In this final section of the Introduction, links have been omitted because they would have been more distracting than useful.
Considerable interest attaches to the study of the manuscripts of Quintilian, the oldest of which may be grouped in three main divisions: (1) the complete manuscripts, (2) the incomplete, and (3) the mixed.
The most important representative of the first class is the Codex Ambrosianus, a manuscript of the 10th or 11th century, now at Milan. As we have it now, it is unfortunately in a mutilated condition, nearly a fourth part of the folios having been lost (from ix. 4, 135 argumenta acria et cit. to xii. 11, 22 antiquitas ut possit). Halm secured a new and trustworthy collation of this MS., distinguishing carefully between the original text and the readings of the second hand.
Although now in the defective condition above indicated, the Ambrosianus must have been originally complete. In this it differs from the representatives of the second family of MSS., the most valuable member of which—the Bernensis—is of even greater importance for the constitution of the text than the Ambrosianus, at least in those parts which it contains. It is the oldest of all the known manuscripts of Quintilian, belonging to the 10th century. The peculiarity which it shares with the other members of its family is that it contains certain great lacunae, which must have existed also in the manuscript from which it was copied, as they are indicated in the Bernensis by blank spaces. The size of the first lacuna varies with the fortunes of the particular codex: in the Bernensis it extends from the beginning to 2 §5 (licet, et nihilo minus). The others are identical in all cases: v. 14, 12—viii. 3, 64: viii. 6, 17—viii. 6, 67: ix. 3, 2—x. 1, 107 (nulla contentio): xi. 1, 71—xi. 2, 33: and xii. 10, 43 to the end.
To the same family as the Bernensis belongs the Bambergensis A, which was directly copied from the Bernensis not long after the latter had been written: it also is of the 10th century. But inasmuch as in the Bambergensis the great lacunae were, at a very early date, filled in by another hand (Bambergensis G73), this manuscript may now rank in the third group, where it became the parent, as I hope to show below, of the Harleianus (2664), and through the Harleianus of the Florentinus, Turicensis, and an innumerable company of others. Besides reproducing Bambergensis G, these MSS. follow for the most part the readings introduced by a later hand (called by Halm b) into the original Bambergensis A. A recent examination of the Bambergensis has suggested a doubt whether the readings known as b, which are often of a very faulty character, can have been derived from the same codex as G.
Halm’s critical edition of Quintilian is founded, in the main, on the manuscripts above mentioned, with a few examples of the 15th century for the parts where he had only the Ambrosianus and the Bambergensis G, or the latter exclusively, to rely on. Since the date of the publication of his text (1868) great progress has been made with the critical study of Quintilian. In 1875 MM. Chatelain and le Coultre published a collation of the Nostradamensis (see below), the main results of which have been incorporated in Meister’s edition (1886-87). And in a critical edition of the First Book of the Institutio (1890) M. Ch. Fierville has given a most complete account of all the continental manuscripts, drawing for the purpose on a previous work in which he had already shown proof of his interest in the subject (De Quintilianeis Codicibus, 1874).
There can be little doubt that Halm’s critical instinct guided him aright in attaching supreme importance to the Bernensis (with Bambergensis A), the Ambrosianus, and Bambergensis G. But much has been derived from some manuscripts of which he took no account, and there is one in particular, which has hitherto been strangely overlooked, and to which prominence is accordingly given in this edition. Before proceeding to deal with it, I shall annex here a brief notice of the various MSS. which figure in the Critical Notes, grouped in one or other of the three divisions given above. An editor of the Tenth Book of the Institutio is especially bound to travel outside the rather narrow range of Halm’s critical edition, as so much of the existing text (down to 1 §107) has been based mainly on Bambergensis G alone. In addition to collating, for the purposes of this edition, such MSS. as the Ioannensis at Cambridge, the Bodleianus and Balliolensis at Oxford, and the very important Harleian codex, referred to above, I have also carefully compared eight 15th century manuscripts in the hope (which the Critical Appendix will show to have been not entirely disappointed) of gleaning something new. This part of the present work may be regarded as supplementing, for this country, what M. Ch. Fierville has already so laboriously accomplished for the manuscripts of the Continent.
Of the first family, the outstanding example is the Ambrosianus. The resemblances between it and Bambergensis G are sufficient to show that the manuscript from which the latter was copied probably belonged to the same class. But this manuscript, which must have been complete (like the Ambrosianus originally), has altogether disappeared: one of the great objects for extending the study of the MSS. of Quintilian beyond the limits observed by Halm is the hope of being able to distinguish between such examples as may seem (like the Dorvilianus at Oxford) to preserve some of the traditions of the family, and those whose origin may be clearly traced back to Bambergensis A and G. For all the complete MSS. of Quintilian in existence must be derived either from this family or from the mixed group of which the Bambergensis, in its present form, seems to be the undoubted original.
In the second group we must include, not much inferior to the Bernensis, the Parisinus Nostradamensis (N) Bibl. Nat. fonds latin 18527. It is an independent transcript in all probability of the incomplete MS. from which the Bernensis was copied, and as such has a distinct value of its own. It is ascribed to the 10th century. For the readings of this codex I have been able to compare a collation made by M. Fierville in 1872, with that published by MM. Chatelain and le Coultre in 1875.
Then there is the Codex Ioannensis (in the library of St. John’s College, Cambridge), a recent examination of which has shown me that the account given of it by Spalding (vol. ii. pr. p. 4) must be amended in some particulars. In its present condition it begins with constaret (i. 2, 3), but a portion of the first page has been cut away for the sake of the ornamental letter: originally the MS. must have begun at the beginning of the second chapter, like the Nostradamensis, the Vossiani 1 and 2, the Codex Puteanus, and Parisinus 7721 (see Fierville, p. 165). Again, the reading at xi. 2, 33 is clearly multiplici, not ut duplici, and in this it agrees with the Montpellier MS. (Pithoeanus), which is known to be a copy (11th century) of the Bernensis (see M. Bonnet in Revue de Phil. 1887). A remarkable feature about this MS. is the number of inversions which the writer sets himself to make in the text. These I have not included in the Critical Notes, but some of them may be subjoined here, as they may help to establish the derivation of this manuscript. The codex from which it was copied must have been illegible in parts: this is probably the explanation of such omissions (the space being left blank) as tum in ipsis in x. 2, 14, and virtutis ib. §15. It is written in a very small and neat hand, with no contemporary indication of the great lacunae, and may be ascribed to the 13th century. It agrees generally with the Bernensis, though there are striking resemblances also to the Pratensis (see p. lxiii and note). Among the inversions referred to are the following:—x. 3, 1 sic etiam utilitatis, for sic utilitatis etiam: ib. §30 oratione continua, for continua oratione: 5 §8 alia propriis alia translatis virtus, for alia translatis virtus alia propriis: 7 §21 stultis eruditi, for stulti eruditis: ib. §28 solum summum, for summum solum. Some of these peculiarities (e.g. the inversion at 5 §8) it shares with the Leyden MSS.—the Vossiani i. and iii., a collation of which is given in Burmann’s edition: these codices M. Fierville assigns to that division of his first group in which the Nostradamensis heads the list (see below, p. lxiv). I may note also the readings viderit bona et invenit ( 2 §20), which Ioan. shares with Voss. iii.: potius libertas ista ( 3 §24) Ioan. and Voss. i.; ubertate—for libertate—( 5 §15) Ioan. Voss. i. and iii.
To the same family belongs the Codex Salmantinus, a 12th or 13th century manuscript in the library of the University of Salamanca. M. Fierville, who kindly placed at my disposal his collation of the Tenth Book, thinks it must have been indirectly derived from the Bernensis. He notes some hundred variants in which it differs from the Nostradamensis (most of them being the errors of a copyist), and some thirty-seven places in which, while differing from the Nostradamensis, it agrees with the Bernensis and the Bambergensis. This MS. also gives ubertate in 5 §15 : it agrees in showing the important reading alte refossa in 3 §2 : and resembles the Ioannensis in certain minor omissions, e.g. certa before necessitate in 5 §15 : idem before laborandum in 7 §4 : et before consuetudo in 7 §8 : cp. subiunctura sunt for subiuncturus est 7 §9 . For other coincidences see the Critical Appendix.
In the same group must be included two MSS. of first-class importance for the text of Quintilian, for a collation of which (as of the Codex Salmantinus) I am indebted to the kindness of M. Fierville. They are the Codex Pralensis (No. 14146 fonds latin de la Bibliothèque nationale), of the 12th century, and the Codex Puteanus (No. 7719) of the 13th. The former is the work of Étienne de Rouen, a monk of the Abbey of Bec, and it consists of extracts from the Institutio amounting to nearly a third of the whole. There are eighty sections, of which §76 (de figuris verborum) includes x. 1 §§108-131; §77 (de imitatione) consists of x. 2, 1-28; §78 (quomodo dictandum sit) of x. 3, 1-32; and §79 (de laude scriptorum tam Graecorum quam Latinorum) of x. 1, 46-107. The importance of this codex arises from the fact that it is an undoubted transcript of the Beccensis, now lost. The Beccensis is supposed by M. Fierville (Introd. p. lxxvii. sq.) to have belonged to the 9th or 10th century, in which case it would take, if extant, at least equal rank with the Bernensis. That it was an independent copy of some older MS. seems to be proved, not only by the variants in the Pratensis, but also by the fact that both the Pratensis and the Puteanus (which is also a transcript of the Beccensis) show a lacuna after the word mutatis in x. 3, 32. This lacuna must have existed in the Beccensis, though there is no trace of it elsewhere. Guided by the sense, Étienne de Rouen added the words correctum fuisse tabellis in his copy (the Pratensis): the text runs codicibus esse sublatum.
The general character of the readings of the Pratensis may be gathered from a comparison of passages in the Critical Appendix to this volume. Among other variants, the following may be mentioned,—and it will be seen that certain peculiar features in some of the MSS. used by Halm (notably S) probably arose either from the Pratensis or from its prototype, the Beccensis. At x. 1. 50 Prat, gives (like S) rogantis Achillen Priami precibus, while most codd. have Priami before rogantis: ib. §53 eloquentie (so Put. S 7231, 7696) for eloquendi: ib. superatum (so Put.) for superari: §55 aequalem credidit parem (as Put. S Harl. 2662, 11671): §67 idque ego (as Put. S) for idque ego sane: §68 qui fuerunt and also vero, omitted (as in Put. S): so also tenebras §72, valuerunt §84 (as 7231, 7696), and veterum §97: at §95 Prat, gives et eruditissimos for et doctissimos, and hence the omission of erudit. in S. On the whole, the study of the text of the Pratensis seems to give additional confidence in the readings of G: for example §98 imperare (as Put.): §101 cesserit (Put. 7231, 7696): ib. nec indignetur. Étienne de Rouen carefully omitted all the Greek words which he found in his original, and this strengthens the contention that φράσιν in 1 §87 (see Crit. Notes, and cp. §42) was originally written in Greek. At 2 §20 quem superius institui for quem institueram in libra secundo is an indication of the fact that Étienne de Rouen was making a compendium of the Institutio, and not transcribing the whole treatise. This probably detracts from the significance of those readings which seem to be peculiar to the Pratensis, among which may be noted 1 §48 putat for creditum est (where Put. has certissimum): §59 ad exemplum maxime permanebit (ad exitum Put. and S): §78 propinquior for propior: §80 mediocri for medio: §81 assurgit for surgit: §109 in utroque for in quoque. Peculiar readings which Prat. shares with the Puteanus (and which were therefore probably in the Beccensis) are §46 in magnis for in magnis rebus: §49 innuit for nuntiat: §50 excessit for excedit: §54 ne virtus for ne utrius (neutrius): §57 ignoro ergo (S) for ignoro igitur: §63 plurimumque oratio: §68 in affectibus communibus mirus: §79 discernendi for dicendi: §107 nominis latini for latini sermonis. At 1 §72 Prat. has qui ut a pravis sui temporis Menandro (Put. ut pravis), and this became in S Harl. 2662 and 11671 qui quamvis sui temp. Men. There are frequent inversions, e.g. dicendi genere §52 (Put.): Attici sermonis (Put.) §65: plus Attio (Put.) §97: cuicumque eorum Ciceronem (as Put. 7231, 7696) §105: sit nobis §112: est autem (as Ioan.) §115: forum illustrator (as Ioan.) §122: creditus sum §125: dignis lectione 2 §1: possumus sperare §9: nemo vero eum §10: aliquo tamen in loco aliquid §24: scientia movendi §27: ipso opere 3 §8: se res facilius §9: desperatio etiam §14: vox exaudiri §25: praecipue in hoc §26: possunt semper §2874.
From the list of readings given above, it will be seen that the Codex Puteanus is in general agreement with the Pratensis, each being an independent copy of the same original. The variants given by this MS. will be found in the Critical Appendix for the part of the Tenth Book collated by M. Fierville, 1 §§46-107. At times it is even more in agreement than the Pratensis with the later family, of which Halm took S as the typical example: e.g. 1 §61 spiritu: ib. merito omitted: §72 possunt decernere (for possis decerpere—possis decernere Prat.).
In the arrangement introduced by Étienne de Rouen in the Pratensis, the last two sections (§§79 and 80) consist respectively of x. 1 §§46-107, and xii. 10 §§10-15. These portions of the Institutio must have formed part of the mutilated original from which the Beccensis was copied, and they have been reproduced separately along with 1 §§108-131 in two Paris MSS. (7231 and 7696), a collation of which has been put at my disposal by M. Fierville. The first is a mixed codex of the 12th century, containing nine separate works, of which the extracts from Quintilian form one. The second, also of the 12th century, belonged to the Abbey of Fleury-sur-Loire, and comprises five treatises besides the Quintilian. In both the title runs Quintilianus, libro Xº Inst. Orator. Qui auctores Graecorum maxime legendi. M. Fierville states (Introd. p. lxxxv.) that of forty-five variants which he compared (x. 1 §§46-68) in the Pratensis, Puteanus, 7231, and 7696, twenty-eight occur in the two former only, eight in the two latter, and nine in all four. He adds that the Vossiani i. and iii. resemble the two former more nearly than the two latter. Both 7231 and 7696 agree in giving the usual collocation at §50 illis Priami rogantis Achillen: at §59 the former has ad exim, the latter ad exi: at §61 both give eum nemini credit, omitting merito (as Put. and S): at §68 namque is et sermone (as Prat.: namque sermone Put.): ib. in dicendo ac respondendo (Prat. Put. in dicendo et in resp.): §72 (apparently) ut pravis sui temporis iudiciis: §82 finxisse sermonem (as Prat. Put. and most codd.): §83 ac varietate: §88 laudandus partibus (laudandis part. Prat. Put. Harl. 2662, 11671): §91 visum (visum est Prat. Put.): §98 senes non parum tragicum (Prat. Put. Harl. 2662, 11671): §107 Latini nominis: §121 leve (Prat. N). In §98 Thyestes is omitted in both (also in Prat. Put.): is this a sign that the name was written in Greek in the original? In 7231 I have noted two inversions which do not seem to appear in 7696: dedit exemplum et ortum 1 §46: proximus aemulari §62.
M. Fierville classifies the various members of the whole family of MSS. which has just been reviewed in five sub-divisions. The first includes the Bernensis, Bambergensis A, Ambrosianus ii., Pithoeanus (these two are direct copies of the Bernensis), Salmantinus, three Paris codices (7720, 7722 and Didot), and probably the Ioannensis. In the second he ranks the Nostradamensis, Vossiani i. and iii., and a Paris MS. (7721): in the third the Beccensis, Pratensis, and Puteanus: in the fourth a codex Vaticanus, referred to by Spalding: and in the fifth the fragments just dealt with (7231, 7696). Of these he rightly considers the Bernensis, Bambergensis, Nostradamensis, Pratensis, and Puteanus to be of greatest importance for the constitution of the text.
At the head of the third group of the manuscripts of Quintilian must now be placed the Codex Harleianus (2664), in the library of the British Museum75. This manuscript was first described by Mr. L. C. Purser in Hermathena (No. xii., 1886); and to his notice of it I am now able to add a statement of its history and a pretty certain indication of the relation it bears to other known codices. As to date, it cannot be placed later than the beginning of the 11th century. There are in the margin marks which show clearly that at an early date it was used to supply the great lacunae in some MS. of the first or incomplete class; one of these should have appeared in the margin of the annexed facsimile, a being used at the beginning and b (as here x. 1, 107) at the end of the parts to be extracted. The manuscript contains 188 folios and 24 quaternions, and is written in one column. At the beginning the writing is larger than subsequently, just as the first part of the Bambergensis is larger than G, which the Harleianus (H) closely resembles. On fols. 90-91 the hand is more recent, and the writing is in darker ink: fols. 61-68 seem to have been supplied later. There is a blank of eight lines at the end of 161v., where Book xi. ch. 1 concludes; ch. 2 begins at the top of the next page. There is also a blank line (as in Bn and Bg) at iii. 8, 30, though nothing is wanting in the text.
The result of my investigations has been to identify this important manuscript with the Codex Dusseldorpianus, which we know disappeared from the library at Düsseldorf before Gesner’s time. In the preface to his edition of 1738, §20, he describes it, on the evidence of one who had seen it, as ‘Poggianis temporibus certe priorem, necdum, quod sciatur, recentiori aetate a quoquam collatum’: its remarkable freedom from variants and emendations suggests that it must have lain long unnoticed. When Gesner wanted to refer to it, he found it was gone: ‘tandem compertum est mala fraude nescio quorum hominum et hunc et alios rarissimos codices esse subductos.’ It had, in fact, been sold by the Düsseldorf librarian, possibly acting under orders. The diary of Humphrey Wanley, Harley’s librarian, shows that he bought it (along with several other manuscripts) on the 6th August, 1724, from Sig. John James Zamboni, Resident Chargé d’Affaires in England for the Elector of Hesse Darmstadt. Zamboni’s correspondence is in the Bodleian at Oxford; and I have ascertained, by examining it, that he received the Harleian manuscript of Quintilian from M. Büchels, who was librarian of the Court library at Düsseldorf in the beginning of last century, and with whom Zamboni drove a regular trade in manuscripts.
‘The correspondence’ (to quote from what has already been written elsewhere) ‘is of a very interesting character, and throws light on the provenance of several of the Harleian MSS. The transactions of the pair begin in 1721, when Büchels receives 1200 florins (not without much dunning) for a consignment of printed books. Zamboni, who was something of a humourist, is constantly endeavouring to beat down the librarian’s prices: “j’aime les beaux livres,” he says on one occasion, when pretending that he will not entertain a certain offer, “j’aime les beaux livres, mais je ne hais pas l’argent.” The trade in MSS. began in 1724, when Büchels sent a list from which Zamboni selected eleven codices, assuring his correspondent that if he would only be reasonable they would soon come to terms. Early in the year he offers 500 florins for the lot, protesting that he had no intention of selling again: “sachez, Monsieur, que je ne vous achète pas les livres pour les revendre.” Three weeks after it came to hand, he made over the whole consignment to Harley’s librarian. It included our Quintilian and the great Vitruvius—the entries in Zamboni’s letters corresponding exactly with those in Wanley’s diary. In the end of the same month Zamboni is writing to Büchels for more, protesting that his great ambition is to make a “très jolie collection” of MSS. (Bodl. MSS. Add. D, 66).’
What the history of the Harleianus may have been before it came to Düsseldorf, I have been unable to ascertain. The only clue is a scrawl on the first page: Iste liber est maioris ecclesiae. This Mr. Purser has ascribed, with great probability, to Strasburg. The Codex Florentinus has an inscription showing that it was given by Bishop Werinharius (the first of that name, 1000-1029?) to the Cathedral of St. Mary at Strasburg; and Wypheling, who made a catalogue of the library there (circ. 1508), says of this bishop: ‘multa dedit ecclesiae suae praesertim multos praestantissimos libros antiquissimis characteribus scriptos; quorum adhuc aliqui in bibliotheca maioris ecclesiae repositi videntur.’ This shows that there was a greater and a less church at Strasburg, to the latter of which the MS. may formerly have belonged. And if, as is now generally believed, neither the Florentinus nor the Turicensis can be considered identical with the manuscript which roused the enthusiasm of the literary world when Poggio discovered it in 1416, it is not impossible that we may have recovered that manuscript in the Harleianus, if we can conceive of its having migrated from Strasburg to St. Gall.
The following paragraph appeared in the book as a single-sheet Addendum labeled “Place opposite p. lxvi.” Its original location was therefore at the point “...the insertion at a wrong place in the // text...” in the second paragraph after the Addendum.
Writing in the ‘Neue Heidelberger Jahrbücher’ (1891, p. 238 sqq.), Mr. A. C. Clark, of Queen’s College, Oxford, supplies some very interesting information in regard to Zamboni’s purchases. It seems that Zamboni was able to tell Lord Oxford’s librarian that the MSS. which he was selling to him had originally belonged to Graevius; and by comparing the Zamboni correspondence in the Bodleian Library with the posthumous catalogue of Graevius’s library, Mr. Clark has now discovered that Büchels was offering to Zamboni the entire MSS. collection of that great scholar, which in this way ultimately found a home in the library of the British Museum. Graevius died in 1703, and the Elector Johann Wilhelm bought both his books and his manuscripts. The former he presented to the library of the University of Heidelberg: the latter he retained in his own library at Düsseldorf. In regard to the Harleian codex of Quintilian, Mr. Clark’s theory is that it belonged formerly not to Strasburg, but to the cathedral at Cologne, which is more than once referred to as ‘maior ecclesia.’ Gesner must have been in error when he said that this codex had not been recently collated (cp. Introd. p. lxv); for Gulielmus had seen it at Cologne, and in his ‘Verisimilia,’ iii. xiv, quotes some variants and ‘proprii errores’ from the preface to Book vi, all of which appear in the Harleianus as we have it now. And as Graevius is known to have borrowed from the library of Cologne Cathedral, in 1688, an important codex of Cicero ad Fam. (Harl. 2682), Mr. Clark infers that he got the Quintilian at the same time. He evidently omitted to return them; and after his death they passed, with many other MSS., first to Düsseldorf, and then to London.
It was only after the Bambergensis arrived in the British Museum (where it was sent by the authorities of the Bamberg Library, in courteous compliance with a request from me) that it was possible to form a definite opinion as to the place occupied by the Harleianus in regard to it. At first it appeared, even to the experts, that the latter MS. was distinctly of older date than the former: it is written in a neater hand, and on palaeographical grounds alone there might have been room for doubt. But a fuller examination convinced me that the Harleianus was copied directly from the Bambergensis, possibly at the very time when the latter was being completed by the addition of the parts known as Bambergensis G, and of some at least of the readings now generally designated b. These latter, indeed, the Harleianus slavishly follows, in preference to the first hand in the original Bambergensis: probably the copyist of the Harleianus was aware of the importance attached to the codex (uncial?) from which the b readings were taken. In view, however, of the defective state in which the Bambergensis has come down to us, as regards the opening part, and considering also the mutilation of the Ambrosianus, we may still claim for the MS. in the British Museum the distinction of being the oldest complete manuscript of Quintilian in existence.
The proof that the Harleianus stands at the head of the great family of the mixed manuscripts of Quintilian (represented till now mainly by the Florentinus, Turicensis, Almeloveenianus, and Guelferbytanus) is derived from a consideration of its relationship to both parts of the Bambergensis on the one hand, and to those later MSS. on the other. I begin with a point which involves a testimony to the critical acumen of that great scholar C. Halm. In the Sitzungsberichte der königl. bayer. Akademie der Wissenschaften zu München, 1866, i. pp. 505-6, Halm established the dependence of the Turicensis and the Florentinus on the Bambergensis by pointing out, among other proofs, the insertion at a wrong place in the text of both these codices of certain words which, having been inadvertently omitted by the copyist of the Bambergensis from their proper context, were written in by him in a blank space at the foot of the page in which the passage in question occurs. The passage is ix. 2, 52: circa crimen Apollonii Drepani[tani: gaudeo etiam si quid ab eo abstulisti et abs te] nihil rectius factum esse dico. When the copyist of the Bambergensis noticed his mistake, he completed Drepanitani in the text, and wrote in the words gaudeo etiam ... abs te at the foot of the page, with a pretty clear indication of the place where they were to be taken in. In the Bambergensis the page ends with the words (§54) an huius ille legis quam, and the next page continues C̣ḷọẹlius a se inventam gloriatur. Noticing that in both the Florentinus and the Turicensis the marginal addition (gaudeo etiam ... abs te) is inserted not after legis quam but after Clodius, Halm drew the inference that these codices were copied from the Bambergensis not directly, but through some intervening manuscript. The Harleianus is this manuscript. In it the words referred to do come in between legis quam and (Cloe)lius: indeed, so slavishly does the writer follow the second hand in the Bambergensis, in which the letters C l o e are subpunctuated, that the Harleianus actually shows et abs te lius a se inventa76, exactly as the writer of b wished the Bambergensis to stand. It must be feared that the copyist of the Harleianus did not know enough Latin to save him from making very considerable mistakes. If I am right in believing that this manuscript must take rank above the Turicensis and the Florentinus (and all other MSS. of this family), it is he who must be credited with a great deal of the confusion that has crept into Quintilian’s text. It may be well to mention one or two obvious examples. In ix. 3, 1 the text stands utinamque non peiora vincant. Verum schemata, &c. In the Bambergensis, utrum nam is supplied by b above the line, and in the margin que peiora vincant verum, the words affected by the change being subpunctuated in the text. The copyist of the Harleianus takes the utrum nam and leaves the rest, showing utrum nam schemata: this appears as utrim nam schemata in the Turicensis, and as utinam schemata in the Florentinus and Almeloveenianus. Take again ix. 3, 68-9 in the Bambergensis (G) quem suppli[catione dignum indicaris. Aliter quoque voces aut] eadem aut diversa, &c. The words enclosed in brackets are the last line of a particular column (142 v.); they were inadvertently omitted by the copyist of the Harleianus, and as a consequence we have supplici in Turic. and Flor., supplitia in Guelf., &c. Again at x. 7, 20 a certain sleepiness on the part of the scribe of the Harleianus, which caused him to write Neque vero tantas eum breve saltem qui foro tempus quod nusquam fere deerit ad ea quae, &c., has given rise to the greatest confusion in Turic., Flor., Alm., Bodleianus, Burn. 243, &c. In this H follows exactly the second hand in Bg., except for the remarkable insertion of the words qui foro between breve saltem and tempus: at this point the copyist of H must have allowed his eyes to stray to the beginning of the previous line in Bg, where the words qui foro hold a conspicuous position. Flor. and Tur. repeat the mistake, except that the latter gives eum brevem for eum breve. Again at the end of Book ix, Bambergensis G gives ut numerum spondet flexisse non arcessisse non arcessiti et coacti esse videantur: this reading is identical with that of the Harleianus, except that the latter for arcessiti gives arcessisti, a deviation promptly reproduced by the Florentinus, while the Turicensis shows accersisti. Perhaps the most conclusive instance of all is the following: at iv. 2, 128 the Bambergensis gives (for ἐπιδιήγησις) ΕΠΙΔΙΗΤΗϹΕΙ: this appears in H as ΕΠΙΔΙΗϹΕΙ the seventh and eighth letters having been inadvertently omitted by the copyist. F makes this ΕΠΙΘΕϹΙΕ and T shows ΕΠΙΘϹΙϹ (επιλιησει—Spalding).
The four forms of the Greek word appear in the printed text as:
As the Bambergensis (Bg), in its present state, only commences at i. 1. 6. (nec de patribus tantum), the readings of the Harleianus (H) are for the Prooemium and part of chapter 1 of first-class importance. In the pr. §1 we have pertinerent H, pertinent T: §2 diversas H, divisas T: §5 fieri oratorem non posse HF, fieri non posse oratorem T (as A): §6 amore H, studio F: iτ ingenii H, iter ingenii T, ingenii F: §13 officio quoque H, quoque officio F: §19 summa H (also Bg), summam T: §25 demonstraturi HF, demonstrari T: §27 adiumenta H (a correction by same hand on adiuvante): so Bg F: adiuvante T. In chap. 1 §3 sed plus HT: sed et plus F: hoc quippe viderit H Bg F: hoc quippe (om. viderit) T.
These instances are taken from the introductory part of the First Book, where Bg almost entirely fails us, only a few words being here and there decipherable. Wherever I have compared, in other places, the readings of Bg (and G), H, T, and F, I have found H, if not always in exact agreement with the Bamberg MS. (often owing to the copyist’s ignorance of Latin) invariably nearer the parent source than either T or F. Here are a few instances from the First Book: I §8 nihil est peius Bg H T, nihil enim est peius F: ib. §11 defuerit Bg H T, defuerint F: ib. §12 perbibet Bg H F, perhibet T: ib. §16 formandam Bg H, formandum F T: 2 §18 in media rei p. vivendum Bg (b) H, in med. rei praevivendum T, reip. videndum F: ib. §24 depellendam Bg H, repellendam T: ib. §31 concipiat quis mente Bg H, quis mente concipiat F: 4 §27 tereuntur Bg H T, intereuntur F: 6 §9 dicet Bg, dicit H F, dicitur T: ib. §14 dici ceris Bg (dici ceris),A diceres H, dici F T: ib. §30 aliaque quae consuetudini serviunt Bg H,—in margin of H aliquando consuetudini servit (b): F and T adopt the latter, and give the alternative reading in the margin: 10 §28 haec ei et cura H F, haec et cura ei T: 11 §4 pinguitudine Bg H, pinguedine F T. Among scattered instances elsewhere are the following: ii. 5, 13 dicentur Bg H, docentur T: 5 §26 hanc Bg H, om. T: 15 §8 testatum est Bg H, testatum T. In ix. 363 G has parem (for marem A): H gives patrem and F T follow suit: cp. ix. 4, 8 hoc est G H, id est F: ib. §16 quoque G H, om. T: ib. §32 nesciat G H, dubitet F: dignatur G H, digne dicatur F: viii. pr. §3 dicendi G H, discendi T: ix. 4, 119 ignorabo G, ignoraba H, ignorabam T: ib. §129 et hac fluit G H, et hac et hac fluit T: xii. 11, 8 scierit G, scieret H, sciret T: ib. 2 §18 autem Bg H, om. T: x. 1, §4 numuro quae G H, num muro quae T, numeroque F: ib. §50 et philogus G, et philochus H T, et epiloghus F: ib. §73 porem G H, priorem F T: ib. §75 vel hoc est G H, hoc est vel T: x. 2, 7 posteriis (for historiis) H, posteris F (posterius ed. Camp.): x. 2, 10 discernamus Bg, discernantur b, disnantur H T, desinantur F. Noteworthy cases of the close adherence of T to H are the following: Empedoclena i. 4, 4: vespueruginem i. 7, 12: tereuntur i. 4, 27: flex his x. 1, 2: gravissimus x. 1, 97: ipsae illae quae extorque eum credas x. 1, 110, where both also give trans usum for transversum, and non repe for non rapi: morare refinxit finxit recipit x. 3, 6: nam quod cum isocratis x. 4, 4. In other instances the writer of T has evidently tried to improve on the reading of H: e.g. in the title of Book x, H gives an abbreviation which T mistakes for quo enim dandum: also extemporal facilitas which appears in T as extempora vel facilitas: x. 1, 79 ven iudicis H (in mistake for se non iud.), which is made by T into venit iudicis. Many similar instances could be cited in regard to both T and F; the reading tantum, for instance, in x. 1, 92, which occurs in both, has evidently arisen from H, which here shows something that looks more like tantum than tacitum (the reading of G). Again, in every place where Halm uses the formula ‘F T soli ex notis,’ H will be found to correspond77.
A. (dici ceris) text image showing inserted letters:
With such evidence as has been given above, it is impossible to doubt that the Harleianus must now take rank above both the manuscripts which, before the appearance of Halm’s edition, held so prominent a place in the criticism of Quintilian, the Codex Florentinus and the Codex Turicenis. The former is an eleventh century MS., now in the Laurentian library at Florence. On the first page is this inscription: Werinharius episcopus dedit Sanctae Mariae: on the last Liber Petri de Medicis, Cos. fil.: and below Liber sanctae Mariae ecclesiae Argñ. (= Argentoratensis) in dormitorio. There were two bishops of Strasburg bearing the name of Werner: the first 1001-1029, and the second 1065-1079. M. Fierville (Introd. p. xciv) tells us that the first Werner (of Altemburg or Hapsburg) laid the foundations of the cathedral at Strasburg in 1015, and presented to the Chapter a number of valuable books; and we also know that in 1006 he had attended the Council at Frankfort to promote the erection of a cathedral church at Bamberg. Here then we have the elements of a solution of the problem. Bishop Werner was a patron of letters; and learning that by the addition of what is now known as Bambergensis G a complete text of Quintilian had been secured, he had it copied. The Codex Harleianus was in all probability the first copy, and from it the Codex Florentinus was reproduced. The latter was still at Strasburg in 1372, a fact which (though hitherto it seems to have been unnoticed) is enough to dispose of its claim to be considered the manuscript of Poggio, which he describes as ‘plenum situ’ and ‘pulvere squalentem’ lying ‘in teterrimo quodam et obscuro carcere, fundo scilicet unius turris, quo ne capitales quidem rei damnati retruderentur.’ If so important a MS. had passed from Strasburg to St. Gall within forty years of Poggio’s visit, it is hard to believe that it would have been allowed to lie neglected and unknown. After 1372 we know nothing certain of its history till it reappears in the library of the Medicis at Florence in the latter part of the fifteenth century. It is generally supposed that some time between 1372 and 1417 it must have been transported from Strasburg to the monastery of St. Gall, and that it passed from there to Florence after Poggio’s departure. A similar theory may quite as legitimately be maintained in reference to the Harleianus, which, as I have already indicated, may be the very manuscript which Poggio discovered at St. Gall in 141678.
The Codex Turicensis was long considered to be of older date than the Florentinus, but recent investigations seem to have proved the contrary. Halm attributes it to the second part of the eleventh century, and E. Wölfflin takes a similar view. In the beginning of the eighteenth century it passed into the library at Zürich. Spalding believed it to be the manuscript discovered by Poggio, and M. Fierville is of the same opinion: Halm rejects this theory. The great point in favour of the claim of the Turicensis is that it is known to have come from St. Gall, while we can only conjecture the history of the Harleianus. But the Turicensis cannot have been the MS. which Poggio carried with him into Italy, according to a statement made by Bandini, Regius, and others. It is true that this statement is hard to reconcile with what Poggio himself says in his letter to Guarini, whom he informs that he has made hasty transcripts of his various ‘finds’ (presumably including the Quintilian) for his friends Leonardo of Arezzo and Nicolai of Florence. But Poggio may have had his own reasons for a certain degree of mystery about his good fortune. In the preface to his edition, Burmann speaks of the manuscript of St. Gall, on the authority of the librarian Kesler, as having been ‘honesto furto sublatum’: if it was the Harleianus there is perhaps little need to wonder that nothing has been known till now of its later fortunes79.
The affiliation of other MSS. of this class (which includes also the Almeloveenianus) to the codices which have just been described, may be determined by the application of certain tests. Prominent among such MSS. is the Codex Bodleianus, which has received more attention from editors of Quintilian than its merits seem to me to warrant. It repeats word for word the remarkable error attributable to the Harleianus at x. 7, 20 (see above, p. lxviii): in other places it embodies attempted emendations, e.g. x. 1, 90 nec ipsum senectus maturavit: 2 §7 de metris for dimiteris (see above, p. lxvii, note). It belonged to Archbishop Laud, and must have been written in the fifteenth century.
Of the same age and family are two manuscripts often cited by Halm, the Lassbergensis and the Monacensis. The former was formerly at Landsberg in Bavaria: it is now at Freiburg. The reading atque interrogationibus atque interrogantibus, which Halm gives from it alone at x. 1, 35, I have found also in G and H; this seems quite enough to identify its parentage. The Monacensis was collated by Halm for his critical edition in the parts where he had to rely on A G or on G alone: with no conspicuous results,—‘nihil fere aliud effectum est quam ut docere possemus, ubi aliquot locorum, qui in libris melioribus leviter corrupti sunt, emendatio primum tentata sit’ (praef. viii, ix).
Alongside of these I would place a rather interesting MS. in the British Museum, which has been collated specially for the purpose of this edition, with no result worth speaking of, except to establish its class. It repeats the mistake of H at x. 7, 20: and the fact that the copyist began his work in a hand that was meant to imitate writing of the eleventh century seems, along with the internal evidence, to prove that it is one of the copies of Poggio’s MS. In x. 2, 7 it has posterius for historiis (a mistake in H—see p. lxix): and in the same place it shows (like the Bodleian codex) de metris for dimiteris. This is also the reading of the second hand in the Turicensis. Such differences as exist between it and H F T may be ascribed to attempted emendation: e.g. vertere latus x. 3, 21. Poggio’s letter to Guarini is copied at the end of the volume.
The other MSS. of the fifteenth century, so far as they are known to him, M. Fierville divides carefully into two classes (his third and fourth). The principal features of difference which distinguish them among themselves, and from those already mentioned, are that they incorporate, in varying degrees, the results of the progress of scholarship, and that they are seldom copied from any single manuscript. A detailed examination would no doubt establish what is really the point of greatest moment in regard to them: how far are they derived, through Poggio’s manuscript, from the Bambergensis, and how far from such complete manuscripts as the Ambrosianus and the original of Bambergensis G? Some of them (as well as other fifteenth century MSS., with a description of which I desire to supplement M. Fierville’s Introduction, pp. cii sq.), are of at least as great importance as those referred to above as having been collated in part by Halm.
The Argentoratensis (S), also used by Halm, may be mentioned first: it was collated by Obrecht for his edition of 169880. This manuscript was destroyed in the bombardment of Strasburg, August 24, 1870. Then there are the MS. of Wolfenbuttel (Codex Guelferbytanus), collated for the first time by Spalding: the Codex Gothanus, used by Gesner for his edition of 1738: the Codex Vallensis (Parisinus 7723), which purports to bear the signature of Laurentius Valla (9 December, 1444), whose corrections and marginal notes it contains81. The list of these and several others, all carefully described by M. Fierville, may now be extended by a short reference to various MSS. in this country, hitherto uncollated. The results of my examination of them (as well as of the Bodleianus, and Burneianus 243, referred to above) appear in the Critical Appendix: if few of them are of first-class importance, it may at least be claimed that right readings, with which Spalding, Halm, and Meister have successively credited the early printed editions,—e.g. the Cologne edition of 1527,—have now been attributed to earlier sources. And when M. Fierville had so carefully examined the MSS. of France, Germany, Switzerland, Italy, and Spain, it seemed of some importance that his laborious work should be supplemented by a description of the MSS. belonging to the libraries of this country.
In the British Museum there are eight manuscripts in all of Quintilian’s Institutio: of the most important of these, the Harleianus (H), I have already given an account, and one of two MSS. in Burney’s collection (Burn. 243) has also been mentioned. Of the remaining MSS. two may be taken together, as they are in complete agreement with each other, and show conclusive proofs (as will appear in the notes) of relationship to such codices as the Argentoratensis and the Guelferbytanus. The first of these two MSS. (Codex Harleianus 2662) has an inscription bearing that it was written by Gaspar Cyrrus ‘nationis Lutatiae,’ and was finished on the 25th of January, 1434,—only eighteen years after Poggio made his great discovery. So great an advance is evident in the text, as compared with the readings of H F T, that it seems probable that this MS. owes little to that family. The same may be said of the Codex Harleianus 11,671, a beautiful little quarto, dated 1467: it has the Epitome of Fr. Patrizi attached (see Classical Review, 1891, p. 34). The following cases of remarkable errors will suffice to connect both these MSS. with the Guelferbytanus: x. 3, 12 a patrono suo for a patruo suo: 1 §97 verum for veterum: 1 §55 equalem credidit parem (as also Prat., Guelf., S, and Voss. i. and iii.): 1 §72 quamvis sui temporis Menandro for ut pravis sui temporis iudiciis Menandro: 7 §6 adducet ducetur. Another very interesting MS. in the British Museum is Harleianus 4995, dated July 5, 1470: it contains the notes of Laurentius Valla, which were frequently reproduced at the time, and might be classed along with the Vallensis were it not that a marginal note at x. 6, 2 (where a false lacuna appears in most codices, as Bn. and Bg.), ‘hic deficit antiquus codex,’ makes it probable that the copyist had more than one MS. at his side82. This MS. agrees with the Vallensis and Gothanus in reading cognitioni for cogitationi x. 1, 1: ubertate for ubertas 1 §109: et vis summa §117: eruendas for erudiendas 2 §6: nobis efficiendum ib. §14: decretoriis 5 §20. The other two Harleian MSS. (4950 and 4829) present no features of special interest: I have, however, included them in the critical notes for the sake of completeness. The former was written by ‘Franciscus de Mediolano’: it is often in agreement with the Lassbergensis. The latter finishes with the words ἡ βίβλος τοῦ σωζομένου and the motto ἀγαθῇ τύχῃ. The readings of the Burneianus 244 are also occasionally recorded in the notes. All three are in general agreement with L, and also with the Codex Carcassonensis, a fifteenth century MS. of which M. Fierville published a collation in 1874.
A greater degree of interest attaches to two Oxford manuscripts, one of which (the Codex Balliolensis) is unclassed by Fierville, while the other (the D’Orville MS.) has never been examined at all. The former was used by Gibson for his edition of 1693. It begins at bis vitiosa sunt i. 5, 14, but there are various lacunae, which do not correspond with those of the incomplete family. The MS. is in fact in a mutilated condition. In the Tenth Book we miss its help after the end of the first chapter till we reach iii. §26, where it begins again with the words quam quod somno supererit: it stops abruptly at nostrorumque Hort(ensium) x. 6, 4. It is in general agreement with Harleianus 2662. I may note that in i. 5, 36 it has interrogatione, a reading which Halm says appears for the first time in the edition of Sichardus, 1529: ib. §69 it has e rep with A and 7727, with the latter of which it is in close correspondence (e.g. forte at i. 5, 15, all other codices forsan or forsitan).
There remains the D’Orville MS. in the Bodleian at Oxford (Codex Dorvilianus),—a manuscript which has been entirely overlooked, except for a single reference in Ingram’s abridged edition of the Institutio (1809). Yet it seems well deserving of attention. In some places it shows a remarkable resemblance to the Ambrosianus (e.g. Getae 1 pr. §6: et quantum ib. §8): at 1 pr. §4 it has summam inde eloquentiae (Spalding’s reading, found in no other MS.): destinabamus al. festinabimus ib. §6 (the alternative being a reading peculiar to A). Its most important contribution to the Tenth Book is 7 §20, where it gives the reading which Herzog conjectured and which I have received into the text: neque vero tanta esse unquam debet fiducia facilitatis: in 2 §14 (see Critical Notes) it has quos eligamus ad imitandum, a reading peculiar to itself. For the rest it is in general agreement with the Balliol codex. It is Italian work, of the early part of the fifteenth century,—earlier, Mr. Madan thinks, than the Codex Bodleianus. A marginal note at ix. 3, 2 shows that the copyist must have had more than one MS. before him. In some cases it would appear as if he carefully balanced rival readings: at 1 pr. §12. all codices have quaestio ex his incidat except A, which gives ex his incidat quaestio: the reading in the Dorvilianus is quaestio incidat ex his: again at i. 2, 6 ante palatum eorum quam os instituimus, many codices give mores for os: Dorv. shows quam vel mores vel os.
List of editions, tractates, and books of reference.
Besides the complete editions of Spalding, Zumpt, Bonnell, Halm (1868-9) Meister (1886-87), use has been made of the following editions of Book x.:—
Among the Translations, reference has been made to Lindner’s (Philologische Klassiker, Wien, 1881), Alberti’s (Leipzig, 1858), and Herzog’s (Leipzig, 1829); also to Guthrie’s (London, 1805), and Watson’s (in Bohn’s series).
The following have been used as books of reference:—
Wilkins: Cicero, De Oratore, Books i. and ii. (2nd ed.) |
Oxford, 1888 and 1890. |
Sandys: Cicero, Orator. |
Cambridge, 1889. |
Kellogg: Cicero, Brutus. |
Boston, 1889. |
Wolff: Tacitus, Dialogus de Oratoribus. |
Gotha, 1890. |
Andresen: „ „ |
Leipzig, 1879. |
Reiske: Dionysius Halicarnassensis. |
Vols. v-vi. Leipzig, 1775-7. |
Usener: Dionysius Halicarnassensis Librorum de Imitatione Reliquiae, Epistulaeque Criticae Duae. |
Bonn, 1889. |
Ammon: De Dionysii Halicarnassensis Librorum Rhetoricorum Fontibus: Dissertatio Inauguralis. |
Munich, 1889. |
Volkmann: Die Rhetorik der Griechen und Römer. |
2nd ed. Leipzig, 1885. |
Causeret: Étude sur la langue de la Rhétorique et de la Critique Littéraire dans Cicéron. |
Paris, 1886. |
and Fierville: Quintilian, Book i. |
Paris, 1890. |
The references to Nägelsbach’s Lateinische Stylistik are to the eighth edition (Nägelsbach-Müller).
The periodical literature bearing specially on the Tenth Book of Quintilian has grown to very considerable dimensions within recent years. The following articles and tractates have been consulted:—
Table of places where the text of this edition differs from those of Halm (1869) and Meister (1887).