1

 Index

Page

Introduction 7

Chapter 1. On the question, what are numbers? 11

1.1. A dual misunderstanding 12

1.2. Mathematical and philosophical approaches 14

1.3. Solution to the diversity of approaches 40

Chapter 2. Are there numbers outside of mathematics? 43

2.1 How do we understand the assertions of existence?

2.2. What have others said about the numbers?

2.3. Existence of pre-mathematical numbers

Chapter 3. The mathematical numbers 69

3.1. Mental operation for counting

3.2. Primitive languages for counting

3.3. Manipulation of numerical symbols

3.4. The ontology and epistemology of numbers

Chapter 4. Mathematical explanations on numbers 85

4.1. Numbers seen as magnitudes 118

4.2. Numbers seen as sets 137

Chapter 5. Solution to the diversity of referents 145

Bibliography 156

 Acknowledgements

The original idea of this book I set out for the first time in my doctoral dissertation that
called Mathematics and empirical science: What could be the numbers? UNAM, 1989,
México. However, there I used another strategy to prove the point. So, this book is
inspired by the dissertation, but rather, I used several subsequent works through which I
have advanced in the clarification of some ideas. So part of the material in this book has
previously been published in the form of articles or chapters in collective books, some
in Spanish and some in English. In some cases there have been no substantial changes,
but other changes have been greater. The following are references to paragraphs and
chapters containing previously published material:

- Chapter 1, "On the question, what are numbers?" is a updating of the chapter
“What is Philosophy of Mathematics looking for?” in 18 Unconventional Essays on the
Nature of Mathematics of Reuben Hersh (comp.), Springer 2006, USA.

- The Chapter 2, “Are there numbers outside of mathematics?” is a translation

and updating of the paper “¿Existen números fuera de la matemática?” Theoria, year
VIII, 1993, No. 19.

-Subsection 3.2, "Primitive languages for counting" is a translation and updating

of the article "Reflexiones sobre el origen de los números aritméticos", Analogía
Filosófica, year 11, 1997, no. 1

The rest of the book is completely unpublished at the time of this writing.

My thanks to the following colleagues for contributing their criticism of various

parts of this book from its original gestation in my doctoral dissertation to its present
form: Carlos Álvarez (México), Adolfo García de la Sienra (México), Alejandro
Garciadiego (México), Reuben Hersh (USA), Raymundo Morado (México), Ulises
Moulines (Germany), León Olivé (México), Raúl Orayen (dead), Mark Platts (México),
Juan Antonio Robles (México), Alejandro Tomasini (México), Jonatan García
(Durango) and Damián Islas (Durango). Of course, they are not responsible for any use
or neglect of their comments.
5

 Also thanks to the students of the course entitled "Philosophy of Mathematics",
taught in the School of Mathematics of the University Juárez of the State of Durango,
from 27 February to 27 March 2010. These students were Paola Ríos Alvarado, Andrés
Ceniceros Nájera, Miguel Ángel Ortiz Castañeda, Francisco Javier López Fragoso and
Gerardo Salvador Corral. The course focused on the content of this book, and the
stimulating discussions that occurred throughout the course led me to incorporate
several changes and clarifications.

 Introduction

In looking at Frege’s definition of numbers,1 I discovered that Frege did not define
arithmetical numbers; rather, he built other numbers that resemble and explain
arithmetical numbers. Fregean numbers are similar to arithmetical numbers because
both meet Peano Axioms. In addition, Fregean numbers explain arithmetical numbers,
because Frege suggests that arithmetical numbers are not related directly to the
experience of counting but are only related indirectly with that experience.

While studying Euclid, I also discovered that he had also built certain numbers
that are similar to arithmetical numbers. Euclidean numbers were constructed in terms
of magnitudes. By using these numbers, one can attempt to explain, inter alia, how the
joining of arithmetical numbers gives rise to other numbers.

On the basis of these examples, I came to the idea that there are two kinds of
numbers, arithmetical and meta-arithmetical, with the latter being explanatory paintings
of the first.

This idea generated the following questions: a) What are arithmetical numbers,
ontologically speaking, and how is it that we know them? b) If Euclidean and Fregean
numbers can be viewed as paintings of arithmetical numbers, could we see the
arithmetical numbers as paintings of other numbers that are more directly related to the
experience of counting? c) What is the nature of those numbers that I call “pre-
arithmetical numbers”, if such numbers indeed exist?

1
Throughout this work, by “numbers”, I always mean natural numbers, of which I suggest that
there are three levels. Therefore, I shall not speak of real numbers, imaginary numbers, or other
numbers that are not natural.

 I label “pre-arithmetical numbers” as first-level numbers, “arithmetical
numbers” as second-level numbers, and “meta-arithmetical numbers” as third-level
numbers. I must clarify that the expression “arithmetic” is equivalent, throughout this
work, to the expression “mathematical”; thus, to say “mathematical numbers” is
equivalent to saying “arithmetical numbers”.

I found the ontology of arithmetic numbers by pursuing three routes, all of

which ended in the same place: a) I first pursued a logical analysis of the process of
counting; b) I also engaged in an historical analysis of the first symbols used to
represent numbers; and c) I found in the first two definitions of numbers in volume VII
of Euclid's Elements that are the same definitions that I had discovered by following the
other routes. This led me to define arithmetic numbers as the following: they are the
abstract result of the mental process of counting that is symbolised and manipulated
arithmetically. Thus, arithmetic can be seen as the systematisation of a basic mental
process that symbolises the result of a repeatable action. Subsequently, as these symbols
arose, they were objectivised and manipulated independently of their meaning.

I answered affirmatively the question regarding the existence of pre-

mathematical numbers. I did so by observing the linkage of arithmetic numbers with
things of the world in expressions such as “3 horses plus 2 horses yields 5 horses”. This
expression apparently uses the arithmetic numbers 3, 2, and 5, but is actually talking
about concepts which I call pre-mathematical numbers. The expression means that a
trio, of anything plus a pair of the same adds up to a quintet of those same things. This
is, indeed, a relationship between the concepts "trio", "pair", and "quintet".

As often happens, I will change the order of discovery and start by giving first-
level numbers, then second-level numbers, and conclude with third-level numbers. Prior
to that, we will see in chapter 1 the problem we are trying to resolve with the
proposition that there exist three levels of numbers. I want to stress that throughout this
text, I will use multiple diagrams because I believe that the notion of painting, which is

explained in section 2.1, is central to understanding why I say that third-level numbers
are paintings of second-level numbers and why these, in turn, can be viewed as
paintings of first-level numbers.

 10

 Chapter 1

On the question, what are numbers?

11

1.1. A dual misunderstanding

There is a lack of clarity regarding the question “what are numbers?” First, we must
clarify from which approach we are asking the question; second, we must make clear
the class of numbers to which we refer. Failure to provide these clarifications has often
led to a dialogue marked by miscommunication.

We can pose the question from two approaches: the mathematical and the
philosophical. In mathematical terms, Peano (1889), Dedekind (1893) and Hilbert
(1900) defined numbers in terms of classes, systems (series) or the mutual relations
between numbers. For these authors, a number is defined within mathematics.

For philosophers, however, this definition is insufficient. Russell (1919, p. 10),

for example, said

We want our numbers to be such that they can be used for counting common objects,
and this requires that our numbers should have a definite meaning, not merely that they
have certain formal properties.

Frege defines numbers in relation to concepts from the philosophical point of

view. Accordingly, an affirmation of numbers is an affirmation of concepts and their
relationships. Thus, Frege gives an ontology for numbers that relates them with
something extra-mathematical.

Formally, both mathematicians and philosophers are looking for an answer to

the same question: what are numbers? In reality, they are looking for different responses
as I will explain in 1.2. Mathematicians are looking for the structure of mathematical
numbers, while philosophers are looking for how numbers are possible or what their

12

place is in the world. It is symptomatic that for mathematicians in general, Peano is
more important than Frege, while for philosophers, the opposite is true.

From the philosophical point of view, there is also confusion about numbers and
the symbols we use to represent them. For example, does the symbol “3” mean the same
thing in the expressions “3 horses”, “3 + 2 = 5”, and “3 is a set”? I think not. In this
work, I will try to show that there are three levels of numbers: pre-mathematical
numbers, such as the number mentioned in “3 horses”; mathematical numbers, such as
those mentioned in “3 + 2 = 5”; and meta-mathematical numbers, such as the number
mentioned in “3 is a set”. I designate these numbers as first-, second-, and third-level
numbers, respectively.

In relation to different classes of numbers, Plato had already spoken of the

existence of two kinds of numbers: peer (pure) numbers and unequal (or physical)
numbers (Philebus, Sect. 56d 10). Following this line of thinking, Kline (1968) stated
that the Greek term arithmos does not mean the same as the modern number defined by
Vieta. The first term means a number of things, while the second means a concept.
Buldt, Löwe and Müller (2008, p. 314) say that to "identify the Greek numbers with the
modern axiomatic characterisation seems very strange". Going further, there are those
who, like Watson (1990, p. 283), claim that "the natural number is a cultural construct
formulated differently in different cultures".

In these and other similar cases, we must first ask what kind of numbers we are
talking about and what approach we will consider. That is why in the following
paragraphs, we first discuss different mathematical approaches.

1.2. Mathematical and philosophical approaches

1.2.1. The dialogue of the deaf

The question “what is mathematics?” has had relatively few and, in general,
contradictory answers. I believe that one of the problems is that it is not clear through
which focus it is best to attempt an answer, due in large measure to the vagueness of the

13

question itself. The approaches could be many, but I will concentrate here only on the
mathematical and the philosophical.

In addition to the vastness of opinion regarding the nature of mathematics

(which could be cleared up by analysing facts and reasons in favour of or against one or
another hypothesis), I hold the claim that there is a dialogue of the deaf taking place,
due to the fact that there is no consensus as to what an answer might be.

To recognise this dialogue of the deaf, let us examine some examples. First, we
find that some authors, both mathematicians and philosophers of mathematics, are
apparently talking about the same thing. For example Körner (1967) says:

It is characteristic of mathematical theories that they can themselves become the subject
matter of mathematical theories. It is thus in principle possible for mathematical
theories and philosophical theories about mathematics to be incompatible. (p. 118)

Within the discussion that has taken place between the philosophy of
mathematics and mathematics itself, much has been said about the limitative theorems
of Gödel and others. Rodriguez-Consuegra (1992, p. 446), says:

Gödel proved the existence of propositions true but undemonstrated in a formal system
sufficiently rich for containing arithmetic... It seems to me that the more relevant
philosophical consequences are the following: Once they proved that truth and
demonstrability are different things, then the truth of certain propositions is directly
intuitive.

Körner (1967, 132), going far beyond this, concludes that,

The meta-mathematical discoveries of the present century imply the falsehood of the
common doctrines shared by the classical philosophies of non-competitive
mathematical theories.

14

 Though it is unanimously accepted that mathematics are recursive, the hidden
premise that philosophy, too, is formed by theories is not accepted. In fact, in writing
against this idea, Wittgenstein (1918, 4.112) says that “Philosophy aims at the logical
clarification of thoughts. Philosophy is not one doctrine but an activity”.

Confronted with the idea that mathematics and the philosophy of mathematics
can be rival theories, I believe that Wittgenstein (1967) would have responded in the
following way:

A philosophical problem has the form: “I don't know my way about” (123). “It leaves
everything as it is. It also leaves mathematics as it is, and no mathematical discovery
can advance it” (124). “It is not a matter of the philosophy to resolve a contradiction
through a mathematical discovery” (125).

Even if we agree with Wittgenstein that the philosophy of mathematics is

different from mathematics itself or from meta-mathematics, we still must elucidate
how philosophical work is done and why is it different from mathematical work.
Wittgenstein (1918, 4.112) says that “A philosophical work consists essentially of
elucidations”. However, what is not clear for one person may be clear to someone else.
Concerning this matter, Putnam (1967, p. 296) affirms that “The fact that philosophers
all agree that a notion is ‘unclear’ doesn’t mean that it is unclear”.

We can find other examples of the dialogue of the deaf when we see that, in the
view of some mathematicians, philosophers have attempted to place mathematics within
their own philosophical perspective. However, those attempts are almost always
external to mathematics itself, and mathematicians reject the characterisations that are
made about them by philosophers. For example, Santiago Ramírez (1990, p. 419) tells
us that:

They [Philosophers] have conceived traditionally the relation between philosophy and
mathematics as that in which philosophy, whatever its metaphysical foundation, tries to
subject mathematics to philosophical discourse, or philosophical norm. From
15

 Pythagoras to Analytical Philosophy the question is to exhibit mathematics as a
discipline, discourse, or special kind of knowledge where philosophical or
epistemological hypothesis about existence, about truth, and about method are
confirmed.

In these claims, as Ramirez (1989, p. 318) said, following Cavailles: "the

essence of mathematics is a problem, among others, that philosophy cannot resolve". In
the same vein, Courant and Robbins (1941, p. 7) conclude that "For scholars and
layman alike it is not philosophy but active experience in mathematics itself that alone
can answer the question: What is Mathematics?"

Perhaps the difficulty in resolving this issue is the nature of philosophy itself, as
Ramírez, Cavailles, and Courant Robbins seem to say. That is, philosophy cannot solve
this issue. Or perhaps, as Hersh (1979, p. 34) says, the reason is the fact that "there are
not many professional philosophers who know functional analysis, algebraic topology
or stochastic processes". This position is reaffirmed by Amor (1981, p. II), commenting
on the work of Hersh: "this is a reflection of a mathematician and not of a not-
mathematical philosopher, so it's a true reflection of actual Mathematics".

The first proposition forces us to define the limits and the reach of philosophical
reflection, which we will attempt in 1.2.3. The second proposition forces us to clarify
what we understand by “knowing mathematics”.

Concerning the latter, Maurice Frechet (1955, p. 21) said, “mathematicians do

not know, for example, the whole mathematical analysis”. In contrast, Keneth Ribet (in
Lemonick 1993, p. 13) says that the number of mathematicians who really could
understand completely the arguments of Wiles about Fermat’s theorem could meet in a
meeting hall. Therefore, what is the meaning of the expression “to know mathematics?”
Perhaps it is not enough to be a mathematician. Maybe it would be necessary to be a

16

creator of a part of mathematics. Maybe even this would not be enough because of the
growing and unfinished complexity of mathematics.

Therefore, who, or what discipline, can answer the question “what is

mathematics?” What qualifications are required for such a task? Some philosophers,
such as Plato, Aristotle and Kant, were not mathematicians but contributed important
ideas to the philosophy of mathematics. Other philosophers, such as Descartes and
Leibniz, were philosophers and great mathematicians as well. Others, such as Frege and
Wittgestain, had mathematical training. It would seem then that these circumstances are
not determinant factors concerning the relevance of their ideas regarding mathematics.

It is also the case that some mathematicians, such as Cantor, Poincare and
Frechet, among others, have engaged in reflection on the different ways by which they
approach their own work, thereby providing us with access to their vast psychological
and historical experiences. Yet, are they doing philosophy? Hersh (1979, pp. 34-35)
said the following about this:

But the art of philosophical discourse is not well developed today among
mathematicians, even among the most brilliant. Philosophical issues just as much as
mathematical ones deserve careful arguments, fully developed analysis, and due
consideration of objections. A bald statement of one’s own opinion is not an argument,
even in philosophy.

Accordingly, to understand how we can do philosophy of mathematics and how

can we understand differing claims concerning mathematics, I propose that we should
first clarify what is it that a philosopher and a mathematician look for when each of
them has questions concerning mathematics. That is to say, how does each understand
the question “what is mathematics?”

17

 1.2.2. Mathematical focus
Mathematicians in general are not interested in the question. Only a few among them
have stopped to explore the subject, and they have done so with various kinds of tools.
Nonetheless, it would seem that almost all of them believe that it is an internal affair of
mathematics itself. In Cavailles (1938, p. 172) words: “There is no definition, nor
justification of the mathematical objects, except mathematics itself”.

However, if that is true, how is it possible to reflect about mathematics from the
view point of mathematics itself? It seems there have been three ways: 1) doing meta-
mathematics; 2) examining the history of mathematics, and 3) practicing mathematics.

1.2.2.1. Meta-mathematics
It is generally accepted that mathematics is recursive; that is to say, mathematicians can
mathematically re-work their results. However, if we do not know what mathematics is,
it would be difficult to explain how we can do the mathematics of mathematics. In any
case, I believe that, at least, it is possible to do mathematics axiomatically and non-
axiomatically.

Let us remember that the use of axioms is a technique based on the ideas of
Plato and Aristotle. The use of axioms consists in ordering certain kinds of knowledge
(mathematical or non-mathematical) and finding affirmations (axioms) from which we
can deduce all the other affirmations of that body of knowledge. Knowledge organised
in this way can, in theory, substitute for the original knowledge while gaining in clarity
and precision.

Polya (1957, p. vii) says:

18

 Mathematics has two faces... Mathematics presented in the Euclidean way appears as a
systematic, deductive science; but mathematics in the making appears as an
experimental, inductive science.

In the words of Maurice Frechet (1955, pp. 21, 22):

Mathematics is not a completely logic theory... In spite of the fact that most of the
mathematical works consist in doing logical transformations from propositions admitted
as truthful..., it is not hard to admit that intuition guides the work in a specific direction.

Or, in the recent words of William Byers (2007, p. 10):

This book offers another vision of mathematics, a vision in which the logical is merely
one dimension of a larger picture.

In this sense, there can be axiomatic and non- axiomatic forms of meta-
mathematics. It is characteristic of both of these forms that they depart from some
mathematical theories and construct other mathematical theories. It would seem that the
difference consists in the following. While the axiomatic approach proceeds with the
intention of substituting the primitive theory by another theory that makes theoretical
propositions clearer and more precise, the non-axiomatic approach does not provide a
substitute for primitive theory but rather makes such theory dependent upon another
theory. We can say that both axiomatic and non-axiomatic meta-mathematics explain
and clarify primitive mathematical theory. One does so by reducing primitive
mathematical theory to its constituent parts and the other by relating such theory to
other elements or to other mathematical theories.

With respect to the axiomatic theories (called foundations), I believe that they
are by their own nature meta-mathematical and certainly attempt to clarify the
relationships of mathematical entities. In Gödel’s own words:
19

 The so-called logical or set-theoretical “foundation” for number-theory, or of any other
well established mathematical theory, is explanatory, rather than really foundational,
exactly as in physics where the actual function of axioms is to explain the phenomena
described by the theorems of this system rather than to provide a genuine “foundation”
for such theorems. (in Lakatos, 1978, p. 27)

However, the limitations of formal systems in axiomatic work are well known.
For this reason, axiomatic theories, in general, have not replaced original theories, and
both subsist. Hersh (1979, p. 38) says:

The common presupposition was that mathematics must be provided with an absolutely
reliable foundation. The disagreement was on strategy, on what had to be sacrificed for
the sake of the goal. But the goal was never attained, and there are few who still hope
for its attainment.

It seems that the philosophy of mathematics went into crisis with the appearance
of Gödel’s work. However, his theorems concerning the incompleteness and
inconsistency of arithmetic address formal systems, in Hilbert’s sense. They are
important for axiomatic meta-mathematics (or foundational meta-mathematics). They
demonstrate that an axiomatic picture of arithmetic does not recollect the totality of
arithmetic. This demonstration is interesting for meta-mathematics and mathematics
because it allows us to examine the place and the utility of formal systems in the totality
of mathematics. It is an internal affair concerning the relations of mathematical entities.
That being the case, the crisis was internal to Hilbert’s and similar programs, such as
Logicism and Intuitionism. However, mathematics itself and its philosophy went their
own separate ways.

With respect to non-axiomatic meta-mathematics, we could mention Group

Theory in its first stages and, in general, what Cavailles (1938, p. 173) called “theme”,
that is to say, “a transformation of one operation in element of one superior operational
field: example, the topology of the topological transformations”. If we see primitive
20

axiomatic theories as structures, this meta-mathematical work would be building other
structures formed by simple structures. In other words, it would be constructing
complex structures through which it is possible to study and clarify the properties and
relationships of certain primitive structures and their elements.

Briefly, meta-mathematics is a development of mathematics. In either case, we

can say that both meta-theories attempt to substitute for or subsume primitive structures.
In fact, they explain and clarify primitive theories and tend to make mathematics more
homogeneous.

Consequently, with regard to the question “what is mathematics?” it seems that

some mathematicians understand it as a quest to discover the properties and internal
relations of the elements that form mathematics.

1.2.2.2. Historical analysis

In opposition to meta-mathematics and, in particular, to the axiomatic approach, some
mathematicians, such as R. Thom (1980, p. 27), claim that work in meta-mathematics is
not sufficient because

Formalism denies the status of mathematical to most of that which they commonly have
understood as mathematics and it does not say something new about its development.

For this reason, some believe that an answer that would take more completely into
consideration the fertility of mathematical work must utilise a historical analysis that
clarifies the facts and methods that make mathematical work possible.

Garciadiego (1996, p. 14) says

21

 As historians of mathematics and sciences we are interested in knowing the origins of
problems which men have tried to solve in the past time, the ideas they used as a
starting point, and what they expected as an answer…; that is, how an idea is born,
grows and changes in order to conform the science’s field.

It is generally accepted that history is an intelligible and relevant reconstruction

of data. This description implies that an interpretation of those data is involved.
“History emerges when chronology is selected, organized, related, and explained”
(May, 1974, p. 28). Thus, the history of mathematics is an explanation that provides us
with an image that takes into account not only the final result of a mathematical theory
but also its origin and development. I believe that this image allows us better to
understand how mathematics was born and subsequently developed.

Briefly, the history of mathematics gives us an approach to the subjective

elements that have been present in the conformation of the objective elements of
mathematics. Consequently, it seems that historians of mathematics understand the
question “what is mathematics?” as asking for the origin and development of
mathematical elements.

1.2.2.3. Mathematical practice

Among the practical works, there are at least two important varieties: a) those that only
do mathematics, and b) those that are the result of a psychological or heuristic self-
reflection.

For the first variety, such as that practiced by Courant and Robbins (1941),

What points, lines, numbers “actually” are cannot and need not be discussed in
mathematical science. What matters and what corresponds to “verifiable” facts is
structure and relationship... For scholars and laymen alike it is not philosophy [and

22

 maybe history not either], but active experience in mathematics itself that alone can
answer the question: What is mathematics? (pp. iv, v)

For these and other authors, what is of interest to mathematicians regarding

numbers is not their ontology or how we came to know them. Rather, by taking
numbers for granted, mathematicians are to inquire into their generalities, the existence
of some numbers such as irrationals, transfinites, and ideals, and problems derived from
those numbers, such as the problems of continuity and infinity (see Frechet, 1955, pp.
417-449).

From this perspective, we could perhaps agree with Newton Da-Costa that
“mathematics is all that which is in mathematical books and reviews”. This
characterisation, of course, is insufficient. However, in its defence, we could say that
mathematics is an interminable field and “with respect to the motor of progress, it seems
to escape to investigation” (Cavailles, 1938, p. 175).

With respect to heuristic-psychological works, we can mention Pappus, who lived

at the end of the third century A. D., and, in our day, Polya (1957) and Velleman
(1994), bypassing Descartes, Leibniz, Bolzano, and Poincaré, among others. These
scholars deal, in general, with the conscious and even unconscious ways by which
mathematicians have attempted to solve problems, such as the regressive reasoning
masterfully exposed by Pappus, the method of analysis-synthesis used in book XIII of
Euclid’s Elements, Eleatics’ use of reduction ad absurdum, mathematical induction,
analogous reasoning, and the recourse to drawing a figure,2 among others. It seems that
in all these works, mathematicians are guided by an “aesthetic feeling which all true
mathematicians know... because the useful combinations are, just, the more beautiful”
(Poincaré, 1908, p. 52). Occasionally, they are also helped by the unconscious, as
Poincaré and Polya maintain.

2
Which De Lorenzo (1993) calls “figural work.”

23

 Certainly, there are many mathematicians even in our times, such as Cavailles
himself (in some of his works) as well as others, who try to characterise mathematics by
reference to their own mathematical practice. This is the case for J. De Lorenzo (1992,
pp. 447, 448), when he says:

It is a myth that all mathematical work is a logic syntactic work... In mathematical

practice, axioms are not the starting point, they are not the key of the knowledge
process, but nuclear concepts are, and some times, hypothesis or conjectures.

In the recently published book How Mathematicians Think (2007), W. Byers

examines the role of ambiguity, and even the contradictions and paradoxes that are to be
found in mathematical creativity.

These authors attempt to describe how mathematicians work and, in so doing, to

describe how mathematics became what it is now. Thus, it seems some mathematicians
understand the question “what is mathematics?” as a question regarding mathematical
practice, that is, as an interrogation of what mathematicians do.

1.2.2.4 Synthesis-approach mathematicians

To sum up the mathematical approach and apply it to the question of numbers, we can
say that many mathematicians understand the question “what are numbers?” as an
inquiry into the properties of those entities and their relations with other mathematical
entities, as necessitating an exploration of their origin and development within
mathematics, or as requiring an examination of the ways mathematicians use them.

1.2.3. Philosophical focus

Philosophers, even when they appreciate and use the reflections discussed above, are in
general not satisfied with such materials. They can agree with certain meta-

24

mathematical or historical projects, but it seems that philosophers3 ask other questions
as well.4 But for what are philosophers, or mathematician acting as philosophers,
looking? That is, what constitutes for them an acceptable, or even controvertible,
answer to the question “what is mathematics?”

1.2.3.1 What is philosophy?

When we ask about philosophy, it is a genuine question. I do not have a definitive
answer, and I believe that we cannot find one easily. Nevertheless, because the present
investigation requires a precise understanding of what we mean by “philosophy”, I will
make my view explicit.

I think that we can see philosophy as a Socratic dialogue within a particular

community,5 which starts with doubts about inherited knowledge and grows through
time in an ongoing attempt to clarify concepts or to build new concepts through which
that community thinks about the world. As Hegel said, "Philosophy is to think about
thinking", and I add “by means of which man has questioned thought itself, particularly
its limits and capabilities, and pondered its fruits”. We can sum up these questions in
Kantian terms. How is thought possible? Under which set of assumptions does thinking
make sense? What is the place of thought in the world?

3
This distinction, of course, is not exact. In fact, there have been mathematicians and
historians who worked on central philosophical subjects. We can mention, for instance, the
strong philosophical concerns of Gödel (1949), which, it seems, guided him to his technical
works. On the other hand, I believe that it would be unjust to classify Wilbur Knorr’s work as
only historical because, I believe, it is also philosophical.
4
As Shapiro (1994, p. 157) says, “Philosophers have their own interests, beyond those of their
colleagues in other departments”.
5
I understand philosophy here as a dialogue within Western culture. Other cultures, such as
that of China or India, have their own internal dialogues. That circumstance makes it difficult,
but not impossible, for a dialogue to exist between different cultures. Within Western culture
there are also various subcultures, each with its own internal dialogue.

25

 Ontological and epistemological inquiries point in precisely the same direction
because the first kind of inquiry attempts to clarify concepts by which we think about
the world and the second seeks to clarify how our thinking is related to the objects
under study at a given time. In other words, how can we know objects?

This characterisation is tentative and only intends to pick out some of the more
general characteristics of philosophy to clarify the philosophy of mathematics. There
are, of course, other characterisations, though not an excessive number because only a
few philosophers have turned their attention to that question. We will comment on two
very common characterisations.

The first characterisation says that philosophy is the mother of science. For
instance, Cornman, Lehrer, and Pappas (1992, p. 5) say,

Philosophy was once construed so broadly as to cover any field of theoretical inquiry.
Any subject matter for which some general explanatory theory might be offered would
have been a branch of philosophy. However, once a field of study came to be dominated
by some main theory and developed standard methods of criticism and confirmation, then
the field was cut off from the mother country of philosophy and became independent.

In support of this thesis, we can mention the fact that, in the past, many
philosophers were also scientists, for example, Aristotle and Descartes. Some scientists,
such as Newton, for example, in turn called their scientific work “natural philosophy”.
Nevertheless, even and when from Tales philosophical work has closely resembled
scientific work, that does not give us license to say that philosophy and science address
the same subjects and attempt to discover the same things.

I believe that we can explain from my perspective why some people see
philosophy as the mother of science. In fact, I believe that it would not be hard to accept

26

that critical and dialogical works of philosophy addressing inherited knowledge can
provoke new research in some scientific fields. For example, the critical work of
Berkeley (1948-57) on Analysis provoked Weierstrass’ mathematical work (cf. Robles,
1993, pp. 219-227). However, such examples do not prove that mathematics was born
of philosophy, only that both disciplines influence one another.

Mathematicians usually do not care about philosophical work. Some of them

even openly reject such work, as do Courant and Robbins (1941). However, sometimes,
philosophical work has had an effect on scientific work, even though such was not the
original purpose. For this reason, some philosophers have thought philosophy is the
mother of science. However, scientific work has also led to new philosophical work.
Such was the case when Kant was inspired by Newton's scientific work. I believe that
there is a relationship between these fields, but they look for different things, and each
has its own methods.

Another common idea about philosophy has its roots in phenomenology and
existentialism. We can see this, for example, in the teachings of Ortega and Gasset
(1973, p. 177), where he says, "the radical problem of philosophy is defining that mode
of being, that primary reality which we call our life". This idea is close to what many
people think philosophy must be: a view of the world or a personal cosmological view.

This view is based on the belief that Physics, Chemistry, Astronomy,

Psychology, and so on, are disciplines that can tell us about the world, while philosophy
cannot. However, these disciplines, in turn, could not answer the question: “what is the
meaning of our life?” or, as Heidegger would say, "why existence is, and nonexistence
is not?" People who hold this view think that philosophy could answer such questions
and believe that philosophy, aided by particular sciences, must give us that general view
about our world.

27

 However, there is a large problem with this conception, in that it requires that
philosophers must understand all the sciences, and attaining such breadth of
understanding is virtually impossible today. Philosophy, from my point of view, is not a
superior judge for the sciences and is not a super-science. I believe that any philosopher
can only know deeply one scientific discipline or, perhaps, only one theory in that
discipline. The only recourse any single philosopher has is to attempt to draw from that
theory its suppositions or the view of the world it implies.

This task is not very different from my characterisation of philosophical activity.

I believe that philosophy attempts to clarify concepts, and when we extract the
assumptions or the general view of one theory, I think we can see most clearly the true
meaning of the concepts that theory uses. In this regard, philosophy does not propose a
worldview but rather informs us of the worldview implicit in the discourse it analyses.
Philosophers perform this operation, for instance, when they analyse the ontology and
epistemology of a theory.

1.2.3.2. Philosophy of mathematics

Consistent with the characterisation of philosophy I proposed above, I think that we can
see the philosophy of mathematics as the dialogue about mathematics by means of
which we try to make clear the ontology and epistemology of mathematics; in other
words, the philosophy of mathematics addresses the following questions. How is
mathematics possible? Under what assumptions does mathematics have meaning? I
believe this position accords with that of Penelope Maddy (1997-2005), when she
reserved the term “philosophy” for metaphysical questions by separating such
questions, as soon as possible, from purely methodological issues.

Perhaps the first philosophers of mathematics were the Pythagoreans,6 when

they said "number is the principle both as matter for things and as constituting their

6
From Burkert (1972), it is thought that the image we have of the Pythagoreans and about
Pythagoras himself was the work of the Neo-Platonists and Neo-Pythagoreans, who built a
28

attributes and permanent states" (in Aristotle, Metaphysics, A.5 986-a-16). In the words
of an important member of the Neo-Pythagorean group, Nicomachus (C 1 A.D., Book
1, Chap. IV),

Which of these four methods must we first learn?... this is arithmetic, not solely because
we said that it existed before all the others in the mind of the creating God like some
universal and exemplary plan, relying upon which as a design and archetypal example
the creator of the universe sets in order his material creations and makes them attain to
their proper ends; but also because it is naturally prior in birth.

For these reasons, it is said, Pythagoras started the systematic study of numbers,
and Nicomachus wrote the important textbook Introduction to Arithmetic, which was
used throughout the Middle Ages. These Pythagorean ideas about the place of
mathematical entities in the world are similar to those of Galileo (Saggiatore S-6): “The
universe is written in mathematical language, being its characters triangles, circles, and
figures” (in Galileo, 1638, p. 29). These ideas are also similar to those of Descartes
when he says that the entire world is composed of only two substances: extension and
thought. He adds that the first must be studied by means of Geometry.

We can say that almost all the empirical sciences that use mathematics rest on
the Pythagorean belief that the world is known only by means of mathematics or, at the
very least, that it is better known using mathematics. I believe that this conception about
mathematics implies that mathematical entities, and the relationships among them, are
the general structure of the world. In the words of Bigelow (1988, p. 13), “Mathematics
is the theory of universals”. This position would hold that mathematics is possible, or it
has meaning, as the study of that which persists under changes. We can agree with this
conception or not, but it certainly presents an ontology and epistemology that give
answers to the following questions: How is mathematics possible? What is the place of
mathematics in the world?

legend around this historical figure and his school. However, there are still some scholars, such
as Kahn (2001) and Riedweg (2005), who defend the idea that there must be some truth in this
image of Pythagoras to have generated such a legend.
29

 Despite the affirmation of Wittgenstein (1967, 124) that “philosophy leaves
everything as it is", I believe that mathematics and the philosophy of mathematics have
influenced each other, just as we said above that there are mutual influences in the
relationship between mathematics and philosophy in general.

The first, and perhaps the most important, influence of philosophy on

mathematics was, according to Szabó (1967, pp. 1-2), the transformation of
mathematics into a deductive science.

Deductive mathematics is born when knowledge acquired by practice alone is no longer

accepted as true...this change was due to the impact of philosophy, and more precisely
of Eleatic dialectic, upon mathematics.

It is not certain that axiomatic mathematics came from Eleatic dialectic because,
according to Proclus (c-V A. D.), Thales and Pythagoras were the first to prove
theorems. However, pre-Greek mathematics was certainly different from Greek
mathematics, which was deductive and was born at the same time as Greek philosophy.
Maybe for these reasons, Plato affirmed in The Republic that mathematics was a
hypothetical-deductive science. Additionally, we ought not to forget the logic of
Aristotle and later that of Frege as a contribution from philosophy to mathematics. The
whole of mathematics is not axiomatic or deductive, but one part of it certainly is.

With respect to the influence of mathematics on philosophy, we can mention the

works of such important philosophers as Kant, Frege, and Wittgenstein, among others.
The mathematical and axiomatic physics of Newton puzzled Kant, who thought that
mathematics expresses universal and necessary knowledge because it refers not to the
changeable world but to fixed forms of our sensibility: space for geometry, and time for
arithmetic. To Kant, mathematics proves that a priori synthetic judgments are possible
and shows the way that other sciences, such as metaphysics, must follow. I think that it

30

is not hard to accept that mathematics is the foundation of Kantian philosophy and,
perhaps, in the foundations of almost all contemporary Analytical Philosophy, which
started with Frege.

Because the philosophy of mathematics and meta-mathematics fields has not

been well delimited, there has been confusion between the work of some philosophers
and some mathematicians. The meta-mathematicians forgot their philosophical
aspirations and began to be part of the mathematics field (see Lakatos, 1978).
Philosophy, meanwhile, had remained silent for awhile. After Gödel’s meta-
mathematical work, we can really only mention the important work of Putnam and
Quine before Benacerraf’s two famous articles of 1965 and 1973. These articles gave
new life to philosophical discussion because, I believe, they focus again on the true field
of philosophy:

They have urged that the central issue in the philosophy of mathematics is to find a way
to identify ontology for mathematics that is compatible with epistemology that does not
invoke mysterious faculties. (Kitcher, 1988, p. 397)

In other words, how is mathematics possible?

Once the goal of the philosophy of mathematics was clarified, there has been a
revival of related philosophical discussion. On one side, there has been a renaissance of
empiricism with P. Kitcher (1984) and others, the realism of Maddy (1990) and
Bigelow (1988), and the structuralism of Shapiro and Resnik. On another front, we can
find the modal mathematics of Hellman and Putnam, and the nominalism of H. Field,
among other interesting conceptions.

1.3. A solution to the diversity of approaches

To the double confusion I mentioned at the beginning of 1.1 relating to the question
“what are numbers?”, which is a particular case of the broader question “what is

31

mathematics?”, the first solution that I propose is to distinguish between the approach of
the mathematician and the approach of the philosopher. I believe that the distinction
established here lets us understand, at least partially, the limits and capacity of either a
philosophical or a mathematical view when they face the question “what is
mathematics?” The explanation of why, on certain occasions, the answers of some are
not important or satisfactory for others can be given by saying that these different views
ask different things with the same question. The mathematical view inquires into the
character and internal connections of mathematical entities. The philosophical view
inquires into the place of mathematical entities in the world; in other words, it asks how
mathematical entities are possible.

Of course, these points of view are complementary, but misunderstanding arises

when scholars believe they are talking about the same thing when they use the same
terms. At that moment, a dialogue of the deaf emerges.

To resolve the confusion relating specifically to these mathematical entities we

called natural numbers, my proposal, which I develop in the chapters 2, 3, and 4, is to
distinguish three classes of natural numbers to which people have been referring and
attempting to define with different approaches.

My approach when I propose these distinctions is philosophical, and therefore, I

am proposing an ontology for natural numbers that not only explains how
mathematicians use numbers within the mathematics but also explains how it is possible
that numeric laws do not depend on any empirical confirmation, yet when we use them
in the empirical world, they are not contradicted by this lack of empirical confirmation.
To achieve this specific point, I propose that there are two kinds of numbers:
mathematical, which we will discuss in Chapter 3 and which are governed by the laws
of the arithmetic, and pre-mathematical, which we will discuss in Chapter 2 and which
can be seen as a bridge between arithmetic numbers and things that exist in the
empirical world.

32

 To address some confusion generated by Benacerraf (1965) and (1973), I
propose that there is a third class of numbers, the meta-mathematical, which we will
discuss in Chapter 4 and which are the fruit of mathematical work when it attempts to
explain numbers mathematically. This distinction between mathematical numbers and
meta-mathematical numbers allows us to suggest an answer to the problem of
Benacerraf, as we will see at the end of the Chapter 5.

33

 34

 Chapter 2

Are there numbers outside of mathematics?

35

2.1 How do we understand the assertions of existence?

The title of the chapter 2 is both disturbing and provocative. The mere suggestion that
there are non-mathematical numbers that share the same nomenclature as arithmetic
numbers suggests an excess contrary to Occam’s razor. My proposal, however, is
realistic because I contend that not only mathematical numbers but also certain pre-
mathematical and other meta-mathematical numbers exist. All of these numbers, I
would argue, exist not only in the minds of people, but also, in a sense, in the empirical
world. To defend this idea, of course, it will be necessary to clarify how I understand
the assertion of existence, such as when I say that the number 3 exists.

2.1.1 The singular object

Apparently, according to common sense, there are only individual objects. However,
how did we arrive at the idea of a stable and separate object?

For Aristotle (Metaphysics, 981a), “experience, indeed, is knowledge of

singulars... all things in the doing, and all generations, are concerned about the
singular”. Accordingly, we perceive this chair, this tree, that wave, and so on. This way
of seeing things I call "the common sense view".

But is that wave, for example, something built with my mind or is it something
outside of my mind and all people perceive it in the same way? It should be
remembered that, after Kant, it is difficult to ignore the participation of our mind in
what we perceive. Something, of course, comes from abroad: lights and shadows that
move in the case of the wave and that our eyes receive as changes in the environment.
But how or why do we isolate one wave from the rest of the sea?
36

 In fact, today, we know that that a tree, for example, is not something separate
and independent but is part of an ecosystem in continuous motion. We also know that
the same tree has parts that make it up and that all are changing. However, we identify it
as "the same", i.e., as an object more or less stable, separated from the rest of the world
and therefore as something that is likely to be counted.

If we focus only on objects that can be sensed, to which we reportedly are first
attracted when we are born(and it is logical to think that that is the case), we find a very
confusing and changing totality of sensations. This could be something like the
following:

37

 What captures our eyes is not the only record. Ernest Mach (1925), for example,
says feelings of emotional pleasure and pain should join such sensations as colour,
sound, heat, pressure, space, and time. These are called sensations. Carnap (1961) called
experiences as that which you are conscious of from outside or inside of ourselves.
Experiences may be visual, auditory, thermal, sentimental, and emotional. Finally,
Bertrand Russell (1914) also mentions as records the results of our own introspection.

However, because we are primarily a visual culture, we will develop the

argument based only on visual perceptions. What is said regarding objects can be
extended by analogy to everything we experience in various ways.

From the drawing above, the Adding Theory (see, for example, Strawson, P.F.
1959) argues that in some way babies go isolating some of those impressions from the
rest; so the order is only as a result of the experience. Some argue that the first things
we can identify are individual things because we interact with them. However, I am
inclined to think that the first thing we can separate is still of a general nature as, for
example, the red, the green, vertical lines, crossed lines, and so on.

My position is that after we perceive a confused and changing set of images, the next
step is to separate “family resemblances”, i.e., to segregate similar sensations.
According to our drawing, we should obtain something like the following in one case:

“What is the colour grey”

38

Or something like:

“That moves to the right”

Or “That which is curved and descends from left to right”

These results can be called universals or features that are repeated again and
again, here and there; that is, they play freely in space and time.

The next step is to identify objects, i.e., to separate something out of the set of
sensations and see it as a stable individuality. If we follow this line of thought, we can
obtain individuals through the intersection of multiple universal features. Vision to
consider individual things as sets of properties begins, in my view, in Plato's Timaeus
(47-53), when he considers that,

Whenever we observe a thing perceptually changing –fire, for example- in every case
we should speak of fire, not as ‘this’, but as ‘what is of such and such a quality’, nor of
water as ‘this’, but always as ‘what is of such and such a quality’; nor must we speak of
anything else as having some permanence, among all the things we indicate by the
expressions ‘this’ or ‘that’, imagining we are pointing out some definite thing... In fact,
we must give the name ‘fire’ to that which is at all times of such and such a quality.

That way, when I say that X is an individual, this will be something that is black
and consists of crossing lines, while the individual ♥ is something that is black and

39

heart-shaped. For these simple examples, I need at least two universals to identify an
individual. Of course, in real life, we use many universals to define a particular.

Similarly for Quine, we perceive inside changing totalities certain fixed points
we call individual things. These are points because they are considered, in some respect,
to be indivisible and fixed because they differ from the rest of the perceived changes.
As Quine (1981, p. 9) says,

Our expressions on external things, our notion of things, constitute only a conceptual
apparatus serving us to predict and control activation of our sensory receptors in the
light of the previous activation of these receivers.

Taking this perspective, we can ask ourselves, for example, in what sense do we
conceive this portfolio I have front of me as a unique object? The notes that define it
are: a) the concept of portfolio, which is something like the "family air" in
Wittgenstein’s sense, and b) a set of Minkowski spaces, i.e., the fact that it was day x at
time y, and day z at time w, and so on when I recognised it as "the same" because of
various features, such as colour and size, which also form a "family air". The concept of
“portfolio” and this particular object can be viewed as having a "family resemblance"
and differ only in the degree of precision because when I designate as an individual an
instance of a concept, I am only adding to the "family resemblance" expressed in the
concept a new "family resemblance" that is represented by this instance and any other
that I would likewise designate.

Thus, according to this perspective, an individual is not very different from a

concept. An individual is something that I identify as "the same" and may observe at
different times and places. Similarly, we can observe the universal “red” in different
places and times, and we can say that it is the same red. There are only degrees of
accuracy. The act of identifying is, according to this way of thinking, to isolate and fix
several general characteristics that have been previously isolated and fixed.

40

 The Gestalt theory of perception of Max Wertheimer (1945), Wolfgang Köhler
(1947) and Kurt Koffka (1927), and in our days Pylyshyn (2007) holds, in contrast, that
children perceive and respond holistically. Only later they are able to perceive the
individual sensations inside of totalities that they initially capture vaguely.

Perhaps we can represent this initial set of sensations by the following picture:

We start by sorting out all the confused and changing totalities that we register,
that is, those images that we do not know in detail and of which we could not enumerate
their constitutive universal features. However, we do identify them as something more
41

or less stable and separated from the rest, that being the tree. Carnap (1961) called these
“experiences”, while Mach (1886) described them as more or less stable complexes of
sensations.

Lakoff and Nuñez (2000, pp. 15-23) affirm that human beings have the ability to
make fast, precise judgments about the size of small collections of objects, which is
different than counting or estimating. That ability is called "subitizing". Apparently,
four-and-a-half-month-old babies, as well as some animals, can distinguish between two
and three element collections and can "say" that two objects less one object is the same
as, or results in, an object.

In my view, we can explain these findings in the study of the cognitive science
as the correlatives of "family airs" mentioned above or as the counterpart of totalities as
suggested by the Gestaltist school.

2.1.2. The assertions of existence

In advance, I must warn that I am not taking the term “exist” as many mathematicians
understand it when they say that one thing exists mathematically if it is consistent with
the interior of mathematics or when they prove the existence of certain entities or
mathematical relations by deducing them from certain axioms or mathematical
principles.

Now, in general terms, when we say that something exists, what do we mean?
Widely differing answers have been given to this question. In philosophy, the
discussion inevitably leads to the first philosophers. Among them, let us remember the
dispute between Plato and Aristotle. For Aristotle (Metaphysics, 981a), only particular
things exist: "experience, indeed, is the knowledge of singulars”. For Plato, existence is
certainly not limited to directly perceptible entities; instead, he posited what we today
call universals. From this tension, we can identify two major streams, each with their
42

respective nuances: a) there are those who claim (e.g., the Nominalists) that there are no
universals, even in individual things, as Aristotle would have accepted (This group
affirms that there are only physical objects, linguistic expressions, and mental events, all
of which are spatially and temporally determined in one Minkowski space); b) there are
those who, with Plato or today with Bigelow (1988, p. 4), defend the existence of
universals, which "play freely with the space and time".

When one inquires into the existence of an individual or a universal, one

questions whether an entity is in the physical world or only in the minds of people. A
unicorn may be in the minds of many people, but no one that has "a robust sense of
reality", as Russell would say, would argue that the unicorn exists. What we have, at
best, is an expression or a mental event without a reference. Thus, I argue that we can
say that something that is in our minds exists only if it has an external representation.

In this understanding of the assertions regarding existence, we can paraphrase

such assertions in terms of the notion of "painting" that was suggested in Plato’s Cratilo
and taken up again in our time by Wittgenstein. It is not my intention to comply strictly
with the notion of painting used by any of these philosophers. Rather, inspired by those
authors, I employ it here to express the intuitive idea of a mental portrait. In Ávila
(1989) and (1992a), I provided some details regarding this idea, and I would like to
emphasise here, for the sake of clarity, what I understand by “mental painting”. It is a
representation in our minds of something that is in the empirical world, and thus, it has
the following characteristics: a) It is a “painting” due to the use that is made of it; b) It
can only be apprehended with the mind; c) It is conceived as an idealisation of
something that is in the world in a more complex way, and therefore, it does not fully
correspond with what it attempts to portray; d) It is considered an interpretation of the
idealised, and hence, it possess do not directly observable elements. In this sense, a
scientific theory, for example one of Newton’s, is a mental painting of certain empirical
phenomena, just as Picasso’s “Guernica” is a painting of the Spanish Civil War. Both
Newton’s theory and the “Guernica” portray empirical phenomena in a simplified
fashion, but they also provide an interpretation or explanation of the phenomenon in

43

question. Of course, a mental painting can be described by means of language, as with
Newton's theory.

If we put ourselves in the realm of language, expressions without referents in the

empirical world speak, in a sense, about opportunities or possibilities, as I believe
Putnam conceives mathematics. We should note, however, that this does not mean
accepting a “Meinongean ontology which indicates that, in addition to existing objects,
there are also objects that have no existence in the world. They include possible objects"
(Orayen, 1970, p. 130). On this point, I agree with Russell and Quine that there is only
one world and that there are not various senses of “to exist”, as Malcolm thinks (see
Herrera, 1976, pp. 86-90). To say that an expression without referents in the empirical
world speaks of possibilities does not imply that the expression gives substance to the
possibilities; it only means that such an expression is a "mental painting" without
reference. In other words, it seems to be a portrait, but actually it is not. It expresses
something that can be understood and could eventually have an empirical referent, even
if that referent is the result of creative action by man. In the end, what can we say with
certainty about that which is outside of our minds, i.e., that which has being objectively
in the world? The architectural drawing of a house, for example, is a pseudo-painting,
which later can become a real painting when the house in question is complete.

However, after Kant, it is difficult to sustain a position that holds that we do not
know the participation of the structure of our perception and the activity of our minds in
the painting that we have of what exists independently of that we know. Therefore,
Quine (1981, p. 32) says:

All objects are theoretical..., [while, even] understanding of a term does not imply a
"designatum", precedes the knowledge of whether the term has or not a "designatum".

In these terms, the so-called physical things are not more real than other things
that are not seen as such. In the end, things, processes, and qualities do more than be as

44

such in the world or even in our perception; they are mental constructs that attempt to
separate and fix some aspect of the continuous and changing totality. Individual things
are such only in the way that we have conceptualised the world. In addition, we have to
remember that this conceptualisation is a cultural product expressed in our natural
language that guides us to "see" things where we can also see processes, qualities, or
totalities. As Orayen (1989, p. 289) said, "Ontology depends on a fairly high degree of
convention". For example, unlike what happens in Western languages,

...a great number of Chinese words do duty for both nouns and verbs-so that one who
thinks in Chinese has little difficulty in seeing that objects are also events. (Watts, 1957,
p. 5)

Therefore, as Quine (1951) says following Duhem and Carnap, "our statements
about the external world face the tribunal of sense experience not individually but only
as a corporate body". Therefore, when testing in general that a linguistic or mental
entity refers to something, that entity should be displayed within a total theoretical-
practical body, such that it directly or indirectly can be seen as an idealisation and
interpretation of something empirical. This is why it is not easy to determine the
references of singular terms because at the beginning and the end of the cognitive
process, we have totalities. It is only in the middle of that process that we have fixed
and isolated entities that could be the references of singular terms.

Therefore, in accordance with what we have said above, a mental painting can
refer to: A) another "mental painting"; B) an empirical universal; or C) an empirical
individual.

More precisely, in the first case (A), we can say that mental painting X refers to
another mental painting Z if: 1) We can establish a homomorphism in the way that
Krantz, Luce, Suppes, and Tversky (1971) propose for all numeric representations.
However, they suggest that there should be a homomorphism that is represented over
the representation. I am proposing that there should be, rather, a homomorphism in

45

reverse to express the idea that one painting is an idealisation of that which has been
painted. Not every element of the item painted is represented in the painting, as Plato
emphasised in the Cratilo; 2) We can show that X interprets Z; and 3) In addition, Z
refers to a universal or an individual empirical.

In the second case (B), we can say that a mental painting Z refers to a universal
empirical S, which is perceptible in certain conceptualisations if: 1) The universal has
empirical instances s1, s2, s3, and so on, i.e., individuals that are partially portrayed by
mental paintings of the universal; 2) We can demonstrate within a successful empirical
theory that Z interprets those instances and based on that interpretation we can predict
certain behaviours of s1, s2, s3, and so on.

We finally say, in the third case (C), that a mental painting Z refers to an
individual empirical M, which is perceptible in certain conceptualisations if: 1) The
painting can establish a homomorphism from Z to M; 2) We can demonstrate within a
successful empirical theory that Z interprets M and based on that interpretation we can
predict the behaviour of M.

2.2. What have others said about the numbers?

Before discussing my proposal that there are three types or levels of natural numbers,
first let us see what has been said both for and against the existence of numbers.

2.2.1. Arguments against

Without seeking to be exhaustive, I will here present three arguments (two by Field
(1980), and another of Quine (1981) that, I believe, are among the most important and
representative.

Within the Nominalistic current, Field (1980) categorically denied the existence
of such entities as numbers. For him, as for many, there are only physical objects,
linguistic expressions, and mental events.

46

 The best argument against Field that sustains the proposition that mathematics
can be seen as true (that is, viewed as if speaking of certain entities which execute
mathematical laws) has been the given by Quine and Putnam when they emphasised
that we need to apply such entities to carry out ordinary inferences about the physical
world and do science. However, against this argument, Field holds that mathematising
an empirical theory only creates a schema (which is not necessary) with the idea of
making deduction easier than in non-mathematised theory.

I consider Field’s strategy of testing for the non-existence of numbers by

showing that the presence of numbers within scientific theories is superfluous because
theoretical predictions can be made without their assistance to be correct. We could say
numbers exist according to what we said above; that is, they only exist if they directly
or indirectly relate to something empirical. In other words, they would exist if we could
test, as per Field’s idea, that there are numeric statements that are empirically falsifiable.
They must be inside a theoretical-practical body that necessarily uses arithmetic.

To Field, an empirical theory (F) for interpreting the domain (R) adds certain
theoretical terms to explain and predict the behaviour of that domain. This
understanding matches the position taken by the Structural Conception in the
Philosophy of Science (e.g., Balzer Moulines, Sneed, 1987). When mathematising the
empirical theory in question, we get a version (F') that does not expand on the theory (F)
and, therefore, is just another way of saying the same thing (F = F').

The example provided by Field (1980, p. 22) for arithmetic begins by replacing
references to numbers (21, for example) by Nominalistic expressions of the type
“∃21x(Cx)”, where

…there are not singular terms or quantifiers for numbers or other abstract entities:
number 21 occurs not as a name, but only as part of a symbol operator.

47

 However, Field says that it is difficult to work with those Nominalistic
expressions. However, if you replace these expressions with mathematised phrases,
such as "the cardinal of the boxes set is 21", which makes explicit reference to the
number 21, you can use mathematics for deductions more easily. Field shows that we
can say the same thing by referring specifically to numbers, or by replacing references
to numbers with Nominalistic expressions, such as those mentioned above.

This strategy leads us to ask ourselves what it means that we can say "the same
thing" in two different conceptual systems: one containing arithmetic and the other not
containing arithmetic but utilising, at a minimum, some form of quantitative logic and n
X. Of course, to say that there are 21 X is not the same as saying that there is the
number 21: the second referent is not spatial and temporal, while with 21 X, we assume
that it is such.

However, to accept the conclusions that Field makes, we should still test to see
whether we accept that there is space-temporality associated with 21 X and not accept
implicitly the same ontology that we use when we accept that number 21 is the cardinal
of a set of space-time entities. To put the point in another way, we can wonder if Field’s
Nominalistic scientific theory is a numeric interpretation of reality or if F does not
contain F'. My intuition is that Field’s Nominalistic translation, indeed, does not remove
all numbers and only prevents us from speaking of them. Given that judgement, if we
can say the same thing in two different conceptual systems, we can ask, is this because
they share the same basic ontology? To change the language we use does not guarantee
the elimination of any entity that is explicitly spoken of in any of these languages. Thus,
to test for an entity, we need to do something more than just identify a way to talk about
it without explicitly mentioning it. As Maxwell has already pointed out in relation to
Ramsey’s statements and Craig’s productions,

48

 We have seen that elimination of theoretical terms, even by explicit definition, would
not necessarily eliminate reference to theoretical (unobservable) entities. We have also
seen that, even if reference to theoretical entities could be eliminated,... the reality
(existence) of the theoretical entities is not thereby militated against... I should say the
likelihood-of the existence of unobserved causes for the observed events would still
remain. (Maxwell, 1962, p. 19)

Field’s (1980, p. 13) other argument holding that mathematics does not talk
about things that exist in some way is as follows:

If mathematics together with a body N of Nominalistic assertions implied an assertion

A which wasn’t a logical consequence of N alone, then the truth of the mathematical
theory would hinge on the logically consistent body of assertions N + no A not being
true. But it would seem that it must be possible, and/or not a priori false, that such a
consistent body of assertions about concrete objects alone is true; if so,… the
mathematics couldn’t be “true in all possible worlds” and/or “a priori true”.

I believe that Field is right in saying that his conclusions imply that mathematics
must be valid in all possible worlds; that is, mathematics does not contain some
interpretation of a particular reality. Indeed, as we affirmed above, Field could maintain
his conclusion that a Nominalistic theory F contains its mathematisation F' if he proves
that F' by itself has no interpretative elements of reality R. Thus, to reject Field’s
position, we must show not only that F contains F' but also that predictions F" of reality
R cannot be obtained from F alone if we remove from it all localisable or equivalent
items to F'.

I believe that both the Field position and the opposite are yet to be tested.
However, against Field, Frechet (1958), and Lakatos (1978) have already suggested that
mathematics is falsifiable, and Hossack (1991, p. 157) recently stated that

49

 The proofs need worlds like our spatio-temporal-causal one, and would fail at less
obliging possible worlds; this implies that the mathematical truths are not necessary
since they do not hold throughout logical space.

Quine (1981, p. 17), in the Reductionist current, holds that

We are left with the ontology of pure set theory, since the numbers and their quadruples
can be modelled within it.

For Quine, as for other reductionists, there are no numbers because every number used
in every context can be replaced by other, presumably more basic, entities, namely sets.
Thus, in attempting not to multiply ontology unnecessarily, Quine (1981, p. 31) only
supports an "unwavering belief in external things,... and also, albeit with less firmness,
in atoms and electrons, and classes". He only reluctantly supports these latter entities
because they permit one to model all mathematics.

This position refers to the situation when, given two conceptual systems, one (in
this case, set theory) contains another (in this case, arithmetic). Set theory does not
explicitly speak of numbers, but if numbers are certain sets, I see no difficulty in saying
that there are numbers, just as there are sets. In fact, for Frege, one of the most
important reductionists, to say that mathematics is logical and that numbers may be
expressed in terms of extensions of concepts (or sets) does not deny the existence of
numbers. It only involves dealing with certain particular sets.

2.2.2. Arguments in favour

To be thorough, I will briefly present three very recent works that defend a realistic
stance on numbers: one of is from John Bigelow (1988), one is from Michael Resnik
(1988), and the last is from Penelope Maddy (1980, 1990, 2005).

50

 Bigelow (1988, p. 3) argues that "numbers and mathematical objects in general
are universals".

The theory of universals will be a systematic collection of statements about the relations
among universals. And that, I claim, is precisely what mathematics is… In mathematics
we study the relations among relations. (Bigelow, 1988, p. 16)

On the basis of the posteriori realism of Armstrong (1978), Bigelow claims,

contrary to the Nominalism of Occam or of Field, that universals exist as individual
objects, but those may not be identified in space or time because they play freely in
space and time. Bigelow (1981, p. 1) says: “Hence universals, too, are physical. That is
to say, the universals which exist are all real physical properties and relations among
physicals things”. He considers, of course, the primary qualities: strength, extension,
figure, movement and number, all of which, apparently, support a mathematical
treatment. The secondary qualities, such as colour, are not in the physical objects, but
that is not necessarily a problem because they are formed from the primaries.

Specifically for Bigelow, natural numbers are mutual distinctions between

objects. The number 3, for example, is the relationship of grouping three mutually
different objects. "I claim: any number n is the n-place relation of n-fold mutual
distinctness” (Bigelow, 1988, p. 52). Accordingly, the origin of the numbers is the
relationship of non-identity or, to use an Aristotelian term, the dyad.

Contrary to this statement, Berkeley (prior to Frege, 1884) argued that "the
number is entirely a mind creature,... because the same thing has a different numeric
designation according to the perspective from which is designed" (in Baum, 1973, p.
184 and Robles, 1993, p. 99). Thus, a house, for example, does not have a single
number assigned. Instead, it would have several because it has, for example, 2 doors, 4
windows, and 200 square yards.

51

 However, to Bigelow, numbers are not properties of physical objects, as claimed
by Mill (1874), nor are numbers part of the preaching about concepts, as Frege (1884)
states; they are relationships between specific objects. According to Bigelow, doors
have a relationship called 2, windows have a relationship called 4, square yards have a
relationship called 200, and the house has a relationship with itself called 1.

“Numbers seen as relationships between specific objects” is a concept supported

by Armstrong’s ideas about the existence of universals. This concept seems to me
interesting and defensible; however, I do not fully adopt that concept because it relies
on a questionable assumption. This concept assumes a widespread belief that the world
is only made of certain physical objects and their relationships. Some argue that the
only things that really exist are objects because they are more or less stable, while
others, such as Armstrong and Bigelow, claim that relationships between specific
objects exist too. However, the basic objects of any ontology are not direct data
experience; rather, they are the product of conceptualisation work, as stated above.

Therefore, even if I share with Armstrong and Bigelow the idea that there are
more objects than those defined spatiotemporally, I do not see the need to maintain the
belief that the world can only be conceptualised from physical objects and their
relationships. Therefore numbers, because they are not in space-time, are not necessarily
relations between objects, as Bigelow concludes.

Resnik (1988, pp. 400, 401), supporting the considerations of Quine, states:

There is no epistemologically significant line between mathematics and science

whereby we can be said to test empirically scientific hypotheses without testing
mathematics... there is no non-arbitrary distinction between the times when a scientist
acts qua a mathematician and when he acts qua physicist.

52

That idea contests the pretence put forward by Field that it is possible to have a fully
Nominalistic theory that does not contain, even implicitly, mathematical items.

Resnik belonged to an important school of thought, as Dedekind, Peano, and

Benacerraf, who conceived mathematics as the study of structures. In fact, about
arithmetic, Resnik adopted the definition of numbers proposed by Dedekind, whereby
they are only "places in any progression". In the words of Resnik (1988, p. 415):

To know the natural number sequence involves knowing that it contains an initial
position (zero), that every position has a unique successor, that the sequence goes on
without end, and that the principle of mathematical induction holds for its positions.

However, Resnik does not conclude there that "there are no such things as
numbers" as Benacerraf did (Benacerraf, 1965, p. 73). Resnik (1997) suggests that
mathematics studies the structures (patterns) that exist in the empirical world, as the
universals of Amstrong and Bigelow.

Penelope Maddy (1980, 1990) supports the ideas of Gödel (1944 and 1947):
there is an analogy between mathematics and physical sciences. This analogy is in
agreement with the reductionists, such as Quine, in defending the idea that mathematics
(and particularly natural numbers) can be reduced to the theory of sets. On this basis,
Maddy was given the task of displaying that the direct perception of sets provides the
foundation to sustain a causal theory of our knowledge of mathematical objects.

Set-theoretic realism is a view whose main tenets are that sets exist independently of
human thought, and that set theory is the science of these entities. (Maddy, 1980, p.
163)

To substantiate this realistic position in mathematics, Maddy (1990, p. 180) states, “all
sets have physical grounding and spatiotemporal location, and all physical objects are
sets”.

53

 To the objections of Benacerraf (1965), that numbers can be identified with
different sequences of sets and that, therefore, numbers may not be sets, Maddy (1990,
p. 179) responds “that neither sequence is the numbers, that numbers are properties of
sets”.

I intend to take here a realistic stance on mathematical objects, particularly in

relation to numbers. Additionally, I agree with Maddy on the idea that we perceive more
than physical individual objects alone, as will be discussed in 2.1.1. However, my
objection to Maddy’s conclusions is that we do not necessarily have to view the
numbers of arithmetic as sets or properties of sets, which I will discuss in Chapter 4.2.

In a recent work, Maddy (2005) distinguishes between Robust Realism, Thin

Realism and Arealism. In previous work, Maddy openly joins Robust Realism, but later
she examines the difficulties of Robust Realism and advantages of Thin Realism,
which, in his view, very closely relates to the Arealism. Thin Realism accepts the
existence of things that are non-spatiotemporal and a-causal, which resembles the
positions of Bigelow and Resnik. Maddy (2005) concludes by leaving the question
open.

2.2.3. Synthesis
According to the ideas put forward in paragraph 2.1, claims about the existence or non-
existence of numbers result in terms of mental paintings, while for some realistic
authors, such as Bigelow, Resnik and Maddy, numbers are real paintings of something.
For other non-realistic authors, such as Field and Quine, numbers are only pseudo-
paintings. For the former, numbers have a counterpart outside of mind, whereas for the
latter, numbers are mental entities but that do not relate to the real world.

According to the way of understanding the assertions of existence described in

paragraph 2.1, the distinction between universals and individuals is irrelevant. This
54

distinction obliged Field to deny the existence of the numbers because they are
universal. To the contrary, defending the existence of numbers obliged Bigelow and
Resnik to defend the existence of universals in general and Frege and Maddy to defend
the idea that numbers are individuals. However, if in the end, if it is possible to perceive
a “family resemblance” of individuals and universals, stating that numbers are
individuals or universals does not allow us to decide on their existence.

Thus, in the case of the natural numbers, we can say, for example, that
arithmetic number 3 exists if we can see it as a mental painting that directly or indirectly
portrays something in the empirical world. In fact, I will attempt to display in the
following chapters that meta-mathematical numbers are mental paintings of arithmetic
numbers, which are mental paintings of pre-mathematical numbers, and these last
numbers refer to empirical events.

2.3. Existence of pre-mathematical numbers

It is undeniable that the Nominalists and the Realists have a point. Indeed, the
Nominalists claim that numbers do not speak, at least directly, about the empirical
world. On the contrary, the widespread use of numbers in empirical science suggests
that, as the Realists argue, there is something in the world that is captured by the laws of
arithmetic.

In this state of affairs, my proposal is to suggest that arithmetical numbers not

refer directly to experience; rather, they refer to experience indirectly through certain
pre-mathematical "numbers". Specifically, I portray, from Avila (1989), arithmetical
numbers as mental paintings that refer to other mental paintings, which refer to the
empirical world. This proposal is close to the idea of García de la Sienra (1990): there is
a way prior to the mathematics of referring to the real world. De la Sienra proposed that
such way of referring to the real world is the philosophy of being outlined basically by
Aristotle, Suarez and Leibniz. I will explain my idea in this respect in this subsection.

55

 First, I will explain that pre-mathematical numbers are the result of a
reclassification of the empirical in pairs, trios, and so on which is possible only after a
primitive conceptualisation of experience. These pre-mathematical numbers, which
multiply and complicate the already questionable ontology of arithmetic numbers, allow
us to build a plausible bridge between the empirical world and pure arithmetic.

With the goal of clarifying how we reach these reclassifications, we start from
the idea that we perceive the world as a changing totality, as demonstrated in 2.1.1.
From that idea, we can imagine that the classification process starts grouping, i.e.,
isolating and settling certain parts or aspects of that changing totality. We call
universals to these parts or aspects. Then we can “see” certain individuals as the cross of
several of those universals. To clarify this idea, I suppose that we have globally
conceptualised the world so that only we "see" the following objects:

Now, once we have these individual objects, we can regroup them by means of the
following concepts:

(A) “Gray triangles”

56

 (B) “Gray squares”

(C) “Black triangles”

(D) “Black circles”

(E) “Black squares”

(F) “Gray circles”

We can see that there is a "family resemblance" between objects that fall under
"gray triangles", "gray squares", “black triangles”, and "black circles". We call this
reclassification “pair”: a pair of gray triangles, a pair of gray squares, a pair of black
triangles, and a pair of black circles. That is, the family resemblance is expressed as “be
a pair”. There is another family resemblance between the objects that falls under "black
square” and “gray circle”. We call this reclassification “trio”: a trio of red squares and a
trio of red circles fall under “trio”. This is the family resemblance “be a trio”.

Thus, I define the pre-mathematical numbers as the concepts "unit", "pair",

"trio", and so on, under which fall, respectively, units, pairs, trios, and so on.
57

 These reclassifications, which I call pre-mathematical numbers, are those that
Russell (1919, p. 14) describes by saying, "it is clear that to conceive numbers is a way
to group... we can assume all pairs in a group, to all trios in another and so on".

My proposal is that arithmetic numbers idealise these reclassifications. In these

terms, the entire scheme would be as follows:

Concept A

Concept C “Pair” 2

Concept F

“Trio” 3
Concept E

The reclassifications “pair” and "trio" are obtained as "family resemblances"

between that which falls, on the one hand, under A and C, and, on the other hand, under
E and F. Thus, "pair" can be seen as a mental painting of something that is in a world
that has been previously, specifically conceptualised. Finally, the arithmetic numbers 2
and 3 appear in this schema as paintings of "pair" and "trio", respectively.

58

 Now, to support the idea of arithmetic numbers linked to the empirical world by
mean of numbers that I call pre-mathematical, next we will discuss the following
questions: A) Can arithmetic numbers be viewed as mental paintings of the pre-
mathematical numbers? B) Can pre-mathematic numbers be seen as mental paintings
with perceptible instances in the empirical world, and in that sense, can we see
arithmetic as a double idealisation and interpretation of something empirical?

A) To prove that mathematical numbers can be seen as paintings of the

reclassifications I call pre-mathematical numbers, we need to complete the following: 1)
find a homomorphism between the former and the latter; 2) show that the former can be
seen as an idealisation and interpretation of the latter; and 3) show that pre-
mathematical numbers can be seen as conceptual paintings with instances in the
empirical world.

1- First, to display the homomorphism between arithmetic numbers and pre-

mathematical numbers suffices to test: a) that for each arithmetic number there is one
pre-mathematical, and if the former are infinite, so also will be the latter; b) that for
each relationship between arithmetic numbers there is a similar relationship between
pre-mathematical numbers, i.e., that the pre-mathematical numbers meet, at least, the
Peano axioms; and c) that there is also homomorphism between each arithmetic number
and its corresponding pre-mathematical numbers, i.e., each arithmetic number can be
seen as an individual portrait of its corresponding pre-mathematical number.

Now, if n1, n2, n3,... are arithmetical numbers and r1, r2, r3,... are the
corresponding pre-mathematical reclassifications, (A-1a) is already true because given
any n natural number, I can find the corresponding reclassification, which would be the
"family resemblance" between that which falls under the concept "things that added get
n” and other concepts with instances that can be one-to-one with the instances of "things
that added get n”. Even with the "family resemblance" corresponding to infinitely large
natural numbers, it is possible to find a corresponding reclassification because the world
does not have a fixed number of empirical objects. It has as many "items" as we are able
to imagine. I understand here “imagine”, not as the creation of fictitious items, such as
59

unicorns, but rather as our ability to split our experiences in multiple ways. For
example, in the above conceptualised world, we have 14 objects: squares, triangles and
circles, black and gray. However, it is possible to have 37 objects, grasped as long, short
and curved lines, or more, if we conceptualise in other ways.

To test (A-1b), let us determine whether the successor and the operation of the
sum that governs between mathematical numbers also operate similarly between pre-
mathematical numbers.

The "sum" of pre-mathematical numbers can be seen as the union of two

reclassifications that overtake different groups of objects that are similar in some ways.
Only we can "add" different groups of entities that are similar in a sense, e.g., apples
with apples or pears with pears. My proposal is to suggest that these expressions
actually "add" pre-mathematical numbers, to which the arithmetic sum is a necessarily
idealised portrait.

Schematically, we can see the sum of pre-mathematical numbers as follows:

XXxxχ IIIII
X x II X x χ III
+ =
TT // TTT OO T T T T T

pair + trio = quintet

If we look only for the “X”, we can see that this "sum" has multiple constraints: the
“Xs” that are grouped by the concept "pair" must be different from the “Xs” grouped by
the concept "trio" (in the picture, this is true because the pairs are black and the trios are
60

gray); each element of the couple and of the trio is a different object, with the exception
of be “X”; to join the two groups, we obtain another group consisting of five different
“Xs”; there must be “Xs” in both groups to carry out the "sum"; in this regard, there
may be pairs without a corresponding trio, as in the case of (//). From this simple
example, we can see that, indeed, an operation analogous to (though with certain
restrictions) arithmetic addition can occur between the so-called pre-mathematical
numbers. With pre-mathematical numbers, we form a new "family resemblance" from
the two "family resemblances" given. It is not difficult to show that this “sum” is both
commutative and associative.

However, to explain the relationship "z is w successor", we must show that some
Peano Axioms are met for all rx entities: namely, that there is a reclassification, say r1
(axiom 1), that is not successor of any other (axiom 8) and rx + r1 = Srx, where Srx is
also a reclassification (axiom 6), and Srx is rx successor.

There is, indeed, r1, which is the "family resemblance" between all the concepts
that may be deemed as "defined descriptions", i.e., that each of these concepts are
satisfied by one object only. However, r1 is not rx successor for any n if we do not
accept r0 as a true reclassification. However, as in the arithmetics, we could accept r0,
extending the notion of pre-mathematical numbers. Finally, for all rx exists Srx because
if rx is a "family resemblance" between certain concepts, there will be a new "family
resemblance" when objects described by those concepts add another object, which is
true if we accept the conditions of possibility of the “sum” of reclassifications, i.e., that
they are different groups of similar objects.

61

 xxxx…n+1
xxx…n
+ = + x o =
ooo…n

/
oooo…n+1

We can finally briefly show that (A-1c) is true: in other words, arithmetic
numbers are individual objects (as was emphasised by Frege and Russell) that portray
other individual objects (i.e., pre-mathematical numbers). We can see that arithmetic
numbers and pre-arithmetical numbers are individual objects because these numbers can
be defined individually. Frege did so with the meta-mathematical numbers that we will
present in Chapter 4, and here, I have attempted to show that pre-mathematical numbers
are closely linked to certain specific concepts; thus, to define each pre-mathematical
number we do not need other pre-mathematical numbers, i.e., we can define, for
example, “unit” with the concepts that encapsulate an element, “pair” with the concepts
that encapsulate two elements, and so on.

2- Second, we will determine how arithmetic numbers can be used to idealise

and to interpret pre-mathematical numbers. The idea that arithmetic numbers idealise
pre-mathematical numbers may be recalled because the arithmetic sum occurs without
restriction, while the "sum" of reclassifications is only possible under the restrictions
that we have noted above.

We can also see arithmetic as an interpretation that allows explanation, at least

in part, of the general behaviour of pre-mathematical numbers. We can see that taking
into account the so-called "mathematical induction" (Peano axiom 9) is exactly the
numbers law that allows certain prediction in the numbers behaviour. For example,
using mathematical induction, we know that for all arithmetic numbers, the sum of two
62

odd numbers or two even numbers produces an even number and that the sum of an odd
number and an even number always produces an odd number. Within the pre-
mathematical numbers, we define as "odd number" (rO) all reclassifications of r1 or the
successor of an "even number" (rE), where rE is defined as the successor of an "odd
number". In effect, that is true for all rO and rE, that rO + rE = rO. This equation is true
because, in general, pairs and trios, which are the entities that fall under the
reclassifications r1, r2, r3, behave as mathematical numbers; that is, a couple of horses,
together with a trio of different horses, produce a quintet of horses. The same is true for
any other pairs and trios.

However, with the “sum” of empirical objects this does not have to always be
the case. It is perfectly conceivable that, for example, the empirical “sum” of the weight
of two bodies be unequal to the arithmetic addition of the respective numbers, perhaps
due to where energy is produced in the union of bodies. In fact, similar events occur in
the subatomic world, for example volumes of water and alcohol, or other quantities,
such as the temperature of two bodies, join together.

On this basis, we can say that arithmetic numbers laws idealise and interpret the
general behaviour of, for example, pairs and trios. Arithmetic is not valid in all possible
worlds, as Field supposed, and we cannot use arithmetic with all pairs and trios in our
world, even though it seems irrefutable by its indirect reference to the empirical world.

Thus, we conclude that arithmetic numbers can be seen as mental images of pre-
mathematical numbers.

3- Finally, to prove that the arithmetic numbers can be seen as real paintings we
subtract show that pre-matemáticos numbers can be seen at the same time as conceptual
paintings with instances in the empirical world. This brings us to the second case (B).

63

 B) To test that a mental painting (pre-mathematical number) refers to a universal
empirical, we need to determine the following: 1) if there are empirical instances of this
universal; 2) if we can show in the interior of a successful empirical theory that mental
painting interprets and explains the behaviour of these empirical instances.

1- Pre-mathematical numbers have empirical instances from the time these

numbers are reclassifications of empirical objects.

2- We subtract show, within a total theoretical and practical body containing

arithmetic numbers, which in effect pre-matemáticos numbers may be as real paintings
of something empirical, and therefore, as the link of arithmetic numbers with the
experience. Thus, we will have to do it seeing if both are essential for a successful
theory, such as the scientific theory of the movement due to Galileo, for example, get
some empirical predictions.

Galileo began by defining uniform movement and uniformly accelerated

movement, which occurs during the fall of bodies on Earth, in the following manner:

By uniform or equal movement I understand that that spaces travelled by a mobile in

times equals, whatever they are, compare equal to each other. (Galileo, 1638, p. 191)
We call movement uniformly accelerated to him that in equal times acquires equal
increments of velocity. (Galileo, 1638, p. 205)

From those definitions, Galileo mathematically figured the corollary that a body in
freefall during times 1, 2, 3, 4, and so on runs spaces that are as the odd numbers 1, 3, 5,
7, and so on. This result, verifiable in experience, astonished Galileo and convinced him
that the laws of nature work mathematically.

Now, how are the numbers 1, 2, 3, and so on used here? In other words, do they
refer to the mental paintings r1, r2, r3, and so on? Also, in turn, does each of these
reclassifications respectively cluster empirical units, pairs, and trios? And, finally, can

64

we bypass them without mutilating the image of the world that we form with this
theory?

Galileo mentions durations and distances and assign numbers to them. However,
what is the meaning of the expression “2 seconds” or “3 centimetres”? The duration is
the perception of relative changes. There are, however, some changes longer than
others, such as the rotation of the Earth around the Sun or the rotation of the Moon
around the Earth. In the case of this experiment, we have a comparative concept that
distinguishes between what is time and what is not while also distinguishing between
times. The next step in the accuracy of the concept, as Carnap (1961) states, is assigning
numbers to different times so that we can say that, for example, the time tn is 2 times the
time tm, etc. When we reach this point, we are able to apply arithmetic laws to the
domain in question. Galileo’s references could be as follows:

Motion of bodies Concepts of first Reclassifications Arithmetical

intention numbers

Movement Spaces travelled

↓
X 1 X2

↵ in seconds
Time “Pair” 2

Y 1 Y2

Spaces travelled

Z1 Z 2
↑
Z3
“Trio” 3

⇐ Time in seconds

S 1 S 2
S 3

65

 We can see in this schema that the conceptual painting of "pair" is instantiated
by a couple of seconds or a couple of centimetres, being that these instances are
perceived by the senses. The same is true for "trio", which is instantiated by different
trios perceptible to the senses. The same would occur with other many reclassifications
of this type.

It is precisely this double idealisation and interpretation that allowed Galileo to

mathematically predict the behaviour of these pairs and specific trios to say that bodies
fall to the Earth with a particular acceleration.

About an object of study, we can have a purely qualitative explanation as that

supplied by Aristotle, or we can quantify the domain and use arithmetic in our
predictions. Thus, we can say that without the quantification of the perceived pairs and
trios expressed by means of Galileo’s definitions, it would not have been possible to
predict specific acceleration of the bodies in free fall. Because, indeed, how would we
find the speed of a body in free fall on Earth without knowing the laws of natural
numbers?

We can see this idea by mean of the following schema:

Aristotle’s study
Distances between of motion Quantification of the
objects and times movement s with the
definitions of Galileo

Common predictions, such Metaphysical theory

as heavy bodies fall to that the bodies return
Earth and gases rise to to its origin
heaven

Precise predictions, such as: "bodies in +

freefall run in times 1, 2, 3, 4… distances 1, 3,
5, 7, 9...." Laws of arithmetic

66

 In the words of Galileo (1638, p. 190):

There is nothing in the nature oldest than movement and there is no shortage of
voluminous books on this subject... In spite of this, many of their properties, very
worthy of unknown, have not been observed or demonstrated so far... Indeed, as far as I
know, no one has shown that a mobile that falls on the basis of rest, iterates in equal
times spaces that keep together the same proportion between successive odd numbers
starting with the unit.

To summarise this chapter: if existence claims are claims of empirical reference;

if empirical, in turn, is conceived as perceptible only through different
conceptualisations; if we accept, with Nominalists, that arithmetic numbers do not
directly refer to anything empirical, but we also accept that arithmetic allows us to
predict some empirical behaviour; then we can accept the existence of certain pre-
mathematical numbers as bridges between arithmetic numbers and certain empirical
objects or phenomena.

67

 68

 Chapter 3

The mathematical numbers

69

3.1. Mental operation for counting
Regardless of the possible source of numbers or the beginning of the process for
counting objects, one might wonder, how do we count today? In other words, what
mental operations are necessary to count? Perhaps analysing these operations will shed
some light on how it was possible that the process for counting arose and how the same
numbers arose.

We now move from the original chaotic state to the formation of more or less
stable individual objects, and we obtain certain objects, such as the following: © ∏ ♦
♠ ♥ ϕ Ξ ♥. When we have obtained these objects, we need to do the following:

a) We group some of these objects in a set; in other words, we take one or more of
the characteristics of what has been identified to differentiate objects that
have these characteristics from objects that do not. Thus, we have a group
of objects that share one or more features. Objects that are black, for
example, will be in a group, or there will be a group of gray, or a group of
objects with the same figure. Thus, we should obtain the following groups,
among others that are possible:

{© ∏ ♥ ϕ Ξ }

{♥♦ ♠ }

{♥ ♥}

b) Then, we take the first object of the first group, i.e., ©.

c) We then deprive © of all its attributes except for the fact that it is (or we see it
as) a stable singular object. Thus, we are left with a sort of ghost without
any attributes, which we denote as . This ghost, in the words of
Euclid (Elements, Book VII, def. 1), “is that with respect to which each
existing thing [and which is to be counted] is called one".

70

 d) We do the same with ∏, then with ♥, ϕ and Ξ. At the end of this process, we
should obtain the following:

{ },

where each of these ghosts is indistinguishable from the other because

they have no features. As Plato (Republic VII (526ª) said, "the unit, as
considered it is equal to any other unit without differ in any way and
without contain itself part".
e) Given that the results of this process are highly abstract, we need to adopt a
symbol for any object already stripped of its attributes, for example, “⏐”.

f) If we represent the ghosts by identical symbols, we would obtain something

similar to the following: ⏐ ⏐ ⏐ ⏐ ⏐, where the symbols, just as the
ghosts that they represent, are also identical to each other.

g) As a next step, we would do the same with the second group, thus obtaining
the following: ⏐ ⏐ ⏐.

h) Finally, we would obtain ⏐ ⏐ from the third group in the same way.

i) We can represent the grouped objects in the first, second, and third group in
the following manner:

⏐⏐⏐⏐⏐

⏐⏐⏐

⏐⏐

Here, in my opinion, the process of counting terminates. The groups of vertical

stripe correspond to the objects previously grouped into the mentioned sets. Stated
another way, the first set has ⏐⏐⏐⏐⏐ objects; the second set has ⏐⏐⏐ objects; and the
third set has ⏐⏐ objects. Therefore, the symbol ⏐ means that we should isolate and
remove all attributes to something that has been isolated; in other words, it represents a
more or less complex mental operation.

71

3.2. Primitive languages for counting
In this book, we try not to investigate the history of the numbers of arithmetic. Our inquiry
is logical. However, at this point, we pursue certain historical data about the primitive
languages with the idea of seeing a form that could conform to the process of the symbolic
representation of the results of accounts.

The first sources of mathematics in Mesopotamia (-3000 AC) consisted of a series

of "bubbles" and clay tablets, reportedly functioning as recorded amounts of goods.
According to scholars that study the era (Ball, 1991; Bell, 1940; Cajori, 1928-29; Powell,
1989; Ritter, 1993 and 1994), numeric signs were organised in a dozen different systems.
In principle, signs only distinguished one and two. All other quantities were called
"multiple". Then, they were used different symbols to represent the same quantities of
different objects or measures. They could use, for example, γγγ… for counting trees;
♣♣♣… for counting flowers; ςςς… for counting houses; and λλλ… for counting
persons. Only later (at beginning of the second Millennium), did they unify the symbols
for counting any thing:

λλλ γγγ ςςς ♣♣♣

⎮⎮⎮

From that time, symbols were established with writing and handling that did not
depended on objects that they represented. Apparently, they went from managing diverse
objects to managing something perfectly homogeneous as points (...), lines (⎮⎮⎮), or
marks (Ú Ú Ú) on stone.

72

 It is interesting to note that in older known numbering systems (see Cajori, 1928-
29), small sums represented numbers, or at least a meeting, of the symbol used to represent
the number one. This can be seen in the following box:

Indo-Arabic 1 2 3 4 5 10

Babylon Ú ÚÚ ÚÚÚ ×

Egyptians

(Hieroglyphic) ⎮ ⎮⎮ ⎮⎮⎮ ∩

(Hieratic) ⎮ γ ψ >

Chinese, Japanese ~ ~ ~ †

Greek (penthouse) ⎮ ⎮⎮ ⎮⎮⎮ Γ Δ

Romans I II III V X

Maya . .. ... - =

Hindu - = ≡

In the Hieratic Egyptian and Chinese languages, there were slight differences
between the symbol for one and the symbols for two and three; regardless, two is
represented by two features similar to the features used to represent one, and the same goes
for the three. The Indo-Arabic symbols, which we use today, stylise Hindu characters: the
"=" shifted to 2, and "≡", written quickly, became 3. On the contrary, the box above shows
that they adopted different symbols for larger numbers, sometimes from the four, five or
ten. This adoption surely was due to the impracticality of representing, for example, the
number 100 by repeating the symbol used for the number one.

73

 Therefore, because the same sign was used to refer to three horses, three stones,
and so on, the sign already did not refer to horses or stones; rather, they referred to
something different than the numbered objects. Now, what do symbols such as Ú Ú Ú or
⎮⎮⎮ refer to? Do those symbols concern, perhaps, specific threesomes? To trios in
general? To the concept of "trio"? Or, perhaps, these marks simply scored the objects as
they were counted? The hypothesis that I will defend is that this last answer addresses the
handling of these symbols better than the other possibilities.

With respect to whether such symbols represented, perhaps, a trio in general, the
concept "trio", or simple marks, we should consider the handling of such symbols.

Regarding the first symbolisations of numbers, we can say that numerical

representation reflected in almost all cases, if not all, the fundamental operation of
summation (see Cajori, 1928-29 and Willerding, 1969). In the Maya, system for example,
‘.' symbolises the one, and '-' symbolises the 5; thus, '.' represents the 6, and '=' represents
10. In some systems, such as the Roman and Babylon, they also used subtraction. In the
Roman, e.g. IV = V - I and IX = X - I. Other systems, such as the Babylonian, also used
the multiplication: If × represented ten and Ú Ø 100, thus × Ú Ø represented ten
hundred. I stress that the passage from summation to multiplication meant a breakthrough
because in the latter, we sum numbers, reflecting the fact that the numbers are taken as
entities with their own existence, insomuch that we can count them.

Symbolisation reflects the tradition in ancient times of managing numbers as

entities that were added, multiplied or subtracted to obtain another entity of the same kind,
e.g., that by adding five to one, we obtain six. This use of numerical symbols, in my
opinion, means that they were referring to very peculiar entities that had a common basis,
e.g., that the number 3 was not very different from the number 2 because from the number
2, I can reach the number 3 by adding the number 1.

74

 This behaviour of the numbers can be explained if we conceive, or simply handle
the number 2 and number 3 as formed by repetitions of the number 1. If so, it can be seen
easily that ⎮⎮ + ⎮ =⎮⎮⎮. This idea would explain how we obtain three from two and one.
In short, primitive languages reflect the widespread sum operation and the relationship of
equality between different groups of numbers, implying a common basis.

Now, based on the above, the Ú Ú Ú or ⎮⎮⎮ symbols cannot represent specific
trios. In a trio, the three elements are different, although possibly similar, while the three
elements of each group that represents the number 3 are exactly the same: it is the same
symbol repeated three times. The oldest languages were pictographic, so if you wanted to
represent a trio, it would be more correct to do so with three different symbols (for
example: λÚ♣). There would be a different symbol for each different trio, and, as stated
above, once numeric languages unified, only one symbol was used for each number.

However, if we identify the symbol ⎮⎮⎮ with a trio in general, i.e., if that symbol
represents all trios or each possible trio, the sum operation, for example, would not be
general. To add any pair with any trio we would need, at least, for the grouped objects in
the pair to be different from those grouped in the trio; i.e., that there were no repeats of any
of the grouped objects because if this is the case, we should obtain three or four objects
instead of five. However, ⎮⎮ + ⎮⎮⎮ = ⎮⎮⎮⎮⎮ does not give any restrictions, provided
that with two and then three vertical lines, will we obtain five vertical lines.

However, the number ⎮⎮⎮, for example, cannot represent the concept that all trios
fall under. I called these concepts that are divided worldwide into pairs, trios, and so on
"first level numbers". Certainly, these reclassifications resemble arithmetic numbers, but in
spite of that resemblance, I do not think that we can identify arithmetical numbers 1, 2, 3,
and so on with the concepts “unit”, “pair”, “trio”, and so on as shown in 2.3.

75

 Finally, assuming that primitive symbols are only marks repeated many times, in
this case it should be noted that I am not talking about those in space-time: these are
always different from each other. I mean, rather, the result always equals of the action of
marking, which can be repeated as many times as desired. There is no difference between
⎮ and ⎮: in fact, they are exactly the same mark drawn twice. Therefore, both symbols
refer to the same item. This item is not the singular act of marking that is also given in
space-time but the result of an action that, as considered in the abstract, can be repeated as
many times as desired. The result of the action of marking is an individual abstract. In
other words, it is the result of an action that is not defined in space-time, but only by
characteristics that make it such object and nothing else.

Hilbert (1922) adopted a similar idea, saying that numbers are a series of trade
marks ⏐, ⏐⏐, ⏐⏐⏐,... "whose form is space and time-independent, and independent of
the circumstances in which were produced". This proposal differs from mine because to
Hilbert, numbers are marks, while I propose that such marks are only symbols. Thus,
primitive languages for counting repeat the same symbol and reference to something that
can be repeated many times. Thus, ⎮⎮⎮ refers to the outcome of the action of counting,
repeated three times. There is no difference between one brand and another, as there is no
difference across a sum of ones.

3.3. Manipulation of numerical symbols

Certain entities represented by the repetition of one symbol (⏐ or other similar) are
generated through the process of counting. When we consider these entities as isolated
objects, which can even be counted, we are generating something that can be
manipulated in itself without any other consideration. In my opinion, in this moment,
arithmetic was born.

76

 The results of the accounts represented, for example, ⏐⏐⏐⏐⏐, ⏐⏐⏐, ⏐⏐,
become autonomous entities as well as any other result that can be represented with
repetitions of ⏐ or another equivalent symbol.

Once we have obtained these symbols, we can forget about their meaning and
manipulate them in multiple ways. Let us look at this situation in some detail:

a) First, we have some groups of things

{© ∏ ♥ ϕ Ξ }

{♥ ♦ ♠ }

{♥ ♥}

b) Hence, we count and symbolise the results with a group of identical

marks:

⏐⏐⏐⏐⏐

⏐⏐⏐

⏐⏐

c) Each of these groups is considered an object in itself:

⏐ ⏐ ⏐ ⏐ ⏐, ⏐ ⏐ ⏐, ⏐ ⏐

d) We then join and compare the marks in different ways:

⏐ ⏐ + ⏐ ⏐ ⏐ = ⏐ ⏐ ⏐ ⏐ ⏐;

⏐ ⏐ ⏐ ⏐ ⏐ > ⏐ ⏐ ⏐ > ⏐ ⏐;

⏐ ⏐ ⏐ ⏐ ⏐ + ⏐ ⏐ ⏐ + ⏐ ⏐ = ⏐ ⏐ ⏐ ⏐ ⏐ ⏐ ⏐ ⏐ ⏐ ⏐;

⏐ ⏐ ⏐ ⏐ and ⏐ ⏐ are pairs because they can be divided into two equal
parts ⏐ ⏐ /⏐ ⏐ and ⏐/⏐;

77

 ⏐ ⏐ ⏐ and ⏐ ⏐ ⏐ ⏐ ⏐ are odds because they do not have two equal
parts ⏐ ⏐/ ⏐ and ⏐ ⏐ ⏐ / ⏐ ⏐.

e) Then, we discovered the laws governing such unions and the general
behaviour of these symbols that represent numbers:

⏐ ⏐ ⏐ +⏐ ⏐ =⏐ ⏐ +⏐ ⏐ ⏐ Commutative Law

(⏐ ⏐ ⏐ ⏐ ⏐ + ⏐ ⏐ ⏐) + (⏐ ⏐) = (⏐ ⏐ ⏐ ⏐ ⏐) +

(⏐ ⏐ ⏐ + ⏐ ⏐) Associative Law

We must emphasise that an important part of the systematisation of the results is

in the use of symbols. In fact, the manipulation of symbols has allowed the study of the
entities represented by them.

In short, the manipulation of symbols that were used originally to count is

nothing other than the systematisation of counting and the combining of the results of
the counts. According to that systematisation, from the process of counting objects
emerged certain peculiar entities represented by repetition of marks or symbols, which
loosened from numbered objects and acquired separate life. Now, how do we move
from these symbols to the creation of the numbers of arithmetic? My proposal is that
that step occurred when objectivity was given to these symbols or, rather, to what they
represent.

In conclusion, we can say that the process of counting objects creates certain
autonomous entities that can be handled. My proposal here is that these entities are fairly
natural numbers of arithmetic.

78

3.4. The ontology and epistemology of numbers

3.4.1. Ontology
Regardless of the historical origins of numbers, which go beyond the aims of the present
study, the question that we want to pose here is what are the numbers of arithmetic? We
know what arithmetic numbers are and what their properties are because the discipline
of arithmetic has reported that knowledge, but we do not know what these numbers are
in terms of their ontology. That is, what type of entities are they? Are they universal or
individual? Are they abstract or concrete?

My proposal, as mentioned above, is that arithmetic numbers are abstract

objects,7 such as musical notes or geometric figures. Also, similar to other abstract
objects, arithmetic numbers result from repeatable actions, where “repeatable” indicates
that the results of these actions are always considered equal.

Numbers are objects because we can distinguish them from one another. The
number 4, for example, has the feature that it is greater than 3, less than 5, and so on.
We say that they are abstract objects because we cannot identify them spatiotemporally,
but they have relationships that are well defined.

Actions that generate musical notes are physical actions that produce sounds.
For example, we can vibrate a rope that produces a "C", and as long as I vibrate the
same rope or other similar rope I get "C". Actions that produce geometric figures are
mental actions because I came, for example, to the triangle by measuring, which is a
mental activity.

7
As Hersh (1997) says, “Frege showed that mathematical objects are neither physical nor
mental. He labeled them abstract objects” (p. 13).
79

 I can reach the number 3, for example, in a process of counting and consider the
balance on the account as an object in itself. In that sense, the 3 is the result of isolating,
removing attributes, and symbolising each entity isolated. Then, I obtain something
such as this: ⏐⏐⏐.

Thus, the ⏐⏐⏐ arithmetic is an abstract individual, as the musical note "C",
which is also, incidentally, the result of a repeatable action. Therefore, ontologically, we
equate numbers to musical notes and to geometric figures and other "entities",
classifying them as abstract objects; i.e., that are not spatial-temporal but have well-
defined mutual relations. For example, the result of issuing musical note "E" shall be
treated as equal, regardless of whether emits a singer or another emits the note, in a one
moment or another. This note will have the same musical value and relations with other
musical notes, such as "C" and "A". The same idea holds true the numbers ⏐⏐⏐ and
⏐⏐: they will have the same value and same relations to each other regardless of their
origin.

As in Platonism, I think that mathematics, in general, is about abstract entities.

However, unlike other Platonist versions, I believe that these entities are not
independent of human beings but are rather a social historical product, as Hersh (1997)
suggests. To sum up, my proposal is that arithmetic, or second-level numbers, are
abstract objects produced during the process of counting. Given the abstract character
and complexity of the mental operations that give them rise, symbols that represent
numbers have been crucial for its study and manipulation. In fact, returning to the
respective origin of symbols in repeatable acts, we can see that the manipulation of
symbols formerly representing them preserves the essential characteristics of the results
of such repeatable acts. This preservation occurs because symbolising numbers with the
repetition of a single symbol represents the fact that isolating and counting always
produces the same results.

80

 Now, second level numbers may be seen as paintings of first-level numbers, as
we established in 2.3. I do not mean that arithmetic numbers arose with the intention to
portray the pre-mathematical numbers, but only that we can see the arithmetic in that
way. We clarify the relationship between the first and second level numbers using the
following scheme:

Numbers
of 1st Pair Trio Quintet
level

Empirical
objects

Mental process of counting

Numbers
of 2nd ⎮⎮ ⎮⎮⎮
level

Handling ⎮⎮ + ⎮⎮⎮ = ⎮⎮⎮⎮⎮

symbols

This schema shows that, based on certain empirical objects (or even abstract
objects), two mental processes can operate: one that reclassifies the objects, giving rise
to the first-level numbers, and another that counts them, which represents the origin of
81

the arithmetic numbers, or second-level numbers, as represented by the repetition of a
single numeric symbol. The schema shows that the manipulation of these symbols is the
source of another symbol of the same class that falls under a number of the first level.

Stated another way, the handling of arithmetic symbols allows us to establish a

relationship between pair, trio and quintet. This relationship is set in general but can be
applied to individual cases under certain restrictions as shown in 2.3. A couple and a trio
form a quintet, provided that:

a) The pair and the trio fall under a single concept

b) The elements of the pair and the trio are different from each other.

Therefore, these assumptions are implicit in the expression: "A pair of horses added to a
trio of horses produces a quintet of horses". The expression 2 + 3 = 5 is an idealised
picture of the first expression.

Therefore, according to the above, we can say that, in fact, arithmetic can be
seen as the systematisation of the mental acts of counting and the combination of the
results of the accounts. If so, arithmetic does not explain a domain as do the empirical
sciences, but arithmetic does systematise certain mental operations. Perhaps, but only
perhaps, the same passes for other formal sciences.

With the aim of better understanding the ontological aspect of arithmetic

numbers, we extend the comparison with musical notes. Thus, let us produce a scheme
similar to the one that we created for the numbers but now with musical notes:

82

Clasification Severe Acute
Symphony X

The movement
of bodies
produce
sounds

Symbolise
The sounds C A

Study of sounds

This scheme shows that the movement of bodies, which can be natural or
produced by human beings, generate certain sounds that, on the one hand, may be
classified as severe, acute, and so on; and on the other hand, can be symbolised and
studied by means of the symbolisation, generating a combination of sounds that, in turn,
will fall under any appropriate classification.

Thus, both the numbers and musical notes can be seen as ideal paintings of
empirical entities or entities close to empirical. Number 3 can be viewed as a painting of

83

the concept "trio", which, in turn, is instantiated by specific trios. The note "C" can be
seen also as a painting of a certain class of specific sounds.

3.4.2. Epistemology

From an epistemological perspective, we can now ask ourselves how it is that we obtain
knowledge of arithmetic numbers without invoking mysterious powers. According to
the ontology we proposed, we can know the properties of these numbers from the
moment we have contact with them through the mental acts that produce them, and we
can repeat those acts.

For example, we know that 2 + 3 = 3 + 2 = 5 because if certain mental act A and all
similar acts give the result ⏐⏐, mental act B and similar acts give ⏐⏐⏐, and mental act C
and similar acts give ⏐⏐⏐⏐⏐, then by making mental act B after the mental act A, we still
have the same result as if we make first the mental act B and then the mental act A.
Additionally, if we make only mental act C, then we obtain what we symbolically
represent with several identical symbols: ⏐⏐⏐⏐⏐.

This method of viewing the arithmetic natural numbers as the result of mental
repeatable acts tends to create a causal bridge between us as connoisseurs and the
abstract numbers that result from certain repeatable mental acts.

Mathematical proofs are internal to the same mathematics and do not need any
empirical confirmation. We can deduce from the Peano Axioms that, for example, "The
sum is commutative". My point here is that we accept that the arithmetic sum is
commutative because in most cases, the sum of things is commutative. This idea is not true
in a few cases, such as the temperature of two bodies together. However, because the law
applies to most of the items that we add, we adopt this law as an idealisation of the
empirical.

84

 Chapter 4

Mathematical explanations on numbers

85

In this book, I propose that the numbers described in Chapter 3 are second-level
numbers that can be viewed as paintings of the reclassifications in pairs and trios
described in Chapter 2, which I called first-level numbers. In this chapter, we will
examine the third-level numbers by means of two concrete examples: the numbers built
by Euclid and the numbers built by Frege.

4.1. Numbers seen as magnitudes

Euclid's Elements presents geometrical objects axiomatically; therefore, he begins by

defining concepts such as point, line, and plane. Book V discusses proportions or
comparisons between magnitudes. Book VI applies proportions to geometric objects.
Books VII to IX discusses numbers. In books X to XIII, Euclid resumes the discussion
of geometric objects. Questions suggested by this order include the following: Why is
arithmetic inserted between the geometric discussions, and why does arithmetic come
after the general discussion of magnitudes? There are those who, like Grattan-Guinness
(1996, p. 362), think that "the order in which the books lie is mysterious". Did Euclid
seek to base arithmetic in geometry? Are the numbers a special class of magnitudes? In
what follows, I will suggest an interpretation of Euclidean arithmetic that offers
responses to these issues.

First, the reader will be presented with the context in which Euclid’s Elements
was done with respect to deductive way of working and with respect specifically to
numbers. Second, an attempt will be made to show that Euclid handles two classes of
numbers: one of these, I call Euclidean mathematical number; the other, I call Euclidean
meta-mathematical numbers. My proposal here is that Euclidean mathematical numbers
are the arithmetic numbers described in the previous chapter.

4.1.1. Context of Elements of Euclid

According to Burkert (1972, p. 417), and Szabó (1978, p. 186), abstract mathematics
making use of proofs was an invention of the Greeks, as opposed to Babylonian
“recipes”. Nevertheless, as Neugebauer (1957, p. 146) states:

86

 To say that Greek mathematics of the Euclidean style is a strictly Greek
development does not mean to deny a general Oriental background for Greek
mathematics as a whole.

Concerning this last point, Gillings (1972, pp. 232-4) defends the notion that the
Egyptians, contrary to what was believed, often gave mathematical proofs as rigorous as
the given Greeks, even though they were presented in non-symbolic form.

However, Szabó, (1967 and 1978), and Burkert (1972, pp. 424-425) propose that
Parminides and Zenon (while investigating the ontology of being) marked the beginning
of purely rational thought, i.e., that which transcends sense perception and should be
disregarded on the grounds that one ought to follow the argument. According to these
authors, the attempt at purely logical argumentation, a systematic progression from one
thought to another, and the advancement of proofs and conclusions in the Euclidean
way are an invention of the Eliatics. From this perspective, we could say that geometry,
as styled by Thales, was reinforced by the contributions of Parmenides and Zenon.

Netz (1999, p. 273), however, affirmed that the most significant contributions go
no further than the fifth century B.C. In fact, for Netz, Greek mathematics could have
been a sudden explosion of knowledge in Plato’s lifetime, starting with Hipócrates of
Chio, who was the first mathematician to write on Euclidean subject matter.

According to Burkert (1972, p. 425), the geometry of Hippocrates of Chios clearly

connected with the logic of Eliatics. His methods for the quadrature of lunes show a
well-developed style of linear argument, demonstrating that he had a clear idea of the
deductive nature of mathematics. However, the sources to confirm this are indirect. We
know that Hippocrates of Chio through the work of Eudemus, a disciple of Aristotle. In
addition, according to Netz (2004, p. 257), Eudemus not only picked up the work of
Hippocrates but also reconstructed it “by transforming it into what he saw as the
canonical form of mathematics” with the intention to place the work of Hippocrates into
a history of unitary geometry according to the patterns of Aristotle. In addition, the text
of Eudemus came through the commentaries of Simplicius on the Physics of Aristotle.
87

However, according to Simplicius, he reproduces Eudemus “word for word, adding
only for the sake of clearness a few things taken from Euclid’s Elements”. Fowler
(1999, p. 390) and Netz (2004) have agreed with that idea.

According to Netz (1999, p. 277), the precision of the definitions and the
characteristics that the deductions must have were adopted from Aristotle, as can be
seen in Posterior Analytics. Under that influence, the theorems that appear in the
Elements of Euclid are proven one after the other, in a deductive chain. Although it is
known from Muller (1981) that the deductive chain in the Elements is far from perfect,
there are, for example, implicit elements, and some theorems are not used in subsequent
proofs.

Notwithstanding, regardless of their roots in the Eleatics or in Platonic era, the

Euclidean Elements is, without doubt, the work that culminates these ideas about
deductive work.

For this reason, the previous context is important here because we must explain
the role of the first two definitions of the Euclidean arithmetic, which seem not to have
any role in the deductive chain. We must also explain, or at least suggest, why Euclid
placed arithmetic within his geometry.

Euclidean arithmetic was believed until recently to be the work of the

Pythagoreans or Pythagoras himself. From Burkert (1972) onwards it has been
discovered that the former image of the Pythagoreans and Pythagoras was the work of
the Neo-Platonists and Neo-Pythagoreans, who built a legend around this historical
figure and his school.

However a generally accepted idea is that Pythagoreans represented numbers by

geometrical figures made with counters or pebbles, and in that way they talked about
square or triangular numbers. As stated by Netz (2002, p. 26), “Geek mathematics is

88

centred on geometry”.8 According to Aristotle (Metaphysic 1092b), the Pythagoreans
“decided what was the number of a thing by imitating its form with pebbles”. According
to this statement, some people (see Knorr, 1975, p. 170) have sustained the claim that
Pythagoreans proofs were made with pebble figures. But, the fact that most of the
Euclidean propositions on numbers, as Becker (1957) showed, can be proved with
simple pebble figures does not prove that the Pythagoreans made such proofs. In fact,
the Neo-Pythagorean Nicomachus, for example, did not use that type of proof, or any
other type, in his Introduction to Arithmetic.

Notwithstanding, the correlation between numbers and figures made with pebbles
could have inspired to Euclid to make the correlation between numbers and magnitudes,
as we show here.

Concerning the notion of numbers, in Greek science, arithmos is the counting of a

number of things. Fowler (1999, p. 13) states that “the very concrete sense of Greek
arithmoi is given by the sequence: duet, trio, quartet, quintet”. Then, “there is no unique
2 or 3, any pair of units is a 2, for example”. (Mueller, 1981, p. 59) Thus, arithmos
indicates, in each case, a definite number of definite things.

In this context, Plato makes the following distinction in the study on numbers. He
affirms in Republic, Georgias and Philebus that there is a theoretical science and a
practical science about even and odd numbers. Following his other distinction between
a pure idea and a sensible object, he says that theoretical science deals with evens and
odds formed of equal (i.e., pure) units, while practical science deals with evens and
odds formed of unequal (i.e., physical) things. In the Republic VII, Plato asks “What
numbers are those that you speak about, in which the units are equal amongst them, do
not differ in the least one from the other, and do not contain in them any part?” The
answer is the mathematical numbers because different units that might not even be
indivisible form empirical numbers. Euclid, we show below, worked with the numbers
that Plato called mathematical or pure.

8
That would also explain that in Euclid's elements, arithmetic (Book VII-IX) is inside of a
treatise of geometry.
89

 I called the concepts "pair", "trio", and so on, pre-mathematical numbers. Indeed,
neither these concepts nor empirical pairs and trios are mathematical numbers in the
Platonic sense. In paragraph 2.3, we discussed the idea of seeing the arithmetic
numbers⎮⎮, ⎮⎮⎮,… as paintings of “pair”, “trio”, and so on. In a similar way, ⎮⎮ is a
painting of each of the objects that fall under "pair". In this case, there is a single
painting for different pairs because this is an idealised painting that only collected the
“family resemblance” between a pair of horses, a pair of trees, a couple of houses, and
so on. Therefore I maintain that pure numbers of Plato can be seen as idealized
paintings of arithmoi formed of physical units.

4.1.2. The Elements of Euclid

Once we are focused exclusively on Euclidean arithmetic, I must emphasise that I will
not attempt a new historical reconstruction of that work. I will not pretend to clarify the
facts and thoughts that originated and developed Euclid’s idea on numbers. I would
rather attempt to determine how Euclidean numbers were possible. I will emphasise, as
Ian Mueller (1981, p. x) does, “philosophical, foundational, and logical questions, rather
than certain kinds of historical and mathematical questions”. However, my major
concern is with ontology, not with the deductive structure of the Elements, as Muller
does, and not with the shaping of the deduction, as Netz does (1999).

Euclid’s Elements expresses his notion of numbers, with 23 definitions and 102
propositions in Volumes VII to IX. The first two definitions are significantly different
than the rest of the definitions and propositions; thus, what is the place of these two
definitions in the deductive chain of the arithmetic?

This remarkable question has been explained in different ways. Some scholars, as
Iamblichus, have suggested that the two first definitions are not Euclid’s. In that regard,
I accept the two first definitions as part of the Euclidean work, like Szabó (1978),
Mueller (1981), Fowler (1999) and others scholars, who have studied the logical
structure of The Elements. And I focus on how that singular fact is possible or under
what suppositions it has meaning.
90

 In the two first definitions, Euclid defines unit and number as follows:

Def. 1. A unit is that with respect to which each existing thing is called one.
Def. 2. A number (arithmos) is a multitude composed of units.

Euclid never uses his definition of a unit in any of his proofs, and he uses his
definition of number only in the following cases:
a) Explicitly as a help in some definitions. For example, we mention definition
15: “A number is said to multiply a number when that which is multiplied is
added to itself as many times as there are units in the other, and thus some
number is produced”.
b) As a guiding idea in some proofs. For example, the proposition 31 of book
VII states: “Any composite number A is measured by some prime number”.
To prove this, he used the idea that a number is composed of units. Euclid
must have used this idea because he says the following: “For, if it is not
found, an infinite series of numbers will measure the number A, each of
which is less than the other: which is impossible in numbers”. He must have
thought that it is impossible to have an infinite series of numbers more and
smaller because numbers were constructed from certain elemental blocks, the
units.
c) As a bridge between numbers and magnitudes, as we will see in the current
chapter.

Euclid did not use those definitions in the rest of his arithmetic; rather, he changes his
strategy and analyses numbers by measuring one with the other.

My solution to this disparity is that the definitions of number and unit (Defs. 1
and 2, book VII) defined the entities that the rest of the arithmetical books discuss; that
is, the definitions of number and unit provide the ontology. In the rest of his arithmetic,
Euclid studied the structural relations of these entities between themselves by means of
representing these entities with the supposedly clearest analogous entities: the
91

magnitudes. This analysis explains why the first two definitions had a different status in
the advancement of proofs and conclusions.

With respect to the ontology expressed in the two first definitions, Reed (1995, p.
65) comments on the definition of unit: “A key point is the use of the phrase “is called”.
The definition of unit is verbal”. He continued, “a unit is that which results from a
particular way of regarding already defined things, namely regarding each of them as an
existing thing”. I believe, similarly to Reed, that the first definition indicated separation
or segregation of one thing from the rest, therefore making it individual.9 In this way,
the unit is a representation of one individualised object. As we described it in 3.1 and as
Plato has said, “each unit is equal to every other unit without the smallest difference and
contains within itself no parts”. (Plato, Republic, Book VII, Sect. 525d-526a).

Some modern interpreters have argued that the Platonic numbers are finite sets of
pure monads, and all equal each other. However, if “arithmos” is a count of pure
monads or of something that is the same, how can I count one and the same thing
several times? Frege (1884 to 1894) clearly saw this difficulty because he states that
monads are identical to each other, but when he counts them, they are different”.
Because of this and other problems, I think that Euclid’s first two definitions describe
arithmetic numbers not as sets of pure monads but as the result of counting various
things. The “arithmos” word means the account of a number of things, which conforms
to the definition of mathematical numbers in chapter 3 as the result of counting. What
counts are not pure monads because Euclidean unit definition refers to the empirical
things. He does this by saying that "everything that exists is called one". Euclidean
numbers are not, therefore, crowds of pure monads, whatever they are, but crowds
formed of things viewed only as existing things at the time that we count them.

9
Kitcher (1984, p. 108) adopts a similar view, taken from J.S. Mill: “arithmetic describes those
structural features of the world in virtue of which we are able to segregate and recombine
objects”.

92

 Other than two first definitions, Euclid’s definitions, problems, and theorems are
given in terms of parts and measures. Euclid’s general strategy was to measure numbers
with each other to prove there are different number types, and then to describe the
syntactic relations among the types. In fact, many definitions and propositions, or their
proofs, include the word ‘measure’ (katametre); i.e., as Mueller (1981, p. 61) suggested,
“The fundamental undefined notion in Euclid’s arithmetic is the notion of
measurement”.

But how can we measure numbers? We can measure one thing with another
when one is an aliquot part, which is the case when the larger thing is equal to n-times
the smaller thing. However, we cannot measure them if the smaller thing does not fit
exactly n-times into the larger thing. In that case, the smaller thing is a non-aliquot part
of the larger thing, and Euclid says that the smaller is ‘parts’ of the larger.
Def. 3. A number is a part of a number, the less of the greater, when it measures
the greater;
Def. 4. But parts when it does not measure it.

Aristotle supported this way of looking at numbers when he said: “Number is a

multitude measurable by one” (Metaphysics, Book X, Sect.1057a).

Euclid’s point of view, as Reed (1995) says, implies that numbers are viewed as
magnitudes, since the things that are measured are seen as magnitudes. I argue, more
specifically, that the Euclidean numbers from the third definition are magnitudes, since
only the magnitudes can be measured, and following Euclid, only the magnitudes can be
exemplified with line segments. As we know, in this part of his arithmetic Euclid
represents the entities that he calls numbers, and even what he calls “unit” (proposition
1 of book VII) by means of line segments.

At this point, these entities are really magnitudes. In fact, Euclid similarly
represents magnitudes by means of line segments in book V, which is dedicated to
magnitudes in general. Nevertheless, the magnitudes of the arithmetical books are a
93

special class of magnitudes. Arithmetic numbers, described in the two first definitions,
"are not a special class of magnitudes", says Grattan-Guinness (1996, p. 370). But the
Euclidean meta-mathematical numbers, described in the books VII to IX with the
exception of the first two definitions, are a special class of magnitudes because Euclid
arbitrarily fixed one as the smallest of the magnitudes he considered, with the condition
that it could measure the other magnitudes. For this reason, Euclid has to prove similar
theorems in books V and VII-IX.

Even though numbers that are defined by means of the first two definitions are
very similar to these magnitudes, arithmetical numbers are not magnitudes; “Numbers
and magnitudes are distinct types of quantity” (Grattan-Guinness, 1996, p. 366). The
peculiarity of Euclidean arithmetical numbers is that every number is formed by
repetitions of the unit. Therefore, we can think of the unit as a basic block for making
any arithmetical number. In this way, the unit is not a number. On the contrary, there is
no basic block for making magnitudes. As Vega (1991, p. 86) says:
They do not say in any part of the Elements that numbers of VII-IX be a kind of
commensurable magnitudes… It seems that numbers are mathematical objects
relatively autonomous and independent from magnitudes of V-VI.

I suggest that the Euclidean arithmetical numbers defined in the two first
definitions are not magnitudes. However, Euclid studied these numbers, and the unit, by
means of certain magnitudes, which I call Euclidean meta-mathematical numbers.

The question, then, is how was it possible that Euclid selected these magnitudes
(which I call third-level numbers) to explain the structural relations of the arithmetical,
or second-level numbers? The answer, I suggest, is because the arithmetical numbers
have structural similarity with these magnitudes. As stated in Grattan-Guinness (1996,
pp. 362 and 370), “structural similarity between them is evident”.

In general, the magnitudes have two main structural qualities: a) each one is
different from any other only because the first is larger or smaller than the others, and b)

94

we can add a magnitude to another magnitude of the same kind (or to itself) and get a
bigger magnitude of the same kind.

The same can be said of arithmetical numbers and units, where one is bigger or
smaller than the other. Indeed, according to the second definition, numbers are made of
units, and each number is different only because it has more or fewer units: II ≠ III ≠
IIII.

On the other hand, we can add a number or a unit to another and get a bigger
number, even by adding several small numbers: II + II + I = IIIII. All numbers are
comprised of units. Here the second definition is used as a bridge between the numbers
defined in the first and second definitions and the magnitudes handled in the rest of the
Euclidean arithmetic.

However, we cannot completely substitute numbers for these magnitudes,

though we can explain the structural relations of one type of these entities by means of
the structural relations of the other type. The structural resemblance between different
types of entities could have been behind the Pythagoreans’ idea of using numbers to
explain musical notes and geometrical figures. Euclid could have had a similar notion
when he tried to explain arithmetical numbers by means of certain magnitudes. As
Muller (1981, p. 63) says:
His arithmetical assumptions are basically generalizations of geometric ones.
This relationship is perhaps made most clear in the case of Euclid’s assumptions
about measurement in the arithmetic books.

The described magnitudes in the arithmetical books of Euclid, which I call meta-
mathematical, or third-level, numbers, can be seen as pictures of the arithmetical, or
second-level numbers. As we saw above, the Pythagoreans represented numbers by
means of pebble figures, which could have allowed them to study many properties of
numbers. The pebble figures ∴ and :: are not difficult to consider as explicative
paintings of some similarities and differences between the numbers 3 and 4. In the same
95

way, Euclid uses the line segments that represent determined magnitudes as explicative
figures of the numbers.

Netz (1999) emphasises the important role that diagrams play in Greek
mathematics. For the geometry of Euclid, Netz (1999, p. 19) shows that the diagrams
frame the reference such that, “there are assertions which are directly deduced from the
diagram”. However, as Netz (1999, p. 267) puts it, “it is clear that the relation between
diagram and object represented by the diagram is much less iconic in arithmetic than it
is in geometry”. In Euclidian arithmetic, segments of lines are treated as variables
because we do not know the concrete numbers they represent. All we can know is that
one is smaller than the other. In fact, “the only numerals found in the earliest and best
manuscripts of Euclid’s works… [are] the three numerals in Elements, Book XIII, 11”
(Fowler, 1999, 222). Consequently, we can deduce very little from the diagram.

Along this line of reasoning, Muller (1981, p.67) asserts that in proposition VII-
14, “One notes that the diagram plays no real role in this proof except possibly as a
mnemonic device for fixing the meaning of the letters”. The same could be said about
many other propositions in the arithmetic of Euclid’s, since the diagrams are
deductively superfluous. The result is that the relationship between the magnitudes and
line segments is less iconic in Euclid. I contend that managed magnitudes of books VII
to IX can be seen as paintings or diagrams of arithmetic numbers.

Finally, I would like to emphasise that Plato, in the Republic, was not in
agreement with the use of diagrams because they added ambiguity to the ontology of
mathematical entities, since it would seem that the mathematicians were speaking of the
specific diagrams they worked with. In contrast to Plato, as Netz (1999, p. 57) says
about their diagrams, “Greek mathematicians need not speak about their ontological
principles”. Greek mathematicians, according to Netz, decided to engage in
mathematical pursuits without considering ontological matters. This decision was, and
continues to be, advantageous to mathematical work.

96

 However, in this essay, I am interested in ontology. Accordingly, I assert that
ontologically, the first two definitions describe the Euclid’s object of study and the
technical work is found in Euclid’s arithmetic beginning with the third definition.
Euclid studies arithmetical numbers, and he elaborates theorems about them in a dual
process by: a) representing them with other entities that are structurally similar, and b)
studying particular cases with diagrams that work like metaphors, applicable to general
and abstract cases (See Netz, 1999, p. 54).

Thus, when Euclid uses the word “number”, sometimes he is referring to

numbers of the second level, and other times he is speaking about numbers of the third
level. When he says that numbers are composed of units, he is referring to the
mathematical, or second-level, numbers; when he says that a certain number measures
another number, he is referring to the meta-mathematical, or third-level numbers.

4.2. Numbers seen as sets

The work of other mathematicians on numbers provides an explanation found on the

Euclidean path. In fact, Frege (1950), Peano (1967), and Dedekind (1963), to name a
few, explain the arithmetical numbers in terms of extensions, classes, chains, sets, or
other mathematical entities that are isomorphic or homomorphic to numbers, just as
Euclid did with magnitudes. Therefore, while we cannot say that numbers are
extensions, classes, or sets, we can say that numbers can be seen as extensions, classes,
or sets, in the same way that Euclid regarded numbers as magnitudes.

According to Frege (1884), numbers are extensions of concepts such that each
number is the extension of one concept.10 Frege writes: “The number which applies to
the concept F is the extension of the concept ‘equinumerous with the concept F’”. For
example, the number 2 is tied to the concept that groups all pairs, and the number 3 is
tied to the concept that groups all trios, and so on.11

10
We can see this most widely in Ávila (1992b).
11
Before Frege, Mill (1974, cap. XI) argued that “Two, for instance, denotes all pairs of things”.
In our time, Bigelow (1988, p. 5) says that “The number three is instantiated by any three
97

 According to Russell, numbers are classes of classes. As Russell affirmed in
Principia, which I analysed in detail (Avila, 1992b), his definition is equivalent to the
definition provided by Frege.

In order to prove that their definitions are correct, Frege and Russell define the
number 0, the number 1 as the successor of 0, and so on. Then, they show that 0 + 1 = 1,
and they describe other mathematical characters, including that extensions or classes
defined as numbers execute Peano’s axioms. By introducing the concept of cardinality,
Frege and Russell’s discoveries enable us to analyse the extra-mathematical use of
numbers. But what is the meaning of executing Peano’s axioms and the cardinality
condition?

Peano’s axioms say the number sequence is a system of elements that stand in a
certain relation to one another. In other words, the elements have certain syntactic
relations. The cardinality condition enables us to analyse that sequence in terms of
counting. But Benacerraf (1965) shows that there could be different groups of sets that
execute Peano’s axioms and the cardinality condition, and he concludes that numbers
are nothing but places in any recursive sequence.

Notwithstanding the arguments of Benacerraf, there are philosophical reasons to

preferring the Fregean or Russellian sets as the true natural numbers. Furthermore, there
are mathematical reasons for identifying the natural numbers with the finite sets of von
Neumann.

From a philosophical point of view, to say that numbers are the extensions or
classes that group all pairs in one group, all trios in another group, and so on, links
numbers with the process of counting different things and of grouping things that have
similar results when they have been counted. This process of differentiation makes a re-
classification of the things in pairs, trios, and so on. The process is a reclassification
because concepts classify the world in horses, dogs, and so on, and once they are

distinct things”. I call the concept that groups all pairs, trios, and so on, numbers of first level
(Avila, 1989, and here 2.3).

98

grouped in a certain way, we can regroup the trios of horses or dogs in a new group, and
the pairs of horses or dogs in another group, and so on. Then, if we see numbers
associated with these re-classifications, we discover an ontology for numbers that is
perfectly clear, and it places numbers in the world of concepts. This process has an
abstract character, but it allows us to relate numbers to the empirical world. For these
reasons, the Fregean and Russellian proposal is very suggestive for philosophers, but it
has some problems that we will summarise below.

On the other hand, from a mathematical point of view, Steinhart (2002) affirms
that mathematicians ordinarily identify the natural numbers with the finite von
Neumann ordinals. This is so because the von Neumann w-series satisfies not only
Peano’s axioms and the cardinality condition as other w-series, but it is the only one
which is recursively defined, it is uniformly extendible to the transfinite, and satisfies
other mathematical conditions, which Steinhart enumerates. Notwithstanding, I think
that the von Neuman w-series have some problems too as we will see below.

In the first place, both the Fregean and von Neumann series seem very artificial
and forced. While we can sum two Euclidean arithmetical numbers so easily, when we
want to sum two Fregean numbers we must complete several complicated operations;
the same is true with the numbers of von Neumann.

With mathematical, or second-level numbers you only need to collect the result
of two accounts (both represented by the repetition of a symbol) and bring together both
results, or representations: ⎮⎮ + ⎮⎮⎮ = ⎮⎮⎮⎮⎮.

But in order to sum the class of all pairs with the class of all trios, we must
choose one element of the first class and another element of the second class, both of
which must execute certain conditions. Basically, both elements, which are really sets,
are disjoined and group elements of the same kind. Once we have selected one pair of
the first class and one trio of the second class, we do the same thing we did with the
Euclidean arithmetical numbers to obtain a new set that groups five element of the same
99

kind. Finally, we form a new class that combines all of the equinumerous sets with the
set of five elements. That class is the result of the sum.

The sum of the numbers of von Neumann is equally complicated. We would

need to complete too many steps to sum {{} {{}}} + {{} {{}} {{} {{}}}}. We cannot
simply join the elements of the set that represents the number 2 with the elements of the
set that represents the number 3.

On similar lines, in Remarks on the Foundations of Mathematics, Wittgenstein

observes that there is a great difficulty in making simple operations with Russell’s
numbers. For this reason, he thinks that Russell’s numbers fail to prove that the result of
an arithmetic operation must be a determined number. Tautological expressions, such as
(∃ a...g) (∃ a...l) ⊃ (∃ a...s), are used by Russell in Principia Mathematica to attempt to
prove g + l = s. But the Russellian tautological expression is nothing, according to
Wittgenstein, but another numeric language like the ones used in the decimal system, or
any other equivalent system that uses the repetition of bars to express different numbers,
such as (⏐⏐, ⏐⏐⏐, ⏐⏐⏐⏐⏐). Each of these languages has its own rules that lead to the
same results. They apparently speak of different entities but, according to Wittgenstein,
mathematical equations do not constitute empirical propositions. Mathematical
equations do not contain ontological presuppositions; rather, they express the rules of a
system.

As Benacerraf says, there is another reason to deny that numbers are really sets
after all: the work of Takeuti (1954), in which he shows that the Gödel-von Neumann-
Bernays set theory is reducible to the theory of ordinal numbers. In his paper, for
example, Takeuti defines ∈ in terms of <. The question is, which is really which?

The Takeuti’s work, among other examples, and the complexity of operations
with the Fregean or Russellian numbers and those of von Neumann, in spite of having
certain advantages, suggests that the arithmetical numbers are not these constructions
that have been made to explain the arithmetical numbers. We must admit the separate
existence of the arithmetic numbers on the one hand, and certain explanatory structures
on the other.
100

 My proposal is that the arithmetic numbers are those to which Euclid refers with
his two first definitions in book VII and we described in 3.4. The arithmetic numbers
have been used in different cultures, and they have been explained with different
resources, resulting in the meta-mathematics numbers. For this cause, the meta-
mathematics numbers can be viewed as paintings of the arithmetic numbers.

101

 102

 Chapter 5

Solution to the diversity of referents

103

We resolved the confusion of approaches to the question, “what are the numbers?” at
the end of chapter 1. We intended to solve the second confusion regarding references of
the numerical terms in chapters 2, 3 and 4, which we summarise below.

We can see the relations between the first, second, and third level of the
Euclidean numbers as follows:

Magnitudes:

Numbers: | || ||| ||||

Reclassifications:

unit pair trio quartet

X ZZ UUU XXXX

0 JJ 000 TTTT

The idea in this picture is that the magnitudes of Euclid, represented with line
segments, can be seen as paintings of the mathematical numbers represented by the
repetition of identical marks, which are also visible as paintings of the reclassifications
in pairs, trios, quartets, and so on.

104

 As we saw in chapter 3, arithmetic numbers are the abstract result of counting
objects. In other words, arithmetic numbers result from isolating certain objects,
removing all of their attributes, and keeping only groups of identical ghosts, each of
which represents a different object. These results are the natural numbers of the
arithmetic.

For better handling, these ghosts can be represented by identical marks, and
thus, we can more easily perform comparisons and operations between different
mathematical numbers, as well as find new mathematical numbers. We can easily see,
for example, that || + || = |||| or that ||| < |||||. Thus we can study structural relationships
of the numbers and forget what the marks represent. But when these marks are replaced
by other less explicit marks such as 7 or 9, we have to take into account that 7 = |||||||
and 9 = |||||||||.

Now, mathematicians can know that their operations are correct when they
prove the operations with the identical marks. But if we want to obtain empirical
certainty, we can return to repeatable acts that caused the mathematical numbers. For
example, if we do certain mental acts over some objects and we get |||, we repeat these
acts over others objects and we get |||||, and then we repeat a third time these acts on
other objects and we get |||, then we can compare the resulting marks and observe that
there were the same number of objects in the first and the third results.

On the other hand, reclassifications in units, pairs, trios, and so on, surely
emerged as a process arising from the original classification because they found a
family resemblance between three horses, three dogs, or three other things. These
reclassifications are linked directly with physical objects, and even with non-physical,
since they arise from a mental process that allows us to organise experiences in different
ways. Yet no matter what historical order in which numbers and reclassifications
emerged, my proposal is that we can see mathematical numbers as paintings of the
reclassifications. This is so because they have a large structural resemblance. Of course,
105

there are differences that we annotated in 2.3, but the similar structure enables us to
think that || is an idealised picture of the concept “pair”, under which fall things like ||,
xx, II, and OO.

Finally, the third level numbers, such as the magnitudes of Euclid symbolised
with line segments, can explain arithmetic numbers using other supposedly more basic
entities. However, this explanation is not the only possible explanation of the
arithmetical numbers. Indeed, we mentioned in subsection 4.2, other explanation
proposed by Frege and Russell, which we can see by mean of the following
representation:

ZZZZ

JJ JJJ
UUUU
Set of sets X X
X
II
0 TTT XXXX
VV

Numbers: | || ||| ||||

Frege’s meta-mathematical numbers are extensions of concepts that group other

equinumeric concepts. In other words, the Freagean meta-mathematical numbers are a
grouping of concepts. According to Russell, these meta-mathematical numbers are
classes of classes, or, if you like, sets of sets. In Avila (1992b) we can appreciate that all
of these definitions are equivalent. In the diagram presented above, we have chosen the
expression with sets because it is easier to represent.

106

 A full diagram would appear as follows:

Euclidean Fregean Von Neumann’s

xx Level
{{} {{}}} 3rd.
II
VV

|| 2nd.

pair

ZZ 1st.

JJ

We have said in this book that the meta-mathematical numbers have been built
from Euclid to explain some of the features of the mathematical numbers because this
last numbers became mysterious since they are abstract objects handled by means of
symbols. My proposal is that Euclid attempted to explain how mathematical numbers
can be combined to create other numbers, and that his explanation was to show that they
behave similarly to magnitudes in general. Frege, for his part, tried to explain how these

107

abstract objects are related to the things of the world. His explanation was to suggest
that numbers apply to the world by means of concepts.

In the previous diagram, you can see that mathematical numbers are unique. As
Frege said, there is only a 2. Therefore, 2 + 3 always have been ‘= 5’, even in different
cultures (see Urton, 1997, p. 7; and Verran, 2000, p. 356). Greiffenhagen and Sharrock
(2006, p. 113) say

The fact that practices of arithmetical computation are predominantly constant or

equivalent across the different examples [or cultures] is not actually denied by
the relativist.

However, different cultures assign very different ontology to the same numbers.

My solution is that mathematical, or second level numbers, have always been

the same, while the ontology attributed to them belongs to the pre-mathematical or
meta-mathematical numbers. This matches, incidentally, with Triplett (1986, p. 445)
when he said: "the arguments of the [relativists] do not belong to mathematics, but
rather to the meta-mathematics or several philosophical theories about mathematics".
For that reason, my proposal is that we need to distinguish between the mathematical,
pre-mathematical, and meta-mathematical numbers since they are often mixed in the
same speech.

On the other hand, the "pair" concept is also unique because it has a precise
meaning. Although we could imagine concepts that group all pairs, where different
meanings grouped the same objects, what interests us is the extension of the concept
rather than its intention. Therefore, these different concepts would be only different
names to group the same objects. In the same way, we can have different symbols for
arithmetical numbers that would not alter their meaning or the operations we can do

108

with them. In short, in the second level, we have only one arithmetical number and only
one set of pairs.

However, with numbers of the third level, it is possible to have multiple

representations of arithmetical numbers as shown in the picture above. One is in terms
of magnitudes, and the other two are in terms of sets of sets. One of these is the Frege-
Russell version, and the other is the version of Von Neumann. With a little imagination,
we could probably have others. This is not unreasonable since on the same tree, for
example, we could make different equally valid paintings. Certainly some of the
paintings could reveal certain aspects of the tree, while others would not, and the same
could happen with different numbers of the third level. The Frege-Russell version
includes the possibility that the second level will connect with the concepts and with the
trios and pairs that can even be physical. The same is not true of the Von Neumann
version.

The third-level Fregean numbers are certainly very similar to my first-level

numbers. The substantial difference is that my numbers are concepts, and the Fregean
numbers are extensions of concepts. However, it is easy to see that the extensions of
first level numbers are precisely the Fregean numbers. So Russell is referring to the
Fregean numbers when he says, "it is clear that to conceive numbers is a way to group...
we can assume all pairs in a group, to all trios in another and so on" (Russell, 1919, p.
14). However, Frege numbers are not a method of grouping, but the same group that
results from some form of grouping.

In summary, my proposal is that if we distinguish among three classes of

numbers, we can avoid many misunderstandings, such as those referred to in the chapter
1. This distinction multiplies the ontology of numbers that is already suspicious, but the
distinction allows us to understand several mysteries of inexhaustible life natural
numbers. The distinction allows us to understand, for example, how the numbers,
without being a product of physical science, apply successfully to the physical world.

109

The distinction also enables us to understand the emergence of efforts to explain the
mathematical numbers by means of other mathematical entities such as magnitudes and
sets.

As a corollary, we can say that this approach provides a solution to Benacerraf’s

much discussed problems regarding the nature of numbers. Benacerraf’s point amounts
to saying that meta-mathematical numbers cannot be arithmetical numbers. In other
words, the arithmetical numbers are not sets, but some meta-mathematical numbers can
be sets; that is to say, we can make a picture of the arithmetical numbers in terms of
sets.12

I defined the actual mathematical numbers as abstract objects resulting from the
mental process of counting. I defined the arithmetic work as the systematisation of the
results of the counting process. Systematisation consists of symbolising results of the
counting process and the manipulation of these symbols once they are freed of their
origin. In these terms, the arithmetic of the natural numbers does not attempt to explain
a physical domain, as does the experimental sciences, but is the systematisation of one
mental operation.

12
This idea was taken from Ávila (1989 and 1992b).
110

 Bibliography

Amor, J.A. (1981). “Prólogo” in Spanish edition of Hersh (1979) published by Facultad
de Ciencias, Universidad Nacional Autónoma de México.

Aristotle (c-IV, BC). Metaphysics. Translated by W. D. Ross. London: Encyclopedia

Britannica, Inc.

Armstrong, D.M. (1978). Universals and Scientific Realism. London: CUP.

Ávila, A. (1989). Las matemáticas y las Ciencias empíricas: ¿qué podrían ser los
números? Doctoral Dissertation, UNAM, México.

Ávila, A. (1992a) “Algunas consideraciones acerca de los números vistos como

pinturas” Analogía Filosófica VI (1): 9-28.

Ávila, A. (1992b). “La definición de número en Gottlob Frege” Crítica: revista

hispanoamericana de filosofía XXIV (72): 73-101.

Ball, W.W. (1993). A short account of the history of mathematics (Fourth edition). Dover
Publication, Inc., New York.

Balzer, W., Moulines, C.U., and Sneed, J.D. (1987). An architectonic for science. D.
Reidel Publishing Company, Holland.

Bell, E. (1940). The development of mathematics. Mc. Graw Hill, NY.

Baum, R. (ed) (1973). Philosophy and Mathematics: from Plato to the present,
Freeman, Cooper and Co., San Fancisco.

Benacerraf, P. (1965). “What numbers could not be” in Paul Benacerraf and Hilary
Putnam (eds.) Philosophy of Mathematics (1983). New York: Cambridge
University Press, pp. 272-294.

Benacerraf, P. (1973). “Mathematical Truth” in P. Benacerraf and H. Putnam (eds.)

Philosophy of Mathematics (1983). New York: Cambridge University Press, pp.
403-420.
111

Berkeley, G. (1948-57). The works of George Berkeley Bishop of Cloyne. A. Luce and T.E.
Jessop (comp.). Thomas Nelson, Londres, vols. I-IX.

Bigelow, J. (1988). The Reality of Numbers: A Physicalist’s Philosophy of Mathematics.

New York: Oxford University Press.

Buldt, B., Löwe, B. and Müller, Th. (2008). “Toward a new epistemology of
mathematics” Erkenntnis 68: 309-329.
Burkert, W. (1972). Lore and Science in Ancient Pythagoreanism. Cambridge: Harvard
University Press.
Byers, W. (2007). How mathematicians think. Princeton: University Press.
Cajori, F. (1928-29). A history of mathematical notations. Dover Publication, Inc., New
York, 1993.

Carnap, R. (1961). Der Logische Aufbau der Welt. 2nd. Edition, Hamburg: Felix Meiner.

Cavailles, J. (1938). Méthode Axiomatique et Formalisme. Paris: Hermann Editeurs des

Sciences et des Arts.

Cornman, J.W., Lehrer, K. and Pappas, G.S. (1992). Philosophical Problems and
Arguments: An introduction. Fourth edition, Cambridge: Hackett Publishing Co.

Courant, R. and Robbins, H. (1941). What is mathematics? Second edition, New York:
Oxford University Press, 1996.

Dedekind, R. (1893). “The nature and meaning of numbers” in Essays on the Theory of
Numbers (1963). Translated by W. W. Beman. New York: Dover Publications,
Inc.

Euclides (c-III, BC). The Thirteen Books of the Elements. Translated by Sir Thomas L.
Heath (1956). New York: Dover Publications.

Field H. (1980). Science without numbers. Oxford: Basil Blackwell.

Fowler, D. (1999). The Mathematics of Plato’s Academy: A new Reconstruction. New

York: Oxford University Press.
Fréchet, M. (1958) Les Mathématiques et le Concret. Paris: Presses Universitaires de
France.

112

Frege, G. (1884). Die Grundlagen der Arithmetik. Translated by Austin in The
Foundations of Arithmetic (1950). Oxford: Blackwell and Mott.

Galileo, Galilei (1638). Consideraciones y Demostraciones Matemáticas sobre dos

Nuevas Ciencias. Edición de C. Solis y J. Sadaba (1981). Madrid: Editora
Nacional.

García de la Sienra, A. (1990). “Estructuras y Representaciones”, Crítica: revista

hispanoamericana de filosofía XXII (64): 3-22.
Garciadiego, A. (1996). “Historia de las ideas matemáticas: un manual introductorio de
investigación” Mathesis XII (1): 3-113.

Carnap, R. (1928). La Construcción lógica del mundo, traduccion de Laura Mues,

México: UNAM, México.
Gillings, R.J. (1972). Mathematics in the time of the pharaohs. New York: Dover
Publications, Inc.
Gödel, K. (1944). “Russell’s Mathematical Logic” in P. Benacerraf and H. Putnam
(eds), Philosophy of Mathematics. Cambridge: Cambridge University Press,
1983, pp. 447-69.

Gödel, K. (1947). “What is Cantor’s Continuum Problem?” in P. Benacerraf and H.

Putnam (eds), Philosophy of Mathematics. Cambridge: Cambridge University
Press, 1983, pp. 470-85.

Gödel, K. (1949). “A remark about the relationship between relativity theory and
idealistic philosophy” in Albert Einstein, Philosopher-Scientist edited by Paul
Schipp, Evanston, Illinois: Library of Living Philosophers, pp. 555-562.

Greiffenhagen, Ch. and Sharrock, W. (2006). Mathematical Relativism: logic, grammar,

and arithmetic in cultural comparison. Journal for the Theory of Social
Behaviour 36 (2): 97-117.
Heijenoort, J.V. (ed.) (1967). From Frege to Gödel. Cambridge Massachusetts: Harvard
University Press.

Herrera, A. (1976). ¿Es la existencia un predicado lógico? México: IIF-UNAM.

Hersh, R. (1979). “Some proposals for reviving the philosophy of mathematics”
Advances in Mathematics 31 (1): 31-50.
113

Hersh, R. (1997). What is mathematics, really? New York: Oxford University Press.

Hilbert, D. (1900). “Acerca del concepto de número“, in C. Alvarez y L. F. Segura,

Fundamentos de la Matemática (1993), Colección Mathema, México: FC-
UNAM, pp. 17-22.

Hilbert, D. (1922). “La nueva fundamentación de las matemáticas”, en C. Alvarez y L.

F. Segura, Fundamentos de la Matemática (1993), Colección Mathema, México:
FC-UNAM, pp. 37-62.

Hossack, K. (1991). “Access to Mathematical Objets”, Crítica: revista

hispanoamericana de filosofía XXIII (68): 157-181.
Kahn, C. (2001). Pythagoras and the Pythagoreans. Indianapolis: Hackett Publishing
Company, Inc.
Kitcher, P. (1984). The Nature of Mathematical Knowledge. New York: Oxford
University Press.

Kitcher, P. (1988) “Introduction”, Revue Internationale de Philosophie 42 (167): 397-

399.

Kline, J. (1968). Greek mathematical thought and the origin of algebra. New York:
Dover.
Knorr, W.R. (1986). The Ancient Tradition of Geometric Problems. Boston: Birkhauser.

Knorr, W.R. (1975). The evolution of the Euclidean Elements. Boston: D. Reidel
Publishing Company.
Koffa, K. (1927). The growth of the mind: an introduction to child psychology. New
York: Harcourt, Brace & Company, Inc.
Köhler, W. (1947). Gestalt Psychology. New York: Leberirght.
Körner, S. (1967). “On the relevance of Post-Gödelian Mathematics to Philosophy”, in
Lakatos (1967), pp. 118-137.

Krantz D., Luce, D., Suppes, P. and Tversky, A. (1971). Foundations of Measurement.
New York: Academic Press.
Lakatos, I. (ed.) (1967). Problems in the philosophy of mathematics. Proceeding of the
International Colloquium in the Philosophy of Science. London 1965.
Amsterdam: North-Holland Publishing Company.
114

Lakatos, I. (1978). Mathematics, science and epistemology: Philosophical Papers, vol.
2, edited by J. Worrall and G. Currie. Cambridge: Cambridge University Press.

Lakoff, G. and Núñez, R.E. (2000). Where mathematics come from. USA: Basic Books.

Lemonick, M. (1993). "Fin al último teorema de Fermat" traducción de R. S. A. Chi en

Revista del Seminario de enseñanza y titulación, IX (85).

Lorenzo, J. (1992) “La matemática, ¿incompleta, aleatoria, experimental?” Theoria VII

(16-17-18): 423-450.

Lorenzo, J. (1993). “La razón constructiva matemática y sus haceres”, Mathesis IX (2):
129-153.

Mach, E. (1885). Analysis of Sensations, and the relation of the Physical to the
Psychical. Translated by C.M. Williams. Edited by Sydney Waterlow.

Maddy, P. (1980). “Perception and Mathematical Intuition” in Philosophical Review, 84

(2): 163-96.

Maddy, P. (1990). Realism in Mathematics. New York: Oxford University Press.

Maddy P. (1997). Naturalism in mathematics. Oxford University Press. Oxford.

Maddy, P. (2005). “Mathematical existence.” In The Bulletin of Symbolic Logic, 11 (3):

351-376.

Maxwell, G. (1962). “The ontological status of Theoretical Entities.” H. Feigl and G.

Maxwell (eds), Minnesota Studies in the Philosophy of Science, vol. III,
University of Minnesota Press, Minneapolis.
May, K. (1974). Bibliography and Research: Manual of the History of Mathematics.
Toronto: University of Toronto Press.

Mill, S. (1874). A System of Logic (8th ed.). New York: Harper.

Mueller, I. (1981). Philosophy of Mathematics and Deductive Structure in Euclid’s
Elements. USA: The MIT Press.
Netz, R. (1999). The shaping of deduction in Greek mathematics. Cambridge University
Press.

115

Netz, R. (2002). “Counter culture: Towards a history of Greek numeracy.” Hist. Sci.,
xl.
Netz, R. (2004). Eudemus of Rhodes, Hippocrates of Chios and the Earliest form of a
Greek Mathematical Text. Centaurus 46: 243-286.
Neugebauer, O. (1957). The exact sciences in antiquity. USA, Providence: Brown
University Press.
Nicomachus of Gerasa (c-I, AC). Introduction to Arithmetic. Translated by M. L.
D’Ooge. London: Encyclopaedia Britannica (1952).

Orayen, R. (1970). “Predicación de existencia y prueba ontológica”, Crítica: revista

hispanoamericana de filosofía, may-sep.
Orayen, R. (1989). Lógica, significado y ontología, México: UNAM.
Ortega y Gasset, J. (1973). ¿Qué es la filosofía? Madrid: Espasa Calpe.

Peano, G. (1889). “The principles of arithmetic presented by a new method” in

Heijenoort (1967), pp. 83-97.

Platon (c-IV, BC). The Dialogues of Plato. 5 vols. Edited by B. Jowett. Oxford: Oxford
University Press.
Poincaré, H. (1908). Science et Méthode. Paris: Flammarion

Pólya, G. (1957). How to solve it? Second Edition, Princeton: Princeton University
press.

Powell, M. (1989). "Measure & weiths" Reallexikon der Assyriologie 7, 457.

Proculs de Lycie (c-V, AC). Les commentaires sur le premier livre des Eléments
d’Euclide, Brujas (1948).

Putnam, H. (1967). “Mathematics without foundations”, The Journal of Philosophy

LXIV, I (19 January).

Pylyshyn, Z. (2007). Things and Places: how the mind connects with the world.
Massachusetts: MIT Press.

Quine, W.V. (1951). “Two dogmas of empiricism”, Philosophical Review 60 (1): 20-43.
Quine, W.V. (1981). Theories and Things. Harvard University Press.

116

Ramírez, S. (1989). El desconocido número 5. Doctoral Dissertation, FFyL-Universidad
Nacional Autónoma de México.

Ramírez, S. (1990). “El reencuentro del cosmos: filosofía del caos”, Mathesis, VI (4):
419-431.

Reed, D. (1995). Figures of Thought. London: Routledge.

Resnik M. D. (1988). “Mathematics from the structural point of view”, Revue
Internationale de Philosophie, 42 (167): 400-424.
Resnik, M. D. (1997). Mathematics as a Science of Patterns. Oxford: Oxford University
Press.

Riedweg, Ch. (2002). Pythagoras: his life, teaching and influence. Translated by S.
Rendall. Ithaca: Cornell University Press, 2005.

Ritter, J. (1993). “Mésopotamie: une énigme résolue?” Courrier de l'Unesco 14.

Ritter, J. (1994). Mathematiques Mésopotamiennes. Iriem Lille.

Robles, J.A. (1993). Las ideas matemáticas de George Berkeley. México: UNAM.

Rodriguez-Consuegra, F. (1992). “El convencionalismo y la analogía entre matemática

y física”, Mathesis VIII (4): 441-458.

Russell, B. (1919). Introduction to Mathematical Philosophy. London: Allen & Unwin.

Russell, B. (1914). Our Knowledge of the external world. Cornell University Library

Shapiro, S. (1997). Philosophy of Mathematics: Structure and Ontology. New York:

Oxford University Press.

Steinhart, E. (2002). Why numbers are sets? Synthese 133: 343-361.

Strawson, P.F. (1959). Individuals, London: Methuen.
Szabó, A. (1967). “Greek dialectic and Euclid’s axiomatic”, in Lakatos (1967), pp. 1-26.

Szabó, A. (1978). The beginnings of Greek mathematics. D. Reidel Publishing

Company. Hungary.

Takeuti, G. (1954). Construction of the set theory from the theory of ordinal numbers.
Journal of the Mathematical Society of Japan 6: 196-220.
117

Thom, R. (1980). Parabole e Catastrofi: intrevista su matemática, science e filosofia.
Milano: Il Sagiatore.
Triplett, T. (1986). Relativism and the sociology of mathematics: Remarks on Bloor,
Flew, and Frege. Inquire 29 (4): 439-450.
Urton, G. (1997). The Social Life on Numbers. Austin: University of Texas Press.

Vega, L. (1991). Introducción. In Euclides, Elementos. Madrid: Editorial Gredos.

Velleman, D.J. (1994). How to prove it? New York: Cambridge University Press.
Verran, H. (2000). Accounting mathematics in West Africa: some stories of Yoruba
number. In H. Selin (Ed.), Mathematics Across Cultures: The History of non-
Western Mathematics, pp. 345-371. Dordrecht: Kluwer.

Watts, A. (1957). The way of Zen. Panthom Books Inc. of Random House Inc. NY.
Willerding, M. (1967). Mathematical Concepts: A Historical Approach, Volumen Five.
Prindle, Weber & Schmidt Publishing.

Wittgenstein, L. (1918). Tractatus Logico-philosophicus. Translated by CK Ogden,

1922. London: Routledge and Kegan Paul. London.

Wittgenstein, L. (1958). Philosophical Investigations. Oxford: Basil Blackwell.

Wittgenstein, L. (1967) Remarks on the Foundations of Mathematics. 2nd. Edition.
Oxford: Basil Blackwell.

Watson, H. (1990). Investigating the social foundations of mathematics: natural number

in culturally diverse forms of life. Social Studies of Science 20 (2): 283-312.

118