In my previous post The War Against the Infinitesimals, I mentioned that there is a clear difference between the indivisibles, whose origins go back to Archimedes (287 BCE - 212 BCE), and the infinitesimals—a term introduced by Nicolaus Mercator (1620-1687)—, used by Gottfried Wilhelm Leibniz (1646-1716) to develop his version of the calculus.
In this post, I will present some key ideas of Bonaventura Cavalieri (1598-1647), who was the first in the seventeenth century to write about the use of indivisibles for the calculation of areas, volumes and centers of gravity. His principal work, published in 1635, was entitled Geometria indivisibilibus continuorum nova quadam ratione promota [A geometry of continua by way of indivisibles advanced in a new manner]. I will be working with the Italian translation, entitled Geometria degli Indivisibile [Geometry of the Indivisibles], published in 1966 by Lucio Lombardo-Radice1.
The idea of indivisibles is to consider a line as a set of points, a surface as a set of lines, and a solid as a set of planes. As I wrote in my series of posts on Archimedes, the latter used this idea productively to derive results, but did not publish proofs of those results until he had generated proofs by exhaustion, which repeatedly used arguments by contradiction. In the last two posts pertaining to Archimedes,
Parmenides and Zeno of Elea Criticise the Pythagorean Definition of Point, Part I
Parmenides and Zeno of Elea Criticise the Pythagorean Definition of Point, Part II
I showed how the attacks by the Eleatics Parmenides (late 6th century BCE - early 5th century BCE) and Zeno (c.490 BCE-c.430 BCE) discredited the Pythagoreans, who had considered the points of a line to be enumerable, i.e., countable: they are not. These attacks were picked up by Aristotle (384 BCE-322 BCE), and codified centuries later by the Roman Church with the Council of Constance in 1415. As a result, philosophers and scientists had to tread carefully when writing about indivisibles. For example, in a note to the First Day of Galileo Galilei’s (1564-1642) Two New Sciences2, Stillman Drake writes:
the proposition that a line might be composed of indivisibles, strongly opposed by Aristotle, had been condemned as heretical in 1415 by the Council of Constance. John Wyclif was exhumed and his body burned for this and other Epicurean doctrines. [p.39, n.21]
Cavalieri studied mathematics under Benedetto Castelli (1578-1643), who himself had studied under Galileo. Cavalieri and Galileo corresponded regularly, and Galileo even supported Cavalieri’s acquiring a chair in mathematics in Bologna. Both were interested in the question of indivisibles, but from different perspectives: Galileo focusing on physics, Cavalieri on geometry. In the future, I will focus on the interaction between these two, and on Galileo’s discussion of the indivisibles in his Two New Sciences.
Cavalieri’s book is a massive tome, 888 pages in the Italian edition, made up of seven books. The second book is where his key principles are presented. I think it is worth examining that book’s introductory paragraphs, written by his translator Lombardo-Radice. Therein, he refers to “Lure”, who corresponds to the Soviet Hellenistic philologist and historian of science Solomon Yakovlevich Lurie (Соломон Яковлевич Лурье, 1891-1964), who wrote a book about the theory of the infinitesimals of the ancient atomists3.
The quotations below are edited versions of Deepl translations. The original Italian quotations are in footnotes.
Lombardo-Radice presents the three possibilities of considering the continuum. First is the atomistic position:
1. Atomistic position. A continuum is composed of its indivisibles, i.e., it is the sum of its indivisibles. This position, which, as does Lurie, we can call ‘Pythagorean’, is an assumption concerning physical space, which does not hold when it comes to geometric space, should it be thought that the points of a segment, etc., constitute a numerable succession. Were that the case, in fact, each point (each line, each plane) would have an immediate predecessor and successor; and this is obviously not the case, since it is always possible to insert another point between two points, another line between two lines, etc.4
Next is the semi-atomistic position. A indivisible is drawn across the object, leaving a trace of an infinite number of the indivisibles inside the object.
2. Semi-atomistic position, which Lurie calls ‘neo-Pythagorean’. A continuum is generated by the movement of one of its indivisibles: the indivisible, as it moves, leaves as a trace of its passage the infinity of the indivisibles existing in the continuum…. What this infinity is then, and whether the ‘traces’ left by the ‘flowing’ indivisible exhaust the continuum or not, is left undefined.5
Finally is the anti-atomistic position, which might be called Aristotelian:
3. Anti-atomistic position. A continuum is not the sum of its indivisibles: you can add as many points as you like, but you will never obtain a segment, however small; the sum of as many segments as you like will never give a surface, etc. An indivisible, in moving, does indeed generate a continuum (of a dimension greater by one than that of the indivisible), but not a succession of indivisibles, ‘vestiges’ of its passage….6
Cavalieri takes the second position:
Cavalieri's position, as is clear from the fundamental definitions I, II and III, is the second, the ‘semi-atomistic’ or ‘neo-Pythagorean’ one. It was the only position that could allow one to move forward without having analysed—as Cantor managed to do at the end of the 19th century—the various types of infinity. On the other hand, it is precisely Cavalieri’s positions (and Galilei’s position, partly discordant but substantially in agreement on this point) that best foreshadow the modern analysis of the (cardinal) infinite number.7
The brief introduction to the second book, written by the translator, is followed by a series of key definitions by Cavalieri. The translations below are my own, and are quite free, retaining the meaning, but without the structure. Given the complexity of the Italian text I was working with, I literally had no success using automatic translators.
The first definition is for the expression “all of the lines”. A plane is moved across a plane figure, and all of the lines corresponds to all of the lines in the plane corresponding to the intersections of the moving plane with the given plane:
I.) Let two tangents of a given plane figure be opposite and parallel to each other, and two planes be extended from those tangents, so that the two planes are either perpendicular or obliquely angled from the given plane, and let one of the two planes be moved towards the other, while remaining parallel, ultimately the first plane being superimposed on the other. At each moment that the first plane moves, it intersects with the given plane in a straight line. When the two planes are perpendicular to the given figure, we refer to all of the lines, or all of the lines of a right transit, of the given figure as the set of all of those intersecting lines, and we retain one as reference. When the two planes are oblique with respect to the given figure, we refer to all of the lines of the oblique transit of the given figure, and we also retain one as reference.8
The second definition defines “all of the planes”. A plane is moved through a solid, and all of the planes correspond to all of the planes in the solid corresponding to the intersections of the moving plane with the given solid:
II.) Let two tangent planes of a given solid be opposite and parallel to each other, and let one of the two planes be moved towards the other, while remaining parallel, ultimately the first being superimposed on the other. At each moment that the first plane moves, it intersects with the given solid in a plane. We refer to all of the planes of the given figure as the set of all of those intersecting planes, and we retain one as reference.9
The next four definitions follow on from the first two: all of the points of a given line, all of the abscissae of a given line, the residuals of all of the abscissae of a given line, the maxima of all of the abscissae of a given line:
III.) Let there be two straight lines traversing a given plane figure or solid, such that the first intersects perpendicularly with internal planes parallel to the tangent planes, the second intersects obliquely with said planes. We call the points in the line intersecting said planes perpendicularly all of the points in the right transit, and the points in the line intersecting said planes obliquely all of the points in the oblique transit.10
IV.) Let there be a straight line with two extremities. For each point on this line, the partial line from one of the extremities to that point is called an abscissa. We call the partial lines from one of the extremities all of the abscissae of the given line, either of right transit, if the latter is perpendicular to the tangent planes, or of oblique transit, otherwise.11
V.) For each given abscissa from the previous definition, the partial line from the point to the opposite extremity of the original line is called a residual of the abscissa. We call the partial lines to the opposite extremity the residuals of all of the abscissae of the given line, once again, either of right transit or oblique transit.12
VI.) If for any of these abscissae of a given line, the same given line is provided, or a line equal to it, we call it a maximum of this abscissa. We refer to the maxima of all of the abscissae of the given line, or simply, the maxima of the abscissae.13
Cavalieri provides a very interesting corollary to the sixth definition:
Corollary. And since there are as many abscissae as there residuals, there are also as many maxima as there are abscissae, therefore there will be as many maxima as there are residuals; that is, for each residual there will be one maximum, as always, either of right transit or oblique transit of the given line itself.14
In other words, there is a one-to-one relation between the abscissae, the residuals and maxima, all of which, of course, are infinite sets. The translator highlights the importance of this result in a footnote:
The concept of a one-to-one correspondence between infinite sets seems to us one of Bonaventura Cavalieri's most important contributions to the development of mathematical abstraction. Cavalieri boldly rids himself (boldly compared to his time) of the problem of the number of elements of the two sets, and is only concerned with seeing whether or not it is possible to establish a law that makes an element of the first set correspond to one and only one element of the second set. In other words, he does not ask himself quot sint [how many are] the elements of the two sets: he is content to show the tot ... quot [as many as]. And this, even though it is a tendency towards anticipations and ‘precursors’, seems to us an idea very close to that of Georg Cantor….15 [my emphasis]
As a reminder, in my first post on Archimedes, The Rediscovery of the Archimedes Palimpsest and His Use of the Infinite, I quoted Reviel Netz16, who related his discovery that Archimedes himself had written about the one-to-one correspondence between infinite sets:
And Archimedes pointed out that the number of triangles of which the prism was made was the same as the number of lines of which the rectangle was made. Surely he meant this to be verified by the fact that there was a one-to-one relationship. [p.201]
So, as we can see, Cavalieri was reviving a concept that Archimedes himself had used. In some sense, this is not surprising, because Cavalieri’s mentor, Galileo, had spent much of his life reviving the Archimedean tradition. But it does raise questions. The Archimedes Palimpsest, containing the passage referred to by Netz, was only discovered by Johan Ludvig Heiberg (1854-1928) in 1906. Is it possible that Galileo and Cavalieri had access to an Archimedean text, with the same information, that is not known today? Or is it possible that this idea came from another, non-European source? Or did Cavalieri simply redo the steps that Archimedes had once made?
If you wish to donate to support my work, please use the Buy Me a Coffee app.
Bonaventura Cavalieri. Geometria degli Indivisibile. A cura di Lucio Lombardo-Radice. Torino: Unione Tipografico-Editrice Torinese, 1966.
Galileo Galilei. Two New Sciences, Including Centers of Gravity & Force of Percussion. Translated, with Introduction and Notes, by Stillman Drake. Madison: The University of Wisconsin Press, 1974.
С. Я. Лурье. Теория бесконечно-малых у древних атомистов. Москва, Ленинград: Издательство Академии Наук СССР, 1935. (S. Ya. Lurie. The theory of infinitesimals of the ancient atomists. Moscow, Leningrad: Publishing House of the Academy of Sciences of the USSR, 1935.)
Posizione atomistica. Un continuo è composto dai suoi indivisibili, è la somma dei suoi indivisibili. Questa posizione, che possiamo con Lure chiamare «pitagorica», è une ipotesi relativa allo spazio fisico, che non si regge quando si tratta dello spazio geometrico, se si pensa che i punti di un segmento ecc. costituiscano una successione numerabile. In tal caso, infatti, ogni punto (ogni linea, ogni piano) dovrebbe avere un precedente e un successivo immediato; e questo ovviamente non è, potendosi sempre tra due punti inserire un altro punto, tra due linee un’altra linea ecc. [p.188]
Posizione semi-atomistica, che il Lure chiama «neo-pitagorica». Un continuo è generato dal movimento di un suo indivisibile: l’indivisibile, muovendosi, lascia come traccia del suo passagio la infinità degli indivisibili esistente nel continuo (per es., tutti i piani di un solido «veluti vestigia plani a basi ad oppositam basim continuo illi [regulae] aequidistanter fluentis quodammodo relinqui intelliguntur», è detto nella Prefazione della Geometria degli Indivisibili). Quale sia poi questa infinità, e se le «tracce» lasciate dall’indivisibile «fluente» esauriscono il continuo o no, si lascia impregiudicato. [p.189]
Posizione antiatomistica. Un continuo non è somma dei suoi indivisibili: si possono sommare quanti si vogliano punti, ma non si otterrà mai un segmento sia pur piccolo; la somma di quanti si vogliano segmenti non darà mai una superficie, ecc. Un indivisibile, muovendosi, genera sì un continuo (di dimensione superiore di uno a quella dell’indivisibile), man non una successione di indivisibili, «vestigia» del suo passaggio. Quest’ultima asserzione può sembrare ‘curiosa’. Io credo debba essere interpretata al seguente modo (mi si permetta un’immagine che non trovo nella letteratura dell’epoca): noi abbassiamo una tenda terminata da un listello lineare. Per quanto breve sia il percorso del listello, viene sempre generata una superficie nel moto (viene abbassata una parte della tenda). Il moto di una linea genera di colpo una superficie, compatta e non risolubile in linee: su di essa potranno essere poi tracciate quante si vogliano altre linee parallele alla linea generatrice, ma il moto della generatrice di per sé non porta al ‘tracciamento’ di altre linee. [p.189]
La posizione del Cavalieri, come risulta chiaramente dalle fondamentali defizioni I, II e III, è la seconda, quella ‘semi-atomistica’ o ‘neo-pitagorica’. Era l’unica posizione che potesse consentire di andare avanti senza avere analizzato—come riuscirà a fare Cantor alla fine dell’Ottocento—i vari tipi d’infinito. D’altronde, proprio le posizioni di Cavalieri (e la posizione di Galilei, in parte discordanti ma su questo punto sostanzialmente concordanti) sono quelle che meglio preludono alla moderna analisi del numero (cardinale) infinito. [p.189]
Se per tangenti opposte di una qualsivoglia figura piana data si conducono due piani mutuamente paralleli, perpendicolari, o inclinati rispetto al piano della figura data, indefinitamente prolungati dall’una e dall’altra parte, dei quali l’uno sia fatto muovere verso il rimanente, [rimanendo] ad esso sempre parallelo, fino a sovrapporsi ad esso—le singole linee rette, che durante tutto il moto sono intersezioni del piano mosso, e della figura data, prese insieme si chiamino: tutte le linee di tale figura, presa come riferimento una di esse; e ciò quando i piani siano perpendicolari alla figura data. Quando invece sono inclinati rispetto ad essa, si chiamino: tutte le linee del medesimo transito obliquo della figura data, [presa] come riferimento del pari una di esse; si potrà tuttavia, quando convenga, chiamare anche le [linee] prima dette di transito retto, così, come [si chiamano] queste [ultime] di transito obliquo; [insomma] precisamente del [transito], che avviene in tale inclinazione dei piani paralleli rispetto alla figura data. [pp.191-192]
Se, dato un solido qualsivoglia, si siano condotti piani tangenti opposti di esso con un riferimento qualunque, prolungati indefinitamente da ambo i lati, dei quali l’uno sia fatto muovere verso l’altro sempre ad esso parallelo, fino a sovrapporsi ad esso, i singoli piani, che vengono racchiusi nel solido data durante tutto il movimento, insieme presi, si chiamino: tutti i piani del solido dato, preso come riferimento uno di essi. [p.192]
Se due linee rette incontrano internamente piani tangenti opposti, l’una perpendicolarmente, l’altra obliquamente, i punti, che sono intersezioni della linea data incidente perpendicolarmente, e dei singoli piani, che presi insieme si dicono: tutti i piani (prolungati tuttavia in modo, che possano secare le [rette]), ovvero i punti, che sono intersezione di essa, e del piano che è stata fatto muovere, generati nel moto complessivo, presi insieme si chiamino: tutti i punti di transito retto della linea data incidente perpendicolarmente; i quali punti nel [case del]la incidente obliquamente si chiamino [invece: tutti i punti] del medesimo transito obliquo di essa. [p.192]
Se prendiamo le linee comprese tra uno degli estremi della linea retta data, e i singoli punti, che insieme presi si dicono: tutti i punti di transito retto, oppure obliquo, di essa, codeste [linee], insieme prese, si dicano: tutte le ascisse della linea data, le quali (anche se non venga detto esplicitamente) supporremo si chiamino di transito retto, se i punti sono di transito retto, oppure del medesimo transito obliquo, se i punti di essa sono di transito obliquo. [p.193]
Le linee retto poi, [situate] sulla linea data della definizione antecedente, comprese tra i medesimi punti e l’estremo rimanente, si dicano: residue di tutte le ascisse della linea data, di transito retto, se i punti sono di transito retto, oppure del medesimo transito obliquo, se i punti presi sono di transito obliquo di essa. [p.193]
Se per una qualsiasi di quelle [linee], che vengono dette tutte le ascisse della linea retta data, si immagini presa una volta la stessa linea data, oppure [una linea] ad essa uguale, codeste [linee] insieme prese si dicano: le massime di tutte le ascisse della linea data, oppure si sottintenderà sempre che siano [le massime] di tutte, anche se si diranno soltanto: le massime delle ascisse. [p.194]
E poiché tutte le ascisse sono tante, quante [sono] tutte le residue, le massime poi di tutte le ascisse sono tante, quante tutte le ascisse (infatti ad una ascissa qualunque corrisponde una delle massime), pertanto le massime di tutte le ascisse di una linea data saranno tante, quante sono anche le residue di tutte le ascisse, quante che siano tutte le ascisse, o le residue; cioè, per una residua qualsivoglia avremo anche una delle massime (prese sempre di transito retto, oppure del medesimo transito obliquo della [linea data] stessa). [p.194]
Il concetto di corrispondenza biunivoca tra insieme infiniti sembra a noi uno dei più importanti contributi di B.C. allo sviluppo della astrazione matematica. Il Cavalieri si libera arditamente (arditamente rispetto alla sua epoca) del problema del numero degli elementi dei due insiemi, e si preoccupa soltanto di vedere se è possibile, o no, stabilire una legge che faccia corrispondere a un elemento del primo insieme uno ed un solo elemento del secondo, e in modo che ogni elemento del secondo sia immagine di uno e[d] un solo elemento del primo. In altre parole, egli non si chiede quot sint gli elementi dei due insiemi: si accontenta di far vedere il tot ... quot. E questa, pur rifuggendo dalla tendenza alle anticipazioni e ai ‘precursori’, ci sembra un’idea davvero assai vicina a quella di Georg Cantor (cfr. del resto quanto si è detto nell’Introduzione e quanto si dirà a proposito della Discussione con Galilei nell’Appendice I). [p.194,n.2]
Reviel Netz and William Noel. The Archimedes Codex: How a Medieval Prayer Book Is Revealing the True Genius of Antiquity’s Greatest Scientist. Philadelphia: Da Capo Press, 2007.
I likey like
Amazing that arguments about indivibles would result in a corpse being dug up and burnt.
Our current insanity has nothing over that!
John- this sentence below makes it look like Aristotle supported the attacks on Pythagoras. But you are saying Aristotle (and the Church) ATTACKED the attacks, no? Aristotle was defending Pythagoras, correct?
“These attacks were picked up by Aristotle (384 BCE-322 BCE), and codified centuries later by the Roman Church with the Council of Constance in 1415. “