The present article examines the textual transmission of the Aristotelian Mechanics, a treatise on mechanical questions now generally ascribed to the Peripatetic School. The treatise was edited three times in the nineteenth century, namely by Johannes van Cappelle (1812), Immanuel Bekker (1831) and Otto Apelt (1888); most recently, an edition was produced in the twentieth century by Maria Elisabetta Bottecchia (1982). Bottecchia's edition is a clear improvement over the previous editions in the extent of its research. Whereas the other editors of the Mechanics altogether consulted a total of nine manuscripts, Bottecchia considered nearly the complete manuscript material for her critical edition of the text. When I started my project I did not expect to find significant new results which would make a completely new critical edition of the text necessary.
However, as I compared Bottecchia's analysis of the manuscript tradition with my own examination of the text and collations of the manuscripts, it became clear that her edition contains numerous errors. It would go beyond the scope of this paper to present a detailed examination of Bottecchia's edition; let me mention here just two examples. First, Bottecchia argues that the manuscript V2 depends on Wa.Footnote 1 This is incompatible with strong evidence concerning the chronology of the manuscripts: there is reason to think that V2 was composed in the first half of the fourteenth century and Wa more than a century later, in about 1465.Footnote 2 Even more important is Bottecchia's failure to recognize the influence exerted on the textual tradition by Georgios Pachymeres, parts of whose Byzantine paraphrase of the Mechanics were incorporated into some manuscripts of the text. Bottecchia regarded Mut. 76 (Mu), which is an apograph of Pachymeres' paraphrase, as an authentic Aristotelian manuscript and included readings of this paraphrase in her critical edition of the text.Footnote 3
A new examination of the manuscripts of the Mechanics is therefore necessary, and will yield a different perspective on the textual transmission of the treatise. The present article offers a survey of my results from collating all 31 manuscripts in which the Mechanics has been handed down to us.Footnote 4 I shall first give an account of the different editions of the text, and then present my own analysis of the manuscript families, resulting in a stemma codicum which illustrates the affiliations of the manuscripts.
I. EDITIONS OF THE MECHANICS
The editio princeps of the Mechanics is the complete edition of Aristotle, printed by Aldus Manutius in Venice (1495–8). We do not possess the printer's copy of the Mechanics, but the edition appears to be based solely upon the manuscript Be, together with influences from the Byzantine paraphrase by Pachymeres.
The nineteenth-century editions of the treatise by van Cappelle (1812), Bekker (1831) and Apelt (1888) are all likewise unsatisfactory, in that they too consider only a small part of the textual tradition and are too much affected by Pachymeres' paraphrase. Van Cappelle's edition contains the Greek text together with comments and a Latin translation.Footnote 5 In his introduction, van Cappelle mentions three manuscripts of the Mechanics collated by him, namely Lv, Par and Pt. The Parisian manuscripts, Par and Pt, both include readings from different textual traditions. Moreover, they appear in places to depart from the authentic Aristotelian text, as a result of influence from the paraphrase of Pachymeres.
In his complete edition of Aristotle, Bekker cites the manuscripts P and Wa as his main sources for the Mechanics. These manuscripts, which represent different branches of the textual tradition, contain valuable readings and, when analysed together with L and Ha, cover the most important textual variants. However, apart from numerous errors in collation, many of Bekker's readings are printed without any justification, not being contained in any of the manuscripts cited by him.Footnote 6 In the course of collation I discovered many agreements between Bekker's edition and variants included in Be, a manuscript written by the Italian humanist and professor of Aristotelian philosophy in Padua, Niccolò Leonico Tomeo. In 1525, Leonico published a Latin translation of the Mechanics with commentary. Probably Bekker consulted this commentary during the process of making his edition of the treatise, and tacitly included some of Leonico's emendations in his text.Footnote 7
Apelt's edition contains the Greek text of the Mechanics and of five further smaller works by Aristotle. Apelt did not examine any new manuscripts, but relies on the manuscripts consulted by Bekker, P and Wa, and also on Par, previously analysed by van Cappelle. In his foreword, Apelt praises the edition by van Cappelle and the ‘excellent’ codex Par. He holds that Par offers many good readings of corrupt passages and fills the lacunae of other manuscripts. Although this is true to some extent, we should be careful in accepting the variants from Par, since many of them were borrowed from the paraphrase by Pachymeres and do not contain the authentic text. Given that Apelt did not add any new manuscripts to the ones collated by his predecessors, his edition does not make a significant contribution to our understanding of the textual tradition.
The Loeb Classical Library edition of W.S. Hett (1936) relies upon the Bekker text, except in a few cases where the edition of Apelt has been employed.Footnote 8 Hett himself has also made some smaller emendations.
The most recent edition of the Mechanics is due to Bottecchia (1982). Bottecchia examined nearly all the manuscripts of the text, which she divided into four manuscript families, namely a, b, c and d. Her text relies primarily upon manuscripts in her families a and b. Ha, L and Oa, all in family a, together with P and V3, both belonging to family b, are the main sources for her edition of the text. However, the last manuscript, V3, is not valuable for the reconstruction of the Mechanics, since it is a copy of the first printed edition of the text by Manutius, and therewith contains many elements from the paraphrase by Pachymeres. Because she neglected to consult the families c and d for her edition, Bottecchia excluded some important variants offered by these manuscripts.
II. THE MANUSCRIPT FAMILIES
Altogether 31 manuscripts of the Mechanics have been handed down to us. The following manuscripts contain the complete text or excerpts from it:
I have collated three new manuscripts, Hd, Nb and Nc, of which the first two are of special interest.
The textual transmission starts fairly late; the oldest manuscripts date from the beginning of the fourteenth century. The dating of the oldest preserved manuscript, Ha, has been a subject of much controversy over the past decades. Some years ago, however, Rashed convincingly argued for a date around 1300.Footnote 9 A transmission that starts that late brings with it many difficulties, since it is hard to assess the earlier phases of the textual tradition with reasonable certainty. The situation is further complicated by the presence of an independent branch in the tradition, that of the paraphrase by Pachymeres, which in turn influenced some of the manuscripts and the previous editions of the treatise. The paraphrase of the Mechanics contained in Pachymeres' compendium of Aristotle's philosophy offers an almost literal quotation of the Aristotelian text. Therefore, it can be considered as a witness of the textual tradition of the Mechanics. Every now and then, Pachymeres makes some smaller emendations or adds an illustrative remark in the text, giving his text an independent position among the other manuscripts.
Different readings led me to divide the writings into three manuscript families, namely a, b and c. The independent tradition of the paraphrase by Pachymeres derives from the same hyparchetype as family a; it is, however, recorded in the stemma codicum as a different branch, since, through Pachymeres' emendations and additions, paraphrastic elements were incorporated into the text. Pachymeres' paraphrase of the Mechanics is important in constituting the text, since it was written at the same time or even shortly before the authentic Aristotelian manuscripts were copied. Finally, there is a group of contaminated manuscripts, which include aspects of different manuscript families. These writings do not all contain the complete text, but only excerpts from it.
Family a
The two manuscripts in this family are among the oldest preserved manuscripts of the Mechanics, and derive independently of each other from the hyparchetype α. They share readings which separate them from the other families, for example at 847b25 διɛστήκατον a: διέστηκɛν cett.; 848b24 πλɛυρῶν a: πλɛιόνων cett.; 850a32 κινɛῖν a: κινῆσαι cett.; 851b18 ὁ τῆς ἁμάξης τροχὸς a: ὁ τροχὸς ὁ τῆς ἁμάξης cett.; 853a25 ὄντι τῷ μόχλῷ a; ὄντι cett.; 853a27 ἐϕ' ᾧ ΔΕΖΗ a: ΔΕΗΖ cett.; 853b32 πρόσκɛιται a: ἔγκɛιται cett.; 854b20 τὴν ἐλάττω διάμɛτρον a: τὴν διάμɛτρον τὴν ἐλάττω cett.; 854b27 πλɛυράν a: πλɛίω cett.; 855a13 ὀξɛίας ὥσπɛρ συμβαίνɛι a: ὀξɛίας συμβαίνɛι cett.; 856a24 κωλύɛι a: κυλίɛι cett.; 856b28 ΒΓ a: ΑΓ cett.; 857b12 οὕτως a: ἐλάττων cett.; 857b38 ἀνίστηται ἀναστῆναι a: ἀναστῆναι cett.; 858b11 καὶ τότɛ μένɛι a: καὶ τιθέαμɛν ɛἰ cett.. Further variants of family a can be found at 848a6, 848a11, 848b16, 849b20, 850a1, 850a24, 850a27, 850a31, 851a7, 851a13, 851a28, 852a6, 853a30, 854a11, 854a14, 854b5, 854b11, 854b20, 855a3, 855a4, 855b3, 855b24, 856a34, 857a2, 857b5 and 858b24.
Now let us have a closer look at the two manuscripts in this family:
Ha Marc.gr. 214, thirteenth–fourteenth century, c. 1290–1303. Parchment, 330 x 230 mm, 238f. Eth. Nic., Metaph., Ph., Gen. corr., De an., excerpt from Rh., Cleomedes, Caelestia; Sens., Mem., Somn. vig., Insomn., Div. somn., MA, Long., Juv., Resp., Col., Lin. ins., Mech. (f.203–10), Spir., Cael.
Whereas earlier scholars dated Ha anywhere from the eleventh to the fifteenth century, Harlfinger was able to determine the date more precisely between 1270 and 1370.Footnote 10 Rashed settled the controversy about Ha definitively, fixing the date between 1290 and 1303.Footnote 11 He identified one of the scribes who worked on Ha as the copyist of another manuscript, Ambr. G51 sup., and argued convincingly that parts of Ha were likely copied from Ambr. The terminus ante quem of Ambr. is 1303, the year in which Conradus Beginus acquired this manuscript and took it along to Kaffa. Therewith, Ha can be dated more precisely, namely after the making of Ambr. and before Conradus' acquisition; these facts speak for a date in the final years of the thirteenth century, possibly in about 1300.
Ha has influenced a group of manuscripts that derives from Oa. Since Oa is copied from a manuscript in family b and contains the variants of family a only after emendation, this group of manuscripts will be presented in a separate section. The second manuscript of family a, L, of which no direct descendants are known for the Mechanics, was copied at the beginning of the fourteenth century.
L Vat.gr. 253, fourteenth century, beginning. Paper, 241 x 164 mm, 269f. Cael., Gen. corr., Mete., De an. Book 3, Sens., Mem., Somn. vig., Insomn., Div. somn., MA, Long., Juv., Resp., Col., Lin. ins., Mech. (f.246v–262v), Spir.
On account of a number of separative errors, a dependence in either direction between Ha and L can be ruled out: for example at 847b25, 848b23, 849b20, 850a25, 851a13, 852b1, 852b22, 852b39, 853a7, 853a12, 853a13, 853a34, 853b9, 853b14, 854a3, 854a13, 854a22, 854b3, 854b10, 854b27, 855b34, 855b38, 856a26, 856b3, 856b14, 857a20, 858a12 and 858b26. Since Ha and L can independently of each other be traced back to the hyparchetype α, the readings of both manuscripts are important in constituting the text. Bottecchia was the first editor to fully consider this fact, and consulted both manuscripts for her critical edition of the Mechanics.
Family b
Family b consists of the following manuscripts: P, Ba, Be, Lv, Mv and D1. P is the oldest and best-known manuscript of this family.
P Vat.gr. 1339, fourteenth century, second half. Parchment, 280 x 210 mm, 460f. Part. an., Gen. an., IA, De an., Sens., Mem., Somn. vig., Insomn., Div. somn., MA, Long., Juv., Resp., Col., Lin. ins., Mech. (f.281v– 294), Spir., Mu., Hist. an.
The value of codex P as a textual witness has been questioned for various treatises from the Corpus Aristotelicum.Footnote 12 Although P shows many traces of contamination, scholars generally agree that this manuscript contains readings which constitute independent textual evidence. For the Mechanics there are agreements between P and family c,Footnote 13 but on account of some relevant textual variants P can be traced back to its own hyparchetype β. Some characteristic readings of P, which are also present in the manuscripts Ba, Be and Lv,Footnote 14 can be found at 847b19 σὺν ἀλλήλοις b: μɛτ' ἀλλήλων cett.; 848a23 ὑπɛναντίους b: ὑπɛναντίως cett.; 848b33 προɛιρημένα πρότɛρον b: προɛιρημένα cett.; 849b2 om. τῷ μɛίζονι b; 849b18 κατὰ ϕορὰν b: κατὰ ϕύσιν a: τὸ κατὰ ϕύσιν c; 852a9 αὐτῶν b: αὐτὸν cett.; 852a17 χωρίω b: χρόνῳ cett.; 854a34 ἀφɛρɛῖ b: ἀϕαιρɛῖται cett.; 854b29 πɛπληρώσθω b: παραπɛπληρώσθω cett.; 855b26 μὴ b: μὲν cett.; 855b27 μɛίζω b: μɛίζονι cett.; 855b39 αὐτὴν b: αὑτοῦ cett.; 856b13 ἴσαι ɛἰσίν b: ɛἰσίν ἴσαι cett.; 857a38 καθιέναι b: καθɛικέναι cett.
The codex P can be attributed to a copyist by the name of Joasaph and was written at some time in the second half of the fourteenth century.Footnote 15 Most of the Mechanics editions acknowledge the importance of P as a direct descendent of the hyparchetype β. Both Bekker and Apelt adopted many readings of P, and Bottecchia too included various readings from it in her edition of the treatise. Although P contains valuable variants, the editor should in every case carefully consider if a peculiar reading is the result of contamination, constitutes an independent variant, or is simply an error.
Two further manuscripts in family b, which share part of the readings of P, are Mv and D1. Both of these manuscripts were copied in Constantinople in the middle of the fifteenth century.Footnote 16 D1 was composed as a cooperation between different copyists. The final pages of the Mechanics were written by Matthaios Kamariotes, by whose hand Mv was also copied.
Although Mv and D1 have many similarities with P, it is not possible that these manuscripts are dependent on P, since P contains many peculiar readings, for example at 848a3, 850b13, 850b17, 852a4, 852a23, 853b7, 854b27, 855a27, 856b20, 857a23, 857b6 and 857b11. Besides, Mv and D1 contain further readings that coincide with variants from family c,Footnote 17 as for example at 850a35 ἀντὶ σπαρτίου γίνɛται c: ἐστι τὸ σπαρτίον a: ἐστι τί σπαρτίου/-τίον γίνɛται b; 854a20 μαλακὴ c: μαλθακὴ cett.; 857a2 τῇ c: τῷ cett.; 858b9 ἐλαττονα c: ἐλάττω cett.; 858b25 τοῖς c: τοίνυν cett. The readings from family c show a proximity to the manuscript Wa: for example at 852b22, 855a5, 855b26 and 855b37. Both manuscripts finally agree in numerous significant errors which are not shared by P, as for example at 850a3, 850a19, 850a28, 851b4, 853a5, 853b7, 853b14, 854b11, 855a11, 855a19, 855b17, 855b34, 856a22, 856b1, 856b2, 856b21, 857a3, 857a29 and 858b15.
Certain text passages show that the copyists of P, Mv and D1 had similar difficulties in reading their source: 856b2 [lac.] om. τὸ P: [lac.] om. ἢ τὸ μὲν μέγɛθος Mv/D1; 856b4 οὔτ [lac.] om. ω διπλασι όπλɛυροι P: οὕτως [lac.] om. διπλασιό πλɛυροι Mv/D1. At the end of D1 Kamariotes even mentions the corruption of his source at f.353v: σημɛιωτέον ὅτι πάνυ διɛϕθορὸς ἦν τὸ ἀντίγραϕον. The exemplar of Mv and D1 must be a corrected or contaminated codex, since many readings from family c are incorporated in these manuscripts. Because of the similar problems the scribes of all three manuscripts had with their corrupt source, Mv and D1 were probably copied from the same hyparchetype as P was, except that it underwent some emendations in the century between the composition of P on the one hand and of Mv and D1 on the other. Since D1 cannot be the exemplar of Mv and vice versa,Footnote 18 both manuscripts are direct copies of the hyparchetype β, although not completely independent of each other. The fact that the hyparchetype was in such a bad condition, makes it reasonable to assume that Kamariotes and the copyists working for him constantly referred to Mv while they were copying D1. This would also explain the high degree of shared errors in both manuscripts.
A highly interesting and influential manuscript from family b, which can be traced back to P, is Be:
Be Bern. 402, fifteenth century, before 1497. Paper, 235 x 155 mm, 144f. Theophr. Ign., Metaph., Lap., Sud., Vert., Lass., Pisc., Vent., Od.; Arist. Xen., Mir., Spir., Mech. (f.99–114), Mu.; Alexander Aphrodisiensis, Fat.
The copyist of this manuscript is Niccolò Leonico Tomeo.Footnote 19 Leonico owned this manuscript and used it for his Latin translation of the Mechanics from the year 1525. My collations of Be confirm the statement of Sicherl that this manuscript, together with some influences from the paraphrase by Pachymeres, provided the basis for the first printed edition of the Mechanics by Manutius in Venice.Footnote 20
Be contains many emendations which are inserted by the same hand both in the margins and within the text.Footnote 21 Some of these emendations agree with variant readings contained in the paraphrase by Pachymeres, for example at 850a13 add. ἔσται, 851a27 add. δὲ, 852a17 add. χρόνῳ, 853a25 add. τῷ μόχλῷ, 855b36 add. καθ' αὑτὸ, and 856a9 add. μικρὸς. Others, however, have no connection with any extant manuscript and are probably based on Leonico's own interpretation of the text. Interestingly, some of the emendations in Be were adopted in the first printed edition by Manutius, and have found their way through the Aldina even into the modern editions of the treatise. Further, Bekker included many other variants offered by Leonico, which are not contained in the Aldina, in his edition of the treatise: for example at 850b25 ᾗ Be: ɛἰ cett.; 851b9 ἀντισπᾶν Be: ἀντισπᾶ cett.; 855a38 ἡ δὲ τοῦ ἐλάττονος ἐλάττων Be: om. cett.; 856a29 μικρὸς Be: μέγας cett.; 857a18 τὸ μέσον Be: om. cett.: ἔστω μέσον Bekker; 858b29 δὲ Be: om. cett.; 858b30 ɛἰς Be: om. cett. As I mentioned above, Bekker, as he made his edition of the Mechanics, probably had access to Leonico's translation and commentary on the treatise, and thereby included Leonico's emendations in his edition.Footnote 22 An important task for a new critical edition of the Mechanics is to distinguish Leonico's activities on the treatise from the authentic Aristotelian manuscript readings.
Oa and its descendants
This group of manuscripts combines readings from two manuscript families. The primary readings in these manuscripts come from family b, whereas variants from family a are only contained in secondary instances. The central member of this group, Oa, stems from the manuscript P and was emended after completion with the help of codex Ha.
Oa Marc.gr. 216, middle of the fifteenth century, 1445. Paper, 220 × 135 mm, 348f. Phgn., Vent., Mir., Mu., Xen., Mech. (f.90–110v), Spir., Plant., Pr.
Oa was copied in the year 1445 by Johannes Scutariotes on behalf of Cardinal Bessarion. Shortly after the emendations from family a were added, Oa provided the main source for the other manuscripts in this group, namely Q (which was also copied on behalf of Cardinal Bessarion, by Johannes Rhosos in 1457), Hd and Vph.
Many variants in Oa have been changed from P's reading of the text into the version as contained in Ha, as can be seen at 850a35 ἐστι τὸ σπαρτίον γίνɛται; 850b7–8 τὸ δὲ ἐφ΄ᾦ τὸ Δ; 850b17 add. τὸ πλɛῖστον; 852b26 ἔνθα μὲν γὰρ; 853a25 ὄντι add. τῷ μόχλῳ; 857b6 add. τὸν κάδον; 858a1 ἴσας ɛὐθɛίας; and 858b11 καὶ τιθέαμɛν ɛἰ add. τότɛ μένɛι. In some cases variants of both P and Ha are extant next to each other: at 852a5 ἔχοντɛς/ἔχοντα; 853b32 ἔγκɛιται/πρόσκɛιται; 854a34 ἀϕɛρɛῖ/ἀϕαιρɛῖται; 854b27 πλɛίω/πλɛυράν; 855a3 πλɛυρὰ/πολλαπλασία; 855b3 γίνονται/γίνɛται; 855b27 μɛίζω/μɛίζονι; and 857b18 ὑποκɛιμένου/ἐπικɛιμένου. In Bottecchia's arrangement of the manuscript families, Oa takes a much more important position than it actually has. Bottecchia recognizes that Oa is dependent on P and contains corrections from Ha, but at the same time she traces Oa back to the hyparchetype. As my collations have proved, Oa does not stem from the hyparchetype, since all primary readings in this manuscript coincide with P and all secondary readings can be explained by means of Ha.Footnote 23
Oa with emendations provided the basis for the other manuscripts in this group: Q, Hd and Vph. The manuscript Hd, which I have collated for the first time, shares nearly all of its peculiar readings with Vph. However, a dependence of one of these manuscripts upon the other can be eliminated: Vph cannot be a copy of Hd on the basis of chronology, and Hd cannot have been copied from Vph, because there are some passages that are omitted only in Vph, for example lines 850b14–15, 853a15, 854b27 and 856a23. It is also not possible that these manuscripts are independent copies of Oa, since they share many peculiar errors that are absent from Oa, and furthermore in every case where Oa provides two different readings, Hd and Vph share one and the same reading: 852a5 ἔχοντα; 853b32 πρόσκɛιται; 854a34 ἀϕαιρɛῖ; 854b27 πλɛίω; 855b3 γίνɛται; 855b27 μɛίζονι; 857b18 ἐπικɛιμένου. It is very unlikely that two copyists working independently of each other made the same decision every time when they were confronted with two different readings. Therefore, it is natural to conclude that Hd and Vph are dependent on the same lost manuscript, which in turn stems from the manuscript Oa.
Family c
Family c, which can be traced back to its own hyparchetype γ, has the following members: Wa, Um, V1, M1, F2, Nb, q, Nh, F1, Par and V2.Footnote 24 This family is distinguished, amongst other things, by the presence of scholia, both contained within the text and inserted in the margins.Footnote 25 The manuscripts in family c are also characterized by the omission of various longer passages and have characteristic readings which are not preserved in the other manuscript families, for example at 848a5 καὶ ɛἰς τὸν ὄπισθɛν c: καὶ τὸν ὄπισθɛν cett.; 848b10 λίγῳ/ὀλίγῳ c: λόγῳ cett.; 850a35 ἀντὶ σπαρτίου γίνɛται c: ἐστι τὸ σπαρτίον a: ἐστι τί σπαρτίου/-τίον γίνɛται b; 850b18 ἀμϕοτέροις c: ἀμϕότɛρα/ἀμϕότɛρον cett.; 852b17 θᾶττον c: μɛῖζον cett.; 854a20 μαλακὴ c: μαλθακὴ cett.; 854b9 ἐστιν μὲν οὖν τὸ c: ἐστιν οὖν τὸ μὲν cett.; 854b14 ɛὕρɛσις c: ἄρσις cett.; 855b16 om. πɛριϕέρɛια c: πɛριϕέρɛια cett.; 856a1 ἡ Α c: ὁ Α cett.; 857a12 ἐπɛιδὴ διότι c: αἴτιον δὲ διότι a: αἴτιον δὲ ὅτι b; 857b13 τῶν ξύλων c: τὸ ξύλον cett.; 858b9 ἐλάττονα c: ἐλάττω cett.; 858b25 τοῖς c: τοίνυν cett.
V1 and Wa derive independently of each other from the hyparchetype γ. The manuscript V1, written by the physician Demetrios Branas about 1430, is the older of the two:Footnote 26
V1 Vat.gr. 905, fifteenth century, c. 1430. Paper, 218 × 146 mm, 160f. Joannes Tzetzes, Εἰς τὴν Ὁμήρου Ἰλιάδα ἐξήγησις; Arist. Lin. ins., Mech. (f.35v–51), Mu., Hist. an.
V1 does not contain all the scholia from family c, but only those at 848a32 and 852a23. In the tradition of the Mechanics, it has one direct descendant, M1, which is an almost perfect copy of its original.
Wa was written some thirty-five years after V1 was composed. It was copied by Georgios Tzangaropulos in an accurate hand:
Wa Urb.gr. 44, fifteenth century, c. 1465. Paper, 293 × 220 mm, 85f. Eth. Nic., Lin. ins., Mech. (f.69–78), excerpt from Iambl. VP.
All the scholia of family c were added in the margins of the text by a different hand, probably in the sixteenth century.Footnote 27 Both V1 and Wa have several longer omissions, although of different passages. Since all manuscripts in family c are characterized by such omissions, it is reasonable to assume that the hyparchetype γ was already corrupted at least to some degree, and may have contained different stages of emendation.
Of the other manuscripts in family c, nearly all contain the whole range of scholia, and all share a great number of longer omissions, as for example at 851a20–2, 855a30–2, 855b6–10, 856a27–9, 857a30–2 and 857b30–2. On the basis of further omissions and peculiar errors, these manuscripts can be split up into two groups. Um, q, Nb and F2 form the first group and are closely connected with each other. Um and q are the oldest of these manuscripts and it is possible that they are the exemplars of the remaining manuscripts in this group. Um was copied in Crete by Giovanni Simeonachis on behalf of Marco Lippomano. The terminus ante quem for this manuscript is 1446, the year in which Lippomano died.Footnote 28 The manuscript q can, on the basis of water marks, be dated with certainty in the first half of the fifteenth century, and some parts of the codex can even be ascribed to the first quarter of this century, about 1420.Footnote 29 Whereas the catalogues agree on a date for Nb in the fourteenth century,Footnote 30 Harlfinger has established that the codex is of a later date in the second half of the fifteenth century, possibly even the beginning of the sixteenth century.Footnote 31 Hence Nb can be discarded as a possible original of the other manuscripts in this group. On the basis of peculiar omissions, especially the omission in Um of lines 847a18 and 855a12, and in q of 848a4–5, 855a35–36 and 856a16, it is impossible to ascertain a dependence of one of these manuscripts upon the other. Therefore, it is likely that Um, Nb and q stem independently of each other from the same lost manuscript. F2 contains all the peculiar omissions and errors of Um, and is thus a copy of this manuscript.
The second group, consisting of Nh, F1 and Par, shares four further omissions, namely of lines 852b23–4, 854b20, 854b26, and 856b25–6. There is a close connection between Nh and Par, which can be seen in the corresponding errors at for example 851a5, 851a9, 851a24, 851b33, 852a2 and 852a9. There are, however, further peculiar errors in Nh that are not contained in Par, by reason of which a dependence of this manuscript on Nh can be eliminated. The codex Par received an elaborate emendation after completion: all omissions were completed and many variants were added or overwritten after erasing the first reading. Par was emended with a manuscript containing the paraphrase of Pachymeres, as is shown by the remarks added at 847b26 and 848a37, and by the variants at 850a35, 850b37, 851a20, 851b33, 852a5, 852a12, 852a18, 853a10, 853a25, 853a27, 853b6, 853b32, 854b1, 854b20, 855a13, 855b39–856a1, 856a35, 856b21, 856b32, 857a24, 857b2, 857b6, 857b38, 858b9, 858b24 and 858b25. The manuscript F1, finally, is a copy of Nh.
Family c offers some valuable textual variants which are not contained in the other manuscript families, as for example at 849a36 ὀρθὰς c: ὀρθὴ/ὀρ[lac.] cett.; 849b18 τὸ κατὰ ϕύσιν c: κατὰ ϕύσιν a: κατὰ ϕορὰν b; 850a35 ἀντὶ σπαρτίου γίνɛται c: ἐστι τὸ σπαρτίον a: ἐστι τί σπαρτίου/-τίον γίνɛται b; 852b17 θᾶττον c: μɛῖζον cett.; 857a28 ἀπὸ τῆς c: τῆς a: [lac.] ἀπὸ τῆς b. Even more important for a new critical edition of the Mechanics is the position of family c, where readings from this family agree with family a against family b. As I mentioned above, the hyparchetype of family b was a corrupted and contaminated manuscript, and the editor should be very careful in adopting readings from this family. On the basis of agreement between families c and a he might reconsider the following variants: 851b1 δɛῖ κινɛῖν b: κινɛῖν δɛῖ cett.; 853b2 ἕλξɛι b: om. cett.; 854a17 τῇ χɛιρὶ μόνῃ ψιλῇ b: ψιλῇ τῇ χɛιρὶ μόνῃ cett.; 856b13 ἴσαι ɛἰσίν b: ɛἰσὶν ἴσαι cett.
Whereas the editors of the treatise from the nineteenth century consulted the manuscripts Wa and Par, family c was neglected in the edition of Bottecchia. In her division of the families, these manuscripts all depend on Ha and cannot be traced back to a separate hyparchetype. Wa and V1 are of especial importance, inasmuch as they stem directly from the hyparchetype γ. Par, on the other hand, which was examined by van Cappelle and Apelt, should be considered with great care, since it contains many emendations by Pachymeres that do not contribute to the textual transmission of the treatise.
III. INDEPENDENT TRADITION OF PACHYMERES
The next group of manuscripts, which directly descends from the hyparchetype α, does not contain authentic Aristotelian writings, but the Byzantine paraphrase of Aristotle's works by Georgios Pachymeres. This paraphrase influenced the first printed edition of the treatise by Manutius and thereby also the manuscripts V3, Nc and u, which are copies of Manutius' edition.
Pachymeres was a Byzantine historian and philosopher who wrote, around 1300, a compendium of Aristotle's philosophy in twelve books. The Mechanics are in the final book, and are less a paraphrase than an almost literal quotation of the Aristotelian text. Because of this, it is complicated to distinguish Pachymeres' writings from the authentic Aristotelian text, with the result that manuscripts containing Pachymeres' paraphrase have often been considered as Aristotelian writings. In Wartelle's Inventaire many apographs of Pachymeres are still listed among the Aristotelian manuscriptsFootnote 32 and, as mentioned above, Bottecchia did not recognize Mu as a manuscript by Pachymeres and included many paraphrastic elements in her edition. The paraphrase, however, is of great importance for the transmission of the Mechanics, since it was written at the same time or even shortly before the authentic manuscripts were copied. We even possess two autographs of Pachymeres' Φιλοσοϕία, Berol.Ham. 512 (Pachb) and Par.gr. 1930 (Pachp), which contain the Mechanics. Footnote 33 Pachp is posterior to Pachb and is a copy of this manuscript.
Pachb Berol.Ham. 512, fourteenth century, beginning. Paper, 308 × 220 mm, 234f. Prooimion; Book I: Org.; Book II: Ph.; Book III: Cael.; Book IV: Gen. corr.; Book V: Mete.; Book VI: Part. an., IA; Book VII: De an.; Book VIII: Parv. nat., De motu an.; Book IX: Gen. an.; Book X: Metaph.; Book XI: Eth. Nic.; Book XII: Col., Lin. ins., Mech. (f.224–34).
The pages near the beginning and end of a manuscript had to endure the most and these are the ones that are often mutilated. This is also the case in Pachb, where the text of Pachymeres breaks off at 857b14 with ϕɛρόντων and is continued with a supplement by a later hand. The copyist of these supplementary pages was identified as Bartolomeo Zanetti,Footnote 34 who was working in Venice when he replaced the damaged pages of the manuscript in about 1530. Zanetti did not have another Pachymeres manuscript at his disposal to use in restoring the last pages of the Mechanics. My collations have proved that the variants in Pachb from 857b14 up to the end of the text perfectly agree with readings from the Aldina: for example at 857b15 om. γὰρ/ἀπέχῃ, 857b16 τὸ, 857b29 ὀρθὸν, 857b33 κνῆμαι, 857b38 ἅμα ἀνίστηται ἀναστῆναι, 858a19 τοῦτο, 858a24 ῥιπτόμɛνα, 858a29 ἀντɛρɛίδɛι, 858a31 μηθὲν, 858b5 ϕέρɛται, 858b9 αὐτὸν, 858b14 δινομένου, 858b21 ὅσω, 858b24 ὑπολɛίπɛσθαι τὸ πᾶν, 858b29 om. δὲ and 858b30 om. ɛἰς.Footnote 35 Printed a couple of years earlier in Venice, the Aldina should have been easily accessible for Zanetti.
The other autograph of Pachymeres, Pachp, which is the work of different copyists, contains for the whole Mechanics the text of the paraphrase. The first pages of the Mechanics are by the hand of Pachymeres himself; at 849a31, however, a change of copyist takes place and the text is completed by an assistant of Pachymeres.
The manuscripts that contain the writings of Pachymeres share many of the readings characteristic of family a, as for example at 847b25, 850a32, 850b7–8, 851b18, 853a25, 853a27, 854b20, 855a13, 856a24, 856b28 and 857b38. Further, there are many peculiar readings not contained in any of the other manuscript families: at 849a4 ἡ ΒΔ πρὸς τὴν ΔΓ Pach.: ἡ βɛ ɛγ cett.; 850a39 βάρη δὲ δύο Pach.: δύο δὲ βάρη cett.; 850b4 μɛίζον Pach.: μɛίζονα cett.; 851a27 δὲ οὗ τὸ Δ Pach.: οὗ τὸ Δ cett.; 851b36 ἐλάττους Pach.: ἐλάττονας cett.; 852b32 ῥᾷον Pach.: θᾶττον cett.; 854a30 μᾶλλον Pach.: ῥᾷον cett.; 854b20 ϕορᾶς Pach.: πλɛυρᾶς cett.; 857a10 ϕέρɛται Pach.: αἴρɛται cett.; 857a29 τὸ ξύλον μακρότɛρον Pach.: μακρότɛρον τὸ ξύλον cett.; 858b14 δυναμένου Pach.: δινουμένου cett. Pachymeres repeatedly adds illustrative remarks and phrases within the text as, for example, in the introduction of the Mechanics where the properties of the circle are explained. It is said by Aristotle that a point along the radius of a rotating circle moves faster the further away it is from the centre. At 848a17 Pachymeres explains: ὡς πολλάκις καὶ ἡμῖν ἐλέχθη, ἐν τῷ πɛρὶ οὐρανοῦ καὶ ἐν ἄλλοις· ὅτι θάττονα ɛἰς κίνησιν τὰ πορρώτɛρα τοῦ μέσου.
Since it is often difficult to distinguish Pachymeres' paraphrase of the Mechanics from the authentic Aristotelian manuscripts, it has happened that the first printed edition by Manutius, and through the Aldina all modern editions, contain variants from his paraphrase: for example at 849a4 ἡ ΒΔ πρὸς τὴν ΔΓ Pach./Aldina: ἡ βɛɛγ codd.; 853a10 ὁ ἡ χɛὶρ αἴρɛι Pach./Aldina: om. codd.; 855b39–856a1 ἀλλὰ τὴν τοῦ κινοῦντος Pach./Aldina: om. codd.; 856a4 ὑπὸ τοῦ ἐλάττονος Pach./Aldina: om. codd. One of the fundamental tasks for a new edition of the treatise will be to remove all these traces of Pachymeres from the text of the Mechanics.
Whereas the manuscript Be provided the basis for Manutius' edition, Pachymeres' paraphrase also exerted influence on the Aldina. Some of the variants offered by Pachymeres found their way into the Aldina through the manuscript Be; other variants were included directly from Pachymeres in the first printed edition of the Mechanics. Bottecchia, however, believes that she has located the exemplar of the Aldina in the codex V3.Footnote 36 As Sicherl pointed out, the relation between these manuscripts is exactly the contrary, and V3 is a copy of the Aldina.Footnote 37 My collations of both manuscripts have confirmed this assumption. The copyist of V3 is Demetrios Damilas, an important and prolific copyist, who is identical with Harlfinger's ‘librarius Florentinus’.Footnote 38 Damilas worked at the end of the fifteenth century in Florence together with the Byzantine humanist Demetrios Chalkondyles, but he was still active at the beginning of the sixteenth century. From 1490 until 1504 his name appears eight times in the register of borrowing of the Vatican Library. Besides, in February 1506 Pope Julius II sent a Motu proprio in favour of Damilas.Footnote 39 Damilas was assigned to copy a certain number of diverse texts for the pontifical library. Thus, a dating of V3 at the beginning of the sixteenth century is very possible,Footnote 40 and allows a dependence of V3 on the Aldina.
Of the two remaining manuscripts in this group, the codex u is a copy of the Aldina, whereas Nc derives from V3.Footnote 41 Although the Aldina and V3 are to a large extent identical, there are some minor errors where V3 and Nc agree against the Aldina: for example at 850a29, 851b31, 852a1, 852a9, 852b2, 852b12 and 857a29.
IV. CONTAMINATED AND INCOMPLETE MANUSCRIPTS
Finally, we must mention a group of manuscripts which are of no great value for ascertaining Aristotle's text. Some of these manuscripts contain a mixture of readings from different manuscript families, as for example Na; others contain only extracts of the Mechanics, for example V2 and Bu.
From the fourteenth century originates the codex V2:
V2 Vat.gr. 2231, fourteenth century, 1317–38. Paper, 206 x 140 mm, 281f. Inter alia Hierocles, in CA; Epictetus, Ench.; Simpl. in Epict.; Theophr. Sign.; Arist. excerpt from Mech. (f. 275–81).
This manuscript contains excerpts from the Mechanics, in which even whole problems have been left out. Thus there are many lacunae: at 848b9–850a29, 851b6–851b14, 851b40–852a8, 852a23–37, 852b11–21, 853a5–b13, 853b22–857a4, 857a22–b20, 857b26–38 and 858a1–12. Still, it is possible to determine with which family V2 is affiliated. The first scholium from family c, at 848a32, is present in this manuscript, and there are further variants that show a proximity to family c: at 850a35, 850b30, 857a12, 858a14, 858b9 and 858b25. This codex also includes modifications and expressions that cannot be found in any other manuscript and are probably inventions by the scribe: for example at 850b4 ἔχɛι V2: γράφɛι cett.; 850b25 μέρος αὐτῆς V2: μέρος cett.; 852b3 λίθον V2: τὸ βέλος cett.; 852b9 θᾶττον καθὼς προɛίρηται V2: θᾶττον cett.; 857b23 διότι V2: πότɛρον ὅτι cett.; 858a19 καὶ τοῦθ’ πάλιν ἕτɛρον V2: καὶ τοῦθ' ἕτɛρον cett.; 858b10 φɛρόμɛνον ὥστɛ ἐν δυσὶ κύκλοις ɛἶναι V2: ϕɛρόμɛνον cett. V2 does not derive from any of the extant manuscripts in family c, since it is considerably older than the other manuscripts contained in this family. It might be a direct copy of the hyparchetype γ, which is possible if we assume that this hyparchetype was composed in the first quarter of the fourteenth century.Footnote 42 Nevertheless, V2 is not a very valuable witness of the textual transmission since, based on its individual character, it is remote from the authentic Aristotelian text.
The codex Na, which was copied in about 1470, is a highly contaminated manuscript containing aspects of family a (for example at 848b24, 856a34, 856b28 and 857a12), family b (at 855b26, 857a38) and family c (at 850a35, 855b16). In addition it has many variants of its own: for example at 851a33 μέσον τι τοῦ κινουμένου Na: τι τοῦ κινουμένου μέσον cett.; 851b9 μὲν οὖν Na: μὲν cett.; 852a16 ὡσαύτως Na: ὁμοίως cett.; 853b32 ἐπίκɛιται Na: πρόσκɛιται a: ἔγκɛιται cett.; 857a5 μικρὰ/ἀπὸ μίκρου Na: μακρὰ/ἀπὸ μακροῦ cett. These readings from different manuscript families and own variants were not added or produced by emendation at a later stage, but are perfectly integrated in Na; therefore this manuscript must stem from a source that was already contaminated.
The manuscript Pt, which was consulted by van Cappelle for his edition of the Mechanics, belongs to a different tradition and is transmitted together with astrological works. It can be dated somewhere between 1370 and 1380, and was possibly copied by the Byzantine astrologer Johannes Ambramios. Pt contains a revised version of the treatise. This is already evident from the title, which is rather paraphrastic: αἰτιολογία τῆς τῶν μηχανικῶν ἐνɛργίας ὥς τινɛς λέγουσιν Ἀριστοτέλους. The peculiar readings, lacunae (at 849b7–852a14 and 855a27–858b31) and paraphrastic character of Pt make it impossible to ascertain its dependence, although there are some similarities with family a: for example at 848b24, 853a27, 853b32, 855a3 and 855a13.
Manuscripts P4 and Ps both contain important variants from family c, but cannot definitely be attributed to this family. P4 contains the greater part of the readings from family c (at 850a35, 852b17, 854a20, 854a36, 854b9, 854b14, 855b16, 856a1, 857a12, 857b13 and 858b9) but none of the scholia or omissions that are particular to this family. The fact that there are no emendations or additions in P4 itself implies that it is based on some earlier manuscript, not preserved, which was already corrected by means of a manuscript from another family.
The other Parisian manuscript, Ps, contains not only the majority of the readings from family c (at 848b10, 850a35, 850b18, 852b17, 854a20, 854b9, 855b16, 856a1, 857a12, 858b9 and 858b25), but also the scholia from this family and its main omissions, which were filled in in the margins. But Ps fits not unproblematically into this family, since some readings from family a are present too. These readings are for one part added in a later stage and suggest an emendation with the manuscript L. For example, at 847b25 a scholium contained exclusively in L has been added in the margins; at 848a11 μαθημάτων has been added; lines 850b7–8 are still absent after completing 850b6–9; at 853a27 ἐϕ' ῷ ΔΕΗΖ has been added in the margin; at 858b11 τότɛ μένɛι was probably written over an erased τιθέαμɛν ɛἰ. For another part these variants from family a do not appear to be later additions: for example 848b24 πλɛυρῶν and 850a1 μέρɛι. The exemplar of Ps might have been an already contaminated manuscript. Further Ps contains many peculiar errors and paraphrastic passages, and is therefore not of special interest for the reconstruction of the text of the Mechanics.
V. STEMMA CODICUM
The results presented above can be illustrated as follows:
Contaminated manuscripts: Bu, Na, Pt, P4 and Ps.
In a new critical edition of the Mechanics, the manuscripts of all three families should be analysed, with special interest in those manuscripts that can be traced back to a hyparchetype, namely Ha, L, P, Mv, Wa and V1. Family b, however, does not have the same status as the other manuscript families. Since the manuscripts in this family have a deteriorated and contaminated ancestor, the editor is not always in a favourable position to recover the authentic reading, when two families agree against the third. It is quite possible that families b and c agree, whereas family a contains the correct variant, e.g. at 848b24 πλɛυρῶν a: πλɛιόνων cett.; 854b27 πλɛυράν a: πλɛίω cett.; 855b24 μένɛιν a: μὲν cett. Although family b is contaminated, it looks back to a different tradition and certainly contains some valuable readings. The editor should, however, carefully consider if a peculiar variant in P or Mv contains an authentic reading or an error.
An analysis of all three families would be a clear improvement upon the previous editions of the treatise, of which Bekker's edition is still the most complete; manuscripts from families b and c provided the main sources for his text, while manuscripts from the oldest family, family a, are accorded only secondary importance. Although Bottecchia studied almost all the manuscript material, she neglected to consult family c for her edition, and thereby overlooked some important readings. She did not consider for example the variant from family c at 850a35 ἀντὶ σπαρτίου γίνɛται. Nevertheless, she included some other readings from family c in her edition, as for example at 849a36 ὀρθὰς; 849b18 τὸ κατὰ; 852b17 θᾶττον; but she did not recognize these variants as belonging to family c and ascribed them to Bekker or the manuscript Na.
The editor might further reconsider the importance of family a. Bottecchia mostly favoured the readings from the manuscript V3, and included the variants from family a in her apparatus. There are, however, still certain passages in the Mechanics where this family alone might contain the correct reading, e.g. at 858a1 ɛὐθɛίας a: ἴσας cett.
A new examination of all three families will most likely favour certain variants different from the ones favoured in previous editions of the treatise, but a far more important task for a new edition is to remove all foreign influences from the text of the Mechanics. The Byzantine paraphrase by Georgios Pachymeres, and to some extent the variants offered by the humanist Niccolò Leonico Tomeo, have influenced all previous editions of the treatise, including the latest edition by Bottecchia. These paraphrastic traces should be individually evaluated, but have to be separated from the authentic Aristotelian text.Footnote 43
