Академический Документы
Профессиональный Документы
Культура Документы
Research on teaching and learning proof and proving has expanded in recent decades.
David A. Reid with Christine Knipping
This reflects the growth of mathematics education research in general, but also an
increased emphasis on proof in mathematics education. This development is a welcome
one for those interested in the topic, but also poses a challenge, especially to teachers
and new scholars. It has become more and more difficult to get an overview of the field
and to identify the key concepts used in research on proof and proving.
This book is intended to help teachers, researchers and graduate students to overcome
the difficulty of getting an overview of research on proof and proving. It reviews the
key findings and concepts in research on proof and proving, and embeds them in a
contextual frame that allows the reader to make sense of the sometimes contradictory
statements found in the literature. It also provides examples from current research
SensePublishers
SensePublishers DIVS
Proof in Mathematics Education
Proof in Mathematics Education
Research, Learning and Teaching
David A. Reid
Christine Knipping
Acadia University, Wolfville, Canada
SENSE PUBLISHERS
ROTTERDAM/BOSTON/TAIPEI
A C.I.P. record for this book is available from the Library of Congress.
No part of this work may be reproduced, stored in a retrieval system, or transmitted in any
form or by any means, electronic, mechanical, photocopying, microfilming, recording or
otherwise, without written permission from the Publisher, with the exception of any material
supplied specifically for the purpose of being entered and executed on a computer system,
for exclusive use by the purchaser of the work.
TABLE OF CONTENTS
Introduction........................................................................................................... xiii
3. Researcher Perspectives..................................................................................... 35
Philosophies of Mathematics........................................................................... 37
Research Based on an a Priorist Philosophy of Mathematics......................... 39
Research Based on an Infallibilist Philosophy of Mathematics....................... 40
Research Based on a Quasi-Empiricist Philosophy of Mathematics ............... 46
Research Based on a Social-Constructivist Philosophy of Mathematics......... 48
Summary ......................................................................................................... 52
Balacheffs Epistemologies of Proof ............................................................... 53
Diverse or Comprehensive Perspectives?........................................................ 54
4. Empirical Results.............................................................................................. 59
Key Studies...................................................................................................... 59
Many Students Accept Examples as Verification ............................................ 59
Many Students Do Not Accept Deductive Proofs as Verification ................... 62
Many Students Do Not Accept Counterexamples as Refutation ..................... 63
Students Accept Flawed Deductive Proofs as Verification.............................. 64
Students Criteria for the Acceptance of Arguments ....................................... 65
Students Offer Empirical Arguments to Verify................................................ 67
Most Students Cannot Write Correct Proofs.................................................... 68
Ideas for Research ........................................................................................... 70
v
TABLE OF CONTENTS
8. Argumentation............................................................................................... 153
Argumentation Versus Proof........................................................................ 155
Argumentation in Accord with Proof........................................................... 158
Argumentation According to Krummheuer ................................................. 161
Summary...................................................................................................... 163
Argumentation in Japan ............................................................................... 164
Ideas for Research........................................................................................ 164
vi
TABLE OF CONTENTS
Part 4: Conclusions
References............................................................................................................ 227
vii
LIST OF FIGURES
ix
LIST OF FIGURES
Figure 32: Sandys symbols for the unknowns and givens in a general
Arithmagon ........................................................................................ 196
Figure 33: Sandys puzzle.................................................................................... 199
Figure 34: Sandys reasoning, including Proof Analysis ..................................... 201
Figure 35: Scientific Verification......................................................................... 202
Figure 36: Surrender............................................................................................ 202
Figure 37: Exception and Monster Barring ......................................................... 205
x
LIST OF PROOFS
xi
LIST OF TABLES
xii
INTRODUCTION
Research on teaching and learning proof and proving has expanded in recent
decades. This reflects the growth of mathematics education research in general, but
also an increased emphasis on proof in mathematics education. This development
is a welcome one for those interested in the topic, but also poses a challenge,
especially to teachers and new scholars. It has become more and more difficult to
get an overview of the field and to identify the key concepts used in research on
proof and proving.
When we met, Christine was working on her doctoral dissertation (Knipping,
2003b). She commented on the difficulty she had making sense of the existing
research. David understood this feeling having had the same struggle when
working on his doctoral dissertation (Reid, 1995b). In the interval the amount of
research to be read and understood had increased, but the relationship between the
work of different researchers was no more apparent. Key terms were used differ-
ently by different authors, disparate theoretical assumptions were made, phenomena
were classified in incompatible ways, all without comment. We wished that
some synthesis of the literature existed that would explain the discrepancies and
make the links we found missing. And having achieved a better understanding
ourselves of the literature as a result of our efforts, we wondered if we could
attempt such a synthesis, perhaps in a journal article. In our discussions it soon
became clear that a longer piece of writing would be needed, and this book is the
outcome.
This book is intended to help teachers, researchers and students to overcome the
difficulty of getting an overview of research on proof and proving. It reviews the
key findings and concepts in research on proof and proving, and embeds them in a
contextual frame that allows the reader to make sense of the sometimes contradictory
statements found in the literature.
The first part provides this frame. It begins with an outline of the history of
proof in mathematics, both as it is usually presented and as it is interpreted by
scholars who take a wider view. Then the various uses of the words proof and
proving in everyday life, science, mathematics and mathematics education are
described and compared. Finally, the various perspectives taken by researchers in
the field are outlined and placed into a structure that allows for comparison.
The second part reviews current research. First, basic findings from empirical
research are summarised. Then important theoretical constructs and classification
systems are discussed in several chapters organised around the themes of the role
of proof, reasoning, types of proof, and argumentation. Finally, several teaching
experiments are described.
The third part focusses on two larger frameworks for examining proving and
argumentation. The first is argumentation processes which are social processes that
occur in classrooms (and elsewhere) through which knowledge changes status.
A method of describing and analysing argumentation processes is outlined which
xiii
INTRODUCTION
xiv
PART I
WHAT IS PROOF?
Reading any research literature can be a challenge at first because most authors
make assumptions about what the reader already knows about the field. This is
necessary as there is never space to include all the background underlying a
publication. In the research literature on teaching and learning proof, assumptions
are often made about the historical context of proof in mathematics, the meanings
of words like proof and proving and about the theoretical perspective of the
author, which is often assumed to be shared by the reader. In Part 1 we consider
these three sets of assumptions and provide a guide to what assumptions might be
made by authors in the field. Unfortunately, but perhaps necessarily, there is not a
single uniform set of assumptions all researchers on proof share. Hence, we will
describe a range of possibilities, without being able to state definitively what
assumptions a given publication is based on. Given an outline of the possibilities,
however, a reader should be able to pick up on the clues in a publication and
identify the assumptions being made.
Chapter 1 concerns the history of mathematics, and presents an outline of the
standard history of proof in mathematics, familiarity with which is often assumed
when proof is discussed. We also present several alternatives to key elements in the
standard history, that some authors refer to in their work.
Chapter 2 discusses the uses of the words proof and proving in mathematics,
mathematics education, logic, science and everyday life. Authors sometimes write
from more than one of these perspectives, which means that their terminology can
shift meaning from one paragraph to the next. Being aware of the possible contexts
and the meanings for proof and proving associated with them will help readers
find their way through this shifting terrain.
Chapter 3 explores the theoretical perspectives of researchers on proof in mathe-
matics education. From within a given perspective, it seems a natural way of seeing
things, and so authors often do not comment on the perspective they take. However,
for communication with the larger community some awareness of these perspectives
is necessary, and for a reader new to the field understanding that different pers-
pectives exists will aid in making sense of what sometimes seem to be contradictory
statements.
1
CHAPTER 1
HISTORY OF PROOF
When one reads a history of mathematics (e.g., Anglin, 1994; R. Jones, 1997; Kleiner,
1991; Kline, 1962), one is likely to encounter a version of the history of proof we
call the standard view. When the history of mathematics is mentioned in research
on teaching and learning proof, it is usually the standard view that is assumed, and
so it has had significant impacts on proof teaching and research. In this section we
will summarise the standard view. In the next section we will introduce some
critiques and alternatives.
3
CHAPTER 1
mathematics: the Greeks established the logical connections among their results,
deducing the theorems from a small set of starting assumptions or axioms.
(Anglin, 1994, p. 14)
A number of authors have speculated on the reason the Greeks began to insist on
proving mathematical statements. Some (e.g., Hannaford, 1998, p. 181; Kleiner,
1991, p. 293) have claimed that the democratic nature of Athenian society created a
context in which logical argument was valued. Others have noted that the existence
of a leisure class meant that there were individuals who had time for philosophical
and mathematical activity without any immediate practical application (e.g., Kline,
1962, p. 45). Kleiner (1991, p. 293) and Arsac (2007, p. 31) also mention the
problem of the incommensurability of the side and diagonal of a square, and
Kleiner adds the need to teach mathematics, as motivations for an emphasis on
proof. Hanna and Barbeau (2002) see the motivation for proving as arising from
the nature of the entities studied in mathematics:
For the early Egyptians, Babylonians, and Chinese, the weight of observational
evidence was enough to justify mathematical statements. But classical Greek
mathematicians found this way of determining mathematical truth or false-
hood less than satisfactory. They saw that, unlike other sciences, mathematics
often deals with entities that are infinite in extent or number, such as the set
of all natural numbers, or are abstractions, such as triangles or circles. When
dealing with such entities, mathematics needs to make absolute statements,
that is, statements that apply to every instance without exception. (p. 36)
Whatever the reason, the origin of mathematical proofs is credited to the Greeks,
whose innovation then spread to other cultures.
4
HISTORY OF PROOF
Proposition 1.
On a given finite straight line to construct an equilateral triangle.
Let AB be the given finite straight line. Thus it is required to construct an equilateral
triangle on the straight line AB.
And things which are equal to the same thing are also equal to one another; [C.N. 1]
therefore CA is also equal to CB. Therefore the three straight lines CA, AB, BC are equal
20 to one another.
Therefore the triangle ABC is equilateral; and it has been constructed on the given finite
straight line AB.
(Being) what it was required to do.
Both the steps of the construction and the proof are justified by references to
common notions, postulates and definitions that are stated earlier in the Elements (see
Table 1 for those referred to in Proof 1). Euclids common notions and postulates
are assumptions that are to be accepted without justification. Nowadays they would
usually both be called axioms. Euclids distinction between them is that common
notions apply outside of geometry, while his postulates are specific to geometry.
Euclids Elements is a structured presentation of the mathematics of that time.
He did not discover any of the theorems he presented, but he did present them as
part of a larger structure. The Elements provided the model for proof in mathematics,
and in other domains, for centuries.
Euclids contribution was the logical organisation of the Elements its
axiomatic structure in which everything is carefully deduced from a small
number of definitions and assumptions. This structure served as a model for
Aquinass Summa Contra Gentiles, for Newtons Principia, and for Spinozas
Ethics. The Elements has been the most influential textbook in history. (Anglin,
1994, p. 81)
5
CHAPTER 1
Common Notion 1 Things which are equal to the same thing are also equal to one
another.
Postulate 1 Let the following be postulated: To draw a straight line from
any point to any point.
Postulate 3 To describe a circle with any centre and distance.
Definition 15 A circle is a plane figure contained by one line such that all the
straight lines falling upon it from one point among those lying
within the figure are equal to one another.
6
HISTORY OF PROOF
7
CHAPTER 1
8
HISTORY OF PROOF
Degrees of formality. It is unlikely that the reader will have encountered many
proofs that meet the formalist definition of mathematical proof (an example is
Proof 27 in Chapter 7). However many proofs are called formal that do not meet
the formalist definition. This can be confusing when trying to interpret a statement
like students of mathematics should understand formal proof (Moore, 1990, p. 57).
Lakatos (1978) distinguishes between three different degrees of formality in
proofs: pre-formal, formal, and post-formal. Lakatos uses formal in the same sense
as the formalists: A sequence of symbols that makes it possible to mechanically
decide of any given alleged proof if it really was a proof or not (p. 62). By pre-
formal he means a proof which is accepted as such by mathematicians, convincing,
but not a formal proof.
[In a pre-formal proof] there are no postulates, no well-defined underlying
logic, there does not seem to be any feasible way to formalize this reasoning.
What we were doing was intuitively showing that the theorem was true. This
is a very common way of establishing mathematical facts, as mathematicians
now say. The Greeks called this process deikmyne and I shall call it thought
experiment. (pp. 6465)
A similar description has been adopted by a group of mathematics education
researchers we refer to as preformalists. We will discuss their work in Chapters 3
and 7. Note, however, that Lakatos refers specifically to informal proofs acceptable
to mathematicians, and the preformalists include proofs that could be made
acceptable. To avoid confusion we will follow the preformalists usage. We will use
semi-formal to refer to informal proofs acceptable to mathematicians, instead of
Lakatoss term pre-formal and use the word preformal (without a hyphen) to
refer to the proofs discussed by the preformalists.
Lakatoss post-formal proofs are proofs about formal proofs. For example, the
proof of the Duality Principle in Projective Geometry Although projective
geometry is a fully axiomatized system, we cannot specify the axioms and rules
used to prove the Principle of Duality, as the meta-theory involved is informal.
(p. 68). Other examples include the consistency and completeness proofs of formal
systems such as Gdels proof of his Incompleteness Theorem.
Most proofs are either preformal or semi-formal. However, the influence of the
formalists has made proofs in general more formal, and within the sub-disciplines
of mathematical foundations and computer science formal proofs are the norm.
In the early years of the twentieth century the mathematical community began to
have confidence that the formal structures they were developing would, for mathe-
matics at least, achieve what Leibniz had dreamed of in the eighteenth century, an
exhaustive collection of logical forms of reasoninga calculus ratiocinatorwhich
would permit any possible deductions from initial principles (Kline, 1980, p. 183).
The standard history of proof suggests that this has, in fact, been achieved.
It is believed that every mathematical text can be formalised. Indeed, it is
believed that every mathematical text can be formalised within a single formal
language. This language is the language of formal set theory. (Davis & Hersh,
1981, p. 136)
9
CHAPTER 1
By formalising mathematics, it was possible to revise the proofs of the past to new
standards of rigour inspired by, but improving upon, Euclids Elements.
During the period from about 1821 to 1908 ... mathematicians restored and
surpassed the standards of rigour which had been established during the
period of classical Greek mathematics. (R. Jones, 1996)
Formalist work at the foundations of mathematics inspired the Bourbaki group in
France to apply axiomatic approaches to algebra and analysis, which in turn inspired
some of the reforms of the New Math curriculum reforms of the 1960s. This brings
us to the present day. In the standard view todays proofs are direct descendants of
the proofs of Euclid, although todays proofs make more use of symbols to make
formalisation easier. Like Euclids proofs they start from axioms and lead to results
that are true within the structure defined by the axioms.
Not everyone accepts the standard view of the history of proof, and alternative
viewpoints have emerged that challenge many claims of the standard view. These
challenges are usually based on historical evidence and sociological analyses. In
this section we will discuss the challenges to several claims in standard view of the
history of proof, including these:
Proof began in Greece and was limited to cultures with an intellectual connection
to Greece, primarily those in Europe. For the early Egyptians, Babylonians,
and Chinese, the weight of observational evidence was enough to justify mathe-
matical statements (Hanna & Barbeau, 2002, p. 36). If one accepts that the
deductive method. ... has always been an essential characteristic of mathematics
(Anglin, 1994, p. 63) then one must conclude that what the early Egyptians,
Babylonians, and Chinese did was not mathematics.
In Euclids Elements everything is carefully deduced from a small number of
definitions and assumptions (Anglin, 1994, p. 81). Euclids proofs are models
of mathematical rigour.
The work of Russell, Frege, etc. re-established mathematics on firm foundations.
In principle, every mathematical proof can be reduced to a sequence of formal
statements, in which each statement follows from previous statements according
to the rules of symbolic logic.
Proofs transmit truth from established axioms to the theorems they prove. The
purpose of proofs is to make this connection from the axioms to the theorems.
We will consider each of these beliefs in turn, but first, it is important to note a
feature of the history of mathematics that makes any discussion of specific practices
problematic.
The history of mathematics is spread over a wide time period and a wide range of
cultures, and in many cases the data available is far from ideal. In the cases of Greek
and Chinese mathematics, the original sources are lost, and most of what we know
about them comes from sources written a thousand years after the originals. In the
case of the Greek texts the copies we have come from Arab sources that had their
own rich mathematical culture, which may have influenced the transcription and
10
HISTORY OF PROOF
translation of the texts they had available. We can see this process in more recent
cases; for example, in Heaths translation of Euclids Elements from Greek to English
his footnotes indicate places where he chose to translate passages into formulations
more accessible to contemporary readers, but differing from the original Greek text
(this occurs often; see, for example, the footnotes to Book I Proposition 4, Heath,
1956, p. 248). Aside from purposeful changes to the texts, there are also accidental
changes and missing sections that have forced later translators, transcribers, and
historians to interpolate material that might not be the same as in the original. With
Egyptian and Mesopotamian sources, we are a bit better off in that some original
papyri and cuneiform tablets have survived, but the interpretation of them is a
challenge for experts as the original languages fell into disuse and had to be
reconstructed.
In addition, we have no way of knowing if the mathematical texts we have
from these cultures are representative of their mathematical practices. If one plucked
a book about mathematics at random from all those printed in the twenty-first
century, the likelihood is that one would end up with a school book, as many
more school books are printed than university texts or specialists monographs.
Such a school book would hardly represent the present level of development of
mathematics.
The lack of original data is one problem facing historians of mathematics, but
the diversity of the data is another. Anyone attempting to look at the whole picture
is forced to work from secondary or even tertiary sources, as the number of
academics with strong backgrounds in both mathematics and history, and able to
read Arabic, traditional Chinese, ancient Egyptian, classical Greek, Sanskrit and
Sumerian (to name only a few of the languages in which important mathematical
texts have been written) is probably very small. As Smoryski (2008) comments:
The further removed from the primary, the less reliable the source: errors are
made and propagated in copying; editing and summarising can omit relevant
details, and replace facts by interpretations; and speculations can become
established fact even though there is no evidence supporting the fact.
(p. 11)
These considerations alone should make one cautious of accepting the standard
view of the history of proof (or any view of the history of proof), and there are also
some other reasons to be wary.
Proof in China
The standard view of the history of proof claims that proof originated in Greece,
and that while Chinese mathematics includes many significant discoveries, the
Chinese did not prove. Mikami [1913] considered the greatest deficiency in old
Chinese mathematical thought was the absence of the idea of rigorous proof
(Needham, 1959, p. 151). Since the 1960s, however, Western scholars have been
aware that there is at least one work of ancient Chinese mathematics in which
proofs play an important role:
11
CHAPTER 1
Proof 2: Figure reconstructed on the basis of Liu Huis commentary on Jiuzhang Suanshu
A right triangle has sides of 8 steps and 15 steps. What is the diameter of its inscribed circle?
12
HISTORY OF PROOF
It must be said, however, that Liu Huis proofs are not like Euclids proofs. For one
thing, Euclid expressed his proofs primarily through words, but Chinese mathe-
maticians made extensive use of diagrams. For example, consider Proof 2, which
shows a visual proof based on the text of Liu Huis commentary on problem 16 of
Chapter 9 of the Jiuzhang Suanshu. The original diagrams are, unfortunately, lost.
Siu (1993) suggests a reconstruction of the figures, which we have changed slightly
to make the argument clearer. In modern terms, the theorem proven is:
Given a right triangle with sides a, b, c, the diameter of the inscribed circle
is 2ab / (a + b + c).
Note that the solution is given in general terms (the words Gou and Gu are
used to refer to the shorter and longer legs of the triangle, instead of using the
numbers given in the problem) and that it is assumed that the reader knows how to
calculate the length of the hypotenuse (the Xian) from the lengths of the legs (i.e.,
that the reader is familiar with the Pythagorean theorem).
This practice of basing proofs on visually convincing diagrams continued when
Euclids Elements was translated into Chinese after it was introduced by Jesuit
missionaries. For example, Mei Wending (16331721) made changes to Euclids
diagrams when he incorporated parts of the Elements into his Jihe bubian (Comple-
ments of Geometry).
He modified the figures to make them immediately readable, although Euclid
operated in the opposite direction, thus making it necessary to resort to deductive
reasoning. (Martzloff, 1997, p. 113)
This preference for readable figures over verbal descriptions is one reason why
Chinese proofs are still not accepted as proofs by some historians of mathematics
(Siu, 1993, p. 345).
The use of diagrams is sometimes rejected entirely and misleading diagrams are
given to support a claim that basing proofs on diagrams is not reliable.
The prevailing attitude is that pictures are really no more than heuristic
devices; they are psychologically suggestive and pedagogically important
but they prove nothing. (Brown, 1999, p. 25)
Philosophers and mathematicians have long worried about diagrams in mathe-
matical reasoning and rightly so. They can indeed be highly misleading.
(p. 43)
One such misleading diagram is shown in Figure 1.
There are also other reasons beyond the use of diagrams for the perception of
the rarity of proofs in Chinese mathematics. Perhaps the most important is the role
of Chinese mathematical texts as textbooks in established schools.
Under the Sui dynasty (518617), and above all under the Tang dynasty
(618907), mathematics was officially taught at the guozixue (School for the
Sons of the State), based on a set of contemporary or ancient textbooks as
written support. (Martzloff, 1997, p. 15)
13
CHAPTER 1
14
HISTORY OF PROOF
Proof in India
The standard history of mathematics makes claims about proof in India that are
similar to those made about China. For example, in describing Hindu mathematics
in the period 2001200 CE Kline (1972) writes:
There is much good procedure and technical facility, but no evidence that
they considered proof at all. They had rules, but apparently no logical scruples.
(p. 190)
Joseph (1992, 1994) critiques this claim, pointing out that, as in China, proofs
(upapatti) were often included in commentaries on mathematical texts, even if they
were not a part of the texts themselves. There is, however, an important difference
between upapatti and Euclids proofs:
The upapattis of Indian mathematics are presented in a precise language,
displaying the steps of the argument and indicating the general principles which
are employed. In this sense they are no different from the proofs found in
modern mathematics. But what is peculiar to the upapattis is that while presen-
ting the argument in an informal manner (which is common in many mathe-
matical discourses anyway), they make no reference whatsoever to any fixed
set of axioms or link the given argument to formal deductions performable
with the aid of such axioms. (1992, p. 194; see also 1994, p. 200)
15
CHAPTER 1
Rigour in Euclids proofs. Do the proofs in the Elements live up to the claims
sometimes made about them, that everything is carefully deduced from a small
number of definitions and assumptions (Anglin, 1994, p. 81)? In fact, they do not.
Euclids proofs make use of assumptions that are never stated, some involve
reference to physical manipulations (as in Liu Huis proofs) and some use specific
cases to justify general conclusions.
Since Euclid still has popularity, and even with mathematicians, a reputation
for rigour in virtue of which his circumlocution and longwindedness are
condoned, it may be worth while to point out, to begin with, a few of the
errors in his first twenty-six propositions. (Russell, 1903/1937, p. 404)
Recall, for example, the first proof in Book I, the construction of an equilateral
triangle (see Proof 1, on page 5). Each step in the construction (lines 512)
indicates the postulate that states that such a construction is possible, and each step
in the proof (lines 1322) indicates which definition, postulate or common notion
justifies that step in the argument.
16
HISTORY OF PROOF
Proposition 4.
If two triangles have the two sides equal to two sides respectively, and have the angles
contained by the equal straight lines equal, they will also have the base equal to the
base, the triangle will be equal to the triangle, and the remaining angles will be equal
5 to the remaining angles respectively, namely those which the equal sides subtend.
Notice, in line 9, the mention of the point C, in which the circles cut one another.
From the diagram it is clear that there are two such points. That Euclid seems to
claim that there is only one is perhaps a minor flaw. He is only trying to prove that
it is possible to construct an equilateral triangle; that the construction might
produce two does not make it invalid.
More significantly, Euclid does not provide a common notion, postulate, or
definition to let us know when we can actually construct a point C, in which the
circles cut one another. While these circles intersect, it has not been established
17
CHAPTER 1
under what conditions the intersections will exist, or that these circles satisfy these
unstated conditions. Of course, from the diagram it is clear that C exists, but the
claim made in the Euclid myth is that Euclid reasons from a small number of
definitions and assumptions not from diagrams.
We turn now to a proof that, like Liu Huis proofs, makes use of physical
manipulations. It is Euclids proof of Proposition 4 of Book 1 (see Proof 3). One
thing that is interesting about the proof (lines 1932) is the lack of references to
common notions, postulates or definitions. In fact, the only such reference (in line 28)
is thought to be a later interpolation (Heath, 1956, p. 249). This is not surprising
when one considers that the whole argument depends on the idea of picking up one
triangle and putting it on top of the other one. The phrase if the triangle ABC be
applied to the triangle DEF (line 19) suggests that ABC be moved so that it
coincides with DEF.
This way of reasoning is not what Euclid is supposed to have done, but it is
quite similar to a way of reasoning used by Liu Hui:
Thus, the argumentation inevitably depends on methods. For example: ...
Recourse to non-linguistic means of communication. This is necessary
because, according to the adage of the Yijing cited by the commentator [Liu
Hui], not all thoughts can be adequately expressed in words ... In place of a
discourse, the reader is asked to put together jigsaw pieces, to look at a figure
or to undertake calculations which themselves constitute the sole justification
of the matter at hand. In each of these cases language is purely auxiliary to
such procedures. (Martzloff, 1997, pp. 7172)
In Euclids proof we are asked to make ABC coincide with DEF in our imaginations,
and then to note the correspondences Euclid points out. This is easy to do, and
quite convincing, but it is not the deductive method as described by Aristotle.
To be fair, Euclid did not reason in this way very often, and it is not, in fact,
possible to deduce this proposition from his common notions, postulates and defini-
tions, so he had to depart from the deductive method, or change his postulates.
Hilbert took the latter approach in his Grundlagen der Geometrie (Foundations of
Geometry, 1899/1921) and added this proposition as an axiom.
Reasoning from visual evidence was a mainstay of Chinese and Hindu mathe-
matics, but fell out of fashion in Greece. There is evidence, however, that it was the
basis for Greek mathematics as well for some time. Euclids proof of Proposition
I.4 is part of this evidence. And, as Martzloff (1997) points out:
The Greek technical term meaning to prove is the verb . Euclid
uses this at the end of each of his proofs. Originally this verb had the precise
meaning of to point out, to show or to make visible. Thus it appears
that the Chinese proofs of Liu Hui and Li Chunfeng were similar in nature to
the first known historical proofs, an example of which is given by Plato
(well-known dialogue in which Socrates asks a slave how to double the area
of a square); moreover, visual elements remained an essential component of
proofs in China for a long time, while in Greece these were abandoned at an
early stage although figurative references were maintained. (pp. 7273)
18
HISTORY OF PROOF
PROPOSITION 20.
Prime numbers are more than any assigned multitude of prime numbers.
Let A, B, C be the assigned prime
numbers; I say that there are more
5 prime numbers than A, B, C.
19
CHAPTER 1
There are two things of interest about this proof. First, the diagram is not needed,
but it is there anyway. This is true in general of Euclids proofs in what we would now
call number theory. Netz (1998) explains the presence of diagrams in these contexts.
He notes that the Greek word diagramma refers to more than a diagram.
It means something much closer to a proposition or a proof (see Knorr,
1975, pp. 6975). This is a notorious fact about Greek practice: it is generally
difficult to tell whether the authors speak about drawing a figure or proving
an assertion, and this is because the same words are used for both. And this
again is because the diagram is the proof, it is the essence of the proof for the
Greek, the metonym of the proof. (Netz, 1998, pp. 3738)
So while Greek proofs are often taken as the model of modern discursive proofs,
for the Greeks themselves they were fundamentally, essentially, associated with
pictures. This explains why diagrams were included in proofs like Proof 4 which
from a modern perspective do not need diagrams. To the Greeks, if it did not have a
diagram, it was not a proof. The presence of diagrams where they are not needed
may also be related to the origins of proofs as visual arguments, noted above.
The second thing of interest about this proof is that it does not, strictly speaking,
establish that there are an infinite number of prime numbers. What it shows is that
if there are three prime numbers, then there must be four prime numbers. This
specific case is used to stand for all cases, a technique which is also common in
Chinese proofs.
Thus, the argumentation inevitably depends on methods. For example: ...
Passage from the particular to the general, based on a specific, well-chosen
example. (Martzloff, 1997, p. 71)
This method of proving, known as using a generic example (see Chapter 7), is
unavoidable if one does not have a method of representing unspecified numbers
symbolically. Euclid had one such method, representing a number as the length of a
line segment, but he did not have a method for representing an unspecified number
of numbers as he had to do in this proof.
To summarise, Euclids proofs do not rigourously use deductive reasoning to
derive propositions from axioms (common notions, postulates), definitions, and
previously established propositions. They use implicit axioms, non-verbal arguments,
and generic examples. This undermines the claim that they establish the proposi-
tions they prove with certainty. Nonetheless, they have been the model and measure
of proofs in the Western mathematical tradition for thousands of years. In Chapter 5
we will revisit the role of proving in mathematics and explore some reasons why
the flaws in Euclids proofs were not considered serious (or even noticed) until the
beginning of the twentieth century, and why they are still being offered as model
mathematical proofs (e.g., by Hanna & Barbeau, 2002).
20
HISTORY OF PROOF
But the severity of these difficulties is sometimes minimised, and the success of the
efforts to solve them overrated. According to the standard view, even if Euclids
proofs are flawed, they were a step in the right direction, and since the work of the
formalists in the early twentieth century, mathematics has once again been placed
on firm foundations. Now, in principle, any mathematical proof can be expressed in
purely formal statements, which can then be checked mechanically, with no chance
of error due to missing assumptions, unclear definitions, use of diagrams, or logical
mistakes. But this is never done, and not only for pragmatic reasons.
An ordinary page of mathematical exposition may occasionally consist entirely
of mathematical symbols. To a casual eye, it may seem that there is little
difference between such a page of ordinary mathematical text and a text in a
formal language. But there is a crucial difference which becomes unmistakable
when one reads the text. Any steps which are purely mechanical may be
omitted from an ordinary mathematical text. It is sufficient to give the starting
point and the final result. The steps that are included in such a text are those
that are not purely mechanical that involve some constructive idea, the intro-
duction of some new element into the calculation. To read a mathematical
text with understanding, one must supply the new idea which justifies the
steps that are written down. (Davis & Hersh, 1981, p. 139)
The missing steps would first have to be supplied before the proof could be
formalised. This would have to be done by an expert in the field, and even if an
expert could be found with the patience for such a task, there would be no guarantee
that the translation of the proof into a formal language would be free of error. The
problem of checking the correctness of the proof becomes the problem of checking
the correctness of the translation into formal language, and that is not formalisable.
The actual situation is this. On the one side, we have real mathematics, with
proofs which are established by consensus of the qualified. A real proof is
not checkable by a machine, or even by any mathematician not privy to the
gestalt, the mode of thought of the particular field of mathematics in which
the proof is located. Even to the qualified reader, there are normally
differences of opinion as to whether a real proof (i.e., one that is actually
spoken or written down) is complete and correct. These doubts are resolved
by communication and explanation, never by transcribing the proof into first-
order predicate calculus. Once a proof is accepted, the results of the proof
are regarded as true (with very high probability). It may take generations to
detect an error in a proof. If a theorem is widely known and used, its proof
frequently studied, if alternate proofs are invented, if it has known applications
and generalisations and is analogous to known results in related areas, then it
comes to be regarded as rock bottom. In this way, of course, all arithmetic
and Euclidean geometry are rock bottom.
On the other side, to be distinguished from real mathematics, we have meta-
mathematics or first-order logic. As an activity, this is indeed part of real
mathematics. But as to its content, it portrays a structure of proofs which are
21
CHAPTER 1
22
HISTORY OF PROOF
process he calls proof-analysis. The next stage in the process is the emergence of
counterexamples to the conjecture. These counterexamples can reveal problematic
definitions and hidden assumptions. Lakatos divides them into three types: The
first is a counterexample to some step of the proof, but not to the conjecture itself
(It is local but not global.). The second is a counterexample to some step of the
proof, and to the conjecture (It is both local and global.). The third type does not
contradict any step of the proof, and yet it is a counterexample to the conjecture
(It is global but not local.). Each type plays a different role in the proof-analysis
(p. 43). A first-type counterexample signals that there is a problem with the proof;
either a hidden assumption must be revealed, or a definition changed, or a new
proof produced. A second-type counterexample is the most important type for
proof-analysis. When a second type counterexample emerges, the next step is to re-
examine the proof to locate the step to which it is a local counterexample, the
guilty lemma.
This guilty lemma may have previously remained hidden or may have been
misidentified. Now it is made explicit, and built into the primitive conjecture
as a condition. The theorem the improved conjecture supersedes the
primitive conjecture with the new proof-generated concept as its paramount
new feature. (p. 127)
The process of proof-analysis is not primarily about proving the conjecture that
was its beginning, but rather improving the definitions and axioms on which it is
meant to be based. Proof-generated concepts are important original contributions
to mathematics. They account for the facts that axioms and definitions frequently
look artificial and mystifyingly complicated and that theorems are loaded with
heavy-going conditions (p. 142).
Counterexamples of the third type exist only if the proof analysis is invalid.
A proof-analysis is rigorous or valid and the corresponding mathematical
theorem true if, and only if, there is no third-type counterexample to it. I call
this criterion the Principle of Retransmission of Falsity because it demands
that global counterexamples be also local: falsehood should be retransmitted
from the naive conjecture to the lemmas, from the consequent of the theorem
to its antecedent. (p. 47)
The Principle of Retransmission of Falsity is very important to Lakatoss thinking,
and sums up what may be the most important critique in his work of the standard
view of proof. In the standard view, truth is transmitted from axioms to increasingly
complicated theorems. Lakatos claims that this is impossible, but more importantly,
that this does not reflect the way mathematics really works. Mathematics progresses
by the retransmission of falsity from conjectures to axioms and definitions. In this
way counterexamples to conjectures reveal problems with the axioms and definitions.
Many concepts in mathematics have existed since before the time of Euclid, but
these concepts are now much more sophisticated, because conjectures based on
them turned out to give rise to counterexamples which forced (because of the
Principle of Retransmission of Falsity) changes to be made to the concepts.
23
CHAPTER 1
SUMMARY
In this chapter we have summarised what we call the standard view of the history
of proof, and described some important limitations and flaws of this view. Most
notably:
While many sources claim that proof originated in Greece and was not a part
of the intellectual activity of other cultures, there is clear evidence of proving
in ancient China and India, and it is possible that proving was part of
mathematics elsewhere, in spite of the absence of evidence.
Euclids proofs are said to be models of rigour, however they make use of
unstated assumptions and evidence from diagrams.
It is believed that mathematical proofs are (or can be made) formal, and that
this means they are absolutely rigourous. In fact, formalisation of most proofs
is not possible, and proofs can only be checked by a qualified reader.
Proofs are said to transmit truth from established axioms to the theorems they
prove, but as Lakatos points out, the process can go the other way; proofs
allow us to locate hidden assumptions and flawed axioms by retransmitting
falsity from a conjecture with counterexamples to the underlying definitions
and axioms.
Euclid himself could never have imagined the consequences of his effort to
systematise the mathematics known in his day, and so it is unfair to blame him for
the confusion resulting from the standard view of proof. As his name keeps coming
up, however, it is convenient to use labels like Davis and Hershs Euclid myth
and Lakatoss Euclidean methodology to describe this point of view. And as long
as we are clear that we are speaking of a particular perspective, held by many
people, even today, and not of a long dead mathematician, we would agree with
Lakatos that:
Euclid has been the evil genius particularly for the history of mathematics
and for the teaching of mathematics, both on the introductory and the creative
levels. (p. 140)
24
CHAPTER 2
The words proof and proving are used in everyday life, mathematics, and mathe-
matics education in a number of distinct ways, usually without comment. For resear-
chers in mathematics education this can lead to confusion and may be a serious
obstacle to future research (Balacheff, 2002/2004; Reid, 2005). Without trying to
establish the right usages of these words, we will outline here some frequent
ones and describe the differences between them.
As you read this chapter you may want to reflect on these questions:
What does proof mean to you?
What should proof mean to students in schools?
How can you determine what an author means by proof ?
EVERYDAY USAGES
Convincing
When we doubt a statement, we may ask, Do you have any proof of that? Can you
prove it? In these questions proof means evidence, and proving means convincing.
When Shakespeares Othello says, Be sure of it; give me the ocular proof (Act III,
scene 1) he means that Iago must convince him of the truth of his accusation by
providing visible evidence. What counts as convincing evidence depends on context,
and may include physical force, verbal abuse, social pressure, or anything else that
persuades someone else. In the Sidney Harris cartoon captioned You want proof ?
Ill give you proof ! the humour comes from a shift in context, as one mathematician
is shown convincing another mathematician by punching him in the nose, which is
an everyday, but not a mathematical usage of proof as convincing.
Testing
Prove is derived from the Latin verb probare, which means to test, to try. The
English verb probe still carries this meaning. Taking prove as meaning con-
vince when it means test can lead to odd interpretations of common expressions.
For example, the expression the exception which proves the rule is often taken in
the paradoxical sense of asserting that the presence of a counterexample establishes
25
CHAPTER 2
the general truth of a rule, which follows if prove is taken to mean providing
convincing evidence. However, the expression is not so paradoxical if prove is
being used to mean test. Then saying the exception proves the rules amounts
to suggesting that examining exceptions closely and reasoning out the way they
occur can lead to a clarification and improvement of the rule. This interpretation
is reminiscent of Lakatoss (1976) process of proof-analysis in which counter-
examples and proving interact to improve theorems in mathematics (see Chapters 1
and 11).
The use of prove to mean test, try can also occur in the noun form; a
proof can be a test or a trial. In some common phrases, proof-read, proof of
the pudding, 100 proof, the word proof is used in this way. Words like
waterproof and fireproof are also based on this meaning; they describe objects
that have been tested and found to be resistant.
SCIENTIFIC USAGES
When one reads an article about a scientific discovery, one might encounter the
words proof and proving used to refer to convincing, but on the basis of
special types of evidence.
Experiments Prove Existence Of Atomic Chain Anchors
Atoms at the ends of self-assembled atomic chains act like anchors with
lower energy levels than the links in the chain, according to new measure-
ments by physicists at the National Institute of Standards and Technology
(NIST).
The first-ever proof of the formation of end states in atomic chains may
help scientists design nanostructures, such as electrical wires made from the
atoms up, with desired electrical properties. (NIST, 2005, italics added)
When scientists prove something they offer convincing evidence, but that evidence
must be of a special type appropriate to science.
MATHEMATICAL USAGES
Godino and Recio (1997, Recio & Godino, 2001) make a distinction between two
usages of the words proof and proving in two areas of mathematics: foun-
dations of mathematics and mainstream mathematics. This distinction is similar to
the distinction made by Douek (1998) between formal proofs and mathematical
proofs, the distinction made by Davis and Hersh (1981) between metamathematics
and real mathematics, and our distinction between formal proofs and semi-formal
proofs which we mentioned in Chapter 1.
In foundations of mathematics, proofs give theorems a universal and intemporal
validity, they rest on the validity of the logic rules used, the use of formal lang-
uages is required, and proving is a way of coming to grips with the theoretical
26
USAGES OF PROOF AND PROVING
Many researchers in mathematics education use the words proof and proving,
in a number of distinct ways. Most use the words in different ways within the same
paper.
For example, consider this sentence:
Even when students seem to understand the function of proof in the mathe-
matics classroom ... and to recognise that proofs must be general, they still
frequently fail to employ an accepted method of proving to convince themselves
of the truth of a new conjecture, preferring instead to rely on pragmatic methods
and more data. (Hoyles & Kchemann, 2002, p. 194, references removed for
clarity, italics added)
27
CHAPTER 2
The use of the singular form proof in the first line instead of the plural proofs
suggests that the word is being used to mean a concept or category. In the second
line the plural is used, suggesting that a set of objects is meant. Finally in the third
line, the verb proving is used. The fact that method of proving seems to include
accepted methods as well as pragmatic methods suggests that proving is used
to mean something different than constructing a proof in this case.
This suggests a starting point for an investigation of the usage of the words
proof and proving in research in mathematics education. Three categories of
usage can be distinguished on purely grammatical grounds:
1) The use of proof in the singular, without an article to refer to a concept.
2) The use of proof with an article or in the plural to refer to an object.
3) The use of the verb prove to refer to an action or process.
Note that proving is a difficult case, as it can be a form of the verb prove but
also a noun: Jim is proving the theorem or Jims proving of the theorem.
Considering word usage in mathematics education research even at the surface
level of the forms of words reveals some striking differences. For example,
consider the frequency of the use of the words proof , proofs, prove (including
proves proven and proved), proving, words beginning with argu+
(argument, arguing, argue, etc.) and reasoning. In Figure 2 the frequency
of the usage of these words in three papers published in Educational Studies in
Mathematics is shown. The left hand column shows an example of an author
(Fischbein, 1999) who uses the verb prove more often than the nouns proof
and proofs. In contrast, the right hand column shows an example of an author
(Uhlig, 2002) who uses the nouns much more than the verb. It is clear from the
centre column that Hanna (2000) uses the word proof much more than prove,
but it is not clear whether she means a concept or an object when she writes
proof . A closer look at the article clarifies this. Hannas use of proof breaks
down into four categories:
proofs in the plural form, 32 occurrences, 18%
proof preceded by a, 17 occurrences, 10%
proof preceded by the, 7 occurrences, 4%
other uses of proof , 122 occurrences, 71%
The final category still contains a few uses of proof to refer to an object (for
example when it is preceded by an adjective, e.g., an explanatory proof), but
most uses of the word refer to a concept.
In the following we will go into more detail about the ways mathematics
education researchers use proof and proving to refer to a concept, an object, or a
process. Note, however, that we do not claim that any researchers usages fall neatly
into a single category, nor that the usages we describe here are themselves disjoint
categories. As the quote from Hoyles and Kchemann at the start of this section
indicates, several usages can occur in a single paragraph. And while the three ESM
articles analysed in Figure 2 show the predominance of some usages over others,
almost all usages appear in all three articles.
28
USAGES OF PROOF AND PROVING
Proof as a Concept
The use of the word proof to refer to a concept is usually clear from the context
or from syntactical considerations, but once one knows that the word is intended to
refer to a concept, does one know to what concept it is meant to refer? Unfortunately,
no. Researchers in mathematics education have a wide range of perspectives on
proof which make it difficult to know what concept they might mean by the word
proof . In the next chapter we will describe some researchers perspectives, but
as many researchers do not provide enough clues in their writing to definitively
identify their perspectives, we can only leave the reader with the advice to be wary.
Proof as an Object
There are a number of different objects proofs can refer to in mathematics education
research, and those objects can be distinguished by their forms or by their function.
The two most common usages are to refer to texts, usually written texts, of a certain
form, or to refer to arguments, spoken or written, with the function of convincing.
Proof-Texts.
The majority of the [high attaining 14 and 15 year old] students were unable
to construct valid proofs in [the domain of number and algebra]. (Healy &
Hoyles, 2000, p. 425)
29
CHAPTER 2
In mathematics schoolbooks and journals, one encounters some texts under the
heading proof . Such proof-texts are characterised by a particular form and style.
The proof-texts of schoolbooks are different from the proof-texts of professional
mathematics journals, but there is sufficient unity in the styles to justify the use
of the same term for both. Writing proof-texts is a goal of recent reform documents:
High school students should be able to present mathematical arguments in written
forms that would be acceptable to professional mathematicians (National Council
of Teachers of Mathematics [NCTM], 2000, p. 58). That students cannot do this is
what is meant when researchers such as Duval (1990), Senk (1985) and Healy and
Hoyles (2000) conclude that students do not understand proof.
Proving can refer to writing a proof-text (e.g., Douek, 1998) but not everyone
who uses proof to refer to proof-texts uses proving in this way.
Proof as a Process
Proof can refer to a psychological process of reasoning, or to a social process, a
certain kind of discourse. In both cases, (as with proof as an object) what process is
being referred to can be determined both by the form of the process, and by its
function.
30
USAGES OF PROOF AND PROVING
31
CHAPTER 2
are voiced and reasons given. Through a process of social negotiation (probably
guided in significant ways by the teacher) an argument is produced that verifies the
truth of the proposition. Note that the criteria for accepting an argument depend on
the class (including the teacher) or more generally on the community. Arguments
might be accepted by some community that would be rejected by others.
32
USAGES OF PROOF AND PROVING
SUMMARY
The words proof and proving can be used in a number of ways, even in an
academic discipline like mathematics education where the exact meanings of these
words would seem to be important. As Herbst and Balacheff (2009) note, ignoring
these multiple usages can lead to a deadlock in efforts to communicate. But the
answer is not to insist on one correct usage.
If the field is in a deadlock as regards to what we mean by proof, we
contend this is so partly because of the insistence on a comprehensive notion
of proof that can serve as referent for every use of the word. ... We have
argued that to make it operational for understanding and appraising the mathe-
matics of classrooms we need at least three meanings for the word. (p. 62)
We have identified a number of usages in this chapter:
A concept of proof
Proof-texts
Convincing arguments
Deductive reasoning
Personal verification
Personal understanding
A social discourse to verify
A deductive social discourse
These usages are not disjoint categories, nor does a researchers use of proof or
proving in one way in one context guarantee that her or his next usage will be the
same. However, being aware that there are different usages is an important step to
being able to decipher mathematics education research.
In our writing we will attempt to use more precise words to say what we
mean, reserving the word proof primarily to refer to a concept. However, to
avoid unnecessary repetitions we may use proof and proving in one of their
33
CHAPTER 2
other senses when the meaning is clear. Similarly, when quoting others we will
clarify how these words are being used if possible and necessary. If the meaning is
sufficiently clear from the context, or if the meaning is so unclear we cannot
determine what it is, we will not attempt to suggest how the author is using proof
and proving.
Word usages can also offer important hints towards larger issues. In the next
chapter we will use three of these usages, proof-texts, reasoning, and discourse, to
distinguish between theoretical perspectives in mathematics education research.
34