This article was downloaded by: [Heriot-Watt University]

On: 27 December 2014, At: 05:50

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Mathematical

Education in Science and Technology
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tmes20

The decimal representation of real

numbers
a
A. Kalapodi
a
Department of Business Planning and Information Systems, T.E.I.
of Patras, M. Alexandrou 1, Koukouli , 26334 Patras, Greece
Published online: 07 Jun 2010.

To cite this article: A. Kalapodi (2010) The decimal representation of real numbers,
International Journal of Mathematical Education in Science and Technology, 41:7, 889-900, DOI:
10.1080/0020739X.2010.486450

To link to this article: http://dx.doi.org/10.1080/0020739X.2010.486450

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 International Journal of Mathematical Education in
Science and Technology, Vol. 41, No. 7, 15 October 2010, 889–900

The decimal representation of real numbers

A. Kalapodi*

Department of Business Planning and Information Systems, T.E.I. of Patras,

M. Alexandrou 1, Koukouli, 26334 Patras, Greece
(Received 17 December 2009)

The representation of natural numbers in decimal form is an unequivocal

Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

procedure while for the representation of real numbers some ambiguities

concerning the existence of infinitely many digits equal to 9 still emerge.
One of the most frequently confronted misunderstandings is whether
0.999 . . . equals 1 or not, and if not what number does this sequence of
digits represent. In this review article, the decimal representation of any real
number is explicitly presented. In particular, it is investigated whether this
representation is unique or not. A condition is given that guarantees the
uniqueness of decimal representation for a subset of real numbers, while for
the remaining numbers two decimal representations exist. It is also
investigated which sequences of digits cannot be accepted as decimal
representations of real numbers. Moreover, analogous results are presented
in the case where a real number is represented in systems with base n, n
being a natural number greater than 1.
Keywords: decimal; unique; representation; real numbers
AMS Subject Classification: 11A63

1. Introduction
The notion of representation is not purely mathematical, since representation
procedures are adopted by many other sciences. An elementary but not naive
representation process can be applied to all linguistic systems, where the letters of an
alphabet are used to express the words of the language. For example, every word of
the English language can be represented by a sequence consisting of some of the 26
letters of the Latin alphabet. It is clear that in this case, the reverse procedure is
meaningless, since a random series of letters does not correspond necessarily to an
acceptable language word.
A first approach to the representation of numbers in mathematical notation,
which is closest to the common way of writing, uses the digits {0, 1, 2, 3, 4, 5, 6, 7, 8, 9}
as alphabet. This alphabet is called a decimal alphabet in order to avoid confusions
about the number system used. The representation problem can therefore be

*Email: [email protected]

ISSN 0020–739X print/ISSN 1464–5211 online

 2010 Taylor & Francis
DOI: 10.1080/0020739X.2010.486450
http://www.informaworld.com
 890 A. Kalapodi

formulated in two questions: Can every real number be expressed using the 10 digits
of the decimal alphabet? Does every sequence of digits represent a real number?
Before answering these questions, it is worthwhile to briefly mention that the
writing of numbers passed through different primitive stages, that can be classified in
three main categories. The first comprises early attempts of recording numbers
through marks, such as knots tied on strings or wooden tallies. In the second, we find
simple grouping systems, such as hieroglyphic Egyptian numerals, Roman numerals
or attic Greek numerals. A typical representative of the third category is the
well-known alphabetic Greek numerals, where a number is expressed using
combination of symbols. All these representations were not appropriate for
calculations, especially when large numbers were involved. This inadequacy has
led to positional number systems, like the Babylonian system, the Mayan system or
the Arabic numerals, systems which gave the potential of overcoming simple
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

arithmetic. A more detailed presentation can be found in [1].

Using modern terminology, we start from the meaning of numbers, a concept
that is not easy to define. As Dedekind says in [2], the answer to the question ‘what is
the nature and meaning of numbers’ is briefly this: ‘numbers are free creations of the
human mind; they serve as a means of apprehending more easily and more sharply
the difference of things’. Consequently, it is necessary to have the ability to describe
theoretically, as well as to symbolize, numbers in the most accurate way. Many
number systems are used in modern mathematics, such as the dyadic or binary
system, the triadic, the vigesimal and the sexagesimal, but undoubtedly the decimal
system is the prevalent.
Decimal notation was not extensively used until the Belgian Simon Stevin used it
in his book De Thiende published in Dutch in 1585 (more known in its French
translation under the title La Disme). Modern decimal notation is due to the Scottish
mathematician John Napier, who also invented logarithms.
In this review article, we aim to present in an explicit way the representation of real
numbers in the decimal system. In many books on Elementary Number Theory, this
problem is confronted either directly [3,4] or under a more general approach [5,6].
We adopt elements from both directions, trying to simplify the presentation. All
theorems and propositions in this article can be found in [3] or in [6].
The following section investigates how an integer can be expressed in decimal
notation, a task which is important for our scope, since the representation of real
numbers is based on the integral part of a number. In the main section, it is firstly
proved that every real number can be represented in the decimal system. We examine
whether or not this expansion is unique. We also explain when a sequence of digits
cannot be accepted as a representation of a real number. In addition, we specify
which expansions correspond to rational numbers. The last section is devoted to
systems with base n, n being a natural number greater than 1, where all the previous
results can be generalized.

2. The decimal representation of integers

In what follows, the number ‘zero’ is not considered as a natural number.
In addition, for a real number x, we denote by [x] the integral part of x, that is the
largest integer which does not exceed x. We mention that for any real number, its
integral part is uniquely determined.
 International Journal of Mathematical Education in Science and Technology 891

Definition 2.1 [6]: A natural number N is expressed as a decimal, or equivalently it

has a decimal representation, if

N ¼ cm 10m þ cm1 10m1 þ    c1 10 þ c0

where
() m is an integer greater or equal to zero,
() cn, n ¼ 0, 1, . . . , m, are integers with the property 0  cn  9 for n ¼ 0,
1, . . . , m, and cm 6¼ 0.
If the natural number N is expressed as a decimal, the sum cm10 m þ cm1
m1
10 þ    c110 þ c0 is called the decimal representation or decimal expansion of N
and it is also denoted as
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

N ¼ cm cm1    c1 c0
The integer numbers {0, 1, 2, 3, 4, 5, 6, 7, 8, 9} are called digits of the decimal system.
Theorem 2.2: Any natural number can be uniquely expressed as a decimal.
Proof: We prove at first that for a given natural number N, there is at most one
decimal representation, by showing that if there exist one, namely N ¼ cm10m þ cm1
10m1 þ    c110 þ c0, then the numbers m and ci, i ¼ 0, 1, . . . , m, are uniquely
determined by N.
In fact, let n denote one of the numbers 0, 1, 2, . . . , m  1. Then
N cn1 cn2 c0
¼ cm 10mn þ cm1 10mn1 þ    þ cn þ þ 2 þ  þ n
10n 10 10 10
and
cn1 cn2 c0 9 9 9
0 þ 2 þ  þ n  þ þ  þ n
10 10 10 10 102 10
1
¼ 1  n 5 1:
10
Thus,
 
N
¼ cm 10mn þ cm1 10mn1 þ    þ cnþ1 10 þ cn
10 n
and similarly
 
N
¼ cm 10mn1 þ    þ cnþ1
10nþ1
   
N N
Therefore cn ¼ 10 n
 10 10 nþ1
for n ¼ 0, 1, . . . , m  1.
Similarly, dividing N by 10m, we conclude that
 
N
cm ¼ :
10m
In addition, since 0  ci  9, i ¼ 0, 1, . . . , m, and cm 6¼ 0, we have

10m  N  9ð10m þ 10m1 þ    þ 10 þ 1Þ ¼ 10mþ1  1 5 10mþ1

 892 A. Kalapodi

thus m  logN 5 m þ 1, that means

m ¼ ½log N :
In order to complete the proof of the theorem it is sufficient to show that for any
natural number N there is at least one decimal representation.
In fact, from the division algorithm, there exist non-negative integer numbers N1
and c0 such that N ¼ c0 þ N110. Replacing N by N1 we obtain the quotient N2 and the
remainder c1 from the division of N1 by 10, that is N1 ¼ c1 þ N210, where N2 and c1
are non-negative integers. Continuing this procedure, it is clear that the quotients
consecutively obtained, when positive, decrease because Nnþ1  N10n . Since those
quotients are non-negative integers, for some k  1 we must obtain Nk ¼ 0. Let m
denote the greatest index for which Nm 6¼ 0. The procedure of repeating divisions
terminates at the m þ 1 step, thus we have obtained the following sequence of
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

equalities:
N ¼ c0 þ N1 10
N1 ¼ c1 þ N2 10
..
.
Nm1 ¼ cm1 þ Nm 10
Nm ¼ cm

where ci, i ¼ 0, 1, . . . , m, are non-negative integers with the property 0  ci  9,

i ¼ 0, 1, . . . , m, and cm 6¼ 0. Hence, we receive a decimal representation of N, that is
N ¼ cm 10m þ cm1 10m1 þ    þ c1 10 þ c0
œ
Remark 2.3: Another proof for the uniqueness of the representation of a natural
number as a decimal can be found in [5] where it is also observed that if we allow the
use of negative coefficients in the decimal representation we lose the uniqueness of
the expansion. Thus, as acceptable decimal representations we consider only those
having the non-negative integers {0, 1, 2, 3, 4, 5, 6, 7, 8, 9} as coefficients, that is the
digits of the decimal system.
In order to extend the Definition 2.1 to integer numbers, it is necessary to observe
that the integer number ‘zero’ has the unique decimal representation 0, that is as a
digit of the system. Moreover, a negative integer can be represented in decimal form
using the decimal expansion of its opposite. In fact, if m is an integer m 5 0, then
n ¼ m is a positive integer, that is a natural number, which has a unique decimal
expansion
n ¼ ak 10k þ ak1 10k1 þ    þ a0

As a decimal expansion of m, we define the following

m ¼ ½ak 10k þ ak1 10k1 þ    þ a0  ¼ ðak Þ10k þ ðak1 Þ10k1 þ    þ ða0 Þ

which is usually called the signed representation of m and it is denoted by

m ¼ ak ak1 . . . a0
 International Journal of Mathematical Education in Science and Technology 893

It is clear that we can summarize the results of this section as follows:

Every integer number n has a unique decimal representation under the
reservation that if n 6¼ 0, then all coefficients in the expansion have the same sign,
that of n.
At this point, we quote a theorem that can be found in [5], which we use in the
next section.
Theorem 2.4 [5]: For integers b 4 1 and n  0 and for any set of integers
{r0, r1, . . . , rn} with b 5 ri 5 b for each i, 1  i  n,
rn bn þ rn1 bn1 þ    þ r1 b1 þ r0 b0 ¼ 0

if and only if
ri ¼ 0 for each i, 0  i  n:
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

3. The decimal representation of real numbers

In order to express a real number in decimal notation, we must extend the
Definition 2.1 by initiating the use of infinite fractions and the properties of
the integral part. We begin our study by non-negative real numbers keeping in mind
that in order to have generalization of previous results, the reservation concerning
the sign of coefficients in a decimal expansion must be still valid.
Definition 3.1 [6]: A non-negative real number x, is expressed as a decimal, or
equivalently it has a decimal representation, if
c1 c2
x ¼ As 10s þ As1 10s1 þ    þ A1 10 þ A0 þ þ 2 þ   
10 10
where
() s is an integer.
() Ai, i ¼ 0, 1, . . . , s and cn, n ¼ 1, 2, . . . are digits of the decimal system, with
the reservation that in case s  1 then As 6¼ 0.
() [x] ¼ As 10s þ As110s1 þ   +A110 þ A0.
If the non-negative real number x is expressed as a decimal, the series
c1 c2
As 10s þ As1 10s1 þ    þ A1 10 þ A0 þ 10 þ 10 2 þ    is called the decimal represen-

tation or decimal expansion of x and it is also denoted as

x ¼ As As1    A0 : c1 c2   
Theorem 3.2: Every non-negative real number has a decimal representation.
Proof: Let x be a real number with x  0 and [x] be the integral part of x. Thus
x ¼ [x] þ x1, where x1 2 R, 0  x1 5 1. We consider [x] and x1 separately.
Since [x] is a non-negative integer, the representation of zero and Theorem 2.2
yield the decimal representation of [x], namely
½x ¼ As 10s þ As1 10s1 þ    þ A1 10 þ A0
where s is an integer, Ai, i ¼ 0, 1, . . . , s digits of the decimal system with the
reservation that in case s  1 then As 6¼ 0.
 894 A. Kalapodi

Passing to x1, we construct the following sequence:

x1 ¼ x  ½x, 0  x1 5 1
x2 ¼ 10x1  ½10x1 , 0  x2 5 1
x3 ¼ 10x2  ½10x2 , 0  x3 5 1
..
.
or equivalently
x1 ¼ x  ½x, xnþ1 ¼ 10xn  ½10xn  for n ¼ 1, 2,    ,
where 0  xn 5 1 for n ¼ 1, 2, . . . . For every n ¼ 1, 2,. . . the non-negative integers
[10xn] are obviously digits of the decimal system and we denote them as cn,
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

n ¼ 1, 2, . . . . Then
c 1 þ x2 c2 þ x3 cn þ xnþ1
x1 ¼ , x2 ¼ , ... , xn ¼ , ...:
10 10 10
Thus,
c 1 þ x2 c1 x2
x ¼ ½x þ x1 ¼ ½x þ ¼ ½x þ þ
10 10 10

c1 c2 cn xnþ1
¼    ¼ ½x þ þ 2 þ  þ n þ n :
10 10 10 10
Since 0  xnþ1 5 1, we have 0  x10nþ1n 5 101n thus lim n!þ1 x10nþ1n ¼ 0. Therefore, we
obtain the following expansion of x into an infinite series:
c1 c2
x ¼ ½x þ þ þ 
10 102
where ci, i ¼ 1, 2, . . . are digits of the decimal system.
Using the decimal notation of [x] we conclude that the real number x has a
decimal representation, x ¼ AsAs1   A0 . c1c2   . œ
Let D be the set of all decimal forms of type
c1 c2
As 10s þ As1 10s1 þ    þ A1 10 þ A0 þ þ þ 
10 102
where s is an integer, Ai, i ¼ 0, 1, . . . , s and cn, n ¼ 1, 2, . . . are digits of the decimal
system with the reservation that in case s  1 then As 6¼ 0.
Proposition 3.3: The function f : [0, þ1) ! D which to every non-negative real
number corresponds its decimal expansion by the algorithm presented in the proof of
Theorem 3.2 is not surjective, but restricted to its image becomes a bijection.
Proof: This function is obviously well-defined. Let x be a non-negative real number
and f(x) ¼ As As1    A0 . c1c2    . It is impossible that every cn, n  m should be 9,
since in that case we have
1 9 9 cm cmþ1 xm 1
¼ þ þ    ¼ m þ mþ1 þ    ¼ m1 5 m1
10m1 10m 10mþ1 10 10 10 10
Thus the function f is not surjective.
 International Journal of Mathematical Education in Science and Technology 895

With the reservation that an infinity of digits of the decimal form are less than 9,
the function f becomes a bijection. In fact, let AsAs1    A0 . c1c2    be a decimal
form with an infinity of digits less than 9. The series As 10s þ As1 10s1 þ    þ
c1 c2
A1 10 þ A0 þ 10 þ 10 2 þ    is convergent and let x be its limit. It is easy to verify that

f(x) ¼ As As1   A0 . c1c2    .

Finally, let x, y be non-negative real numbers with decimal representations
f(x) ¼ As As1    A0 . c1c2   , f(y) ¼ BrBr1    B0 . d1d2    and As As1    A0 . c1c2   
¼ BrBr1    B0 . d1d2   . Then, clearly
As 10s þ As1 10s1 þ    þ A1 10 þ A0 ¼ Br 10r þ Br1 10r1 þ    þ B1 10 þ B0 ,
and
c1 c2 d1 d2
þ 2 þ  ¼ þ 2 þ   :
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

10 10 10 10
Using Theorem 2.4, s ¼ r and Ai ¼ Bi, i ¼ 0, 1, 2,. . . s. In addition cn ¼ dn, for every
n ¼ 1, 2, . . . for if not, let cN and dN be the first pair which differ, so that |cN  dN| 1.
Then
 
X 1
cn X1
dn  1 X1
j cn  dn j 1 X1
9

       ¼ 0:
 1 10n 1
10 n  10 N
Nþ1
10 n 10 N
Nþ1
10n

Thus, since equality must hold, all of cNþ1  dNþ1, cNþ2  dNþ2, . . . must have the
same sign and the absolute value 9. Then, either cn ¼ 9 and dn ¼ 0 for n 4 N, or cn ¼ 0
and dn ¼ 9 for n 4 N, each of these alternatives is impossible. Hence cn ¼ dn for all n.
Thus x ¼ y. œ
Until now, we have presented an algorithm which associates with every
non-negative real number a decimal expansion. We do not know whether or not
there is another decimal representation of a given number. Although, we can
examine the uniqueness of decimal expansion regardless of its construction. In what
follows, we will use the following notation:
Let x be a non-negative real number and a decimal expansion
As As1    A0 : c1 c2   
of x. We put
r0 ¼ As 10s þ    A0 ¼ ½x

c1
r1 ¼ r0 þ
10

c1 c2
r2 ¼ r0 þ þ
10 102
..
.
cnþ1 cnþ2
Thus, x  rn ¼ þ þ    for every n ¼ 0, 1, 2, . . . with
10nþ1 10nþ2
9 9 1
0  x  rn  nþ1 þ nþ2 þ    ¼ n
10 10 10
 896 A. Kalapodi

Lemma 3.4 [6]: Let x be a non-negative real number with decimal expansion

c1 c2
½x þ þ þ :
10 102
There exists natural number ci with ci 6¼ 9.
Proof: If ci ¼ 9 for every i ¼ 1, 2, . . . , then x ¼ [x] þ 1 which is impossible. œ
Thus, any decimal representation should be considered having at least one ci not
9
equal to 9. For example, although the series 10 þ 1092 þ 1093 þ    equals 1 we cannot
accept the form 0.999. . . as decimal representation of 1.
The following two theorems clarify which non-negative real numbers have
precisely one decimal representation and which non-negative real numbers have two
representations.
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

Theorem 3.5: A non-negative real number x which is not a rational number equal to an
irreducible fraction whose denominator is a product of primes each of which is a divisor
of 10 has precisely one decimal representation.
Proof: Let x be a non-negative real number which is not a rational number equal to
an irreducible fraction whose denominator is a product of primes each of which is a
divisor of 10, and
c1 c2
As 10s þ As1 10s1 þ    þ A1 10 þ A0 þ þ þ 
10 102
a decimal representation of x. We know, using the notation presented before, that
0  x  rn  101n , n ¼ 1, 2, . . . . If there exists n such that x  rn ¼ 101n that means all the
digits cnþ1 ¼ cnþ2 ¼    ¼ 9. Then,

1 ½x10n þ c1 10n1 þ    þ cn1 10 þ cn þ 1

x ¼ rn þ n
¼
10 10n
which is impossible. Thus,
1
0  x  rn 5 , n ¼ 1, 2, . . .
10n
whence 0  10nx  10nrn 5 1. Since 10nrn is an integer for n ¼ 0, 1, 2, . . . we have

½10n x ¼ 10n rn and ½10n1 x ¼ 10n1 rn1 for n ¼ 1, 2, . . . :

Hence,
cn
½10n x  10½10n1 x ¼ 10n ðrn  rn1 Þ ¼ 10n ¼ cn for n ¼ 1, 2, . . . :
10n
Thus, all digits cn, n ¼ 1, 2, . . . are uniquely determined. In addition, the unique
definition of the integral part [x] as well as the unique expansion of zero and
Theorem 2.2 guarantee that the non-negative real number x has precisely one
decimal representation. œ
Remark 3.6 [6]: For a non-negative real number x satisfying the requirements of
Theorem 3.5 its unique decimal representation is denoted by x ¼ [x] þ 0, c1c2c3 . . . .
Although we know the formula for the calculation of every digit in this
 International Journal of Mathematical Education in Science and Technology 897

unique representation, the numerical calculations

pffiffiffi are some times extremely
complicated. For example, the 1000th digit of 2 is
pffiffiffi pffiffiffi
c1000 ¼ ½101000 2  10½10999 2
.
Theorem 3.7: A non-negative real number x which is a rational number equal to an
irreducible fraction whose denominator is a product of primes each of which is a divisor
of 10 has two decimal representations.
c1 c2
Proof: Let x be as in the hypothesis, and x ¼ ½x þ 10 þ 10 2 þ    a decimal

representation of x. There exist natural numbers m, l such that x ¼ 10l m . Then,

using the notation presented before, we have 0  x  rm  101m that is
0  l  10mrm  1 and both l, 10mrm are integers. Thus, we have two cases to examine:
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

Firstly, if l  10mrm ¼ 0, then x ¼ 10l m ¼ rm , that is

c1 c2 cm
x ¼ ½x þ þ þ  þ m
10 102 10
So, all digits cmþ1, cmþ2, . . . must be zero. Let m0 be the least natural number with this
property, thus cm0 ¼ cm0þ1 ¼    ¼ 0. If m0 ¼ 1 then x ¼ [x] which is impossible.
If m0 4 1 then cm01 6¼ 0 thus cm0 0 1 ¼ cm0 1  1 is also a digit. Consequently, the
number x also has the representation

c1 c2 cm0 1 9 9
x ¼ ½x þ þ 2 þ    þ m0 1 þ m þ m þ1 þ   
10 10 10 0 10 0 10 0
Secondly, if l  10mrm ¼ 1, then x  rm ¼ 10l m , being possible only in the case where
cmþ1 ¼ cmþ2 ¼    ¼ 9. Let m0 be the least natural number such that
cm0 ¼ cm0þ1 ¼    ¼ 9. If m0 ¼ 1 then x ¼ [x] þ 1, which is impossible. If m0 4 1 then
cm01 6¼ 9 thus cm0 0 1 ¼ cm0 1 þ 1 is also a digit. Thus,

c1 c2 cm0 1 0 0
x ¼ ½x þ þ 2 þ    þ m0 1 þ m þ m þ1 þ   
10 10 10 0 10 0 10 0
c1 c2 cm 1 9 9
¼ ½x þ þ 2 þ    þ m0 1 þ m þ m þ1 þ    :
10 10 10 0 10 0 10 0
œ
Remark 3.8: For example, it is obvious from the two previous theorems that
the rational number 12 can be represented in decimal notation either as 0.5 or
as 0.4999. . . , while the rational number 13 has a unique decimal representation
as 0.3333. . . .
For the following theorem we note that a decimal representation
c1 c2
½x þ 10 þ 10 2 þ   , where x is a non-negative real number, is called finite if there

exists a natural number n0 such that cn ¼ 0 for every n 4 n0, and it is called periodic if
there exist natural numbers s and h such that cnþs ¼ cn whenever n  h. Every finite
representation is considered as periodic. We recall (see Lemma 3.4) that in a decimal
representation it is impossible to have all digits ci, i ¼ 1, 2, . . . equal to 9.
Theorem 3.9: A non-negative real number x is rational if and only if it has a periodic
decimal representation.
 898 A. Kalapodi

Proof: Let x be a rational number equal to an irreducible fraction ml with a decimal

c1 c2
representation x ¼ ½x þ 10 þ 10 2 þ   . Let

x1 ¼ x  ½x
xnþ1 ¼ 10xn  ½10xn , n ¼ 1, 2, . . . :
Then, 0  xn 5 1 for n ¼ 1, 2, . . . and mx1 ¼ l  m[x], that is mx1 is an integer. We
have mxnþ1 ¼ 10mxn  m[10xn] for n ¼ 1, 2, . . . . By induction, it is easy to prove that
all the numbers mxn are integers that satisfy the inequalities 0  mxn 5 m for
n ¼ 1, 2, . . . . If for some n we have xn ¼ 0 then xj ¼ 0 for all j  n. Hence
½10x1  ½10x2  ½10xn1  0 0
x ¼ ½x þ þ þ  þ þ n þ nþ1 þ    :
10 102 10n1 10 10
Thus x has a finite representation. Further, if xn 6¼ 0 for all n ¼ 1, 2, . . . , we have
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

0 5 mxn 5 m for n ¼ 1, 2, . . . . So, the numbers mx1, mx2, . . . , mxn can take only m  1
different values 1, 2, . . . , m  1. It follows that there exist natural numbers h and s
such that h þ s  m and mxh ¼ mxhþs, which proves that xn ¼ xnþs for n 4 h and
therefore cn ¼ cnþs, for n  h. Thus the presentation is periodic.
Conversely, we shall prove that if the sequence of digits c1, c2, . . . is periodic then
c1 c2
the number x ¼ ½x þ 10 þ 10 2 þ    is rational. In fact, there exist natural numbers s

and h such that cnþs ¼ cn whenever n  h. Then,

c1 c2 c1 ch chþ1 ch
þ 2 þ  ¼ þ    þ h þ hþ1 þ    þ hþs þ   
10 10 10 10 10 10

c1 c2 . . . chþs1  c1 c2 . . . ch1
¼ ,
10h1 ð10s  1Þ
which is a rational number. œ
Corollary 3.10 [6]: If a non-negative number x has a non-periodic representation as a
decimal, then x is irrational.
It is easy now to expand a negative real number in its decimal form following the
procedure described for negative integers. That means that if x is a negative real
number we define as decimal representation of x, the series that comes from the
representation of x by changing the sign in all coefficients. All theorems are still
valid, with trivial modifications.

4. Representation in systems with base n

As we have mentioned in Section 1, apart from the decimal system many other
number systems have been used in arithmetical calculations. In the past decades,
some of them have been widely spread due to their application in computer science.
Moreover, number systems appear in theoretical problems not only in Number
Theory but also in Analysis. One of the critical questions, which is answered in
previous sections for the decimal system, is if a number system can sufficiently
express the real numbers. All the results already mentioned for the decimal system
are valid in systems with base n, n being a natural number greater than 1, with slight
modifications.
 International Journal of Mathematical Education in Science and Technology 899

In this section, we quote the new definitions, and we state the corresponding
theorems giving only indications of proofs when necessary. Complete proofs can be
found in [3] or in [6].
Definition 4.1 [6]: Let g be a given natural number greater than 1. We say that a
natural number N is expressed as a decimal in the scale of g if
N ¼ cm gm þ cm1 gm1 þ    c1 g þ c0
where m is an integer  0 and cn, n ¼ 0, 1, . . . , m are integers with the property
0  cn  g  1 for n ¼ 0, 1, . . . , m and cm 6¼ 0. If each number of the sequence
0, 1, 2, . . . , g  1 is denoted by a special symbol, these symbols are called the digits
and N can be rewritten in the form
N ¼ ðm m1 . . . 1 0 Þg ,
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

where  n is the digit which denotes the number cn.

If g  10, the symbols 0, 1, 2, . . . , 9 are taken as the digits in the scale of g.
For example N ¼ (10010)2 means N ¼ 1  24 þ 0  23 þ 0  22 þ 1  2 þ 0 ¼ 18 and
N ¼ (5603)7 means N ¼ 5  73 þ 6  72 þ 0  7 þ 3 ¼ 2012.
Theorem 4.2: Any natural number N can be uniquely expressed as a decimal in the
scale of g, g being a natural number 4 1.
Proof: The formula giving the nth digit of the decimal expansion of N in the scale of
g is
   
N N
cn ¼ n  g nþ1 for any n ¼ 0, 1, . . . , m  1,
g g
and


N
cm ¼ m ,
g
h i
log N
where m ¼ log g . œ
Remark 4.3: Analogously, it can be proved that every integer number has a unique
decimal representation (under the sign reservation) in the scale of g.
Definition 4.4 [6]: A non-negative real number x is expressed as a decimal in the
scale of g, g being a given natural number greater than 1, if
c1 c2 c3
x ¼ As gs þ As1 gs1 þ    þ A1 g þ A0 þ þ 2 þ 3 þ   
g g g
where
() s is an integer
() Ai, i ¼ 0, 1, . . . , s and cn, n ¼ 1, 2, . . . are digits in the scale of g, with the
reservation that in case s  1 then As 6¼ 0.
() [x] ¼ Asgs þ As1 gs1 þ    þ A1 g þ A0.
If the non-negative real number x is expressed as a decimal in the scale of g, the series
As gs þ As1 gs1 þ    þ A1 g þ A0 þ cg1 þ gc22 þ    is also denoted as
x ¼ ðAs As1    A0 : c1 c2   Þg
 900 A. Kalapodi

Theorem 4.5: Let g be a given natural number greater than 1, and x be a non-negative
real number. Then:
(1) The number x has a decimal representation in the scale of g.
(2) In any decimal representation of x in the scale of g, it is impossible to have all
cn, n ¼ 1, 2, . . . , equal to g  1.
(3) If x is not a rational number which is an irreducible fraction with a denominator
such that any prime divisor of it divides g, then number x has precisely one
representation in the scale of g. Moreover, this unique representation is infinite
and has infinitely many digits different from g  1.
(4) If x is a rational number equal to an irreducible fraction whose denominator is a
product of primes each of which is a divisor of g, then x has two decimal
representations in the scale of g. In one of them all cn, n ¼ 1, 2, . . . , except a
Downloaded by [Heriot-Watt University] at 05:50 27 December 2014

finite number are equal to zero, in the other from a certain n onwards all cn are
equal to g  1.
(5) The number x is rational if and only if it has a periodic decimal representation
in the scale of g.
We mention that analogous results hold for number systems with negative base
or even with base that is not a rational number.

References

[1] O. Ore, Number Theory and its History, Dover Publications, New York, 1988.
[2] R. Dedekind, Essays on the Theory of Numbers, Dover Publications, New York,
republication, 1963.
[3] G.H. Hardy and E.M. Wright, An Introduction to the Theory of Numbers, 5th ed., Oxford
Science Publications, Clarendon Press, Oxford, 1993.
[4] A.M. Kirch, Elementary Number Theory: A Computer Approach, Intext Educational
Publishers, New York, London, 1974.
[5] J.A. Anderson and J.M. Bell, Number Theory with Applications, Prentice Hall, Upper
Saddle River, NJ, 1997.
[6] W. Sierpiński, Elementary Theory of Numbers, North-Holland, Amsterdam, 1988.