FROM CAUSES TO RELATIONS: THE EMERGENCE OF A NON-ARISTOTELIAN CONCEPT OF GEOMETRICAL PROOF OUT OF THE QUAESTIO DE CERTITUDINE MATHEMATICARUM Tobias SCHÖTTLER Abstract. Insofar as many Renaissance thinkers regard Aristotelian philosophy of science as the framework for their understanding of mathematics and its proofs, they consider geometrical proofs as syllogisms using causes. Furthermore, they identify geometrical proofs as demonstrationes potissimae, which are a kind syllogism that provides both the cause and the effect of an event. By questioning this assumption, Piccolomini initiates the so-called Quaestio de certitudine mathematicarum. Several scholars agreed with him. Others either maintained that mathematical proofs are demonstrationes potissimae or tried to prove that at least some mathematical proofs satisfy the conditions for being demonstrationes potissimae. Despite their differences in detail, all participants in the debate recognized Aristotelian scientific theory as the norm. Yet even traditionally Aristotelian answers take on a new meaning by virtue of a new context. This marks the birth of a genuinely new debate which has unwittingly left its Aristotelian roots behind. As a result, geometrical proofs are no longer thought of as being based on causes or principles of being, but on the relationship between the different figures. Such a relationalism opens up the possibility of further development of mathematics. Keywords: mathematics, proof, geometry, Aristotelianism, Renaissance, certainty Introduction: Relational and Causal Understanding of Mathematics In his Philosophical Essays concerning Human Understanding (1748), David Hume distinguishes two kinds of objects of human reason, namely relations of ideas and matters of fact. The propositions in geometry, algebra and arithmetic are of the first kind. Unlike the matters of fact, the relations of ideas are demonstratively certain. That the Square of the Hypothenuse is equal to the Squares of the two Sides, is a Proposition, that expresses a Relation betwixt these Figures. That three  Tobias Schöttler, Ruhr-University Bochum, Department of Philosophy, Universitätsstr. 150, 44801 Bochum, email: [email protected] 29 Tobias Schöttler - From Causes to Relations: The Emergence of a Non-Aristotelian Concept of … times five is equal to the half of thirty, expresses a Relation betwixt these Numbers. Propositions of this Kind are discoverable by the mere Operation of Thought, without Dependance on what is any where existent in the Universe. Tho’ there never were a true Circle or Triangle in Nature, the Propositions, demonstrated by Euclid, would for ever retain all their Truth and Evidence.1 What Hume expresses here, I want to call a relational understanding of mathematics, which is characterised by two main features. First of all, mathematical propositions are based on the internal relations between the mathematical objects. Secondly, this implies a flexible stance towards mathematics’ ontological foundation. A historically important counter-model is causal understanding of mathematics. Its point of reference is Aristotle’s theory of science as set out in the Posterior Analytics: “According to Aristotle, full- fledged scientific knowledge of something requires understanding its necessitating causes; this knowledge is produced or best manifested by demonstrative syllogism.”2 Knowledge here means knowledge of the cause in the sense of Aristotle’s four causes (aitia). The four causes can be regarded as four types of explanations why the thing in question is how it is.3 Unlike the relationalism, the Aristotelian understanding of mathematics implies a strict ontological foundation for mathematics and mathematical proofs, insofar as mathematical propositions are based on such causes. More precisely, the Aristotelians regard the mathematical objects themselves (and not the relations between them) as such causes. Via abstraction, mathematical objects are dependent on their instances in the world. The relationalists base mathematical proofs on the relations between the figures and on the particular construction of each figure; the Aristotelians base mathematical proofs on the mathematical objects gained via abstraction. From the 13th century to the Renaissance, Aristotelian philosophy of science is the umbrella concept for the understanding of mathematical method and proof, providing its terminological framework (although scholars in the period combine the Aristotelian concept of science with other approaches).4 In the 17th century, the Aristotelian understanding of mathematics gets replaced by a relational understanding. The latter is a prerequisite for various scientific achievements of modern times. One such scientific achievement is the advent of non-Euclidean geometry. Because Aristotelian mathematics relies on abstraction from experience (and experience shows non-Euclidean geometry to be (psychologically) impossible5), it cannot allow for the possibility of a non-Euclidean geometry. However, with reference to its logical consistency, non-Euclidean geometry is logically possible.6 Thus, in the long run, the relational understanding of mathematics was a preliminary condition for the recognition of the non-Euclidean geometry. Based on the relational understanding of geometry, we can, therefore, accept the logical possibility of non- Euclidean geometry.7 In the shorter term, the relational understanding of mathematics enabled some scientific achievements of the 17th century, particularly the mathematization of non-mathematical sciences and the algebraisation of geometry. Within strict Aristotelianism, all scientific disciplines, however, have their unique 30 Society and Politics Vol. 6, No. 2 (12)/November 2012 subject areas, and hence the methods of one discipline cannot be applied to the subject area of another discipline. Therefore, it was the break with Aristotelian understanding of science and mathematics that allowed for several breakthroughs, not least the so-called “scientific revolution.”8 My paper deals specifically with the break away from the Aristotelian causal theory of geometry and geometrical proof. I want to demonstrate that this break emerged within the scholastic Aristotelianism itself in the second half of the 16th century – in the prehistory of scientific revolution. During the Renaissance there was a growing interest in mathematical method. On the one hand, this was caused by the mathematical problems that craftsmen had to deal with as a consequence of their practical needs. On the other hand, it was a product of renewed interest in non-Aristotelian ancient mathematical writings (such as Euclid’s Elements and Proclus’ commentary on Euclid), which were extensively published in new, translated editions at the time.9 Many actors were important in such developments, but the school of Padua played a decisive role within the discussions about method and mathematics.10 For the Paduan scholars, as for most Renaissance thinkers, Aristotelian philosophy of science (rather its scholastic version than the original one) is the umbrella concept for their understanding of mathematical method and proof. In the early modern period, many philosophers of mathematics either regarded geometrical proofs as syllogisms or thought that they should be reformulated as syllogisms.11 In most cases, geometrical proofs were equated with a specific kind of Aristotelian proof, namely the demonstratio potissima (a kind of syllogism that provides both the cause and the effect of an event; more detailed explanation in chap. 2.2 below). As a critical response to this approach, the Paduan philosopher Alessandro Piccolomini (1508–1579) initiated a debate that came to be called the Quaestio de certitudine mathematicarum. This is the starting point for my investigation of the shift from causes to relations in mathematical thinking. The Quaestio de certitudine deals with three interrelated questions. First, it questions the certainty of mathematics in general. Since mathematical certitude is traditionally justified by the special character of mathematical proofs, the initial argument focuses on the second question: whether or not there is a place in mathematics for the demonstratio potissima. (I use the epithet “initial” in order to distinguish the 16th century quaestio from its revival in the 17th century.) In the course of the debate, these initial questions increasingly lose their relevance. More and more, attention shifts to the third question: “whether the actual procedure of geometers in proving theorems and solving problems could be reconciled with Aristotle’s description of a demonstrated science.”12 Such a procedure was mainly based on Euclid’s Elements. Therefore, the real subject of the discussion is the incompatibility of Euclidean geometry with the Aristotelian understanding of science. The initial debate is a symptom of the contradictions within the traditional preconditions of the understanding of geometry, namely the lack of distinction between Aristotelian proofs and geometrical method.13 In fact, the debate about the certainty of mathematics shows the inadequacy of using the criteria of Aristotelian proof theory to describe mathematical proofs. From a present-day perspective, the 31 Tobias Schöttler - From Causes to Relations: The Emergence of a Non-Aristotelian Concept of … debate is based on completely misguided assumptions. No consensus was reached, and conflicts were not resolved. However, although all participants in the debate remained firmly within the Aristotelian framework, a new concept of geometrical proof ex negativo emerged in the discussion. The initial discussion has a scholastic nature, admittedly, but as it progresses, the Aristotelian understanding of science grows increasingly vague. This lay the groundwork for a non-Aristotelian understanding of geometry with attempts to reformulate Euclidean geometry by means of Aristotelian tools. As understood by an Aristotelian, geometry is supposed to explain single geometric figures in terms of their unique causes. The new concept of geometrical proof, however, emphasizes the relations between figures and focuses the debate on the practice of geometrical constructions. This concept of geometrical proof carries changes in the understanding of geometry in general within itself. When scholars such as Gassendi, Wallis, Hobbes and Barrow take up the debate again in the 17th century, they are no longer interested in the question of whether there are demonstrationes potissimae in geometry, and they are equally uninterested in rescuing the Aristotelian framework. Insofar as they adopt some questions and arguments of the initial debate, they take on the relational understanding of geometry. I am going to trace the shift from causes to relations in three steps. In the first, I will outline the Aristotelian background of the debate, in particular, the traditional justifications of mathematical certainty. More specifically, I shall set out the features of the demonstratio potissima. In the second step, I am going to sketch the main positions in the Quaestio de certitudine mathematicarum. In the third, I want to display the new concept of geometrical proof which looms up in the discussion. In the course of this, I will sketch out the quaestio’s revival in the 17th century and contextualise it: I will consider the relational understanding of geometrical proof in the context of the overall development in mathematical thinking of this period. 2. The Background: Aristotelian Justifications of the Certainty of Mathematics In the philosophy of mathematics during the early modern period, we can identify two justifications for mathematical certainty.14 The first strategy justifies the objective certainty of mathematics by means of the ontological status of their entities (2.1). The second strategy deduces the subjective certainty of mathematics from the character of mathematical proofs (2.2). The second strategy is the more important for the purposes of this paper. Hence, after giving a short explanation of the first, I will concentrate on the second strategy. 2.1. The Ontological Status of Mathematical Entities The first strategy is used by, for example, Thomas Aquinas.15 Following Aristotle, he regards mathematical entities as abstractions based on sense experience. From the ontological status of mathematical entities, Thomas concludes that mathematics is more certain than both natural philosophy and theology. This is so, because, unlike natural philosophy, mathematics does not deal with matter and motion; unlike theology, mathematics considers entities which are given to the senses 32 Society and Politics Vol. 6, No. 2 (12)/November 2012 and to the imagination. To put it in a nutshell, insofar as mathematical entities are created by abstraction, they have the highest level of clarity and evidence. 2.2. The Use of the Demonstratio Potissima The second strategy bases mathematical certainty on the characteristics of the proofs used in mathematics. In the Posterior Analytics, Aristotle envisions science as true knowledge gained via reasons or causes.16 Furthermore, he maintains that mathematical disciplines produce proof by use of a syllogism in the first figure,17 which has the following form: maior: middle term (M) – predicate (P). minor: subject (S) – middle term (M). conclusio: subject (S) – predicate (P). In the early modern period, only a few philosophers of mathematics drew a distinction between geometrical proof and the Aristotelian syllogism.18 Most of them classify geometrical proofs as demonstrationes potissimae, which are regarded as the highest and most certain type of proof. I will explain this type of proof by comparing it with the two other types, namely the demonstratio quia and the demonstratio propter quid.19 All three types of proof are regarded as a syllogism in the first figure. The demonstratio quia infers the cause from its effect. This kind of proof can be illustrated by an example from the Posterior Analytics.20 maior: Non-twinkling heavenly bodies (M) are near earth (P). minor: Planets (S) are non-twinkling heavenly bodies (M). conclusio: Planets (S) are near earth (P). Its middle term is the unique effect (effectus proprius). It signifies the (observed) effect, namely that these heavenly bodies do not twinkle. By rearranging this syllogism, we get the demonstratio propter quid. maior: Heavenly bodies which are near earth (M) do not twinkle (P). minor: Planets (S) are heavenly bodies which are near earth (M). conclusio: Planets (S) do not twinkle (P). The demonstratio propter quid infers the effect from its proximate cause.21 Its middle term signifies the proximate cause of the effect, in our example the proximity to the earth. (Of course, the middle term of such a proof is understood as one of the four causes in the Aristotelian sense.) At this point, Aristotle leaves us. He distinguishes only these two kinds of proof. But following Averroes, Aristotelians assume a third type of proof, namely demonstratio potissima.22 Such a proof infers the effect (esse) and the cause (the propter quid effectus) from fundamental premises.23 It is a syllogism that provides both the cause and 33 Tobias Schöttler - From Causes to Relations: The Emergence of a Non-Aristotelian Concept of … the effect of an event by using a middle term which specifies the proximate cause of the effect in a unique way.24 The primary, as well as the secondary literature, usually content themselves with giving an abstraction explication of the demonstratio potissima, but they do not provide an actual example of a demonstratio potissima.25 There are two possible interpretations. Following the first interpretation, we get the demonstratio potissima by another rearrangement of the syllogisms above. According to the second interpretation, the demonstratio potissima is just a variant of the demonstratio propter quid. There is only one possibility for further rearranging the syllogisms above: maior: Planets (M) are near the earth (P). minor: Not twinkling heavenly bodies (S) are planets (M). conclusio: Not twinkling heavenly bodies (S) are near the earth (P). Indeed, the resulting syllogism does not fulfil all requirements for being a demonstratio potissima. Therefore, it would be more reasonable to assume that the demonstratio potissima is nothing but a variant of the demonstratio propter quid which is only accidentally distinguished from the demonstratio propter quid in the proper sense.26 If so, it would be a demonstratio propter quid in which the effect is unknown. Being a demonstratio potissima would depend on the previous knowledge of the recipient. Despite the problems concerning the interpretation of this kind of proof, Aristotelian philosophers of mathematics base the certainty of mathematics on its use of the demonstratio potissima as the most certain type of proof. They regard it as the most certain type of proof because it provides us with the cause and its effect at once. However, the equivalence of mathematical proof and demonstratio potissima was essentially contested.27 It was this that led Piccolomini to initiate the debate, Quaestio de certitudine mathematicarum. Within the framework of this debate, even the traditionally Aristotelian answers take on a new meaning by virtue of a new context. This marks the birth of a genuinely new debate which has unwittingly left its Aristotelian roots behind. I am not interested in the result of the initial Aristotelian debate, not least because there was no real final solution. Rather, my interest lies in the relational understanding of geometry and geometrical proof which looms in the background of the debate. 3. Main Positions in the Initial Debate About Certainty in Mathematics The subject of the Quaestio is the question of the certainty of mathematics. Nevertheless, the initial debate focuses on the question of whether geometrical proofs can be identified as demonstrationes potissimae. From a logical point of view, three possible positions can be distinguished in the debate.28 The first group defends the identification thesis: according to the traditional position (e.g., Hieronymus Balduinus, Jacob Schegk (1511–1587)), all geometrical proofs are demonstrationes potissimae. The second group denies the identification of geometrical proofs and demonstrationes potissimae (2.1). The critics (e.g., Alessandro Piccolomini (1508–1579), Simon Simonius (1522–1602), Benedictus Pererius (1535–1610), the Jesuits of Coimbra, Martin 34 Society and Politics Vol. 6, No. 2 (12)/November 2012 Smiglecius (1562/1564–1618)) claim that no geometrical proofs are demonstrationes potissimae. The moderate defenders (e.g., Franciscus Barocius (1537–1604), Joseph Blancanus (1566–1624)), as the third group, maintain that at least some geometrical proofs are demonstrationes potissimae (2.2). After Piccolomini’s commentarium de certitudine mathematicarum disciplinarum (1547), the first position maintained very few proponents.29 The debate mainly takes place between the critics and the moderate defenders. To characterize these positions in more detail, I will focus less on their specific arguments and more on their assumptions and the implications of their arguments. I will start with the positions of the critics. 3.1. Critics: No Mathematical Proofs are Demonstrationes potissimae As one of the critics, Piccolomini’s arguments define the debate.30 The critics of the identification thesis often further refine his arguments and investigate their implications. Piccolomini shows that geometrical demonstrations are no proofs by any of the four causes.31 a) Geometrical demonstrations are not proofs by efficient cause (causa efficiens) because mathematics does not deal with action.32 Many of Piccolomini’s arguments are based on the assumption that the geometrical objects understood as pure quantity (quantitas) have no relation to action (actio).33 Simonius (Antischegkianorum Libernus, 304 and 310) follows this argument in a very peculiar way. He believes that there are demonstrationes potissimae used in mathematics with the formal cause as middle term. But he regards only the proofs from the efficient cause as the most perfect.34 b) Geometrical demonstrations are not proofs by final cause (causa finalis).35 Of course, mathematics does have purposes, insofar as it is useful for various applications. But there are no final causes within mathematics. Piccolomini argues that only activities have purposes and therefore final causes. But mathematical objects are immutable. Where there is no change, there can be no purpose (of change). c) Geometrical demonstrations are not proofs by material cause (causa materialis) because there is no real matter (materia realis) in mathematics.36 Mathematics just deals with intelligible matter (materia intelligiblis) created by abstraction. d) Geometrical demonstrations are not proofs by formal cause (causa formalis). Since Piccolomini attacks the traditional view here, its refutation takes up the largest room within his arguments by far. Looking at the progress of the debate, we can identify two influential arguments against the use of formal causes in geometrical proofs. To begin with, the middle term of every demonstratio potissima has to be the definition of the subject or of its property (definitio vel subiecti vel passionis).37 Piccolomini shows that geometrical proofs do not use such a middle term, by referring to Euclid’s demonstration that the angle sum in a triangle equals two right angles (see below fig. 1).38 The demonstration (Euclid I.32) in short is as follows. AB is parallel to CE. Therefore, the alternate angles BAC and ACE equal one another and the corresponding angles ABC and ECD equal one another. Accordingly, the angle ACD equals the sum of the angles BAC and ABC; and thus the sum of the interior angles equals the sum of the angles ACD and ACB. Since the sum of the angles ACD and 35 Tobias Schöttler - From Causes to Relations: The Emergence of a Non-Aristotelian Concept of … ACB equals two right angles, the sum of the interior angles equals the sum of two right angles. Fig. 1 Obviously, this demonstration makes use of the exterior angle. Proclus already questions whether this proof uses (real) causes.39 Following Proclus, Piccolomini shows that the exterior angle is neither a definition of the triangle itself nor of one of its properties. The exterior angle is not part of the triangle’s definition. Even if the exterior angle did not exist, it would still be a triangle. Furthermore, the demonstratio potissima requires a middle term which signifies a unique and proximate cause.40 Piccolomini emphasizes that there is no hierarchy of priorities between the mathematical properties with respect to their dependencies. Instead, one theorem can be proven by different premises.41 One middle term used in a proof is neither more unique nor more proximate than another possible middle term. While Piccolomini does indeed justify the certainty of mathematics by means of the nature of its entities,42 the critical part of his argumentation was significantly more influential. Several scholars agreed with Piccolomini in one respect or another.43 Alongside various refinements of Piccolomini’s observations, two main arguments evolved. Each is deeply connected with the other. The first argument is based on the distinction between the principle of Being (principium essendi) and the epistemological principle (principium cognoscendi).44 Geometric proofs do not use principles of Being insofar as they do not use real causes. Instead, they use only the second one, the epistemological principles, in the sense that the proofs rely on reasons for understanding. We understand or comprehend a figure’s properties by its construction. Indeed, a strict Aristotelian does not regard its construction as the cause of its properties. The construction only provides us with a principle of understanding. By contrast, a principle of Being of a figure’s properties would be its cause in an ontological sense. The second argument is based on the distinction between the essence of a geometric figure and its relations to other figures. Many of Euclid’s proofs demonstrate properties of one geometric figure by using its relations to other figures. But a figure’s 36 Society and Politics Vol. 6, No. 2 (12)/November 2012 relation to other figures does not belong to its essence; nor are any of these relations unique and proximate causes, since there is no hierarchy of priorities between mathematical properties with respect to their dependencies. Thus, geometric proofs do not follow from the essences of figures. I am going to illustrate this argument in reference to Euclid’s construction of an equilateral triangle (see below fig. 2).45 The task is to construct an equilateral triangle on a given finite straight line AB. To do this, we have to describe the circle BCD with centre A and radius AB and the circle ACE with centre B and radius BA. Their point of intersection C has the same distance AB to A and to B. Therefore, the triangle ABC is equilateral. Fig. 2 Euclid uses here the circle, or its definition, in order to construct the equilateral triangle and to demonstrate its properties. In this respect, being equilateral is not proven by the essence of the triangle but by its relation to other figures. Smiglecius (1562/1564–1618) refines this argument. In my English translation, I make its syllogistic structure explicit. My primary interest is not the conclusio but the minor. [maior:] In the demonstratio potissima, the cause of the property or characteristic is the essence of the subject from which the properties originate. [minor:] In Mathematics, the properties are not derived from the essence of the subject, but from the relations to other figures. [conclusio:] Geometrical proofs do not demonstrate properties by using the real cause of the essence or Being.46 Pererius, among others, radicalized Piccolomini’s theses and arguments by denying mathematics the status of science.47 Geometrical proofs do not prove by real causes (in the sense of principles of being), and in Aristotelian philosophy of science, proving by causes is a requisite for being a science. Geometrical proofs do not fulfil this condition and therefore mathematics is not a science in an Aristotelian sense. This is the point at which the moderate defenders enter the debate. 37 Tobias Schöttler - From Causes to Relations: The Emergence of a Non-Aristotelian Concept of … 3.2. Moderate Defenders: Some Mathematical Proofs are Demonstrationes potissimae Many philosophers in the 16th and early 17th centuries did not want to accept the consequence that mathematics (the prime example of a science) should not be a science at all. So, the more moderate defenders, such as Barozzi (aka Barocius) and Blancanus (aka Biancani) tried to prove that at least some mathematical proofs satisfy the conditions for being a demonstratio potissima.48 In order to do this, they have to show that the middle terms of these proofs signify the unique and proximate cause of the property in question. Commonly, they regard the middle terms as definitions which denote formal or material causes.49 Blancanus discusses the two paradigms of geometrical proofs in the debate and comes to the following result. In his opinion, the equilateral triangle (Euclid I.1) is proven by formal cause, insofar as he regards the definition of the circle as a formal cause.50 He takes the proof of the triangle’s angle sum (Euclid I.32) as a proof by material cause, insofar as it is a conclusion from the parts to the whole.51 His arguments are based on a very peculiar notion of definitions in geometry. While the critics usually conceive of such definitions as being nominal, Blancanus argues that in geometry definitions are nominal as well as real at the same time.52 But furthermore, he points out that these definitions denote the reason (ratio) or cause (causa) of the figure in question.53 He labels them as causal definitions (definitiones causales), that is, genetic definitions.54 Prior to Blancanus, definitio causalis was a definition of an attribute as an equivalent to the real definition of the subject term. Blancanus’ use of the term definitio causalis is beyond the scope of its traditional use, insofar as he accentuates the constructive function of such causal definitions. His example is the definition of a square: he takes it to be the definition that designates the cause for being a square. In many places, Blancanus blurs the distinction between the definition and the construction of a figure. By assuming causal definitions, Blancanus introduces causes into geometry and into geometrical proofs. With his peculiar view of geometric definitions and constructions, Blancanus undermines the two major presuppositions of the critical objections. On the one hand, he calls into question the distinction between principles of Being and epistemological principles. On the other, the distinction between the essence of a geometric figure and its relations to other figures becomes debatable. For example, some critics object that the demonstration of the equilateral triangle proceeds from the definition of the circle, which is not part of the essence of such a triangle. In contrast to this, Blancanus regards the whole construction as part of the concept of the figure.55 With this in mind, the Being of geometrical figures is their construction, and Blancanus subverts the two distinctions. Within Blancanus’ theory, both distinctions cannot be applied meaningfully to mathematical objects because the geometric figures depend less on abstraction but more on definition and construction. In this perspective, principles of Being and epistemological principles coincide in geometry, insofar as there are no geometrical figures beyond their construction. 38