = 15◦ .
2
C D
45° 60°
x x y
y
2y
30°
A B E
2y
Figure 5.65
Determine the Fraction of the Rectangle that Is Shaded
In Figure 5.66, we show a rectangle with two tangent semicircles emitted

from two consecutive vertices, C and D, of the rectangle. The common
tangent to the two semicircles is BGH. Our challenge is to find the fraction
of rectangle ABCD which is shaded, namely, quadrilateral ABHD.
B F
C
A
E D
Figure 5.66
Solution: In Figure 5.66, we see that from the point H there are three
tangential segments, HC, HG, and HD, which all have the same length.
Therefore, H must be the midpoint of CD. The area of the rectangle whose
length and width are BC and CH, is twice the area of triangle BCH. Hence,
the area of BCH = 14 area rectangle ABCD, whereupon it follows that the
area of the shaded region is 34 area rectangle ABCD.
Length of a Common Tangent
A semicircle and a quarter circle are inside a square, as shown in Figure 5.67.
The radius of the semicircle is 5, and the radius of the quarter circle is 8.
We need to determine the length of the common tangent whose endpoints
G and F are on the opposite sides of the square.
K L C
D
A B
N
Figure 5.67
Solution: In Figure 5.68, we find that point K is the midpoint of the diameter
DL of the semicircle. And GF is the common tangent whose length we are
seeking. Since BE is perpendicular to GEF and KE is perpendicular to
GEF, being radii to the points of tangency, we know that KEB is a straight
Geometric Gems 269
line. Therefore, the length of BK is equal to the sum of the radii which is
5+8 = 13. When we draw a line through C parallel to GF intersecting AD at
point H , this creates parallelogram GFCH, which then gives us CH = GF.
Furthermore, CKB + KCH = 90◦ and CKB + KBC = 90◦ , therefore,
KCH = KBC. This enables us to show that CDH ∼ = BCK. Thus,
since BK = CH and CH = GF, we then have GF = BK = 13, which
satisfies our challenge.
K L C
D
E
H
A N B
Figure 5.68
Two Circles and an Angle Bisector
In Figure 5.69, we see that circle c is internally tangent to a larger circle k

at point P. We select any point Q on circle c. The tangent line to circle c
at point Q intersects circle k at the points A and B. Quite unexpectedly, we
note that PQ bisects APB. Our challenge here is to verify this appearance.
Figure 5.69
Solution: In Figure 5.70, we see that point E is the intersection of circle k

and PQ. Also, RP is the diameter of circle c, whose center is point M, and
PR extended intersects circle k at point G. Furthermore, the point D is the
intersection point of AB and EM. Then we focus on the shaded quadrilateral
MRQD, and in particular at its angles. If we can prove that MDQ = 90◦ ,
then we would have completed the proof because EF is a diameter of k and
this implies AD = BD or in terms of arcs AE = BE, which then justifies
the equality of the two inscribed angles, that is, APE = BPE.
Geometric Gems 271
Figure 5.70
Fortunately, in quadrilateral MRQD, it is easy to see that MDQ = 90◦ ,

which we will now pursue. Let α be RPQ. Related to circle c we note
that RQD = RPQ = α as they are both half the angle related to the arc
QR. The exterior angle QRM of triangle QRP is equal to 90◦ + α. Since
EMG is a central angle, we have EMG = 2α. Therefore, since EMP
is supplementary to EMG, we get EMP = 180◦ − 2α. Thus, the angle

QDM = 360◦ − α − (90◦ + α) − (180◦ − 2α) = 90◦ . This then produces
AD = BD which, in turn, gives us arc AE equals arc BE, which finally —

via the inscribed angle theorem — provides our desired result, namely, that
APE = BPE.
Two Tangent Quarter Circles in a Rectangle
The bold line shown in Figure 5.71 is a common tangent of the two tangent
quarter circles with centers at opposite vertices in a rectangle. We need to
determine what part of the rectangle is the shaded area between the common
tangent and the diagonal.
Figure 5.71
Solution: In Figure 5.72, let the width of the rectangle ABCD with center M
be 1. Then we can see that the triangles AMD and BMC are equilateral
with side lengths 1, since the radii of the quarter circles are equal. Also,
B M = DM√and AM = CM, as the diagonals bisect each other. This
yields a = 3 for the length of the rectangle, which is comprised of two
altitudes containing the point M of these two equilateral triangles. Since
the radius is perpendicular to the tangent at the point of contact, we have
◦ ◦
right triangle DME √ (with angles 30 and 60 ) so that we √ can compute
DE = √3 = 3 3, and this means since DC = a = 3, then EC =
2 2
√ √ √
DC−DE = 3− 23 3 = 13 3Ä =ä 13 a. Hence, the area of the shaded triangle
MCE is 13 area DMC = 13 14 area ABCD = 12 1
area ABCD, and since
Ä ä
1
we have two such congruent triangles, the shaded area is 2 12 area ABCD
or 16 of the rectangle’s area.
D C
E
1
M
1
F
A a B
Figure 5.72
Geometric Gems 273
Alternative solution
In Figure 5.73, let the width of the rectangle ABCD with center M be
1. Then we can see that the triangles AMD and BMC are equilateral
with side lengths 1 since the radii of the quarter circles are equal. Also,
BM = DM, and AM = CM as the diagonals bisect each other. From
that we can conclude that the triangles AFM and MCE are congruent
isosceles triangles (ASA) with base angles 30◦ , and we also have FMG
as a 30◦ –60◦ –90◦ -triangle. Then reflecting the right triangle FMG in MF
yields triangle FMG with the right angle FG M at point G , which is the
midpoint of AB (note: MFG = 60◦ = BFM). Therefore, area AFM =
2
3
area AG M = 23 · 18 area ABCD = 12 1
area ABCD,
Ä and sinceäwe have two
such congruent triangles, the shaded area is 2 12 area ABCD , or 16 of the
1
area of the rectangle.
Figure 5.73
A Very Special Triangle
For triangle ABC, point D is the intersection point of the angle bisector of
angle BAC with the side BC. The circumcenter, O, of the triangle ABC
coincides with the incenter, I , of the triangle ADC (Figure 5.74). Our
challenge is to find the measures of the angles of the triangle ABC.
O=I
D
A B
Figure 5.74
Solution: In Figure 5.75, let BAC = α and the incenter I of ADC and
circumcenter O of ABC coincide. Then, since AO bisects CAD, we know
BAO = 3 α, and AO = CO, we know that AOC is isosceles with base
4
angles of α4 , and the same holds for BOC. We then can conclude that
ABO must be an isoseles triangle with base angles of 34 α, where α is the
measure of the vertex angle. (Note: At O = I , we have twice an angle
of α2 ; these are exterior angles of the triangles AOC and BOC; as a
consequence, C, O/I , and the midpoint H of AB are collinear.) Since the
angle sum in ABO equals 180◦ , we have α + (2 × 34 α) = 180◦ from which
we get α = 72◦ . Hence, the triangle ABC must be a golden triangle,3
since it is an isosceles triangle with base angles of 72◦ and a vertex angle
of 36◦ , which is a triangle that provides many further wonders!
3 The golden triangle provides a plethora of amazing relationships which can be further
investigated in The Glorious Golden Ratio by A.S. Posamentier and I. Lehmann (Prometheus
Books, 2012).
Geometric Gems 275
Figure 5.75
Chapter 6
For Experts: Stepwise Solutions to Challenging Problems
Problem solving is a time-honored and important issue in mathematics.

Properly presented, problem solving should be an integral part of math-
ematics instruction. One can significantly benefit from being exposed to
tasks for which the problem-solving technique may be new, insofar as it is
not merely a repetition of previously exposed techniques. Of course, these
tasks should be interesting and challenging, but on the other hand, they must
not be too difficult because otherwise frustration will result. They should
be enlightening!
The exposure to new problem-solving techniques should provide one
with an understanding of what is meant by the process of doing mathematics
and further appreciating the mathematical results by justifying them.
It is very important to find tasks that are “problems” in the sense that they
promote creativity and spirit, motivate, are not limited to simple exercises,
but are also not too tricky so as to provide one with a realistic chance at
success. One can become self-confident and be further motivated by having
successful experiences. We are considering motivating and interesting tasks
that contain “real mathematics” that fit into the realm of problem solving and
that provide a broad range of readers with an appetite for more mathematics.
Problem solving should not necessarily be treated as a mathematical
topic as one would consider geometry, algebra, and calculus. Introducing
problem solving should not be restricted to a series of techniques in which
one is taught how to solve problems from a list of heuristic strategies or
tell the students what the possible phases in a problem-solving process are.
Ideally one should try to solve problems by oneself even though initially
277
it might be difficult to work independently. It might take quite some time

for the average reader to have some success, but in the long run this time
seems well spent towards experiencing the feeling what doing mathematics
means.
We do not propose that all mathematics should be dedicated to problem
solving since mathematics has many facts to be learned and many proce-
dures to be practiced.
To fully appreciate problem-solving techniques, we present several chal-
lenging problems (listed as Problem 1, Problem 2, . . .), each of which is
partitioned into preparatory problems (listed as Problem 1a, 1b, 1c, … for
Problem 1 etc.) so that problem solvers can better appreciate the stepwise
solution of the original problem.
In this final chapter, the problems are sometimes more advanced and of
a somewhat higher mathematical level. Yet, we believe a motivated reader
will be able to grasp the material quite well.
The Problems
Problem 1: Given a triangle ABC with incenter I , let D be the midpoint of

the side AC and let G be the point of tangency of the incircle and the side
AC. When D I is extended to meet GB, the point of intersection H is the
midpoint of GB, as shown in Figure 6.1. Why is that true? (The analogous
statement holds for the other two sides AB and BC as well.)
Figure 6.1
For Experts: Stepwise Solutions to Challenging Problems 279
Before we employ a purely geometric solution without algebra and var-

ious calculations (which we will demonstrate a bit further on), we will
present a solution that combines geometry and algebra.
The following series of problems will lead to a solution of Problem 1. The
principle of considering a series of preparatory problems, sort of partitioning
a problem, or a proof, into several parts, is often a good strategy for solving
challenging problems.
Problem 1a: One can establish the formula for the inradius ρ of a right
triangle ρ = a+b−c
2
, where a and b are the lengths of the legs and the
hypotenuse has length c, by an algebraic approach or just by inspecting
appropriately Figure 6.2. It should be easy to explain by merely inspecting
the diagram.
Figure 6.2
After drawing the radii of the inscribed circle from its center, I , to
the points of tangency on lines BC and AC, we can easily see that since
the tangent segments from each of the triangle vertices are equal, we get
a + b = c + 2ρ, which is the equivalent of ρ = a+b−c
2
.
Problem 1b: Analogously, one can establish the formula for the distance z
of the vertex C from the points of tangency with the incircle z = a+b−c
2
by
an algebraic approach or just by once again inspecting Figure 6.3. This is
analogous to Problem 1a, the only difference is that in the general case this
distance does not equal the inradius.
Problem 1c: Consider the situation shown in Problem 1, where triangle
ABC has a right angle at vertex C, as seen in Figure 6.2.
Figure 6.3
To prove that the extension of DI meets GB at its midpoint, we approach

the problem in another fashion. As shown in Figure 6.4, we assume that H
is the midpoint of GB and J denotes the intersection point of the horizontal
line through I and the vertical line through H . We then prove that the
triangles JIH and GDI are similar (that is to say athat the slopes of DI and
−ρ ρ
IH are equal). That means we have to show that 2 ρ = b −ρ holds.
2 2
Figure 6.4
This can be rewritten as a quadratic equation in ρ:

Å ã Å ã Å ã
a b ρ2 1 a b ab
−ρ · −ρ − = 0 ⇔ ρ2 − + ρ+ =0
2 2 2 2 2 2 4
ab
⇔ ρ 2 − (a + b) ρ + =0 (*)
2
√
a+b± a 2 +b2
Using the quadratic formula, we get the solution: ρ = √ 2 . Apply-
ing the Pythagorean theorem to triangle ABC, we have a + b = c and
2 2
because ρ < a+b+c2

we can say that the above quadratic equation (∗ ) is
equivalent to ρ = a+b−c
2
. Our aim was to use this result also for the general
case without further algebraic calculations: We wanted to show that the
points H (as the midpoint of GB), I , and D stay collinear if the point B is
moved horizontally in the height a = h b “above” b, but, unfortunately, we
did not succeed. Now let us move towards the general case using a different
approach with the next problem.
Problem 1d: Given a triangle with the sides a, b, c. Find a formula for the
distance x from C to the foot F of the altitude h b on the side b or on its
extension, as shown in Figure 6.5.
Figure 6.5
Let us first assume that F lies “to the right side of C,” as shown in Fig-
ure 6.5. Applying the Pythagorean theorem twice to the two right triangles,
BCF and ABF, yields: (b − x)2 = c2 − (a 2 −x 2 ), which then enables us to

h 2b
have x = a +b2b−c . If F lies to the left of vertex C (Figure 6.6), then this
2 2 2
value of x would be negative and we would get | x | = c −a2b−b .

2 2 2
Figure 6.6
Now, we are ready to solve Problem 1 using Figure 6.7. Here we assume
that F lies to the left of vertex C (the other case would work analogously).
Figure 6.7
With the same technique as in Problem 1c and with the well-known

and easy-to-prove formula for the inradius ρ of a triangle, we have
ρ = As = 22sA = a+b+c
b·h b
, where A denotes the area and s denotes the semi-
perimeter. Now we need to show the following (applying similar triangles
or using the concept of equal slopes):
hb
−ρ ρ
2
= (**)
|x|+ a+b−c
2
b
2
− a+b−c
2
2
At first glance, this seems rather complicated, but it is not; by cross

multiplying we get a linear equation for |x|:
Ä äÄ ä
hb
−ρ b
− a+b−c
a+b−c
2 2
ρ
2
= |x| + ⇔
2
2
Å ã
hb a+b−c
|x| = − 1 (c − a) −
2ρ 2
, we get after simplification |x| = c −a2b−b , which we

2 2 2
Using ρ = a+b+c
b·h b
know from Problem 1d. If one accepts that x is negative as in case of F

lying on the left side of C, there is no need for distinguishing the two cases.
In the equation (∗∗ ), one would have to write −x instead of |x|.
Now consider a completely different, yet more elegant, synthetic solu-
tion to Problem 1, where we will use some preparatory problems.
Problem 1e: Here we will prove that for triangle ABC the two distances
of the tangency points of the incircle and the excircle from the two vertices
A and C are equal, which we see in Figure 6.8, where we will prove that
CG = AG1 .
Let G and G 1 be the two tangency points (on the side C A) of the incircle
and the excircle, respectively, as shown in Figure 6.8. Then since the tangent
segments of a circle from an external point are equal, we have BR = BS
and BP = BQ, and furthermore by subtraction PR = Q S(∗ ).
Furthermore, PC = CG, PR = PC + CR = PC + CG + GG1 = 2CG +
GG1 , and similarly, we have QS = QA + AS = GG1 + G 1 A + AS =
2AG1 + GG1 . Together with (∗ ) we can conclude that CG = AG1 .
Figure 6.8
Problem 1f: Let G be the point of tangency of the incircle on the side
C A; we will consider G as the “south pole” of the incircle and denote
its “north pole” as G . Our challenge is to prove that in this situation the
points B, G , G 1 are collinear (Figure 6.9), where G 1 denotes the point of
tangency of the excircle.
Figure 6.9
We will use triangle similarity, and let G 1 be the intersection point of

BG with AC. We then have to prove that G 1 is the point of tangency
of AC with the excircle. From G 1 , we draw a perpendicular to AC and
intersect it with the angle bisector B M1 of the angle at B, which yields M1
(Figure 6.10), and we must show that M1 is the center of the excircle. The
triangles BMG and BM 1 G 1 are similar, where M denotes the incenter

of ABC. We define k = BBMM1 . Then we have G 1 M1 = k · G M (here ρ
ρ
represents the inradius, G M = SM, of triangle ABC), and also the triangles
BMS and BM 1 S1 are similar, as well are triangles BMR and B M1 R1 . Hence,
we have also S1 M1 = k · S M and R1 M1 = k ·
R M . This means that
ρ ρ
G 1 M1 = S1 M1 = R1 M1 (= k · ρ), and because we have right angles at
R1 , G 1 , S1 , we, therefore, know that M1 is the center of the excircle.
Figure 6.10
Now we are prepared to solve Problem 1 in another way, synthetically,

without algebra: see Figure 6.11.
Figure 6.11
Since D is the midpoint of GG 1 (Problem 1e) and I is the midpoint of

GG , we see that D I ||G 1 G . And because of the collinearity of B, G , G 1
(Problem 1f), considering triangle BGG 1 , the extension of D I meets BG
at its midpoint (H ).
Another example for the principle of the stepwise solution is the follow-
ing. Here algebraic calculations do not play a major role as this problem is
largely geometric.
Problem 2: Let ABC D be a cyclic quadrilateral and S, T, U, V the incen-
ters of the four triangles, ABD, ABC, BCD, and ADC, which result
from the two diagonals of quadrilateral ABCD. Prove that STUV is a rect-
angle, which is shown in Figure 6.12.
This problem has been used for training students for mathematics
Olympiad competitions and is probably a bit of a challenge for a normal
geometry class. But again, one could think of solving a series of subsequent
problems — better accessible for students and easier to solve — which will
lead to a solution of Problem 2.
Here is a series of problems (2a, 2b, 2c) which prepare us for the solution
of Problem 2.
Figure 6.12
Problem 2a: Cloverleaf figure in a circle

On a circle c, shown in Figure 6.13, where we select the circular arcs AB
with center M1 and BC with center M2 are drawn which intersect at a point
S in the interior of circle c. Let M3 be the midpoint of the arc AC on circle
c, which does not contain M1 , M2 . Then we can also consider the arc CA
with center M3 running in the interior of circle c, as shown in Figure 6.13.
The result is a figure which seems to consist of three clover leaves (in other
words, three circular arcs intersecting each other at S). All three clover
leaves with the vertices A and S, B and S, and C and S consist of two
intersecting circular arcs which determine equal angles at both vertices due
to symmetry. The angles between the corresponding tangents are shown as
dotted lines. These angles we will call clover angles of the clover leaves.
We shall prove the following statements:
(1) The straight lines M1 C and M2 A pass through S.

(2) The straight line M3 B and the arc CA with center M3 pass through S.
Figure 6.13
(3) S is the incenter of the triangle ABC.

(4) The sum of the clover angles at the vertices of the clover leaves 1
A, B, C is 180◦ .
Following are the proofs of the four above statements:
(1) First, we will prove that M1 C passes through S. Since M1 is the
midpoint of ABC ˘ and M2 is the midpoint of BSC, ¯ we have M1 S =
M1 B and M2 S = M2 B. Hence, the quadrilateral SM 1 BM 2 is a
kite, and thus we have BM 1 M2 = SM 1 M2 . We also know that
BM 1 M2 = CM 1 M2 (from the equal arcs BM ¯2 and ˘ M2 C) and this
yields SM 1 M2 = CM 1 M2 which then determines the collinearity

of M1 , S, C, as shown in Figure 6.14. Analogously, one can prove

the collinearity of M2 , S, A.
(2) Since M2 is the midpoint of BC, ˜ we know that ˘ M2 B = ˘ M2 C
and the angles inscribed in these equal arcs are are equal, so that,
CAM 2 = BAM 2 , in other words AM2 is the angle bisector of
CAB containing S (note the collinearities of 1). Analogously, we
see that CM 1 is the angle bisector of ACB containing S. Therefore,
1 This item is not necessary for Problem 2; it is merely offered as further enrichment.
Figure 6.14
S must be the incenter of ABC and it must also lie on the angle
bisector of angle B, which proves both the collinearity of M3 , S, B
and that establishes that S is the incenter of triangle ABC.
(3) Point S was just shown to be the incenter of triangle ABC in the
above point.
(4) Due to symmetry, each of the three leaves has equal clover angles
(more precisely angles between the respective tangents) at both
vertices. At point S, we find that the vertical angles of the inter-
secting tangents produce equal angles as shown in Figure 6.13.
Furthermore, at point S, these angles have a sum of 360◦ , that is,
the three clover angles and the three equal corresponding vertical
angles between the leaves. Thus, the three clover angles at S have
a sum of 180◦ . Since the three clover angles at the respective other
vertices ( A, B, C) are the same, their sum is also 180◦ .
Problem 2b: We know that two parallel lines cut off equal arcs on a circle
as shown in Figure 6.15. The challenge here is to prove that two chords e, f
of a circle are perpendicular if, and only if, the sum of opposite arc lengths
is equal: e⊥ f ⇔ a + c = b + d (see Figure 6.16).
Figure 6.15
Figure 6.16
We begin with the chords that are two perpendicular diameters and shift
first one diameter to a parallel chord and then the other to another parallel
chord. In the first situation of two perpendicular diameters, all four arc
lengths are equal (a quarter of the perimeter), hence, we have a+c = b+d. If
one diameter is moved to a parallel chord (Figure 6.17), a, b become bigger
by the same amount as c, d become smaller. Thus a + c = b + d still holds.
Using the same idea, we get that after the movement of the second diameter
to a parallel chord a + c = b + d also holds true (Figure 6.18).
Figure 6.17
Figure 6.18
Shifting chords in a parallel manner, as shown in Figure 6.18, does not

affect the sums a + c and b + d. Conversely, if a + c = b + d holds, we
can move (shift) both chords to the center of the circle and here we see that
the corresponding diameters must be perpendicular.
Problem 2c: Let ABCD be a cyclic quadrilateral with circumcircle c.

These four vertex points divide circle c into four arcs with midpoints
M1 , M2 , M3 , M4 , as shown in Figure 6.19. With these points as centers,

we draw four arcs in the interior of circle c that have the intersection points
S, T, U, V . We shall prove that these points S, T, U, V are the incenters of
the four large triangles which are determined by the diagonals of the cyclic
quadrilateral ABCD, namely, triangles ABC, BDC, ABD, and ADC.
Figure 6.19
As we can see in Figure 6.19, T is the intersection point in the circle

interior of the arc AB centered at M1 and the arc BC centered at M2 . From
Problem 2a, we know that T is the point of intersection of the angle bisectors
of triangle ABC, which is then the incenter of triangle ABC. The same holds
true for the other three points S, U, V .
Now we are prepared to solve Problem 2 referring to Figure 6.20.
First, we note that M1 M3 ⊥M2 M4 because the sum of two opposite inter-
cepted arcs is in both cases a + b + c + d = 180 ◦ (from Problem 2b). More-
over, we have two equal arcs DM ˘3 = CM ˘3 and due to the inscribed angle
relationship we can conclude that M1 M3 is the angle bisector of CM 1 D.
Furthermore, T is on M1 C and S is on M1 D, hence, we can also state that
M1 M3 is the angle bisector of T M1 S. The triangle T M1 S is isosceles, and
Figure 6.20
thus, we have M1 M3 ⊥ST . We can repeat this procedure with the other sides
of the quadrilateral STUV , which then completes the proof.
In the following, we introduce several other such challenging problems,

which are treated in a similar way: Formulating a series of problems whose
solutions cleverly lead to the solution of the initial problem. These problems
should be interesting and provide some interesting insights and should be
easier to solve by considering the stepwise approach we used previously.
Problem 3: The circle inscribed in quadrilateral ABCD is tangent to its

sides at points X, Y, Z , W , as shown in Figure 6.21. We then find that the
lines XZ and WY pass through the point M, which is the point of intersection
of diagonals AC and BD.
Problem 3a: The circle inscribed in quadrilateral ABCD is tangent to its

sides at points X, Y, Z , W , as shown in Figure 6.22. Then the line XZ divides
the diagonal AC in the ratio CANN = CAXZ .
Figure 6.21
Figure 6.22
Let N be the point of intersection of AC and X Z . Since an angle formed

by a tangent and a chord is one-half the intercepted arc, we have AXN =
1
2
arc XZ = NZD. Furthermore, supplementary angles have equal sines,
hence
sin AXN = sin NZD = sin NZC (1)

We then can conclude the following:
2 · area(AXN) = (AN)(XN) sin ANX = (AX)(XN) sin AXN
For NCZ we have analogously
2 · area(NCZ) = (CN)(ZN) sin CNZ = (ZN)(ZC) sin NZC
Taking the quotient of the last two lines yields, because of (1) and AN X =
C N Z (vertical angles), ( AN )(X N) = ( AX )(X N) . And from that we conclude
(C N )(Z N) (Z N )(Z C)
immediately CANN = CAXZ , as claimed.
Figure 6.23
Problem 3b: Use the solution of Problem 3a to prove Problem 3.

Using an analogous argument, now assume that AC and W Y intersect
at some point, say N , as shown in Figure 6.23. Considering the triangles
AN W and N Y C, we can prove that W Y divides AC in the ratio CAMM
= AWCY
.
However, AW = AX and CY = C Z because tangent segments from the
same point to a circle are equal. In other words, the lines X Z and W Y divide
AC in the same ratio, and that means N = N . We see that the diagonal
AC passes through the point of intersection of XZ and WY . And if we use
the same procedure with the diagonal BD instead of diagonal AC, we can
see that the same holds for the diagonal B D. This completes the proof of
Problem 3.
Problem 4: Let c1 and c2 be two circles intersecting at points A and B

and a straight line through A (other than AB) is drawn, intersecting the two
circles at points M and N , as shown in Figure 6.24. Furthermore, let K be
the midpoint of MN, P the intersection point of the angle bisector of MAB
with circle c1 , and R the intersection point of the angle bisector of BAN
with circle c2 . We need to prove that PKR = 90◦ .
Figure 6.24
This problem seems not to be too complicated, but despite its elementary
appearance, the problem proves to be deceptively hard and may resist several
approaches.2 In the end, there are many different solutions. We offer the
shortest solution using another problem, but for some problem solvers, it
may be difficult to realize that these steps have something to do with the
original problem.
Problem 4a: Let ABC be a right triangle, and congruent right triangles
CDE and BFG similar to ABC, as shown in Figure 6.25. We have
CDE and BFG, which are placed so that their corresponding sides are
2 For details, see De Villiers, M., Humenberger, H. (2021). “Ghosts of a problem past.” At
Right Angles, Issue 9, 105–111.
parallel. Then we will prove that triangle AFE is also a right triangle and
similar to ABC.
Figure 6.25
For a proof of Problem 4a, as shown in Figure 6.25, let CAB = α and
observe that AFB and AEC are similar, as they have equal angles at B
and C, namely, 90 ◦ + β, and then the ratio of their corresponding sides is
AB
equal: AC = k = CEBF
. Therefore, also AF
AE
= k and EAF = α, and thus, the
claimed similarity is proved.
Problem 4b: In a triangle, the perpendicular bisector of a side and the

angle bisector of the opposite angle intersect at a point on the circumcircle,
as shown in Figure 6.26 (since S is a sort of south pole in the figure, one
could call this result south pole theorem).
Let S be the intersection point of the circumcircle with the perpendicular
bisector of AB, then we have equal arcs SA = SB; with equal arcs, the
corresponding inscribed angles must be equal, that means ACS = BCS,
hence, CS must be the angle bisector.
Figure 6.26
Problem 4c: We will use the solution of Problems 4a and 4b to solve

Problem 4.
The plan here is to construct triangle MBN and the midpoints H and
J of the line segments MB and BN, respectively. We then focus on the
triangles BRJ and PBH. One just has to use the result of Problem 4a once
(see Figure 6.27). In retrospect, things often seem to be very easy, but to
find these simple relations is often not trivial, as Problem 4 is definitely a
really hard problem from the perspective of an initial solver.
Now, for the proof, consider Figure 6.27, where H is the midpoint of
BM and J is the midpoint of BN, and let α = BRJ in the right triangle
BRJ. (Recall that BRJ and PBH are right triangles because of Problem
4b.) Then KJ is equal and parallel to BH. Because BRNA is a cyclic quadri-
lateral, we have BAN = 180◦ − 2α, and from this we get (from the cyclic
quadrilateral PBAM) MPB = 180◦ −2α, and this yields HPB = 90◦ −α.
Then we draw QJ parallel and equal to PB, and QK parallel and equal to
PH, in order to get the third right triangle (QJK, legs KJ and QK), all
of them with angles 90◦ , α, 90◦ − α. And by using Problem 4a, it follows
immediately that PKR = 90◦ .
Figure 6.27
Problem 5: A bicentric quadrilateral is one which has both an incircle and

a circumcircle. Show that the area of √
a bicentric quadrilateral with sides
a, b, c, d is given by the formula F = abcd.
Remark: This formula is not only valid for bicentric quadrilaterals, but
also rectangles have this area formula.
Problem 5a: We will show that for a bicentric quadrilateral, as shown in

Figure 6.28, the following equation holds: (ab + cd) cos B = ab − cd.
Figure 6.28
From the well-known property of tangential quadrilaterals, a + c =

b + d, (note that tangential segments from an exterior point to a circle are
equal) we immediately get a − b = d − c, and if we square both sides of
this equation, we get
a 2 − 2ab + b2 = c2 − 2cd + d 2 (*)
Drawing the diagonal AC and applying the law of cosines 3 to triangles ABC
and AC D yield
a 2 + b2 − 2ab cos B = c2 + d 2 − 2cd cos D (**)
The subtraction of the equations, (∗∗ ) − (∗ ), yields 2ab(1 − cos B) =
2cd(1 − cos D), and because cos D = − cos B, we have ab(1 − cos B) =
cd(1 + cos B) and this is equivalent to ab − cd = (ab + cd) cos B, which
is what we sought to prove.
Problem 5b: Show that for a bicentric quadrilateral, the following equation
holds: 2F = (ab + cd) sin B (where F denotes the area of the quadrilateral)
This is a consequence of the well-known triangle area formulas:
ab
2
sin B = F = cd2 sin D and sin B = sin D. (Note: The measures of
the angles at B and D have a sum of 180◦ because ABCD is cyclic.)
Problem 5c: Use the solution of Problems 5a and 5b to solve Problem 5.
From Problem 5b, we get 4F 2 = (ab + cd)2 (1 − cos2 B) and this

sin2 B
yields 4F 2 = (ab + cd)2 − (ab + cd)2 cos2 B; now, with the help of
Problem 5a, we substitute (ab + cd)2 cos2 B by (ab − cd)2 and get
4F 2 = (ab + cd)2 − (ab − cd)2 = 4abcd. Finally,√dividing by 4 and
taking the square root gives us the desired result: F = abcd.
Problem 6: Take an arbitrary initial natural number N0 . We state the fol-
lowing algorithm: Every digit of this natural number is multiplied by 111
and then the sum of all these products is taken.
Example: N0 = 4367, then 4·111 + 3·111 + 6·111 + 7·111 = 444 + 333 +
666 + 777 = 2220 = N1
3 The law of cosines states that in triangle A BC with sides a, b, c, the following equation
holds: c2 = a 2 + b2 − 2ab cos C.
Then the same occurs to this number 2220:
N1 = 2220, then 2·111+2·111+2·111+0·111 = 222+222+222 = 666 = N2
And then to this number the same procedure is applied:
N2 = 666, then 6·111 + 6·111 + 6·111 = 666 + 666 + 666 = 1998 = N3
And another time:
N3 = 1998, then 1·111 + 9·111 + 9·111 + 8·111

= 111 + 999 + 999 + 888 = 2997 = N4
Repeating this procedure with 2997, will yield 2997, so that the following
pattern of results evolves as:
4367 → 2220 → 666 → 1998 → 2997 → 2997 → · · ·
We must prove that all positive integers converge to 2997 by applying this
algorithm (not necessarily within four steps).
Problem 6a: First we prove the following: If one takes an arbitrary positive
integer N0 with at most three digits, the iteration of the algorithm will always
end up at the number 2997 after at most four steps.
If the above-mentioned algorithm is considered as a function f , one can
say N1 = f (N0 ).
We can write f in another way: f (N ) = 111·D S(N ), where DS denotes
the digit sum. Because of the given start condition, we know for all three-
digit numbers D S(N0 ) ≤ 27.
From N1 = 111 · D S(N0 ), we know N1 ≤ 111·27 = 2997. What can

≤27
be said about DS(N1 )?
There are only a few numbers ≤ 2997 with digit sum >27: 1999,
2899, and 2989. But these numbers cannot serve as N1 because they are
not divisible by 111 (not even divisible by 3). Therefore, we can say
D S(N0 ) ≤ 27 ⇒ D S(N1 ) ≤ 27, analogously, in the further steps: All
Ni stay ≤ 2997 and all D S(Ni ) stay ≤ 27.
On the other hand, the following statements hold:
• 3 divides N1 (independent of N0 ) because N1 = 111 · D S(N0 ). Since

the first factor of this product, 111, is divisible by 3, also the product N1
itself must be divisible by 3.
• 3 divides N i ⇒ 9 divides Ni+1 because Ni+1 = 111 · D S(Ni ) and we
know that the first factor 111 is divisible by 3 and the second factor
D S(Ni ) is also divisible by 3 due to the precondition that 3 divides N i .
• 9 divides Ni ⇒ 27 divides Ni+1 because Ni+1 = 111 · D S(Ni ) and
we know that the first factor 111 is divisible by 3 and the second factor
D S(Ni ) is divisible by 9 due to the precondition that 9 divides N i (note
that Ni is divisible by 9 if, and only if, its digit sum DS(Ni ) is divisible
by 9).
That means that if N0 is not a multiple of 3, every step of the transitions

N0 → N1 → N2 → N3 produces one more factor of 3 in the number.
Therefore, N3 = 111 · D S(N2 ) is surely divisible by 27 due to the second
factor D S(N2 ) and by 111 due to the first factor 111. This means N3 is
surely divisible by 999, so there are only three possibilities for N3 namely,
N3 = 999, 1998, 2997, and in the first two cases, the next step produces
2997.
If N0 itself is divisible by 3 or even by 9, one ends up at 2997 even
earlier.
Problem 6b: Prove that the same holds for arbitrary N0 with at most
10,010,010 digits.
Let N0 have at most 10,010,010 digits, then we have D S(N0 ) ≤
90,090,090 and we can conclude N1 = 111·DS(N0 ) ≤ 111·90,090,090 =
9,999,999,990. Thus, N1 has at most 10 digits with DS(N1 ) < 90. Analo-
gously, we can conclude N2 = 111 · DS(N1 ) < 111 · 90 = 9,990, thus, N2
has at most four digits, that is, D S(N2 ) < 36, and we know that 9 divides N2 ,
which is equivalent to 9 divides D S(N2 ), that means D S(N2 ) = 9, 18, 27.
Thus, there are only three possibilities for N3 : 999, 1998, and 2997 and N4
is certainly 2997.
We can use the solution of Problems 6a and especially 6b for solving
Problem 6, which is now provided.
Proof for Problem 6: All integers with at least five digits become smaller
using f (most of them significantly smaller), thus, they eventually end up
in the above-mentioned region, that is, numbers with at most 10,010,010
digits, as per Problem 6b:
Let N be a positive integer with n digits (n ≥ 5): 10n−1 ≤ N < 10n ,
D S(N ) ≤ 9 · n.
Therefore, f (N) < 10N·n ≤ 10
103 ·n
3
n−1 = 10n−4 and that is <1 for all n ≥ 5,
n
N
thus, f (N ) < N as claimed.
We leave to the reader another challenging problem, and one worth
trying, that is, to find a natural number, which needs five or six steps to
arrive at 2,997, and probably a much more difficult problem to find the
smallest such numbers.
Problem 7: In Figure 6.29, we are given a non-cyclic convex quadrilat-
eral ABCD. Then the perpendicular bisectors of the sides are constructed.
The two perpendicular bisectors adjacent to the vertex A (namely, A B
perpendicular to AB, and A D perpendicular to AD) intersect at point A .
Similarly, for the other vertices generating quadrilateral A B C D , which
we could call “Pb4 -quadrilateral” to ABCD. In the same way A B C D
Figure 6.29
4 From “perpendicular bisectors.”

results from the intersections of the perpendicular bisectors of quadrilateral

A B C D . We need to prove that A B C D is similar to ABCD with par-
allel corresponding sides. (Note: In Figure 6.29, perpendicular bisectors are
not extended to their perpendicular lines here in order to keep the figure as
clear and simple as possible.)
Problem 7a: We need to prove that in the context of Problem 7, the exterior
angles of A B C D are equal to the interior angles of ABCD, as shown in
Figure 6.30.
Because of the right angles at the midpoints of the sides AB and BC,
the angle at B (interior angle of A B C D ) is supplementary to the angle
at B (interior angle of ABCD, shown in Figure 6.30). Since the sum of
interior angles in every convex quadrilateral is 360 ◦ . Angles X and Y are
right angles of quadrilateral BXB Y , the remaining two angles B YB and
XBY are supplementary. Similarly, the angles at C and C of quadrilateral
CZC X are supplementary, as are the pairs of angles at D and D as well
as at A and A . Therefore, the exterior angles of A B C D are equal to the
interior angles of ABCD, as claimed.
Figure 6.30
Problem 7b: Prove that in the context of Problem 7, the diagonals of ABCD
have the same angles as the diagonals of A B C D .
We will show that the diagonals of A B C D are perpendicular to the
diagonals of ABCD as follows: The distances from C to C and from C to
D are equal, and the point C is also equidistant from points B and C, since
the points on the perpendicular bisector of a line segment are equidistant
from the endpoints. Thus, the distances of point C to points B and D must
be equal as well. Analogously, one can argue that A is equidistant from
points B and D, and thus A C ⊥B D because A and C are points on the
perpendicular bisector of BD. In the same way, one can show that B D ⊥ AC,
which yields the claimed perpendicularity of the diagonals of A B C D
to the diagonals of ABCD. And this, in turn, means that the diagonals of
ABCD have the same angles as the diagonals of A B C D , as originally
claimed.
Problem 7c: Use the solution of Problems 7a and 7b for solving Problem 7.
From Problem 7b we know that corresponding diagonals and sides (of
the quadrilaterals ABCD and A B C D ) are perpendicular, and this, in turn,
means that corresponding diagonals and sides in the quadrilaterals ABCD
and A B C D are parallel. Hence, corresponding triangles consisting of
two sides and a diagonal of the quadrilaterals ABCD and A B C D are
similar to each other with the same factor of similitude because the two
triangles with the diagonal AC (ABC and ACD) have this diagonal as a
common side. Thus, the quadrilaterals ABCD and A B C D are similar,
as claimed.
Problem 8: In cyclic quadrilateral ABCD, where the opposite sides are not
parallel, let P and Q be the intersection points of the extensions of pairs
of opposite sides. Then the angle bisectors at P and Q are constructed and
intersect the sides of ABCD at the intersection points E, F, G, and H , as can
be seen in Figure 6.31. We shall prove that AC||FG||EH and BD||EF||GH.
Problem 8a: Prove that in the context of Problem 8, the two angle bisectors
at points P and Q are perpendicular, as shown in Figure 6.31 (we had
already discussed this problem in the section “Perpendicular Bisectors of
Extended Sides of a Cyclic Quadrilateral” on page 230).
Figure 6.31
Let S be the intersection point of the two angle bisectors and A, B, C, D

denote not only the vertices but also the corresponding interior angles of
the quadrilateral ABCD. We begin by considering the concave quadri-
lateral PBQS, seen in Figure 6.31, and triangle DPC, where we have
DPB = 180◦ − (C + D). Then, 1 DPB = SPB = 90◦ − C+D . Analo-
2 2
gously, we have SQB = 90◦ − A+D 2
. Then for quadrilateral PBQS, PBQ =
360◦ − B. We then have
Å ã Å ã
C+D A+D
PSQ = 360◦ − (360◦ − B) − 90◦ − − 90 − ◦
2 2
180◦

A+C
= B + D+

− 180◦ = 90◦ .
2
180◦
Note that because ABCD is cyclic, A + C = 180◦ = B + D.

Problem 8b: Prove that in the context of Problem 8, the quadrilateral EFGH
is a rhombus.
The triangles PSE and PSG are congruent (ASA), hence, ES = SG.
Analogously, the triangles QSF and QSH are congruent, then FS = SH.
That means that in the quadrilateral EFGH, the diagonals are perpendicular
and they bisect each other, thus, it is a rhombus.
Problem 8c: Use the solution of Problems 8a and 8b for solving Problem 8.
For AC||F G we have to show ab = dc (Figure 6.32).
Figure 6.32
Recall that an angle bisector of a triangle divides the opposite side in the
ratio of the adjacent sides, therefore, for the two angle bisectors of triangles
ABP and BCQ, we have ab = ij and dc = kl , respectively. For our goal ab = dc ,
we have to show ij = kl . This relation holds on the one hand because of the
law of sines5 , and on the other hand due to sin(180◦ − ϕ) = sin(ϕ) which
is shown in Figure 6.33.
5 The law of sines says that in triangle ABC with sides a, b, c, the following equation holds:
a b c
sin A = sin B = sin C .
Figure 6.33
Applying the law of sines to triangle APB yields ij = sin B

sin A
, and applying
the law of sines to triangle BCQ, we get k = sin C , and since sin A = sin C,
l sin B
both ratios are equal. This proves ab = dc so that in the triangle ABC we get
the desired parallels, namely, AC||F G. Applying the result of Problem 8b
we also know AC||E H .
For BD||GH, one can argue in the following way: First, we have ef =
g+h+i
d+c+ j
(an angle bisector of a triangle divides the opposite side in the ratio
of the adjacent sides), and since the triangles PBA and PCD are similar we
g+h+i
can conclude d+c+ j
= ij . As we established above, ij = kl , and applying
again that an angle bisector of a triangle divides the opposite side in the
ratio of the adjacent sides, we get kl = dc . Therefore, we have shown ef = dc
and this proves BD||GH. Using a result of Problem 8b, which established
that EFGH is a rhombus, we also know BD||EF. Thus, we have shown that
all the lines indicated in Problem 8 are parallel.
Problem 9: In triangle ABC, shown in Figure 6.34, where should point P

in the interior of the triangle be placed so that it maximizes the product x yz
of its (perpendicular) distances to the sides of the triangle?
Figure 6.34
Problem 9a: Which point P on segment AB, shown in Figure 6.35, maxi-
mizes the product yz, which is the product of its perpendicular distances to
the other two sides of the triangle?
Figure 6.35
Let us denote AB = c (constant), AP = t, PB = c − t, or expressed

another way 0 ≤ t ≤ c is variable as can be seen in Figure 6.36.
Figure 6.36
From the similarity of triangles APF and ABL, we have zt = qc ⇒

z = qc ·t and for triangles PBE and ABK, we have c−t
y
= cp ⇒ y = cp ·(c−t).
Hence, y · z = cp · qc ·t · (c − t), and we want to know for which value

constant!
of the variable t does this product become a maximum. Since the constant
factor cp · qc does not affect which value of t yields the maximum of the
product, as it only affects its own maximum value, we can omit this constant
factor, and we have to maximize t · (c − t) with variable t. This is a very
well-known situation: The maximum is achieved for t = 2c (the square
has maximum area of all rectangles with a given perimeter, formulated
algebraically: The product of two non-negative numbers with constant sum
c achieves its maximum, if both factors are equal, namely, 2c ). Thus, the
midpoint of AB is the solution.
Problem 9b: Use the solution of Problem 9a for solving Problem 9.
To solve Problem 9, we can apply the finding of Problem 9a to triangle
A B C, where the line A B is at a distance x above AB, as shown in Fig-

ure 6.37. Then the optimal position of P is at the midpoint of A B . That

means (x again considered as variable) P must lie somewhere on the median
m c since the median for triangles ABC and A B C is one line. Analogously,
P must also lie on the other medians, thus, the solution of Problem 9 is the
centroid, which is the intersection point of the three medians.
Figure 6.37
Problem 10: Prove that in a 2m × 2n distorted chessboard, as we show

it in Figure 6.38, the sum of the shaded areas equals the sum of the white
areas. The meaning of “distorted” more precisely is as follows: On each
side of the convex chessboard, we mark equidistant points (2m or 2n equal
segments), then these equidistant points of opposite sides are joined.
Figure 6.38
Problem 10a: Prove that in a “distorted” 2 × 2 chessboard, shown in

Figure 6.39, the sum of the shaded areas equals the sum of the white areas.
Here both pairs of opposite midpoints of the sides are joined.
Figure 6.39
We will discuss two solutions:

• In the first, shown in Figure 6.40, the connections of the interior
intersection point, I , to the four vertices are drawn, then one can see
immediately that on every side, two triangles with equal area arise (they
are not congruent): one white and one shaded. These triangles have equal
bases and equal altitudes, such as AEI is equal in area to EBI.
Figure 6.40
• For a second possibility, draw the parallelogram by joining the midpoints

of the sides of the quadrilateral, as shown in Figure 6.41. Then one can
see immediately that in the interior of this parallelogram, there is an
area balance between shaded and white (in a parallelogram the diagonals
bisect each other) and the remaining white (triangles AEH and FCG) and
shaded triangles (triangles EBF and HGD) can be “pushed together,”
respectively, to a white and a shaded quadrilateral, which is in both cases
similar to the original quadrilateral, thus, also these quadrilaterals have
equal areas and in this case, they are congruent as well.
Figure 6.41
Problem 10b: Prove that in a “distorted” 2m × 2n chessboard, the sum of

the shaded areas equals the sum of the white areas, which can be seen in
Figure 6.42.
First, we deal with the case where we have a 4 × 4 portion of the chess-
board as we can see in Figure 6.42. Initially, the point labeled 1 can be fixed
as the midpoint of HF and EG. Furthermore, E, F, G, and H are the mid-
points of the sides. As the next step, the point 2 is the midpoint of E1 and PK ,
which is obtained by joining the midpoints of the sides of the quadrilateral
ABFH. Analogously, the points 3, 4, 5 are also midpoints. Subsequently, we
can conclude that the points 6, 7, 8, 9 are the midpoints of the corresponding
line segments (diagonals of a parallelogram bisect each other). Altogether
we see that all the interior intersection points are the points of quartering
the corresponding joining line segments. With that knowledge, it is easy to

argue for the equality of shaded and white areas in this case. In the quadri-
lateral AE1H, we have balance between shaded and white, as well as in the
other three quadrilaterals (EBF1, 1FCG, and H 1GD). Hence, overall the
same exists in the quadrilateral ABCD. But in order to conclude this, it is
essential to know that the points 2, 3, 4, 5 are the respective the midpoints of
the line segments joining the point 1 with the sides’ midpoints E, F, G, and
H . A similar way of arguing can be made in the case of an 8 × 8 chessboard
and even more generally in the case of a 2n × 2m chessboard.
Figure 6.42
Problem 10c: Prove that in a “distorted” m × n chessboard resulting from

any convex quadrilateral, all the interior intersection points divide all the
corresponding joining lines equally, that is, the equality of the line segments
on the sides transfers to the interior net of joining lines.
The corresponding “reduced” form of this problem is shown in
Figure 6.43. If in a convex quadrilateral the opposite sides AB and DC are
divided by the points S, and T in the same ratio qp , and the opposite sides
BC and AD by the points U and V in the same ratio rs , then the intersection
point P of VU and ST also divides these line segments in the ratios qp and
r
s
, respectively. In other words, the equal ratios on opposite sides transfer to
the interior ratio in which the intersection point P divides the joining lines
VU and ST .
Figure 6.43
We can easily establish a “formula” for the points S, T, U , and V

with respect to the points A, B, C, and D. For example, for S we get
S = A + p+qp
(B − A) = p+qq
A + p+q
p
B.
Analogously, we get
q p s r
T = D+ C, U = B+ C,
p+q p+q r +s r +s
s r
V = A+ D
r +s r +s
q
And with that it is easy to verify that p+q V + p+q
p
U = r+s
s r
S+ r+s T holds
true, and this, in turn, yields the claimed proportions that the intersection
point P of lines VU and ST divides these line segments in the ratios qp and
r
s
, respectively.
Problem 10d: Use Problems 10a, 10b, and 10c to solve Problem 10
(Figure 6.44).
Figure 6.44
As a result of Problem 10c, we can divide the case 2m × 2n into m × n

smaller quadrilaterals as in the case 2 × 2 (Figure 6.44), and as a result
of Problem 10a, we know that in all these cases there is balance between
shaded and white, and hence, overall in the given quadrilateral.
Problem 11: Consider four equilateral triangles A B C, A B C,

AA A , and BB B , as shown in Figure 6.45.
Figure 6.45
We begin by noting that point G is the centroid of triangle ABC. Then

we construct two concentric circles c1 and c2 each with center at G, where
circle c1 contains point A, and circle c2 contains point B. We then draw two
equilateral triangles AA A and BB B with vertices on the respective circles,
where A denotes the point rotated around G by 120 ◦ counterclockwise,
and analogously, B denotes the point rotated around G by 120 ◦ counter-
clockwise. Prove that also the triangles A B C and A B C are equilateral,
which we show in Figure 6.45.
Problem 11a: As we have stated above, ABC has point G as its centroid.
Now we construct two concentric circles, c1 and c2 with center G, one
through A and the other through B, and two equilateral triangles AA A
and BB B with vertices on the respective circles (here A denotes the point
rotated around G by 120 ◦ counterclockwise, and the same with point B ).
Prove that A A and BC bisect each other and that the same holds for B B
and AC, as we can see in Figure 6.45.
Extend AG beyond G by half of its length, then the new endpoint will
determine the median for the equilateral triangles A A A and ABC, which
implies that this endpoint is the midpoint of A A and also the midpoint of
BC. Therefore, A A and BC bisect each other. If we extend BG beyond G
by half of its length, then we can see, analogously, that B B and AC bisect
each other.
Problem 11b: Use Problem 11a to prove Problem 11.

From Problem 11a, we can conclude that the quadrilateral A BA C is a
parallelogram, since the diagonals A A and BC bisect each other. Anal-
ogously, we can show that quadrilateral B CB A is also a parallelogram.
Hence, B A = CB and BA = A C. But we also know by 120◦ rotations
with center G that B A = B A = BA and this means that CB A is equi-
lateral. And analogously, we can prove that CA B is equilateral.
Conclusion
The principle of creating a series of preparatory problems is not always pos-

sible with every complex problem but, as we have seen, it is often useful.
In some cases, it is not easy to give hints without divulging the solution.
On the one hand, good hints should not eliminate all the challenges of the
problem; they can lead problem solvers to think creatively about the problem
at hand. It is always desirable to experience mathematics as a process done
by oneself. It is quite possible that this principle — looking for and creating
a series of clever and elaborated preparatory problems — could be applied
more often than we think. Problem solving in mathematics usually demon-
strates the power and beauty of mathematics, in most cases a clever problem
can be solved directly, yet in some cases, the stepwise procedures that we
have demonstrated in this chapter provide a useful alternative. This could
also be especially important for teachers presenting problem-solving strate-
gies, as dividing a problem into smaller parts may significantly increase the
chances for students to discover a clever solution.
Index
A arithmetic
absurdity, 210–211, 217 calculations, 1–39
addition, 1–3 phenomenon, 97
symmetry, 6 relationships, 1
algebra sequence, 2
approach, 279 technique, 16
mistake, 201 wrong, 82
angle automorphic numbers, 79
clover, 287 axioms, 220
constant, 224–226
determination, 265–267 B
exterior, 304 Bell, E. T., 2
measure of, 156–157 bicentric quadrilateral, 299–300
of parallel lines, 222 binomial coefficients, 190
on clock, 221–222 bottom-left vertex, 178–179
relationship, unusual, 229 bottom-right vertex, 182
sum of, 141, 264–265
supplementary, 294 C
supplementary opposite, 153 calculator, 216–217
angle bisector, 305 circle, 153, 158, 170–171
intersection point of, 296–298 and angle bisector, 269–271
of triangle, 307–308 area of, 236–238, 260–261
two circles and, 269–271 concentric, 170
angle sum, 264–265 cloverleaf figure in, 287–289
in star polygons, 222–224 overlapping quarter, 243–244
arbitrary initial natural number, tangent quarter, 271–273
300 circular areas, comparison of, 234–236,
arbitrary positive integer, 301 238–240
arithmagons circumferences, 169
with four vertices, 177–181 complex numbers, 209–210
with three vertices, 181–182 consecutive natural numbers, 174–176
319
consecutive odd numbers, sum of, 90–91 F

culture arithmetic skills, 1 Fibonacci, 8–9
cyclic quadrilateral, 230–232, 286, 298, Fibonacci numbers, 94
305 figures, 219
forms, 219
D four-digit number, 102
data organization, 125–127, 160 fraction, 190, 192
digits 1–9 number pyramid with, 188–190
as mixed-number fractions, 45 of rectangle, 267–268
favorable numerical arrangements of, product of, 75–76
44–45 reducing, 83–86
order of, 60–61
remain in use, 67–68 G
unexpected sum, 41 Gauss, Carl Friedrich, 1–2
distorted, meaning of, 311–313 Gauss’ method, 3
divisibility geometry, 219–221
by 7, 18–20
by 11, 16–17 H
by 13, 20–21 heptagram, 224
by 17, 21 hexagon, 116–117, 238
by larger prime numbers, 22
by prime factors of 10, 18 I
division, 3–4, 46–47, 195–196 inequalities, 192, 208
determining remainders of, 60 solving, 213–214
symmetry, 6–7 infinity, 144
by zero, 193–198 interest, raising, 35–37
dots, lattice of, 118
K
E
Karprekar constants, variations of,
eight-digit numbers, non-repeating, 11 102–103
Elements (Euclid), 219 kernel, 140
equal number of numbers, with equal
sums, 71–72 L
equation, 211–212 law of cosines, 300
simultaneous, 201 law of decay, 37
system of, 198 law of growth, 37
unusual, specific natural numbers, law of sines, 307
176–177 length
equal line segments, unexpected, 263–264 chord, 259–260
equation system yields, 186 of common tangent, 268–269
equilateral triangle, 117, 122–123, 136, Leonardo of Pisa, 8
246, 264, 316–317 linear equation, 283
Euclid, 219 line segments, 313–315
Euler number, 37 logical reasoning, 122–124
Index 321
logical thinking, 120, 128 splitting, 73–74

logic system, 220–221 starting, returning to, 47–48
unusual number 9, 8–10
M number 3999 , units digit of, 121–122
magic square, 88–89 number 9
mathematical formula, 3 magic of, 11–12
mathematical support, shopping with, 30 non-repeating eight-digit numbers and,
mathematics 11
history of, 1 novelties with, 10
process of, 277 secret, 12–13
missing area, discovering, 229–230 numerical peculiarity, 56–57
mixed-number
curiosity with, 184–185 O
fractions, using all nine digits 1–9 as, 45 odd numbers, surprising pattern of, 67
multiples of 3, 46 Olympiad competitions, 286
multiples of 9, 53–54 organized thinking, 120
multiplication, 4, 6
by 11, 14–16 P
by multiples of 9, 51–53 palindromic numbers, 62–66
cross, 197 common divisor of, 67
Russian peasant’s method of, 24–26 parallel lines, 199, 201, 221
simple, 67 parallel manner, shifting chords in, 291
technique for special numbers, 29–30 parallelogram, 160, 313
squares on sides of, 251–252
N pattern, 70–71, 219
natural numbers, 59–60 discovering, 154–155
unusual equation in search for, 176–177 recognition, 128–130
nine-digits, 42 subtraction, 6
number yields, 41 unexpected, 42–43
to make 100, 5 paved paths, 92–94
number pentagram, 223–224
1–9, 48 perimeters, comparison of, 242–243
9 and 11, 50–51 periodic rate, 36
1–100, 114 perpendicular
1089, 103–109 bisector, 230–232, 297–298, 303
6174, 101–102 diameters, 290
142857, 98–100 powers
1,000,000, 87 mistake with, 206
friendly numbers, 76–78 sums of, 58
larger, 192–193 prime factors of 10, 18
multiplication technique for, 29–30 prime number, 49–50
patterns, 68–70 divisibility by, 22
positive, 209 prime relationship, unexpected, 43
peculiar large, 90 problem-solving, 277–278, 318
pyramid with fractions, 188–190 products of 91, 48
Q more squares, sum of squares equals,

quadratic 57
equation, 281 multiple of 5, 7
equations yields, 186 numbers from 1 to 125, 7–8
formula, 143 numbers, magic of, 78–80
quadrilateral on sides of parallelogram, 251–252
bicentric, 299–300 on two parallel lines, 262–263
cyclic, 230–232, 286, 298, 305 spaces, 122
interior angles of, 306 sum of, 23, 177
non-cyclic convex, 303 vertical edges of, 242
region, 149 squared equation, 205
tangential, 255–256, 300 square root, 210
quarter circle, semicircle in, extraction, 202, 204–205
227–228 star polygons, angle sum in, 222–224
quick multiplication technique, subtraction, 155
28–29 pattern, 6
symmetry, 6–7
R successive percentages, 31–35
radius, 170 symmetry, 158–159, 188
Ramanujan, Srinivasa, 58 between division and subtraction, 6–7
random division, 74–75 division and addition, 6
real numbers, 210
rectangle, 162 T
diagonal properties of, 94–96 tangency
equal areas, 232–234 of incircle, 284
five congruent triangles in, 247–248 point of, 284–285
fraction of, 267–268 quadrilateral, 255–256, 300
shaded area in, 258–259 theorems, 220
two tangent quarter circles in, 271–273 binomial, 207–208
rhombus, 306 Pythagorean, 236, 250, 281
right-angle vertex, 149, 152 south pole, 297–298
rule of 72, 38–39 three-digit number, 104, 107–108
top-left vertex, 180
S top-right vertex, 178
semicircle, 256–258 triangle, 145–147, 154–155, 158,
arc, 259 161–163, 226–227, 273–275,
in quarter circle, 227–228 278–279, 281–283, 296–297, 310
sequence, surprise in, 185–187 angle bisector of, 307–308
shapes, 219 area of, 244–245
space, 219 five congruent, in rectangle,
square, 149, 152, 232–234 247–248
area of, 240–241, 249–250 overlapping congruent right, 248–249
arrangement, 114–115 similarity, 285
circular area with, 235–236 unexpected similar, 253–254
curiosity, 23–24 triangular numbers, 70
Index 323
twin primes, 50 V
two-digit number, sum of, 50 variables, 183
two parallel lines, square on, Venn diagram, 124–125
262–263 vertices, arithmagons with, 177–182
typographical errors, 81–82 visual representation, 124–125,
166–167
U
unit fractions, sum of, 187–188 Z
unusual method, two numbers, 26–27 zero-length radius, 170

Sharpening Everyday Mental

Uploaded by

Copyright:

Available Formats

Sharpening Everyday Mental

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sharpening Everyday Mental

Uploaded by

Copyright:

Available Formats

Problem Solving in Mathematics and Beyond

Print ISSN: 2591-7234

Series Editor: Dr. Alfred S. Posamentier

There are countless applications that would be considered problem solving in

Vol. 30 Sharpening Everyday Mental/Thinking Skills Through Mathematics

Vol. 29 Introduction to Chaos, Fractals and Dynamical Systems

Vol. 28 Seduced by Mathematics: The Enduring Fascination of Mathematics

Vol. 27 Mathematics: Its Historical Aspects, Wonders and Beyond

Library of Congress Cataloging-in-Publication Data

British Library Cataloguing-in-Publication Data

Copyright © 2024 by World Scientific Publishing Co. Pte. Ltd.

For any available supplementary material, please visit

Desk Editors: Balasubramanian Shanmugam/Rosie Williamson

Typeset by Stallion Press

Alfred S. Posamentier is currently a Distinguished

Hans Humenberger is currently a full Professor

Vienna in 2005 to accept his current professorship for mathematics at the

About the Authors v

Chapter 1. Arithmetic Calculation Surprises 1

Chapter 2. Numerical Novelties 41

Chapter 3. Challenging Problems with Surprising Solutions 113

Chapter 4. Algebra: A Unique Problem-Solving Tool 173

Chapter 5. Geometric Gems 219

Chapter 6. For Experts: Stepwise Solutions to

Chapter 1 appropriately presents a wide variety of arithmetic calcula-

technique, which requires a bit of patience as it deals with problems in

Arithmetic Calculation Surprises

Although in today’s culture arithmetic skills have been relegated to com-

Some numbers lend themselves to a very beautiful symmetric division,

Surprising Multiplication Results

An analogous pattern can be obtained by multiplying the number 76923

Using All Nine Digits to Make 100

The challenge here is to use all 9 digits 1, 2, 3, 4, 5, 6, 7, 8, and 9 to create

69258 7524 5643 3546 5823

Consider the following addition and multiplication patterns. This is another

Consider the following multiplication and subtraction pattern:

Consider the following symmetry involving division and addition:

Lastly, consider a symmetry between division and subtraction:

Consider the challenge of squaring the number 85 mentally. Here is a

Squaring Numbers from 1 to 125

This can be easily shown with the following examples:

The Unusual Number 9

The first occurrence in Western Europe of the Hindu–Arabic numerals we

these are the nine figures of the Indians 9, 8, 7, 6, 5, 4, 3, 2, 1. With

For 734: 7 + 3 + 4 = 14; then 1 + 4 = 5

For 56,589 5 + 6 + 5 + 8 + 9 = 33; 3+3=6

The product of the remainders: 5 × 6 = 30 → 3 + 0 = 3, compares

The same procedure can be used to check the likelihood of a correct

More Novelties with the Number 9

Although it is a bit cumbersome, especially when compared to a calcu-

Non-Repeating Eight-Digit Numbers and the Number 9

Once again, the number 9 plays an important role in our arithmetic

The Magic of the Number 9

This unusual feature can be easily explained, since n × 9 =

A Secret about the Number 9

The Technique for Determining When a Number is

There are moments in everyday life situations where it can be useful to

N = 104 a + 103 b + 102 c + 10d + e

which implies that divisibility by nine of the number N depends on the

Multiplying by 11 can be significantly simpler than using a calculator, and