Switching Style towards Ease in a Medieval Textbook of Medicine: Revison of Constantine the African’s Pantegni, Theorica in Twelfth-Century Manuscripts

Abstract

Of north-African origin, the Benedictine monk Constantine the African (d. by 1098/9) was the conduit through which Arabic medical works were first transmitted to the Latin West. Having migrated to Salerno in Southern Italy by 1076/7, he began a remarkable translation program, which he subsequently continued in the monastery of Monte Cassino.¹ His most important translation, the Pantegni, was based on the Kitāb al-malakī by the Persian physician ʿAlī ibn al-ʿAbbās al-Mağūsī (commonly Latinized as Haly Abbas, died after 978).² The original Arabic work consisted of two individual parts: Theorica, an intro duction to medical theory including discussion of the four humours, and Practica, with remedies and therapeutic strategies for the physician.

Each part contained ten books. The Latin translation was modelled on this same structure, but was not an entirely faithful translation, especially in the second half of the text, only partially translated by Constantine.³ The Pantegni rapidly gained a pivotal status as the central textbook at the first European medical schools and universities. Other Salernitan medical authors and later encyclopaedists frequently cited and quoted Constantine’s work, and it had influence on the works of, for example, William of Conches.⁴ Around 130 manuscripts of the Pantegni survive—a remarkable number—and the work was printed twice in the sixteenth century.⁵

Yet in spite of its durable success, only a few decades after Constantine’s death his Pantegni was challenged both by a new translation of the original Arabic text and also, it would seem, by a revision of his own translation. In 1127, Stephen of Antioch heavily criticized Constantine for plagiarism and excessive abridgement and revision of Haly Abbas’s original Arabic text, and put together an entirely new, more literal (and complete) translation: this is known as the Liber regalis, or the Regalis dispositio.⁶ Constantine’s Pantegni, or more precisely its section Theorica, was joined by the work of another author, who circulated an emended text: a revised version of the Theorica, with numerous additional short passages originating mostly, it appears, in Haly Abbas’s Arabic, emerged in the first half of the twelfth century at the latest.⁷ What is more, this version also involved a stylistic revision of the text, the focus of the present paper. In the course of time, although the version did not gain a firm foothold as such, many of its readings made their way into a subsequent branch of the transmission, to a degree that varies from book to book and chapter to chapter.⁸ Ultimately, one of Theorica’s two sixteenth-century editions, that printed at Lyon in 1515 (hereafter the Lyon edition),⁹ reproduces features of the revised version, as will be demonstrated below. The ambition of the present paper is to demonstrate and assess the nature and extent of the stylistic revision and to illustrate the redactor’s approach to the text: their editorial strategies, and their reasons for embarking on what amounted to a large-scale revision of Constantine’s magnum opus.

Previous studies

Of some 130 copies of the Pantegni, around eighty, whole or partial, manu scripts convey the Theorica. Some of them also include the Practica or parts of it, while others contain only the Theorica.¹⁰ My collations of a total of fifty-nine Theorica manuscripts and of the two sixteenth-century editions (Lyon 1515, Basel 1539)¹¹ have shown that three early manuscripts of the twelfth century stand out from the others in having a peculiar style and short additional passages that are missing from the more typical textual version (hereafter referred to as the standard version).¹² The manuscripts in question are Basel, Öffentliche Bibliothek der Universität, MS D.III.17 (s. xiimed., Italy?); Paris, Bibliothèque nationale de France, MS lat. 7042 (s. xii1–xiii1); and Rome, Biblioteca Universitaria Alessandrina, MS 171 (s. xiiex.). These manuscripts and their textual version stand at the centre of the present study. Even though the standard version appears to have been domi nant also in the thirteenth and fourteenth centuries, the text of the Theorica was transformed by the effect of the twelfth-century revision. In a branch of the standard version, expressions deemed better as well as passages adding substance were picked from the revised version and merged.¹³ The revised ver sion retained significance until the point in the sixteenth century when it was empanelled in print, in what is still a standard edition of the text: the Lyon edition of 1515 reproduces the revised version in some chapters and combines and edits readings from both versions.¹⁴

In identifying the revision of the Theorica, my study of another Pantegni manu script, and especially its marginal additions, has been crucial. The manu script in question is Helsinki, National Library, MS E.ö.II.14 (s. xii3).¹⁵ Principally, the manuscript represents the standard version of the Theorica (or, for Book V, an incomplete ‘Ur’-version).¹⁶ Yet its text was supple mented with additions in the margins. The Helsinki manuscript was copied by two contemporary scribes, who drew on two (or more) different exemplars: the second scribe took over from the first in the middle of Book IX of the Theorica and wrote to the end. The second scribe also made additions here and there in the margins to earlier portions of the text. These additions number 105 in total, and their length varies from one word to a pas sage of more than seventy. In previous studies I have shown that the additions derive mainly from, and are typical of, the twelfth-century revision, for these passages are missing from the standard version.¹⁷ Further collations of the additions with Stephen of Antioch’s literal translation and, for Book V, directly with Haly Abbas’s original text in Arabic, indicated that most additions stem from the Arabic original.¹⁸ Constantine’s habitual method of translation, apparent also in his other works, was to abbreviate and paraphrase: he often omitted individual words and passages, while retaining the meaning.¹⁹ In contrast, it appears that our redactor restored passages that Constantine had omitted. Constantine’s original omissions were usually intended to avoid repetition and redundancies.²⁰ Also, in these cases the redactor’s aim seems to have been to bring the text of the Latin Theorica closer to Haly’s original text. In the end, both the restoration of pas sages and the stylistic revision point in the same direction: the redactor’s ambition to clarify Constantine’s expression.

The different stylistic features of some witnesses of the Theorica is some thing that has been noticed in previous work. The extent, dating, and the signifi cance of this textual variation has not, however, been discussed. Gudrun Vuillemin-Diem noted the differing style of MS Paris lat. 7042 in her col lation of Theor. III.13 De oculis with it and the Basel edition of 1539, and another manuscript from Paris (BnF, lat. 14393). According to her, ‘[MS Paris lat. 7042] bietet eine zwar inhaltlich entsprechende, aber stilistisch relativ stark abweichende Redaktion’.²¹ The volume edited by Charles Burnett and Danielle Jacquart in 1994 included many important individual contri butions, two of which are especially pertinent to this paper.²² Mark Jordan observed that the Basel edition descended ‘from a more typical exemplar [than the Lyon edition]’, and that the Lyon edition depended ‘more importantly on inferior and idiosyncratic manuscripts’.²³ This showed awareness of the existence of one or more revised versions alongside the standard version. Jordan did not, however, identify the earliest manu scripts containing the revised version, attributing the revisions instead largely to the Lyon editor, Andreas Turinus de Piscia (1473–1543).²⁴

In the same volume, Charles Burnett edited Theor. IV.19 De spiritibus from manuscripts representing two major branches of transmission. He described the style of the second branch as a ‘revised … more flowing and “open” style of Latin’, noting, for example, ‘perfect passive participles … changed into finite verbs’.²⁵ This resonates very well with my own observations on the style of the revised version of the Theorica.²⁶ Burnett did not, however, consider the three twelfth-century manuscripts at the centre of the present study, but he drew on later manuscripts from the thirteenth and fourteenth centuries. These were MSS London, British Library, Sloane 3481 (s. xiii); Paris, BnF, lat. 6885 (s. xiv1) and nouv. acq. lat. 1485 (s. xiv), and the Lyon edition. My collation of that same chapter (Theor. IV.19) with the said three twelfth-century manuscripts, Basel, Paris lat. 7042 and Rome, showed that the principal differences noted by Burnett in his second branch against the standard version had already emerged in the course of the first half of the twelfth century.²⁷

Hence, Burnett’s results are demonstrative of the impact of the twelfth-century revision on thirteenth- and fourteenth-century transmission and on the Lyon edition. What is also significant, Burnett pointed out a passage that is not yet present in the twelfth-century manuscripts: a long quotation from the pseudo-Galenic work De spermate, inserted in Theor. IV.19 in Burnett’s second branch.²⁸ As far as the manuscripts bear it out, there was an interest in emending the Latin Theorica with the introduction of new materials in the thirteenth century, possibly also in the fourteenth. This study concen trates, however, on stylistic revisions made within some decades after Constantine’s death.

In what follows I will first discuss the extent of the stylistic revision of the Theorica, and introduce the manuscripts and the two sixteenth-century editions collated for this study. Then the evidence for the propositions outlined above will be presented through collation of the standard version, i.e. the version authored by Constantine himself, and the revised version of a chapter of the Theorica. In order to demonstrate that the editorial strategies observable in the selected chapter were not exclusive to it, I offer selected parallel examples from elsewhere in the Theorica; these are collated in the same fashion. Finally, I will discuss the authorship of the revision and its intended use.

The extent of the stylistic revision of the Theorica

The evidence for revision is stronger in the first half of the Theorica than in the latter half. This applies to both the ‘extra’ passages as well as to the gram matical streamlining. As to the extra passages, all three of the twelfth-century manuscripts at the centre of the present study include them in the first half of the work (Books II to V).²⁹ In Book VI, MS Paris does not include as many of the additional passages as MSS Basel and Rome. In Books VII to IX, only Rome includes additional passages (the manuscript is mis sing most of Book IX, and Book X completely).³⁰ Also, in the case of the stylistic revision, sample collations between our three twelfth-century manu scripts and the standard version showed that the evidence for revision is weaker and less consistent in the latter half of the Theorica (even though the measure of adaptation can vary from chapter to chapter in the first half as well). This can indicate either contamination in our manuscripts, or that the revising of the latter half was never completed.

For these reasons, study of the revision is better focused on the first half of the Theorica. The chapter selected for collation derives from Theorica’s Book V. The reason for selecting this particular book is firstly that it is present in all of the three manuscripts which preserve the revised version (Basel, Paris and Rome), as well as in MS Helsinki, the insertions in which have been the starting point for my study of the revision. Helsinki lacks Book I, whereas Basel lacks Book II in part and Book III entirely.³¹ Secondly, Book V has been at the centre of my research, as it bears evidence for an earlier revisional stage besides the twelfth-century revision, namely, the comp letion of the preliminary stage (the Ur-version).³² Book V may have been the first book of the Theorica Constantine translated, and it appears to have received the most revision. The book deals with one of the most important themes of the work, the res non naturales, ‘the non-natural things’, affecting health.³³ It is the second longest book of the Theorica (only Book IX, dealing with various ailments in different parts of the body, is longer). The non-natural things are described as the external or behavioural factors that can affect the mixture of humours and qualities in the body, whether for good or ill: air and environment, food and drink, motion and rest, sleep and wakefulness, evacuation and repletion (e.g. baths, inter course, and digestion), and emotions.³⁴ The chapter whose collation I present is Theor. V.34 De somno et vigiliis, ‘On sleep and wakefulness’.³⁵ The text-critical value of collating one full chapter of the Theorica is that it consti tutes a narrative and as such it demonstrates in a straightforward way how the revision affected Constantine’s text. As a result, one can more accurately appreciate the editorial approaches of our anonymous redactor.

The manuscripts collated: the standard version and the revised version

For the standard textual version of Theor. V.34 De somno et vigiliis, The Hague, Koninklijke Bibliotheek, MS 73 J–6 will always serve as our base text (cited here by the siglum Ha). Written in the monastery of Monte Cassino between about 1075 and about 1085, the volume is not only the oldest known survivor but it also stands in close proximity to Constantine himself. Erik Kwakkel and Francis Newton have argued persuasively that the manuscript was produced under Constantine’s own supervision, and as such its text can be taken to represent his authorial voice more closely than other manuscripts.³⁶ The Hague manuscript is of good textual quality, in that obvious errors are very few, abbreviations are not numerous, and the punc tuation is consistent and sound. The original eleventh-century unit contained the Theorica almost to the end of Book X. In the middle and in the second half of the twelfth century, the text of the Theorica was completed and recipes and excerpts from other medical texts added.

In order to demonstrate how the standard version was transmitted in the twelfth century, variant readings have been collected from two early manu scripts, both of high textual quality. These are London, British Library, MS Add. 22719 (cited here as L) and Cambridge, Trinity College, R. 14. 34 (906) (cited here as C).³⁷ Of the two, L is earlier, datable to c. 1110–1130. It con tains the Theorica and the so-called Ur-Practica, i.e. Books I–II and the first part of Book IX,³⁸ as well as other shorter texts and some pharmaceutical recipes. According to Charles Burnett, the recipes have an English origin, and so does the manuscript: it was possibly written at Bath abbey (Somerset). The manuscript was carefully produced, and the rubrication and the use of colours, such as green dots, direct the reading consistently.³⁹ C was in the possession of Bury St Edmunds abbey (Suffolk) in the twelfth century (as was another copy of the Pantegni, now lost).⁴⁰ The text is finely writ ten, and it was apparently compared and corrected with at least one other exemplar. The manuscript contains the Theorica only.

Another source for variant readings that demonstrate the standard version’s transmission is the Basel edition (1539) (siglum ba).⁴¹ My colla tions have shown that its text follows the Hague text and other early manu scripts closely.⁴² Mark Jordan noted in the 1990s that ‘The Basel edition … agrees with readings in a much larger family of manuscripts [than the Lyon edition]’, and that ‘the Basel edition descends from a more typical exemplar’,⁴³ thus corroborating my own view that the Basel edition can be seen as representing basically the standard textual version of the Theorica.

For relations between HaCLba, it will be sufficient to state the following. L contains a text very similar to Ha, devoid of only those few errors that our base text has, and committing a few errors and choices of its own. C differs from Ha more than L does. It is free of a few of Ha’s errors, yet it shows trans positions of words and phrases and other minor differences, such as varia tion between singular and plural (e.g. ut cibus digeratur HaLba against ut cibi digerantur C; and corpus cibis exinanitum HaLba against corpus cibo exinanitum C), or chooses a different word (e.g. illa HaLba, alia C; and ex noctium prolixitate HaLba, ex noctiua prolixitate Cp.c.). As to our collated chapter V.34, in three cases the scribe of C, or a corrector, changed the word order ‘back’ to correspond with the word order in HaLba. The correcting hand also writes alternative singular/plural forms or verb moods supra lineam. As to ba, its text is very close to that of Ha, correcting, however, the few errors found in Ha and making a few errors and choices of its own (different to those of L).

As to the revised version, our base text is Basel, Öffentliche Bibliothek der Universität, MS D.III.17 (cited here by the siglum B). It is the oldest witness for the revision of our sample chapter: the manuscript, in a protogothic hand, belongs to the first half or middle of the twelfth century. The prove nance is almost certainly continental Europe, probably Italy.⁴⁴ The lack of capitula numbers and rubrication indicates that the manuscript is unfinished.⁴⁵ The manuscript contains only the Theorica of the Pantegni. Rome, Biblioteca Universitaria Alessandrina, MS 171 (siglum R) was made towards the end of the twelfth century, the work of three scribes. The manu script contains the Theorica of the Pantegni, ending, however, at Book IX, chapter 2. The upper half of the volume is badly damaged from roughly the half-way point of Book VII to the end, and several folia are misplaced. The manuscript lacks chapter headings and capitula numbers.⁴⁶

The third manuscript witness to the revision is Paris, Bibliothèque nationale de France, MS lat. 7042 (siglum P).⁴⁷ The codicology is more compli cated than is the case with the other manuscripts. P was written by three or more hands datable to the first half or middle of the twelfth century. The first hand betrays features of late Caroline minuscule; the others can be categorized as protogothic. In addition, a later hand of the beginning of the thirteenth century has written one replacement quire, covering the latter part of Book V and about half of Book VI. Our collated chapter V.34 De somno et vigiliis (fol. 51r–v) happens to be part of the stint of that later hand. None the less, it properly represents the revised version. The impli cation is that the exemplar was the same, or else it belonged to the same branch of the transmission.⁴⁸ What is perhaps most significant is that the first two hands may be even older than the hand of the Basel manuscript, thus pushing the terminus ante quem of the Theorica’s textual revision still earlier into the first half of the twelfth century.

The fourth source for variant readings for the revision is the Lyon edition (1515) (siglum lu).⁴⁹ It reproduces sometimes the standard version, some times the revised, depending on book and chapter, and it also combines and edits readings from both versions.⁵⁰ As to our collated chapter V.34, the text of the Lyon edition is very much that of the twelfth-century revision.

As regards the relations between the witnesses for the revision, i.e. BPRlu, a brief observation will suffice here. Our base text B commits errors and omis sions which the other three witnesses do not reproduce. P is the most original in its readings. R is closer to B on the one hand, but sometimes R deviates from all the others and agrees with the standard version (i.e. HaCLba). It seems that the copyist(s) of R used at least two exemplars: the text basically represents the revised version, but at times readings deemed better were chosen from the standard version. As to lu, it is closest to P, agreeing with it and frequently standing against BR.

Textual variation within the group BPRlu implies that none of the witnes ses descended directly from another. It must be emphasized, however, that in cases where the revised version differs most significantly from the standard version—i.e. in the cases which are the most relevant to my argument—these four witnesses are in agreement with each other.⁵¹ The fact that they all share the elements of the stylistic revision suggests that the revision was the result of a systematic reworking by a single redactor, rather than being the accumulation of various layers of scribal interventions over time in the course of transmission.

Theor. V.34: the standard and the revised version

The text of V.34 is given in Table 1 in both versions. Version 1 is given from Ha in the left-hand column, Version 2 from B in the right. Variant readings are supplied in the case of Version 1 from L, C and ba, for Version 2 from P, R and lu. In each case, the orthography of the base text has been followed. Obvious errors (which are only a few in the case of Ha) have been corrected with the help of the other witnesses; such interventions are indicated in the apparatus. Differences of word order in each version are noted, whereas ortho graphical variation goes unrecorded. For Version 1, punctuation and capital letters to indicate new sentences follow the practice of Ha, except for a very few instances in which I have slightly modified them to make the text flow better, at least for a modern reader. For Version 2, there is rather more in the way of modification of punctuation and use of capital letters for new sentences. The principal differences between the standard and the revised version are highlighted in boldface. Section numbers have been added to facilitate navigation and to provide for referencing.

Textual and stylistic analysis

In general, the revised version is more verbose than the standard version: the edited chapter in the standard version has 436 words, the revised version has 475. There are only two instances in which the revised version adds new substance, absent from the standard version: in multum dormientibus [4], and quod in dormiente maxime facit [6], counting as eight words altogether.⁵² All the remaining extra words of the revised version belong to different stylistic and grammatical expression, such as adding discourse connectives (such as aut, autem, sed, uero), or streamlining the syntax by loosening compressed grammatical constructions (such as turning present participles into subordinate clauses). In terms of expression, the revised version is simpler and flows more easily. To encapsulate the editorial strategy of our anonymous redactor, one might describe it as simplification by amplification. In what follows I discuss some of the principal features that differ between the two versions.

• (1) Aut–aut constructions

Perhaps the most striking difference between the Theorica’s standard and revised version is the abundant use of discourse connectives in the latter. The first category is aut–aut (‘or’) constructions, which abound in the revision when introducing lists or alternatives. The standard version prefers uel (‘or’), or other particles, or no word at all. For example, secundum quantitatem temporis, secundum magnitudinem materię (the standard version), as against autsecundum quantitatem temporis,autsecundum magnitudinem materię (the revised version) [5]; si cibusuelhumores sint indigestibiles, as against si cibumauthumores inuenit indigestibiles [6].⁵³

It is true that the standard version does not entirely lack aut–aut constructions (see, for example, the beginning of section 2); neither does the revised version totally abandon the use of uel and siue. Yet the use of aut in the revised version is so substantial that it indicates a conscious preference for that conjunction when introducing lists or alternatives. Our collated chapter has five occurrences of aut in the revised version against two occurrences in the standard version (see sections 2, 5, and 6). Theor. IV.7 De causa mortis ‘On the cause of death’, for example, has around thirty-five occurrences of aut in the revised version against none in the standard. Aut acts almost as a signalling word in the revised version, designating lists or the introduction of alternatives, thus outlining the text and bringing clarity.

• (2) The substitution of the word unde

Another feature of the revision in relation to discourse connectives is the change from unde (‘whence’, ‘hence’, ‘therefore’, ‘for that reason’) to et propter hoc (‘and because of that’, ‘and for that reason’), or some other formulation. Unde is not entirely absent from the revised version, but its field of application is narrower than in the standard version. In our sample chapter, the standard version has five occurrences of unde [sections 2, 3, 4, 6, and 7]. Two of these are replaced in the revised version: to et propter hoc [6], and to postque [7].

The redactor preserved the word unde in three instances when it was used in the sense of ‘hence’ [2, 3, and 4]; in such a case the operative function is merely to bind the sentences together. The redactor also allowed the word unde to stay when its grammatical and/or semantic antecedent was sufficiently explicit. For example, Neque etiam causas ignoret [medicus], unde mutetur aer (‘A doctor should not ignore causes from which the air changes’) [V.2]; Magnum est medicamentum qualitates aeris […] cognoscere […] Vnde ab aere incipiendum (‘It is a great remedy to know the qualities of the air […] From which [reason] we shall start from the air’) [V.2]. In contrast, the redactor appears to have substituted the word unde, when the causal sense of the word and its relation to what has been written above is not sufficiently transparent. In these cases, more explicit formulations, such as et propter hoc (‘and for that reason’), and et ita (‘and therefore’),⁵⁴ are used in the revised version. This semantic variation is sometimes only very slight, and, in the first instance, the word unde carries the meaning ‘for that reason’. Yet the frequent substitution of the word unde reveals something about our redactor’s sense of narrative: by substituting unde he or she again attempted to clarify syntactical relations of the sentences and to make the course of Constantine’s argumentation as lucid as possible.

• (3) The augmentation of other discourse markers

There are many other conjunctions and particles which have been added to the revised vesion here and there. As already stated, the increased word count of the revised version comes mainly from these kinds of editorial decisions. For example, Naturalis [somnus] huic nostro attinet tractatui, as against Sednaturalis [somnus] huic nostro attinet tractatui [2]; Primo, ut cerebrum cum sensibus […], as against Primo, utuidelicetcerebrum cum sensibus […] [3]; Actiones somni diuerse sunt duobus modis, as against Accionesquoquesomni diuerse sunt duobus modis [5]; and Similiter febricitantes, as against Etsimiliter febricitantes [6].⁵⁵

The use of autem (‘but’), tamen (‘however’, ‘yet’), and vero (‘but’, ‘however’, ‘whereas’) is notably more frequent in the revised version than in the standard version. In our sample chapter, the revised version has one occurrence of tamen, whereas the standard version uses sed (‘but’) instead: Vigiliis naturalibus uirtus naturalis atque corpus fatigantur,seduirtus animata confortatur, quia calor naturalis corpus egreditur, as against Naturales igitur uigilie uirtutem naturalem fatigant atque corpus. Virtustamenanimata confortatur, quia calor naturalis corpus egreditur [8]. Here the redactor obviously lightened and clarified the grammatical structure of the sentence. The ablative of cause or agency (Vigiliis naturalibus) was modified to serve as the subject, and the reflexive passive voice (fatigantur) was turned into the active (Naturales uigilie […] fatigant). The sentence was divided in two, so that sed from the beginning of the second sentence was omitted and the word tamen inserted to preserve the required concessive effect and underline the train of thought.

As to vero, the revised version has four occurrences in our edited chapter (see sections 3, 7, and 8), whereas the standard version has none.⁵⁶ The function is again to clarify the idea by adding an adversative particle with relation to what has been said earlier:

[6–7] Version 1: si cibus uel humores sint indigestibiles, […] Si corpus cibis exinanitum sit, calorque naturalis in interiora corporis redierit […] Si somnus cibos et humores inueniat temperatos, […]

Version 2: si cibum aut humores [somnus] inuenit indigestibiles […] Si uero⁵⁷ corpus cibis exinanitur calorque naturalis interiora corporis ingrediatur, […] Si uero somnus cibos et humores inueniat temperatos, […]

Here, the word uero in the revised version really adds clarity, especially when sections 6 and 7 are read as a whole. In contrast, the addition of uero in section 8 does not make a similarly successful impact: Vigilię igitur refrigerant corpus intrinsecus, calefaciunt et exiccant extrinsecus, as against Vigilię ergo corporis interiora refrigerant, exteriorauerocalefaciunt et exiccant. The word-pair intrinsecus–extrinsecus in the standard version is effective and does not require any discourse markers to support the thought. Yet the redactor changed the adverbs intrinsecus and extrinsecus to substantives, corporis interiora–exteriora. With this change, some of the effectiveness of the original formulation was lost; symptomatically, the particle uero was brought in, in order to emphasize the juxtaposition.

It is clear that the abundant use of discourse markers in the revised version was meant to add emphasis to narrative connections. The addition of discourse connectives is perhaps the most important strategy employed by our redactor for clarifying Constantine’s expression.

• (4) Participles as against subordinate clauses and finite verbs

The redactor had further strategies for editing Constantine’s text. A significant difference between the Theorica’s standard version and the revised version is that the former favours (present) participles, whereas the latter prefers subordinate clauses and finite verbs. In our sample chapter, the switch is seen on two occasions: de somno et uigiliis hęc [sc. cibos et potus] comitantibus, as against de somno et uigiliis, que ea [sc. cibos et potus] comitantur [1]; animalis uirtutis actiones in somno sunt quiescentes, as against animalis uirtutis quiescunt actiones, cum somnus adest [3].

Participial constructions such as these may be seen as typical of more elegant, classicizing Latin, while subordinate clauses simplify grammatical stuctures by splitting sentences into smaller units. The use of subordinate clauses provides the sentences with markers which make the sentences easier to comprehend. The marker most frequently used by the redactor is, as one would expect, a relative pronoun, such as que ea comitantur (section 1). Yet conjunctives were applied to the same effect.⁵⁸ The dismantling of a participial construction, which makes a text more straightforward to read or listen to, usually leads to an increase in the word count. The result is a simpler and more streamlined style of Latin.

In section 3, the present participle is connected with the verb esse in the standard version: animalis uirtutis actiones in somno sunt quiescentes. The redactor substituted the construction with the finite verb quiescunt, and turned the temporal noun construction into a temporal subclause introduced by cum: animalis uirtutis quiescunt actiones, cum somnus adest. The result hardly flows better or is simpler than the original formulation. It may be noted that a resembling construction with a present participle + esse follows the previous one almost immediately. This was retained in the revised version with slight modification: Actiones autem spirituales et naturales in suo cursu sunt permanentes / Acciones uero spirituales et naturales in suo cursu sunt manentes [3]. Perhaps the redactor substituted the previous present participle + esse construction for stylistic reasons, to avoid two similar constructions appearing too close to each other. Participles have also been preserved as such here and there, especially when they are used as subjects.⁵⁹ This indicates that careful editorial consideration was used in their substitution.⁶⁰

• (5) Passive as against active morphology

In our collated chapter, the passive (or deponent, or reflexive passive) verbs or constructions are several times turned into the active in the revised version. For example, ut cibus digeratur, humores excoquantur, as against ut cibum [somnus] digerat et humores excoquat [4]; melius digeritur, as against melius digerimus [4]. The turn can also be from the active morphology in the standard version to the passive (or deponent, or medio-passive) in the revised version: sicut uidemus, as against sicut uidetur [6]; [uigilie] calorem augmentant, as against calor augmentatur [8]. The evidence does not indicate consistency in the use of active or passive forms in the revision. On occasion the redactor seems to have wished to dispose of difficult passive morphologies. For example, cibamur ‘(we) take food, eat’ in the standard version was substituted with cibamus (see section 2). The operative denotation is the same but the active mode is somewhat more straight-forward, which also had the same meaning and was lexically perhaps more approachable.⁶¹ On the other hand, digerere ‘to digest’ appears in the passive as well as in the active morphology in both the standard and the revised version.⁶²

Yet the passage in section 8, discussed already above, betrays a conscious effort to streamline the text by dismantling a (reflexive) passive construction. The ablative of cause or agency (vigiliis naturalibus) and the (reflexive) passive voice (fatigantur) were turned into subject and active voice (Naturales uigilie […] fatigant). Irrespective of the redactor’s occasional attempts to clarify the syntax by dismantling passive constructions, the switch from the passive to the active does not constitute an overarching strategy: a great deal of passive morphology has also been preserved intact, and at times such morphologies were even added.

In one instance, a switch from an active construction in the standard version to a passive construction in the revised version had a drastic effect on the meaning: cum calor intret naturalis, materiarum magnitudines uincit et extinguit, as against cum calor intret naturalis, […] magnitudine materie uincitur et extinguitur [6]. The concept of the calor naturalis ‘the innate heat’ is central to the Pantegni and Galeno-Arabic medicine. It is connected with the complicated concept of different virtues and spirits controlling life processes, which are discussed in detail in Theorica’s Book IV.⁶³ In relation to the theme of our sample chapter, i.e. sleep, the innate heat functions by entering the interior parts of the body and digesting the previously eaten food when one is asleep. In section 6, the standard version asserts that if the amount of the food to be digested is bigger than what the natural virtue requires, the innate heat, when entering the body, surpasses and extinguishes the material [of the food to be digested]. This in fact contradicts what comes next: ‘Therefore, those who eat much, should not sleep before they can feel [at least] some amount of digestion of the food [eaten]’.⁶⁴ The revised version is more logical here:⁶⁵ if the amount of the indigested food is bigger than what the natural virtue requires, the innate heat, when entering the body, is surpassed and extinguished by the material [of the indigested food]; this should be avoided by not sleeping until at least some digestion occurred. The redactor emended the incorrect reading of the standard version by changing the active construction into passive (calor naturalis … magnitudines uincit et extinguit, as against calor naturalis … magnitudine materie uincitur et extinguitur), and thus reversing the meaning of the sentence.⁶⁶ The switch was not about active/passive morphology in itself, but about emending an incorrect reading.

• (6) Changes in wording

In revising Constantine’s text, the redactor sometimes made different word choices. In section 4, ‘the natural virtue’ in the standard version was changed to ‘the innate heat’ in the revised version: Naturalis autem uirtus intus operari comprobatur, as against Testatur naturalem calorem operari intus. As discussed above, the two concepts are connected. In our collated chapter, calor naturalis is mentioned for the first time in the same section 4, stating that the innate heat enters the interior parts of the body, aiding digestion in wintertime, when people sleep more (than in summertime). Then follows the passage mentioned above on how the operation of the natural virtue / the innate heat inside the body can be proven. As calor naturalis has been mentioned just before, the reading of the revised version, i.e. naturalis calor, is more logical than the reading naturalis uirtus in the standard version; the revision provides a more coherent text.

Also in section 6, the change from quantitate to magnitudine complies with the text’s internal logic more accurately than does the standard version. The preceding section reads Actiones somni diuerse sunt duobus modis: secundum quantitatem temporis, secundummagnitudinemmaterię quam inuenit in corpore (‘The actions of sleep differ in two ways: according to the quantity of time [of sleep], [and] according to the amount of material which the sleep finds in the body’) [5]. The redactor’s change in section 6 from ex quantitate materie to ex magnitudine materie makes the text consistent with what was written in the preceding section.

In section 8, the change from vigilie non naturales (‘non-natural wakefulness’) to vigilie extra naturam (‘wakefulness outside the nature’) demonstrates that the redactor could proceed with care, attentive to the wider contextual meaning. In the Theorica, the discussion of factors affecting health and illness is divided into res naturales, res non naturales, and res extra naturam, i.e. the natural and the non-natural things and the things external to [the individual patient’s] nature, complying with the Galeno-Arabic concepts. The division occurs in the Theorica for the first time in Book I.3–4. The natural things were the internal components of the human body, such as the body parts (anatomy), the qualities and humours and their complexions, and the virtues and spirits mentioned above; all these are discussed in Theorica Books I to IV. As already mentioned, the non-natural things were the external factors affecting health, such as air, food and sleep among others things, discussed in Book V. The things outside the nature were illnesses with their causes and effects, discussed in Books VI to X. The three categories sometimes overlap in ways which may seem illogical: for instance, sleep and wakefulness were non-natural things, whereas our sample chapter affirms that there are two types of sleep, natural and non-natural, and that the non-natural sleep ‘must be placed among the things outside the nature’ [2]. Wakefulness is similarly divided into natural and non-natural in the standard version. Here the redactor changed the text: instead of non-natural wakefulness, they contrasted natural wakefulness directly with wakefulness outside the nature: Vigilie similiter quedam naturales et quedam extra naturam. De his, que sunt extra naturam, tunc disputabimus, cum de causis dicemus accidentibus (‘Similarly, wakefulness can be natural or outside the nature. We will discuss wakefulness outside the nature when we deal with causes [and] effects [of illnesses]’) [8].

It appears that the terms naturalis and non naturalis were used in the Theorica on more than one semantic level. The res non naturales are also sometimes confused with the res extra naturam, and sometimes even with the res naturales.⁶⁷ Scribal errors evidently added to this confusion in the course of transmission. By replacing the non-natural wakefulness with the wakefulness outside the nature the redactor tried to resolve the tension inherent in the overlapping categories. Given that the things outside the nature were illnesses with their causes and effects, wakefulness outside the nature must also be seen as pathological. It made more sense to the redactor to contrast natural wakefulness with wakefulness extra naturam than with ‘non-natural’ wakefulness, to make the case more apparent.

There are a few more cases in our collated chapter where the redactor changed the wording: for example, somnus est necessarius, as against fit somni necessitas [4]; si cibus uel humores sint indigestibiles, as against si cibum aut humores [somnus] inuenit indigestibiles [6]; Quę [sc. uigilie] si modum excesserint, as against Que [sc. uigilie] si sint plusquam oporteant [8]. Preferences for a different word did not always affect the meaning dramatically. Yet changes of negligible semantic effect play a significant role in outlining different textual versions and illustrating the editorial measures taken by the redactor.

• (7) Additions of new textual material

In our sample chapter, there are two short phrases in the revised version which do not have counterparts in the standard version. The additional phrases are in multum dormientibus ‘in those who sleep much’ [4], and quod in dormiente maxime facit ‘what happens especially when one is asleep’ [6].⁶⁸ I have argued elsewhere that the redactor of the Theorica added numerous short passages from the Kitāb al-malakī, originally omitted by Constantine. Collations with the Kitāb al-malakī and with Stephen of Antioch’s literal translation, the Regalis dispositio, demonstrated that Constantine usually succeeded well in compressing the essential content of his Arabic source and pruning repetitious elements.⁶⁹ As to the passage in section 4, collation with a manuscript of the Kitāb al-malakī shows that the phrase in multum dormientibus, ‘in those who sleep much’, also appears in the original Arabic text.⁷⁰ A corresponding phrase is also in the Regalis dispositio, which has somnus si prolixior fiat, ‘if the sleep is prolonged’.⁷¹ Where Constantine decided to leave the passage out, the redactor restored it. The addition makes a slight change to the meaning of the sentence: the standard version remarks that the extreme parts of the body cool in sleep, without further specification. The revised version adds that this applies to those who sleep much.

For the second passage, quod in dormiente maxime facit, ‘what happens especially when one is asleep’, there seems to be no counterpart in the Kitāb al-malakī, nor in the Regalis dispositio.⁷² Without critical editions of the Kitāb al-malakī and the Regalis dispositio, we cannot determine whether a corresponding phrase was or was not included in the manuscripts of the Kitāb al-malakī which Constantine, Stephen, or our redactor were using. Whatever the case, it is perhaps a better possibility that the phrase originated with our redactor, as an explanatory addition. It is stated earlier in the text that the innate heat enters the internal parts of the body in sleep: Calor enim naturalis interiora tunc [i.e. in somno] petit corporis [4]. It appears that the redactor wanted again, in section 6, to remind the reader/listener and to specify that the entering of the heat takes place particularly in sleep: cum calor intret naturalis, quod in dormiente maxime facit.

These two phrases added to Theor. V.34 can be seen as illustrating two different editorial aspects of the revision process: new passages were incorporated, but the editing also included formulating short, additional phrases when necessary.⁷³ The fact that the interpolations and all the other revisions are intertwined in all the three manuscript witnesses suggests that a single party was responsible for all these revisions. We need to ask in what historical context the redactor might have operated, and what were the premises and motives for their project.

The purpose and authorship of the revision

The manuscripts preserving the revised version of the Theorica, i.e. Basel, Paris, and Rome, do not identify the redactor, but they do ascribe the translation to Constantine.⁷⁴ In order to place our anonymous author in context, one must seek to understand the nature and purpose of the revision, and which premises and characteristics the redactor was working with to have produced such a version. In this context it is useful to discuss a parallel case. Mary Wack has discussed the authorship of the Liber de heros morbo, a text attributed to Constantine and embedded among other Constantinian works in the two manuscripts preserving the text. Wack argues that the Liber de heros morbo is not only an adaptation of a chapter of Constantine’s Viaticum, but a retranslation of the original Arabic text, the Zād al-musāfir by the North-African physician Ibn al-Jazzār. Yet a text, which is clearly an adaptation of the Viaticum, is also preserved in one twelfth-century manuscript. Wack made comparisons between these three pieces, the Viaticum in its two versions and with the Liber de heros morbo (as well as with the Arabic text). She concluded that ‘medieval medical texts were occasionally modified by their users and adapted to practical or pedagogical needs, sometimes resulting in the appearance of a “new” text’. Wack demonstrated that the adapter of the Viaticum simplified the vocabulary and added explanatory material on occasion. She proposed that such explanatory additions ‘almost certainly derive from an early reader’s attempt to make sense of Constantine’s sometimes cryptic and confusing Latin’.⁷⁵

This conclusion resonates with our case. As argued above, the changes made by the redactor were intended to simplify Constantine’s syntax and clarify his train of thought. Our collated chapter also includes an ‘explanatory addition’, to use Wack’s term, namely quod in dormiente maxime facit, discussed above. By way of comparison, the revised version of the Theorica emerged in a context characteristic of that in which the Viaticum was adapted: users needed a more approachable text. It has been noted that ‘more than other works, medical treatises were living texts, used by doctors, and required regular adaptation to constantly changing environments’.⁷⁶

However, unlike the Viaticum, or, for example, the Practica of the Pantegni, the Theorica was not actually used by doctors when dealing with sick patients. Instead, the work was studied and quoted by scholars ‘to conceptualize broader philosophical theories’; evidence of such scholarly use exists from the medical school of Salerno and the cathedral school of Chartres from the first half of the twelfth century onwards.⁷⁷ Despite this different contextual emphasis, it seems that the need for a simplified text existed in the case of the Theorica as it did for the Viaticum. Many syntactic features of the revised version, categorized and analysed above, are such that they render the text more comprehensible for a reader and—for so it would seem—still more so for a listener. Perhaps the revision was intended for the classroom? For example, the aut–aut constructions work by arranging different aspects of a phenomenon into separate and countable units, which makes it easier for the listener to conceptualize the line of thought, something that would have assisted the taking of notes. Also the abundant use of conjunctions and particles in the revised version would have helped listeners in particular to orient themselves in the text and guide them from one thought to the next. It should be noted that conjunctions are not scarce in the standard version of the Theorica. But when compared to the standard version, the revised version seems in places even ‘over-supplied’ in this respect. Conjunctions and particles abound, as in the following: Accionesquoquesomni diuerse sunt duobus modis,autsecundum quantitatem temporis,autsecundum magnitudinem materię [5]; Primo, utuidelicet […]; uisusscilicetet auditus [3]. In the context of oral delivery in a classroom, however, frequent conjunctions and particles would have benefitted reception. By emphasizing how various terms and sentences were related to each other and by slowing down the pace of argumentation, these small insertions would have helped students to better follow the train of thought. The redactor was someone who was wanting to create a simplified version of the Theorica, I would argue, for educational purposes.⁷⁸

As we turn to consider the identity of this redactor, the first point to observe is that (s)he was patently well versed in medicine and knew Constantine’s works well enough to carry out a systematic edition of the Theorica. The new version shows ambition beyond what might be expected of a mere ‘adaptation’: the revision of the syntax was a comprehensive and systematically executed project, whereas adaptation usually aims at more local textual effects.⁷⁹ Presuming that the interpolated, additional passages stem directly from the Kitāb al-malakī,⁸⁰ it would follow that the redactor was also capable of reading Arabic and worked with Haly’s text in front of them. Given the early date of the revision, Constantine’s connections with the medical school of Salerno and the attested use of the Theorica and his other works there from the first half of the twelfth century onwards, the most natural candidate for this redactor would be someone operating within that school. A strong candidate would be the Cassinese monk Joannes Medicus, usually identified with Johannes Afflacius, who was born sometime between 1055 and 1070 and who died in or after 1114. According to the Cassinese biography De viris illustribus from the twelfth century, Johannes was Constantine’s pupil, and was himself a translator of Arabic medical texts. Constantine dedicated five of his works to Johannes, whom he addressed as carissime fili in the prefaces. Peter the Deacon, the Cassinese biographer, records that Johannes died in Naples, in possession of copies of all the works of Constantine.⁸¹ Monica Green, Mary Wack, and Raphaela Veit have investigated the relationships between works attributed to Constantine or Johannes and their Arabic originals, the Kitāb al-malakī¸ among others. It is possible that Johannes was the compiler, or one of the compilers, of the Practica of the Pantegni.⁸² Mary Wack has suggested that Johannes was also the translator of the Liber de heros morbo. He would have been capable of working from the original Arabic text and from the Viaticum, and his authorship is possible on chronological grounds. Referring to comparisons made by Rudolf Creutz in 1930, Wack regarded Johannes’ style as simpler and clearer than that of Constantine. In the same vein, she demonstrated that the Liber de heros morbo simplifies those passages which, on the basis of the Viaticum’s manuscript tradition, seem to have been difficult to receive. Moreover, the Liber de heros morbo contains a passage from the Arabic omitted in Constantine’s Viaticum. Such features apply to the revision of the Theorica.⁸³ The period of Johannes’ activity, c. 1080–c. 1114, would match the revision’s time-frame.

While it is tempting to conclude that Johannes Afflacius was the person most probably behind the Theorica’s revision, it has to be noted that Johannes’ connections to any particular texts (besides the Liber aureus⁸⁴) still need to be confirmed. Furthermore, attribution of texts is hampered by the lack of comprehensive stylistic analyses. The authorship of the Theorica’s revision cannot be solved, at least until we have critical editions of the works of writers associated with Contantine—including Johannes— and comparative stylistic analyses. At the moment it is probable, although not certain, that the redactor of the Theorica worked within the medical School of Salerno or affiliated circles. What is certain, however, is that the redactor believed that he (or she) could improve on Constantine’s text. To take the trouble of collating and editing Constantine’s magnum opus indicates the perception of a need for a simplified version. Such a comprehensive and systematically executed project shows a strong fidelity to Constantine, of a kind which the assembler(s) of the Practica of the Pantegni also showed when putting the text together from bits and pieces.⁸⁵ These enterprises, together with Stephen of Antioch’s new, literal and complete translation from Haly Abbas’s original Arabic work, show the state of ferment the Pantegni was in in the early twelfth century and beyond. The revised version of the Theorica provides a significant insight into the Pantegni’s early reception: the text embodies active engagement with Constantine’s work, modifying it in line with the expectations and practical needs of the work’s audience. Further analyses, historical, stylistic, and textual, are needed, and it is to be hoped that they will stimulate fresh work towards a critical edition of this fundamental text.

I wish to thank the following people and institutions: Samu Niskanen, Matti Haltia, Heikki Solin, Monica Green, Charles Burnett, Reijo Pitkäranta, Timo Korkiakangas, Ilkka Lindstedt, Outi Merisalo, Katja von Schöneman, Päivi Salmesvuori, Kalle Korhonen, and James Willoughby, who also revised my English; the Cultural Heritage Fund of the National Library of Finland and the anonymous donor, the Signe and Ane Gyllenberg Foundation, the Finnish Society of Sciences and Letters, and the University of Helsinki. All remaining errors and shortcomings are, of course, my own.

Footnotes