Heavens of Knowledge: The Order of Sciences in Dante’s Convivio

Abstract

The essay focuses on Dante’s divisio scientiae presented in the second book of his Convivio. As a first step, it offers a fresh reading of Dante’s description of knowledge and the cosmos, emphasizing its deeply visual nature. The article then presents an overview of the Medieval tradition of the divisio scientiae, culminating with divisions that are geographically and chronologically close to Dante’s Florence. In relation to Brunetto Latini’s Rettorica and Tresor, new evidence is provided to elucidate his peculiar division of logic. Ultimately, Dante’s divisio and its objectives are reassessed in light of their historical background, underscoring their cosmological and totalizing scope. Adopting an interdisciplinary approach, the findings consider cognitive practices, such as diagrams, and examine a wide array of sources within their historical and institutional context, highlighting their transmission and dissemination.

Felices animae, quibus haec cognoscere primis

Inque domus superas scandere cura fuit!

(Ovid)1

1. A Visual Hypothesis

In the second book of the Convivio, Dante elaborates a double commentary—literal and allegorical—on his canzone Voi che ’ntendendo il terzo ciel movete. The literal explanation of the ‘third heaven’, of where it is and which angelical intelligence moves it, involves an extensive, apparently disproportionate discussion of the celestial spheres and their number, of the primum mobile and the empyrean, and of angelical hierarchies (Conv. II, i–vi).2 Further on, the allegorical interpretation delves into a lengthy description of the system of knowledge and of its various branches (II, xiii–xiv). As is well known, at the beginning of the allegorical exposition, which conveys the “true” meaning of the poem, Dante tells us how, by reading the consolatory texts of Cicero and Boethius and by attending the ‘schools of the religious men and the disputations of the philosophizers’, he developed a profound love for philosophy and started to explore new intellectual perspectives (II xii, 7).

In the broader context of the treatise as a whole and of its didactic purposes, Dante’s claim of a solid educational background is pivotal to establishing his authority as a trusted source of knowledge and wisdom: as Albert Ascoli puts it, personal events ‘are presented as reflecting the biographical sine qua non for the transfer of knowledge to others’ (Ascoli 2008, p. 71). This biographic account is immediately followed by the description of the various branches of knowledge, each one corresponding to a celestial sphere, which represents its ultimate meaning (‘dico che per cielo io intendo la scienza e per li cieli le scienze’, II, xiii 2). In earlier chapters, Dante had already established other correspondences, making clear that the nine angelical intelligences preside over the nine astronomical spheres and form triads that contemplate the three Trinitarian Persons and their attributes: the potency of the Father, the wisdom of the Son, and the charity of the Holy Spirit (II, v. 6–8).

Despite a long and at times intricate explanation, the system that emerges from Book II is quite elegant and effective. Knowledge is framed within a cosmological system presided over by angelic hierarchies that contemplate the Trinity. Following the ascending architecture of the heavens, the series of sciences lays an ideal path leading to the highest form of knowledge, the ‘divine science’ corresponding to the empyrean. A visual schematic rendition of it could possibly look like this (Figure 1):

With the notable exception of rhetoric (Ricklin 2002), in each pairing, Dante provides a variable number of quotations and authorities to sustain his claims: Horace to state the temporariness of grammatical structures; Alfraganus for the diameter of Mercury; Pythagoras (through Aristotle) for the numerical definition of natural phenomena; Ptolemy and Aristotle to explain the heat on Mars, along with Albumasar and Seneca on its astrological influence; Ptolemy and Euclid for Jupiter and Geometry; and Aristotle on the nobility of astrologia (astronomy) alongside Ptolemy. As regards physics and metaphysics, Aristotle is mentioned within a varied pantheon, including Ovid, the Pythagoreans, Anaxagoras, Democritus, Avicenna, and Ptolemy. Thomas Aquinas, as commentator of the Nicomachean Ethics, is invoked to assert the foundational role of ethics in studying, while Solomon declares the exclusivity and perfection of theology. Despite their interest, the irregularity of the associations of sciences and authorities, the lack of a uniform criterion, and the pervasive presence of some authors, such as Aristotle and Ptolemy, suggest refraining from including them in the general scheme.3

Indeed, such a diagram could be viewed as a mere visual summary devised ex post by a reader somewhat familiar with medieval cosmology and diagrammatic traditions. Yet, I would rather suggest that Dante envisioned something akin to this, articulating his explanation accordingly, and that he expected his reader to conjure a comparable mental image while engaging with the Convivio. I would go so far as to say that the most vital outcome of Book II and its major legacy to the structure of Paradiso lies in this comprehensive and markedly vertical image of the cosmos and knowledge.

In the last few decades, highly refined scholarship has demonstrated the pervasiveness of medieval diagrammatic traditions, unveiling their underlying logic, their epistemological scope, and their relational structure, which make them ‘cognitive artefacts par excellence’ (Kupfer 2020, p. 11). More specifically, cosmological diagrams received attention as crucial in developing an ‘iconographie du monde physique’ (Obrist 2004, p. 19; see also Obrist 2020, 2022) and in establishing a spherical and geocentric image of the universe. Rooted in the commentary tradition of Plato’s Timaeus, representations of the celestial spheres allowed for a proper ‘visuelle Weltaneignung’, a visual appropriation of the world, as defined by Kathrin Müller (Müller 2008). By combining graphic elements and captions, such diagrams were able not only to provide a geometric representation of the cosmos but also to convey further information in a cohesive framework that highlighted relationships between data.4 It is not uncommon to find the Zodiac signs or the angelical hierarchies listed in the celestial spheres, usually occupying the highest heavens (i.e., in Gautier de Metz’ Image du monde, BnF fr. 1444 (https://gallica.bnf.fr/ark:/12148/btv1b90590400/f219.item, 13 April 2024), f. 217r).

At the same time, the encyclopedic tradition dating back to Isidore of Seville’s Etymologiae developed other kinds of concentric circular tables schematizing series of concepts and their combinations, such as the series of septenaries (see the De quinque septenis by Hugh of Saint-Victor in the MS Houghton Library, Typ 584 (https://library.harvard.edu/sites/default/files/static/onlineexhibits/booksinbooks/584.html, 13 April 2024), c. 1235, or the seven septenaries in MS BnF fr. 9220, f. 11v (https://portail.biblissima.fr/fr/ark:/43093/ifdataf1e78fd97040b047d9f98948cf4c9070cb15cab3, 13 April 2024), 13th–14th c.). Further examples are a rota showing the characteristics of different beings in the Dragmaticon by William of Conches (e.g., Montpellier, BU Médecine, H 145, c. 1250–1275 (https://iiif.irht.cnrs.fr/iiif/France/Montpellier/B341722104/DEPOT/IRHT_141938_2/full/800,/0/default.jpg, 13 April 2024)) or the ages of the world in the Liber floridus by Lambert of Saint-Omer (Wolfenbüttel, Herzog August Bibliothek, Guelf 1, f. 31r) (http://diglib.hab.de/?db=mss&list=ms&id=1-gud-lat&image=00067, 13 April 2024); (Heitzmann and Carmassi 2014). One of the most famous representations of philosophy and the liberal arts in the Hortus deliciarum by Herrade of Landsberg (c. 1180) also features a circular pattern (Courcelle 1967, pp. 79–80; Figure 2); interestingly enough, poets and magicians are outside the rota.5

In some cases, the two traditions could overlap to create a composite diagram, wherein the structure of the cosmos serves as a scaffolding for other concepts, as exemplified in certain copies of Isidore’s Etymologie, where some planetary skies (with the earth at the center) are associated with moral features: the moon is coupled with oratio lingue, Jupiter with iracundia and ardor, and so on, whilst the heavens are framed between natura and ratio (Figure 3).

It would be easily argued that nothing in Dante’s text or in its tradition supports my hypothesis. Yet, Dante shared with his contemporaries the Aristotelian principle that ‘there is no thinking without images’, which were perceived as essential to knowledge acquisition and reasoning. In his De memoria et reminiscentia, while delving into this concept already explained in the De anima, Aristotle provides a revealing example involving a geometric figure. When drawing a triangle, he says, we inevitably describe a finite figure with precise dimensions, which are nonetheless arbitrary. The same happens in thinking of a finite object:

[…] intelligere non est sine fantasmate. Accidit enim eadem passio intellectui que quidem et in describendo: ibi enim nulla utentes quantitate trigoni determinata, tamen finitam secundum quantitatem describimus; et intelligens similiter, etsi non intelligat quantum, ponitur ante oculos quantum, intelligit autem non secundum quod quantum est.

(De mem. et rem. I, 449b–450a)6

As Jeffrey Hamburger fittingly pointed out (and as the example of a geometric figure reinforces), Aristotle establishes an analogy between thinking and drawing diagrams. To some extent, it is the very ‘procedure of producing a diagram that resembles the process of thought’ (Hamburger 2020, p. 65). Actually, a diagram does not even require a graphic rendition to exist; it ‘can also be an entirely ideational entity’ (Hamburger 2022, p. 54).

In my view, it is not inappropriate to think that the reading of Dante’s explanation in the second book of the Convivio calls for and even requires mental mapping. Two examples in Dante’s texts, both involving cosmological and geographical descriptions, can substantiate such a claim. The first one is in Purgatorio 4. In clarifying the location of Mount Purgatory, Virgil asks Dante to imaginare, that is, to ‘draw in his mind’, a schematic map featuring Jerusalem and Purgatory sharing the same horizon and the trajectory of the sun in relation to the equator. This thought experiment is effective to the point that Dante declares a perfect clarity of vision:

“… se ’l vuoi poter pensare,

dentro raccolto, imagina Siòn

con questo monte in su la terra stare

sì che … vedrai come etc.

se l’intelletto tuo ben chiaro vada”.

“Certo, maestro mio”, diss’io, “unquanco

non vid’i’ chiaro sì com’or discerno”.

(Purg. 4, 67–77; my emphasis)

The second example is from the Convivio. In order to better visualize the movements of the sun, Dante engages the readers in an analogous and quite long mental mapping, which requires them to imagine three different points on earth and a series of related trajectories:

Imaginando adunque, per meglio vedere, [che] in questo luogo ch’io dissi sia una cittade e abbia nome Maria; dico ancora che se dall’altro polo, cioè meridionale, cadesse una pietra, ch’ella caderebbe in su quel dosso del mare Occeano ch’è a punto in questa palla opposito a Maria. … E qui[vi] imaginiamo un’altra cittade, che abbia nome Lucia. … Imaginisi anco uno cerchio in su questa palla … Segnati questi tre luoghi sopra questa palla, leggiermente si può vedere come lo sole la gira etc.

(Conv. III, v, 10–13; my emphasis)

Both of these cases exemplify not only Dante’s attitude toward mental drawing but also his acquaintance with cosmological diagrams as cognitive tools.7 What remains to be explained is why and how such an image of the universe intersects with the classification of sciences.

While specific divisiones have been variously suggested as possible models for the Convivio, I will reassess Dante’s proposal in the context of the extensive and substantial tradition of the divisio scientiae as a whole. Such tradition shaped late-medieval intellectual habits and influenced both Dante and his readers, who negotiated their vision of knowledge within the same episteme. Accordingly, in what follows, I will first provide a necessary overview of the medieval tradition of the divisio scientiae; then, I will highlight divisiones developed or available in Dante’s Florence and discuss previous pairings of heavens and sciences. In light of this panorama, it will be possible to reassess Dante’s attitudes, choices, sources, and, more crucially, his intellectual stance.

2. The Medieval Tradition of the divisio scientiae

The description of disciplines that Dante provides aligns with the long-standing tradition of the so-called divisio scientiae (or divisio philosophiae), which has its roots in late Antiquity and persists throughout the Middle Ages (Weisheipl 1978; Weijers 1986–1987, 1996, pp. 187–96; Dahan 1994). Besides texts entirely devoted to this aim, it is possible to find divisiones scientiae almost everywhere in medieval doctrinal texts; not by chance, in his own divisio, the Florentine Dominican friar Remigio de’ Girolami labels this scholastic practice as ‘frequenter recitanda’ (Div. sc. 1, in Panella (1981)). Topical loci for the divisio scientiae are: commentaries to the Isagoge and to the second book of Boethius’ De Trinitate (d’Onofrio 2001; Zonta 2001, pp. 65–67); commentaries and quaestiones on the Metaphysics, as well as on other Aristotelian works; biblical exegesis and the commentary tradition on Virgil’s works. Of course, further topical places were the entries devoted to philosophy in lexicons and encyclopedias, whose indexes also reflect a general structuring of knowledge (Dahan 1994, p. 23). Finally, we can typically find divisiones scientiae in the introductions to philosophy and in the student guides developed in the Faculties of Arts (Lafleur 1988; Weijers and Holtz 1997).

Typically, divisions offer a comprehensive view of knowledge and contextualize disciplines within a broader framework, serving various purposes. Despite exceptions, what they typically do not represent is an ordo or cursus studiorum, which is tied to pedagogical criteria; generally, overlaps between study programs and divisions of science are incidental. As Copeland and Sluiter argue, divisions ‘represent a constant development and refinement of an epistemology or theory of knowledge rather than, necessarily, actual programs of study’ (Copeland and Sluiter 2009, p. 5).

The subdivision of knowledge is rarely subjective or arbitrary, being instead rooted in objective principles, in line with the Aristotelian postulate that sciences are defined on the basis of their object of inquiry:‘scientia secatur in res’ (De anima III, 8, 431b 24–25). Hence, the divisio is always a more or less detailed description of what is presented as a matter of fact.

The basic models of medieval divisiones scientiae, variously modified and integrated with each other, are essentially two. The first one (Figure 4), deriving from the Stoics and attributed by Augustine to Plato (De civ. Dei VIII, 4; P. Hadot (1979)), divides philosophy into moral, natural, and rational (or sermocinalis) (Augustinus Hipponensis 1955).8

This division—which still underpins the extremely complex De reductione artium ad theologiam by Bonaventure in the thirteenth century—is illustrated and popularized by Seneca in an epistle to Lucilius dedicated to the classification of philosophy.9 By establishing a correspondence with the universe, Seneca posits that it is not possible to encompass philosophy in its entirety, hence the necessity to describe it in its parts (Ep. ad Luc. 89, 1). After briefly discussing the various opinions held by the philosophical schools of his time, he proceeds to further divide the three parts according to their different objects of inquiry: ethics deals with impulses, actions, and the distribution of goods among human beings; natural philosophy with corporeal and incorporeal substances (with further distinctions); logic with both the oratio continua, the domain of rhetoric, and dialogue, the domain of dialectics (§§ 14–17).

The second model of medieval divisiones scientiae is a schematic summary of what Aristotle partially states in Metaphysics (VI, 1, 1025b–1026a) and divides philosophy into theoretical and practical (or speculative and active).10 Once again, these main branches are further divided on the basis of their object of inquiry (Figure 5). According to Aristotle, in dealing with what is unmoved, eternal, and separate from matter, metaphysics (also defined as ‘theology’ or ‘first philosophy’) is the noblest science (Met. I, 1, 980a–982a).11

This second model has been embraced and transmitted by many authors, including the commentary to the Aeneid by Bernardus Silvestris.12 Boethius and the tradition that stems from his oeuvre stand out for their importance and influence. In the Consolation of Philosophy, the Π and Θ letters on the lower and upper hem of the robe, which Lady Philosophy herself wove (‘suis manibus ipsa texuerat’, Cons. Ph. I, pr. 1, in Boethius (2005)), refer to practical and theoretical philosophy and are connected by a ladder with an unspecified number of steps. According to the commentary by William of Conches, these steps symbolize the individual sciences in the order in which they should be studied: ethics, economy, and politics within the realm of practical philosophy; mathematics, physics, and theology in the domain of theoretical philosophy (it is worth noting that William’s commentary was the most widely circulated until the early fourteenth century and was probably read by Dante).13 Yet, throughout the Middle Ages, the steps of the ladder were also interpreted as the seven liberal arts, as seen in an illustration of a French manuscript of the Consolatio dating back to the second half of the twelfth century (Figure 6).

Here, the dual system devised by Aristotle and endorsed by William—whose commentary is copied within the same manuscript—is replaced by the traditional septenary of the liberal arts, articulated along the ascending pattern of the ladder to represent a ‘pédagogie des échelons successifs’ (Courcelle 1967, p. 33; see also p. 78).14 Reading downward, the liberal arts are: astronomy, music, geometry, arithmetic, rhetoric, dialectics, and grammar.

In the numerous classifications elaborated by Boethius—particularly in his commentary on the Isagoge—it is possible to discern two crucial developments. Firstly, the association of mathematics with the quadrivium.15 Secondly, the extension to logic, which coincides with the trivium (Marrou 1969, pp. 18–19). While Boethius does not extensively address the age-old issue of the status of logic—whether it was a tool or a part of philosophy—he includes it in his division, effectively transforming the Aristotelian bipartition into a highly successful tripartite model capable of subsuming all the liberal arts in an organic manner (Figure 7).

This model substantiates various works: it is notably mentioned in Isidore’s Etymologiae and in the eleventh-century Elementarium doctrinae rudimentum by Papia, and resurfaces in authors associated with the so-called ‘School of Chartres’, such as Clarembaud d’Arras and William of Conches. What is more, it features in the most influential division of knowledge of the twelfth century, elaborated by Hugh of St. Victor in his Didascalicon. Hugh maintains this tripartite structure, adding a fourth category devoted to the seven mechanical arts.16

Starting from the twelfth century and even more so in the thirteenth, with the massive introduction of Aristotelian works and the flourishing of university faculties, the two fundamental systems undergo all kinds of discussions, contamination, and expansion (the Florentine Dominican friar Remigio de’ Girolami rightly lamented an excess of fallacious subtlety).17

The partition in theory and practice shapes the influential twelfth-century De divisione philosophiae by Dominicus Gundissalinus (Gundissalinus 1903),18 which was based on a paraphrasis of al-Fārābi’s De Scientiis. Since al-Fārābi’s work has been influentially pointed out by Bruno Nardi as the model for Dante’s divisio scientiae (Nardi 1944, pp. 213–14), it is worth examining it in more detail.

In the De scientiis, al-Fārābi examines various fields of knowledge, ranging from Arab language to logic (including rhetoric and poetics), mathematics (including the quadrivium, mechanics, and optics), physics, and metaphysics. The final section addresses politics, law, and theology, with the latter being based on Islamic sacred texts. The result is a list of disciplines that are thoroughly described in their contents and parts but are not systematized within a comprehensive framework.19 The literal translation provided by Gerard of Cremona, to which Nardi referred, faithfully mirrors this original organization (Zonta 2001; al-Fārābi 2005, 2013).

Al-Fārābi’s work was enormously influential, establishing the classification of sciences as an autonomous doctrinal discourse, thus laying the foundations for a broader encyclopedia (Zonta 2001).20 Its list of disciplines significantly overlaps with Dante’s series of sciences, particularly in the juxtaposition of civil sciences and theology as the climax of the series. Yet, as Alfonso Maierù crucially pointed out, Dante’s approach differs from that of al-Fārābi’s. Most of all, the De scientiis is less a divisio than a syllabus, outlining disciplines to study in a certain order (Maierù 2004).21 Also, Dante’s list of disciplines reflects university graduate programs, where training in the liberal arts was foundational to advanced studies in natural and moral philosophy, law, and theology. As Fioravanti authoritatively states, Dante’s description of the system of knowledge ‘corrisponde, più che ad astratti modelli classificatori, alla strutturazione effettiva della formazione universitaria, almeno per anni intorno alla metà del ‘200’ (Fioravanti in Alighieri 2014, ad Conv. II, xiii, 7–8). It is not coincidental that Thomas Aquinas offers a similar program of study, wherein trivium and quadrivium are preparatory stages to philosophical disciplines, with logic preceding everything.22 Yet, I will argue that, while Dante’s framework may result in a similar educational structure, its aim differs fundamentally. His goal is to offer an objective description of knowledge rather than outlining a study program.

In basing his work on the De scientiis, Gundissalinus reframed al-Fārābi’s series of disciplines within the Latin tradition of classification of sciences, reorganizing them into a multi-layered taxonomy (Fidora 2021). In the De divisione philosophiae, he posits a fundamental partition between wisdom (divided into theory and practice) and eloquence (the Arts of trivium, poetics included),23 a concept that can be traced back to Cicero24 and is found in Isidore’s’ Etymologie, as well as in various twelfth-century texts, including the works of John of Salisbury and an anonymous Tractatus de philosophia copied in the MS BnF lat. 6750 (Dahan 1982). Its rationale is explained by William of Conches in his commentary on the Consolatio: wisdom and eloquence are two species of sciences, where one represents proper knowledge and the other allows for its articulation and transmission.25 More crucially, in Gundissalinus, the wisdom-eloquence partition is nested under a broader and more fundamental division that categorizes the honesta scientia into divina and humana, with the former coinciding with the Word of God contained in the Bible and the latter further subdividing in wisdom and eloquence (Figure 8).26

In the thirteenth century, the division between divine and human science formed the framework of many divisiones developed by both masters of the Parisian Faculty of Arts and members of the Dominican Order. Most prominently, it features in Vincent of Beauvais’ Speculum doctrinale (Morelli 1999–2000, pp. 176–77) and the influential De ortu scientiarum, composed around the mid-century by Robert Kilwardby and primarily inspired by Gundissalinus and Hugh of Saint-Victor’s Didascalicon (Gundissalinus 2007, pp. 40–43; the De ortu was probably available in Santa Croce in the MS Laur. Plut. 27 dex. 9). According to Kilwardby, divine science is essential for the salvation of the soul and is grounded in Revelation and the study of the Bible; it leads to wisdom (sapientia) and does not coincide with metaphysics (Alessio 2001, p. 111). Hence, metaphysics is a divine science only to the extent that ‘it speaks of God as substance par excellence’ (Maierù 2013, p. 360; Panella 1981), but it depends on reason rather than Revelation.

3. The divisiones scientiae in Dante’s Florence

3.1. Santa Maria Novella

Not surprisingly, the division between divine and human science also underlies the Divisio scientiae of the Florentine Dominican friar Remigio de’ Girolami. Likely dating back to the years of teaching in Santa Maria Novella between approximately 1285 and 1295 (Panella 1981),27 the Divisio responded to a request from fellow friars and turned out to be much longer than expected (‘prolixius deducta est quam a principio cogitarem’, Div. sc. 21). In the prologue, Remigio overtly addresses many of the problems involved in the classification of science, ranging from the choice of a two- or three-folded system through to the difference between theology and metaphysics. As for divine and human scientia, Remigio specifies that divine knowledge can be understood both as the knowledge that is in (and coincides with) God (‘que est in Deo, que quidem est idem quod ipse Deus’) and as the knowledge that comes directly from God (‘immediate a Deo’).28 In turn, knowledge from God is divided into modum inclinationis and modum cognitionis. In the first case, it coincides with one of the seven gifts of the Holy Spirit, and in the second with Holy Scripture, ‘which is called theology’ (‘sacra scriptura que theologia vocatur’; see Figure 9). On account of its certainty, theology is above any other science (‘super omnes alias scientias’).29 We will see to what extent this idea resonates in Dante.

In Santa Maria Novella, a further divisio scientie is found in a booklet, which was in the convent between the end of the thirteenth and the beginning of the fourteenth century, as a possession of a friar named Raynerius (Florence, BNC, Conv. soppr. G.III.451, ff. 1–8; Pegoretti 2020, pp. 126–32). These few folios contain a short version of the Compendium historiae in Genealogia Christi, a popular work by Peter of Poitiers. This work is based on the Historia scholastica by Peter Comestor and consists of intertwined texts and tree-map diagrams detailing the genealogy of Christ from Adam and Eve onwards. In the Florentine manuscripts, the Compendium is accompanied by a tree-map diagram of biblical books and an elaborated menorah, which is also present in other copies of the same work.31 The series of diagrams is concluded by a bipartite tree depicting virtues and science, both originating from God (Figure 10). Science is divided into eloquence, corresponding to trivium, and wisdom (sapientia), subdivided into theorica, practica, and meccanica [sic]. The fundamental partition between divine and human science is ignored.

A similar diagram is found in a copy of the Liber Introductorius by Michael Scot, BnF, Nouvelles Acquisitions Latin 1401, f. 30r (http://diglib.hab.de/?db=mss&list=ms&id=1-gud-lat&image=00067, 13 April 2024), a manuscript of Italian origins dating back to 127932 (on the verso of the same folio, another tree-map describes the vices originating from pride). Actually, Michael Scot composed a Divisio scientie largely based on Gundissalinus’s, whose fragments survive in the Speculum doctrinale by Vincent de Beauvais (Gundissalinus 1903, pp. 364–68, 398–400; Weijers 2005, p. 127). As we will see, in the same book, the liberal arts are also listed with the heavens. Yet none of the peculiarities of Michael’s division, such as his partition of practical philosophy into civilis and vulgaris, seems to emerge from the Parisian or Florentine diagrams, which reflect the twelfth-century tradition.

A comparison between the two manuscripts reveals some errors in the Florentine diagram, in which the branches devoted to the cardinal virtues of fortitude and temperance do not originate from virtus but directly from God. Moreover, it is worth noting how in the Florentine manuscript beneficia are nested under iustitia/liberalitas. This reminds us of the passage in which Dante presents his Convivio as an act of mercy, which is ‘mother of benefit’ (‘misericordia è madre di beneficio’, Conv. I, i, 9), relying on the Augustinian doctrine according to which mercy was related not to charity, but to justice (Gentili 2005, pp. 141–46).33

3.2. Brunetto Latini (and Santa Croce)

Brunetto Latini also articulates a divisio philosophiae both in his Rettorica and in the Tresor.34 Since chapter 17 of the Rettorica has no parallel in the fragmentary Ars rethorice from which Brunetto’s commentary derives (Alessio 1979), it is to be considered original work by the sponitore and stems from the need to position rhetoric within the broader system of knowledge. Its scheme—which is developed through a series of visual diagrams planned and realized as part of the text—divides into theory, practice, and logic, dealing respectively with divine and human things, behavior and action, and their motivation. Far from being tied to logic, rhetoric and the trivium as a whole exceptionally align with government and politics, which constitutes the third part of practical philosophy:35

pensando che lla scienza delle cittadi è parte d’un altro generale che muove di filosofia, sì vuole elli dire un poco che è filosofia, per provare la nobilitade e l’altezza della scienza di covernare le cittadi. Et provedendo ciò ssi pruova l’altezza di rettorica.

(Rett. V, 17, 5, in Latini [1912] 1968, p. 29)

As a result, dialectics is duplicated, appearing both as part of trivium and as a component of logic (Beltrami 1993, p. 119), together with sophistry and epideictics (Figure 11).

This scheme is replicated at the beginning of Tresor I, 1–5 (Latini 2007, pp. 4–17; Beltrami 1993). Here, the customary subdivision of theoretics into theology, physics, and mathematics is accompanied by definitions of each discipline adhering to the principle of the materiality or non-materiality of the object of study. According to this criterion, theology deals with separate substances, but, in reality, it gives access to the articles of faith: knowledge of God, faith in the Trinity, and faith in the law of the Church.

So far, Brunetto’s subdivision of logic into dialectics, epideictic (from which efidica/fidique derives: apodictica > epidictica > efidica/fidique), and sophistry has been detected only in an anonymous brief treatise contained in a manuscript housed in München, BSB Clm 331 (Beltrami 1993, p. 125; Latini 2007, xiv).36 Actually, I can point out that it also features in the widespread Elementarium doctrinae rudimentum by Papia (Dahan 1992, pp. 236–37) and in the rare Liber de philosophia Salomonis (Dahan 1985, pp. 150–52).37 In the copy of Papias’s lexicon housed in Santa Croce in Dante’s time (Speranzi et al. 2021, pp. 389–90), MS Laur. Plut. 27 sin. 3, f. 200v, it features as part of a divisio that the hand copying the text adds in the lower margin (Figure 12). Here, logic corresponds to ‘dialetica’, ‘epithetica’, and ‘sophystica’ (where the form ‘epithetica’ reveals further difficulties with this term).

A more refined image of Lady Philosophy illustrating the same divisio is in the MS BnF lat. 18275, f. 20r (https://portail.biblissima.fr/ark:/43093/ifdatac5eaf983a4079851b7c620b81b8faa7239b17c4c, 13 April 2024), dating back to the mid-eleventh century and including works by the Virgilian exegete Fulgentius (Albi 2021, pp. 32–33).

Brunetto’s innovative positioning of rhetoric and the trivium as a whole under the heading of politics (or scientia civilis) hardly derives from the tradition of the divisiones scientie. Rather, it seems influenced by Cicero (Bartuschat 2015, p. 46) and, more prominently, by the Aristotelian tradition, which promoted the “architectural”—i.e., foundational and framing—quality of politics (Suarez-Nani 1994, pp. 135–36). In the second book of the Tresor, which provides a translation of a compendium of the Nicomachean Ethics, Brunetto states that ‘li ars qui enseigne la cité governer est principal et soveraine et dame de toutes ars, por ce que desoz lui sont contenues maintes honorables ars, si come est rethorique et la science de fere ost et de governer sa menee’ (Tresor II, 3, 1); in the third book, the author adds that ‘rethorique est desouz la science de cité governer, selonc Aristotes dit en son livre qui est translaté ça’ (III, 2, 2). In order to promote politics to the status of a leading science, Brunetto integrates a further, albeit rare, subdivision into the Aristotelian-Boethian tripartite system, assigning to the trivium a position strategically aligned with his aims. Moreover, according to Beltrami (1993, pp. 125–26), the strong correlation he establishes between logic and practical philosophy marks a shift from a tripartite to a de facto bipartite system, where dialectics, rhetoric, and logic are intended as the ‘art of argumentation’ serving the practice of government (Imbach and Koenig-Pralong 2016, p. 58).

By overstepping the criterion that usually presided over the hierarchy of sciences, which is the importance and dignity of their object, Brunetto favored their societal role. The traditional preeminence of theoretical sciences and metaphysics is overcome in favor of a goal that the author shared with his readers: a ‘supranational lay audience’ (Beltrami in Latini 2007, viii), including men who served in government roles. As Imbach and Koenig-Pralong effectively point out, by focusing on his lay readership, Brunetto entrusts philosophy with a new practical function (see Imbach 2003, p. 48; Imbach and Koenig-Pralong 2016, pp. 57–61), which exceeds the boundaries of the schools and their teachings.38

4. The Liberal Arts and the Heavens

Indeed, the tradition of the divisio scientiae was pervasive, and Dante could not help but engage with it. Yet, when compared with it and with its closest counterparts in Florence, the poet’s depiction of the system of sciences appears radically different. Unlike coeval divisiones, which usually developed a treemap-like taxonomy with multiple exits, the poet crafts a linear hierarchy rooted in cosmological and theological principles (the heavens, angelic hierarchies, and the Trinity). As all commentators point out, the pairing of sciences and heavens is far from original and counts several antecedents. Rooted in the longstanding allegorical representation of the liberal arts, it reflects a minor tradition of enumeration of sciences found in some twelfth- and thirteenth-century texts, some of which are geographically and chronologically proximate to Dante.

In his encyclopedic work De naturis rerum, Alexander Neckam correlates the planetary heavens and the seven liberal arts (Johnston 1930): the Moon with grammar; Sun with dialectics; Mercury with rhetoric; Venus with arithmetic; Mars with instrumental music; Jupiter with geometry; and Saturn with astronomy.39 While the planetary order differs from Dante’s, even more interesting is the fact that Alexander establishes a rationale for each pairing: for instance, Venus resembles arithmetic because of their beauty, instrumental recalls armies and Mars, and so on. The spreading of this encyclopedia is testified by Dante’s contemporary and fellow-citizen Francesco da Barberino, who explicitly quotes Alexander’s list of sciences and planets (Francesco da Barberino 1982, vol. 3, pp. 280–81).40

As Thomas Ricklin pointed out, a similar list is presented by Michael Scot in his Liber Introductorius (Alighieri 1996, vol. 2, ad Conv. II, xiii, 8). Yet, Michael’s list is not identical to Dante’s, as there is a swap between geometry and arithmetic. More interestingly, however, the reasons for pairings are recognized but left to readers, who, by exercising their sharpness and knowledge in both the arts and astronomy, will clearly discern them:

Artes scripture sunt 7 causa planetarum 7 que sic eis attribuuntur: gramatica lune, dialetica mercurio, rhetorica veneri, arismetica soli, musica marti, astronomia jovi, et geometria saturno. Qui vero subtiliter discusserit naturam et speculationem tam artium quam planetarum, et gentium qui ipsas adiscunt […] omnino inveniet dictas artes planetis rationabiliter attributas.

(München, Bayerische Staatsbibliothek Clm 10268, f. 29v)41

A comparison according to properties is offered in the Tuscan vernacular treatise on the Composition of the world by Restoro d’Arezzo, which sketches a parallel in a chapter devoted to the calculation of time. The trivium is interrupted by the intrusion of music into the third heaven, while the remaining spheres are dismissed.

E anco saranno sette arti liberali e non più, sì che ciascheduno planeto avarà la sua: la più vile, come la gramatica, sarà de rascione del più vile planeto, com’è la luna, e la dialetica sarà de Mercurio, e Venere avarà la musica, e così ciascheduno avrà la sua.

(La composizione del mondo, II.viii, 6, Restoro d’Arezzo 1976)

Although faint, Restoro’s text exemplifies the extent to which the parallel between the seven arts and planets was widespread enough to surface in vernacular Tuscan encyclopedias.

While all these models are limited to the seven planetary spheres, Helene Wieruszowski pointed out that, in his Rhetorica novissima, Boncompagno da Signa extends the comparison further. In a long and intricate passage—which is ‘introduced as a cosmic “vision” revealed to Boncompagno, “the prophet”’ (Wieruszowski 1946, p. 221), and features as an example of transumptio42—up to sixteen sciences are paired with a likewise number of rotae and rotulae, which substantiate the ‘machina mundialis’. The liberal arts are followed by physica, ius civile, and ius canonicum. The series is closed by theology, with each domain corresponding to a rota; five rotulae follow, including alchimia, negromantia, and the like. Yet, many relevant differences with Dante are to be noted. If the reversal of geometry and music is not particularly concerning, more problematic is the fact that, from physics onwards, disciplines are labeled as ‘professions’, not to mention the total number of sixteen disciplines and the complicated system of rotae and smaller rotulae, whose roles and names are never explained. The comparison with the heavens is limited to the mention of the machina mundi, but it is not properly developed. Moreover, as Wieruszowski observes (Wieruszowski 1946, p. 223 fn. 27), the description of physics corresponds to medicine. We might add that its grouping with professions in civil and canonical law reminds us of the curricula at the University of Bologna, where Boncompagno used to teach. From this point of view, the classification of sciences by al-Fārābi, singled out by Nardi as Dante’s model, appears much closer and more relevant.

5. Cosmic Knowledge: Dante and the System of Sciences

As the survey above has hopefully made clear, any attempt to align Dante’s divisio with traditional treemap-like divisiones seems futile, if not misleading. Moreover, existing comparisons between the sciences and the heavens fail to fully elucidate the poet’s proposal. Indeed, for his divisio scientie, Dante drew inspiration from various precedents but did not fully endorse any of the available models. Moving forward, I will explore the central tenets of Dante’s system of knowledge, their sources, and their scope, beginning with the Empyrean and divine sciences.

5.1. Divine and Human Science

The definition and profile of theology in Dante’s Convivio ignited intense debates among scholars, mainly resulting in its identification with faith and biblical Revelation.43 The Divisio scientie by Remigio de’ Girolami, inspired by Gundissalinus and Kilwardby, is of paramount importance in resolving this interpretative crux. Dante’s conceptualization of theology as a divine science (‘la scienza divina, che è Teologia appellata’, Conv. II, xiii, 8) closely mirrors Remigio’s view of ‘scientia divina’, which is equated with the ‘sacra scriptura que theologia vocatur’ (Div. sc. 2, in Panella 1981). According to Remigio, this divine science, distinct from human science, coincides with both Revelation, conveyed by the Bible, and scientia as a gift of the Holy Spirit. Similarly, a thorough examination of Dante’s definition of the ‘divina scienza’ and of its equivalence with divine peace within their evangelical context—as explicitly referenced by the poet—unveils the profound interconnection of Revelation and the Pentecostal bestowal of the Holy Spirit in the Convivio. As I already argued elsewhere, by referring to the Gospel of John’s account of the Last Supper, Dante postulates a clear, albeit oblique, connection between peace, divine teaching, and the Holy Spirit, which enables the Apostles to comprehend the doctrine of Christ (Pegoretti 2018, pp. 173–85).44 Indeed, Remigio and Dante seem to share the same complex and multifocal conception of divine science, one that binds together as consubstantial the divine message and the Pentecostal charisma.

Furthermore, the partition of scientia into divina and humana ensures that Remigio’s peculiar ‘theology’ holds a status apart, removed from any kind of human intervention. From this point of view, the association that Dante establishes between divine science and the motionless Empyrean (‘lo Cielo empireo per la sua pace simiglia la divina scienza’) is particularly illuminating. The Empyrean—which the poet considers an immaterial sky, ultimately coinciding with God himself45—stands apart from the series of the astronomical heavens, epitomizing a separate and divine level of knowledge, remote from any controversy. As Thomas Ricklin states, ‘contrairement à la philosophie morale du ciel cristallin, qui est la condition nécessaire pour l’exercise des autres sciences, la théologie de l’Empyrée n’a aucun rapport actif avec les autres domaines du savoir humain’ (Ricklin 2002, p. 134).46

As prominent scholars have pointed out, in promoting moral philosophy as the highest amongst human sciences, Dante aligns himself with Brunetto (Imbach 2003, p. 48; Imbach and Koenig-Pralong 2016, p. 67), whose teachings enhanced the architectural role of politics, outlined by Aristotle and emphasized by numerous contemporary thinkers, including Albert the Great and Thomas Aquinas (Suarez-Nani 1994, pp. 135–36; Fioravanti 2014, pp. 60–61). Yet, a contradiction emerges in the fact that in Book III, Dante notoriously advocates for the excellence of metaphysics: ‘la quale, perché più necessariamente in quella [la Filosofia] termina lo suo viso e con più fervore, [Prima] Filosofia è chiamata’ (Conv. III, xi, 16). Indeed, the two statements seem to serve different purposes. In the wake of Étienne Gilson and Francis Cheneval, Gianfranco Fioravanti pointed out that the structural primacy of ethics declared in Book II underpins the Convivio’s underlying project of reorienting the culture of Scholasticism towards a new audience and a mission of political renewal. Such a strategy does not automatically undermine the absolute and traditional preeminence of metaphysics and of individual contemplative life (Fioravanti in Alighieri 2014, p. 60).

Within this framework, however, we should bear in mind the nature of metaphysical speculation in the Convivio and the extent to which Dante emphasizes the limits of metaphysics, ultimately being unable to grasp the essence of God. According to Luca Bianchi, ‘there is a deep relationship between [Dante’s] project of vernacularization of philosophy and his sceptical attitude towards the idea that philosophy is a form of wisdom culminating in the metaphysical analysis of God’s mind and the contemplation of his quiddity’, to the point that the ‘bread’ he serves in his banquet ‘has a low metaphysical content’ (Bianchi 2013, pp. 353–54). In this regard, the separate placement of scientia divina allows for the incorporation of metaphysics within the realm of scientia humana and a clear separation between the two forms of knowledge.47

5.2. Heavens of Knowledge

As we have observed, the tradition of associating sciences with the heavens was not insignificant and was well established in Dante’s Tuscany, as evidenced by Restoro d’Arezzo and Francesco da Barberino. Although none of the available models perfectly align with Dante’s sequence, they offer not only a general framework but also some distinctive characteristics. Albeit feeble, reasons for each pairing are provided by Neckam, while Dante seems to actually articulate those subtle rationales that Michael Scot expected his learned readers to discern. By alluding to the cosmological system, Boncompagno offers an intermediate model extending knowledge up to the Empyrean. Yet, as far as I can see, none of the predecessors integrate further components, such as the angelical hierarchies and the Trinity, within the scheme, a point that scholars seem to overlook and that further reveals Dante’s ambitious program.

Francis Cheneval disclosed the extent to which the structure of the first three books of the Convivio mirrors the introductions to philosophy traditionally delivered at the Faculties of Arts to inaugurate courses. The divisio scientie was a mandatory part of them, together with the discussion of the impediments to knowledge and a praise of philosophy (commendatio philosophiae), respectively developed by Dante in the first and third Book of his work (Cheneval 1998, pp. 359–62).48 In realizing his ‘translatio philosophiae’ (p. 354) from university settings to a lay audience, Dante models the structure of his discourse on institutional practices, innovating from within.

When compared with the prevailing and longstanding tradition of divisio scientiae, well attested also in Florence in both the religious schools and lay culture, Dante’s adoption of a minor tradition associating sciences and heavens appears as conservative as deliberate. The effort to align ten heavens, nine angelical hierarchies, and eleven disciplines is particularly noteworthy, especially considering that, as Étienne Gilson already pointed out, none of the sequences was stable in the late Middle Ages (Gilson [1939] 2016, pp. 100–13). Dante himself acknowledges the diversity of opinions on the number and the order of the heavens (‘del numero delli cieli e del sito diversamente è sentito da molti’, Conv. II, iii, 3). Even the order of the planetary spheres was subject to debate: while the Caldean sequence (Moon, Mercury, Venus, Sun) was widely accepted, the Egyptian one (Moon, Sun, Mercury, Venus)—endorsed by Plato, Aristotle, and Macrobius—persisted (Eastwood 2007; Lerner 1997). Notably, in his Documenti d’Amore, Francesco da Barberino adopts both sequences in two different diagrams (Egidi 1902, p. 83). Also in this case, Dante’s choice appears well-informed: ‘[Aristotile] credette che lo cielo del Sole fosse immediate con quello della Luna, cioè secondo a noi. […] Veramente elli di ciò si scusa nel duodecimo della Metafisica’ (Conv. II, iii, 4). As regards the angelical hierarchies, the Gregorian sequence presented in the Convivio is notoriously replaced by the one by Dionysius in Paradiso 28. Also, the order of the liberal arts was variable.49

Indeed, Dante’s arrangement offers some advantages. As Thomas Ricklin observed, it allowed Dante to reconfigure the astrological influence, which presided over his love for the donna gentile (Voi che ‘ntendendo il terzo ciel movete) as a philosophical system: ‘[Dante] reconstruit le système cosmique qui, par la planète Vénus, était originairement responsable de son amour pour la donna gentile […] sous la forme d’un système des sciences’ (Ricklin 2002, p. 134).50 What is more, it paved the way for a different assessment of disciplines. During the twelfth and thirteenth centuries, the rise of the Aristotelian epistemological framework generally confined the liberal arts to a preliminary learning step or fragmented their subjects into more specialized programs (Delhaye 1969; Kibre 1969; O’Donnell 1969; Roos 1969; Bianchi 1997). We have also seen how trivium and quadrivium were nested within rational philosophy and mathematics. On the contrary, Dante’s scheme, with each liberal art marking a distinct step in the hierarchy of knowledge, emphasizes their role, possibly in polemics with the vision of knowledge conveyed by universities. From this point of view, the proximity of Dante’s heavens with the steps of the arts on the ladder of Lady Philosophy, outlining an ascending path through disciplines (as seen in some readings of Boethius’ Consolation of Philosophy), is remarkable. It is not inappropriate to interpret such a move as deliberately antagonistic to institutional knowledge and teaching practices.

If we delve further into this choice, profound implications become evident. In dealing with the institutional practice of the divisio philosophiae, Dante adopts a minor and rather eccentric tradition that allows him to reframe knowledge in a cosmic setting. Before articulating each pairing, Dante provides a justification to the general association between science and the heavens, on the basis of three reasons: (1) both heavens and sciences move around a firm point; (2) both ‘illuminate’; (3) both bring realities to perfection (Conv. II, xiii, 2–6). Notably, he is the first to do so (Fioravanti in Alighieri 2014, p. 311). Moving forward, each pairing is justified on the basis of the intrinsic qualities of both disciplines and the heavens. According to the aforementioned principle that ‘scientia secatur in res’, and that sciences do not demonstrate their object (Anal. Post. I, 9, 76a, 16–17), divisiones were usually based on the intrinsic qualities of sciences and their domain. In explaining the first reason of his comparison, Dante alludes to this same principle (‘nulla scienza dimostra lo proprio subietto, ma suppone quello’, Conv. II, xiii, 3). Yet, the association of sciences with the skies adds an intrinsic and essential quality to their disposition. To some extent, Dante presents his divisio not as a classification, but rather as a description of factual data. For the sake of argument, we could even say that, for disciplines to switch, the universe and the heavens should change. As Suarez–Nani argues, correspondences ‘aboutissent à un “système” du savoir légitimé par le systeme cosmologique de référence: l’ordre des sciences reproduit des lors l’ordre de l’univers’ (Suarez-Nani 1994, p. 133). With their spherical, enclosed, and harmonious arrangement, the celestial spheres provide both a perfect framework for the organization of the various disciplines and a global, cohesive, and unifying image of knowledge and wisdom; they epitomize the enkyklios paideia, the ‘circle of knowledge’, as Giuseppe Mazzotta effectively condensed (Mazzotta 1993), as opposed to the professionalization of learning and teaching. Indeed, such recognition aligns perfectly with Theodore Cachey’s insightful observation about the ‘underlying cosmological framework of the Convivio’ (Cachey 2018, p. 61), emphasizing its structural significance and its ‘cosmically inflected anxiety’ (p. 65), raised by the displacing circumstances of the exile.

Many attempts have been made to trace signs and remnants of the correspondences between the sciences and heavens in Paradiso.51 Yet, the results are too discontinuous, and no evidence emerges of a systematic implementation of the scheme from the Convivio. Only in a few instances do episodes and explanations in different heavens implicitly remind us of certain liberal arts, such as geometry in the heaven of Jupiter or, possibly, rhetoric in the heaven of Venus (e.g., Mazzotta 1993, passim). Most crucially, in dealing with the Commedia, a reader who would not know the (unfinished and unpublished) Convivio had no way to detect such a systematic vision of knowledge and the cosmos. Actually, I think that the most vital outcome of the divisio scientiae in Convivio II and its legacy to the poem is precisely its general framing of human knowledge within the cosmos, as depicted above (Figure 1). While in his autocommentary, Dante painstakingly aligns the system of knowledge with the cosmos and its metaphysical and theological components, in Paradiso, he literally jumps into the picture, describing it from within.

According to Cachey, when Dante moved on to the Commedia, ‘he shifted not so much the content of his cosmology as his approach to cosmology’: by drawing on twelfth-century cosmological poetry, he was able to ‘participate directly in the cosmos by poetic means rather than describe it from the outside in a scientific mode’ (Cachey 2018, p. 73). This interpretation aligns with the transition from the cosmic system of knowledge presented in the Convivio to the flight of the pilgrim Dante across the heavens (pp. 75–76). A flight that echoes the chariot of Phronesis in Alain de Lille’s Anticlaudianus, built and escort through the heavens by the personifications of the Liberal Arts alongside Reason and Theology (Arcoleo 1969). On the same line, Martianus Capella’s De nuptiis Philologiae et Mercurii Philology represents the knowledge of all the arts—including liberal, but also theurgic and mantic arts—that can lead the soul to re-join the divine reason.52

Building on the legacy of cosmological poetry and driven by greater ambitions, Dante overcomes mechanical correspondences to project himself as a celestial traveler who dynamically gains vast knowledge about himself and the world while flying to the ultimate source of everything, the ‘principio fontale’ (Conv. III, xiv, 5). His cosmic flight involves the whole of the cosmos and knowledge, whose profoundly unifying and interconnected fabric is devised and weaved by a poet through poetry.53