Textual Evidence

Shipley on

Excerpt from: F.W. Shipley,
Certain Sources of Corruption in Latin MSS, (London, 1904)

Page Index

Last Updated: Feb 24, 2010

Shipley: Sources of Corruption

    I. Introduction

    II. CODEX REGINENSIS 762: detailed background

    III. Word Division Errors: adding spaces

    IV. Dittography: (Additions) Accidental repetitions
       (1) Single Letters
       (2) Words, Syllables

    V. Omissions : Accidental and deliberate drops of letters/words
       (1) Ordinary Haplography: omission of letters, syllables, words etc.
       (2) Homoeoteleuton/-arcton: Similar endings or beginnings in a line
       (3) Particles: conjunctions, prepositions
       (4) Difficulties: unintelligible, illegible, lost text

Return to Index

Shipley on Dittography

Scribal Errors

Although Shipley here discusses manuscripts which are not directly of interest for NT textual criticism, his thorough introduction gives a clear idea of the state of scholarship and the changes taking place in both understanding and methodology in textual criticism generally in the early 20th century.

In addition, the choice of manuscripts is particularly apt in that they are well-known, and reveal a great deal about scribal habits and copying practices in the Middle Ages.

Shipley's survey gives a rich picture of both copying generally, and of the Latin MSS transmission stream in particular.

Taken from: F.W. Shipley,
Certain Sources of Corruption in Latin MSS,
CODEX PUTEANUS (P 5th cent.),
(London, 1904) pp. 23-33

Headings have been added for clarity and navigation purposes.


The tendency of Latin textual criticism has in late years been more and more in the direction of a conservative adhe- rence to the authority of manuscripts, wherever possible. This may be seen in the gradually diminishing number of emendations and conjectures in the critical apparatus of recent editions of the Latin texts. Scholars now hesitate much longer about marking a word or an expression as corrupt merely because it is unusual. Confidence in all but very late manuscripts is on the increase. Recent years have seen the reinstatement of not a few manuscript readings whose place had long been taken by conjectures. A knowledge of palaeography is more and more becoming an essential factor in textual criticism, and, except in the case of texts which depend wholly upon manuscripts of the fourteenth and fifteenth centuries, one of the chief tests of an emendation is coming to be, — Is it capable of palaeographical explanation?

This tendency to place textual criticism more nearly upon a palaeographical basis has not been accompanied by a corresponding change in the character of the illustrative material used in books and manuals upon the subject. The collections of examples now placed before the student are not without their value, but they fail along the lines in which textual criticism has made the greatest advance. These examples consist for the most part in

(1) a comparison of the corrupt reading of a manuscript with a conjectured emendation of a scholar, or

(2) in a comparison of the readings of two or more manuscripts of the same author, of which the relationship is generally uncertain, or at least remote.

Illustrations chosen according to either method are often misleading to the student, even granting that, in the first method, the scholar's conjecture is what the author actually wrote. A great many corruptions to be found in manuscripts of all periods are no longer in their initial stages, but are the complex result of several distinct processes of growth. The student, with nothing before him but what the author is supposed to have written and the corrupt reading of, let us say, a thirteenth century manuscript, may be dealing only with a corruption in a late stage. All the earlier steps are missing, and certainty with regard to them is out of the question. Such an illustration has little value for him, leading as it does to no conclusion which is surely right, and possibly to one which is wholly wrong. Likewise, neither of these methods keeps clearly before the student the character of the errors common to certain styles of writing and certain periods of time. Both of them are lacking in palaeographical details.

To be of the greatest practical value, illustrations of corruptions should fulfil the following conditions : (1) the two extremes which are compared should not be too widely separated ; (2) neither of them should be based upon conjecture ; (3) each illustration should present but a single stage in the progress of an error, or at any rate should present but one stage at a time ; (4) the cause of the error should be reasonably certain ; (5) each example should keep distinctly before the student the periods of time and the palaeographical conditions involved. Material for illustrations which would answer all these conditions is not entirely wanting, though little use has heretofore been made of it. It is to be found in a class of neglected manuscripts whose readings have no place in the critical apparatus of the text editions, namely, direct copies of originals which are still extant. The circumstance which renders such copies useless for the constitution of the text of a given author makes them of the greatest value in throwing light upon the history of the texts in general. By comparing such a copy with its original it is possible, as it were, to look over the shoulder of the mediaeval scribe as he sits at his task. One may follow his hand and eye as he copies letter by letter and word by word. The difficulties with which he has to contend either in the script or the text of his original are clearly revealed. It is possible to see exactly how he performed his work, whether faithfully or carelessly, whether he has adhered closely to his text or altered freely, and, when he has made errors, how and why they came to be made. The extent to which the text sujffered in his hands is thus made clear in every detail. Illustrations taken from the readings of two such manuscripts, original and copy, would enable the student to draw his own conclusions with full data before him, — the style of the script of the original, the date of each manuscript, the conditions under which the copy was made, and the knowledge that, in the case of corruptions, he is dealing with but a single stage. By this method it is possible to see exactly what, in the copying of a given manuscript, actv/illy happened^ and then to turn the information to account in considering the texts of other manuscripts produced under the same conditions, the originals of which are now lost.

Examples chosen by this method are as nearly as possible upon a palaeographical basis, and offer the student definite illustration. The comparison of a single pair of representative manuscripts and the errors arising from a single process of transcription would serve to give him a clear idea of the tendency to corruption at a given period. A study of the errors exhibited in four such sets of copies and originals, each set representing a distinct phase of the history of Latin texts, would give him a more definite conception of the whole field than he can possibly get from the more or less random examples of the manuals. For instance, a ninth century copy of an original of the fourth or fifth century, -an eleventh century copy of an original of the ninth, a thirteenth century copy of an original of the eleventh, and a fourteenth or fifteenth century copy of an original of the thirteenth, would serve respectively to illustrate the tendencies of the periods which they represent, and collectively the entire history of Latin texts in so far as extant manuscripts make this possible.

For the last three of these four periods there is no lack of illustrative material of the nature indicated. A search in the libraries would probably disclose an abundance of neglected copies of extant manuscripts. The first of these periods, which is in many respects the most important, is represented, so far as I am aware, only by the single pair of manuscripts which form the subject of the present article. These are (1) the famous codex PuteanuB (National Library, Paris, 5730), of the fifth century, which contains the third decade of Livy's history, and (2) a ninth century copy of it now in the Vatican library, and catalogued as Meginenais 762.

That the significance of these two manuscripts may be properly understood, let me first point out some of the characteristics of the period to which they belong, and the representative nature of the manuscripts themselves. The epoch from the fifth century to the ninth is one which is unique in the history of the texts of the Classical Latin authors. It is marked by a period of almost total inactivity in the multiplication of copies of their works, and is followed by one of unparalleled activity. Almost all of our extant manuscripts of these writers that are earlier in date than the very end of the eighth century are the capital and uncial manuscripts of the fourth, fifth, and the beginning of the sixth centuries. The interval from the middle of the sixth century to the closing years of the eighth is represented by very few existing manuscripts of any but the Church writers. This fact would seem to indicate that, while there was no lack of activity in the reproduction of the writings of the Christian fathers, the copying of the works of the pagan Latin writers was almost totally arrested for over two centuries.

( Of the authors who wrote before the official victory of Christianity the following works are, to the best of my knowledge, the only ones which have been preserved in manuscripts surely belonging to this period : the Agrimensores, 8. VI- VII; Apuleius (?), de Herharum Medicaminibus, s. VI- VII ; Ovid, ex Ponto (fragment), s. VI-VII ; the Pandects, s. VI- VI I ; Probus (?), Catholica, s. VI-VII, VII-VIII, VIII-IX ; excei-pts from Pliny and Apicius, s. VII- VIII ; Censorinus, s. VII ; La^tantius, s. VII ; Sacerdos, s. VII-VIII ; Commodianus, Carmen Apologeticum, s. VIII ; Notae Tironis et Senecae^ s. VIII ; the Anthol- ogy of the Codex Salmasianus, s. VII-VIII.

Of these, Commodianus and Lactan tins were Christian writers ; Probus, Censorinus, and Sacerdos were writers on Grammar ; the works of the Agrimensores, the above-mentioned work on Materia Medica ascribed to Apuleius, the excerpts from Pliny and Apicius, the Pandects, and the Notae Tironis et Senecae were all of a technical or semi-tech nical nature. It would seem probable, then, that with few exceptions such manuscripts only were copied in the seventh and eighth centuries as, from the nature of their subject-matter, did not conflict with the doctrines of the Church.

The active production of copies of the works of the pagan writers begins anew with the revival of learning under Charlemagne. To this new and wonderful activity, which arose with the closing years of the eighth century and continued through the tenth, we are indebted for the preservation of a large proportion of our Latin texts.

The oldest manuscripts of a large proportion of the extant literature from Plautus to the official victory of Christianity are of the ninth and tenth centuries. The following is a list of the works of which the text is based upon manuscripts of this period (viz., the ninth and tenth centuries, and the last decade of the eighth) : Plautus (the Codex Vetus for portions not contained in the Ambrosian palimpsest) ; Lucretius ; Catullus, c. 62 ; Caesar; Sallust; his torica ad Herennium ; the following works of Cicero : Pro Fonteio, pro Flacco, post reditum in senatit, post red. ad Quirites, de domo sua, de haruspicum responsis^ pro Sestio, in Vatinium, pro Caelio, de provinciis consularihus^ pro Balho, in Pisonem, pro Marcello, Philippics, Bhetorica, de Oratore, Brutus, Orator, Part. Orat., Topica, ad Familiares, de Legihus, Paradoxa, Academica Priora, Tusc, Disp,^ de Natura Deorum, Cato Maior, de Divinatione, de Fato, Timaeus^ de Amicitia, de Ojficiis; the Culex^ Copa, Aetna^ and Moretum formerly ascribed to Virgil ; Bernese scholia on Virgil ; Horace ; Ovid, Amoves, Heroides, de Medicamine Faciei, Ars Amatoria, Bemedia Amoris, Fasti, Meta- morphoses, ex POnto (for the greater part) , Halieutica ; Grattii Cynegetica ; Livy, first decade, with the exception of books III- VI, and the Periochae ; Justinus ; Seneca Rhetor ; Verrius Flaccus {Epitome Pauli) ; Hyginus ; Vitruvius ; Ara- tea Germanici ; Manilius ; Phaedrus ; Seneca, Tragoediae (excerpta), ^AitokoXo- kOvtuxtis, Dialogues, Epistolae, de dementia^ de Beneficiis ; Valerius Maxim us ; Curtius Rufus ; Persius ; Lucan ; Quintilian ; Calpurnius Flaccus ; Ilias La- tina ; Petronius ; Valerius Flaccus ; Statius ; Martial • Juvenal ; Tacitus, Annals 1-VI ; Pliny's Letters ; Pliny the elder ; Pomponius Mela ; Celsus ; Columella ; Apicius ; Marcellus ; Frontinus (except the de Aquis) ; Siculus Flaccus; Nemesianus, Cynegetica; Disticha and Monosticha Catonis ; Sueto- nius ; Minucius Felix ; Florus, Bell. Bom. ; Apuleius, de Platone et eius dog- mata ; Calpurnius Flaccus ; Terentius Scaurus ; Gargilius Martialis ; Flavins Caper ; Aero ; Porphyrio ; Gellius ; Maecianus ; Cyprian ; Tertullian ; Q. Sere- nus Sammonicus ; Scriptores Historiae Augustae ; Arnobius.

Roughly speaking, this list includes considerably more than half of the extant classical literature, and the ratio of ninth century authoritative manuscripts upon this list to those of the tenth century is about 3 to 1. This shows how important a place the ninth centuiy holds in the preservation of the texts.

The task of copying was performed by monks. The usual practice in the scriptoria of the various monasteries in the ninth century seems to bave been to secure, for the purpose of making a copy, the oldest available manuscript of a given author either preserved in the library to which the scriptorium belonged, or borrowed from that of another monastery. The oldest available manuscripts were, in the case of the pagan writers, those of the fourth or fifth century in capital or uncial writing. (Capitals were reserved for favourite authors, Virgil in particular.)

Consequently the three hundred years from the end of the fifth century to the beginning of the ninth represent but a single link in the history of the texts of those Latin authors whose writings are preserved in manuscripts not earlier than the ninth century. In all that time the text of such an author has passed through but a single stage in the process of corruption. The errors which have crept into the text in the making of the ninth century copy constitute the only difference between the trustworthiness of a ninth century manuscript of a given author and that of its archetype of the fifth.

This single process of transcription marks what is perhaps the most critical period in the history of Latin texts. It is not in itself characterized by intentional alterations such as are common in manuscripts of the fourteenth and fifteenth centuries. The corruptions are diie almost entirely to ignorance or carelessness on the part of the scribes, much more rarely to attempts at emendation. But, although serious corruptions are few, this period is rich in the germs of future corruptions, on account of a new factor in the making of manuscripts. This factor is the division of the text into words, which now for the first time comes into general practice. In the manuscript of the fifth century there was ordinarily no word-division whatever. The fifth century scribe, if he did not understand the meaning of tlie text before him, was able to conceal his ignorance and evade all difiiculties resulting therefrom by copying letter by letter, a process in which the text of the author suffered but little. The scribe of the ninth century, on the other hand, was forced to make words out of the undivided text of his original, and, with only a superficial understanding of the sense of what he was copying, it is not surprising that he often divided wrongly. The errors thus made are not in themselves difficult to emend, but, simple though they were, they frequently became magnified into grave corruptions in the efforts of the scribes of a later age to restore sense to the passages thus distorted.

Of this important process of transcription no manuscript could be more thoroughly representative than the Meginensis 762.

The Puteanus from which this copy was made, is a typical manuscript of the 5th century, in uncial script, with words undivided. (A fuller description of the manuscript is given in chap. II.)

The Meginensis itself is one of the very best examples of the work of the French monasteries at the beginning of the ninth century.^ The centre of the new activity in the production of manuscripts, and of the reform in writing which began with the closing years of the eighth century and spread over a large part of Western Europe, was the monastery of St. Martin at Tours. Its abbot Alcuin, who was Charlemagne's minister of education, was the pioneer of the new movement. It was in this monastery that the Heginensis was produced, a few years after Alcuin's death. It is written in the reformed script known as the Caroline minuscule. This manuscript is not the work of a single scribe. Eight monks were simultaneously engaged upon it, who, to judge from their orthography, were of more than one nationality. It consequently represents the concentrated efforts of the scriptorium of St. Martin's at the time when this monastery was the centre of the ninth centur.y revival, which was then at its height.

In the present article I have endeavoured to illustrate the tendencies to corruption which characterize the transcription from manuscripts in majuscule writing into Caroline minuscule with divided words, by means of actual examples of scribal errors drawn from a comparison of the readings of the Puteanus with those of its copy the Heginensis, Since collating the manuscripts in 1896 I have made use of the material collected, in giving courses in Latin palaeography. The results encourage me to believe that the collection may prove helpful to other students and give a clearer idea of the errors common to this important period than they now get from the illustrations in the manuals, which are drawn from miscellaneous sources. The starting-point of the error is in every case certain, being simply the reading of the Puteanus, The cause of each error is therefore in the majority of cases beyond dispute. As I have already said, the Heginensis is the work of eight scribes, seven of whom have each copied a quota amounting to considerably more than a book of Livy's text. Among them more than one nationality is probably represented. For this reason the errors here given represent the tendencies of the period much more fairly than if they had been drawn from the work of a single scribe.

For convenience of reference, the examples in the following chapters are arranged in categories, with a minimum of commentary on each variety of error. The reading of the Puteanus is usually given first, followed by the erroneous reading in the Heginensis, Henceforth the Puteanus will be represented by the letter P, and the Reginensis by the letter R.

(This letter is used by A. Luchs to denote another manuscript of this decade of Livy, in the Spirensian group. No confusion, however, can result, inasmuch as in the present article we are concerned with but three manuscripts, all of which belong to the Puteanus group.)

To illustrate several stages in the progress of an error I have occasionally made use of the corrections found in both manuscripts. Corrections in P are designated as follows :

P1, if the correction is by the scribe,
P2, if by the first corrector,
P3, if by the second corrector.

(All the corrections in P designated by Luchs as P*, P^, were made after R had been copied.)

Correctors in R are much more difficult to discriminate, owing to the variety of inks and hands (see chaps. II and XI). For our present purpose it will suffice to designate the corrections made by the scribe as R1 and all later corrections as R2. For the purpose of tracing a stage further the corrections begun in the Heginensis, I have added in a limited number of cases the readings of Codex Mediceus (Florence, Laurentian Library, LXIII, 20), which is in turn a copy of R. This manuscript will be indicated by the letter M. In all the examples in which I have given several steps in the progress of an error I have tried to keep each stage distinct.

(Although M is in the main a copy of R, its readings seem to indicate, especially in the early portion of the manuscript, that the scribe who copied it had also before him either P or some copy of it other than R. This is a matter which I hope to deal with in another paper.)

The following exposition is intended, in a general way, to cover all the points of consequence afforded by the study of the two manuscripts. Accordingly, while this paper contains certain new facts and points of view, it of course has also to deal with many that are already familiar — with the special advantage, however, as already shown, that the two points of comparison in each case are actual examples drawn from two extant manuscripts, of which one is the direct copy of the other.

II. Codex Reginensis 762

I have already given in the preceding chapter a general account of R, but my description of it was there confined merely to such points as directly concerned the purpose of the present article. Before proceeding to deal with the errors of its scribes, I shall first give a more detailed account of this manuscript and its making, concerning which a great many more data are known than is usual in the case of manuscripts of so early a date.

Being a copy of an existing original, and more mutilated [It begins with the words velut caeci evadunt (XXII, 6, 6), and ends with deinceps continua amplexus (XXX, 5, 7). ] than the original both at the beginning and at the end, it is of no value for the constitution of the text of the third decade of Livy, though the readings of M, its eleventh century copy, are to be found in the apparatus of the critical editions for the beginning of Book XXI, which is missing in P. But from a purely palaeographical standpoint it is much more interesting than either the Puteanus (P), or its own copy, the Mediceus (M), and has been the subject of various articles by palaeographical scholars, among whom are Wolftlin [Philologus, XXXIII, 1874, pp. 186-189.], Chatelain [JRevue de Philologie^ vol. XIV, 1890, p. 79 ; PaUographie des Classiques Latins, 9« livi'aison, 1895, with facsimile. ], and Traube. [L. Traube, JSitzungsberichte der Munchener Akademie, 1891, Heft 3, p. 426. ]

Its interest lies not merely in the fact that it is one of the best examples of the developed calligraphy of Tours, but also in the interesting data furnished by the signatures at the end of the various quaternions, which throw no little light upon the method of procedure in manuscript-making in the Middle Ages.

These signatures occur regularly at the end of each quater- nion, as follows : Gyslaij^, fol. 6 ; Aldo, fol. 16, 24, 32, 40, 48, 52 ; Fredeg, fol. 60, 68, 76, 84, 92, 97 ; Nauto, fol. 103, 111 ; Theogrm or Theogrimii, fol. 119, 127, 135, 141; Theodegri, fol. 157, 165, 173, 179, 185 ; Ansoald, fol. 193, 201, 209, 217, 228 ; Landemarus, fol. 236, 242, 250 ; each folio being signed upon the verso side. The manuscript was therefore the work of eight different scribes, each of whom, to judge from the amount done by those whose work is preserved to us in its entirety, copied about forty-four folios of the text, with the ex- ception of Nauto and Theogrimh, who together copied that number. (The work of Nauto stops abruptly about two-thirds of the way down a page [fol. 112 recto] and his part was completed by the scribe whose signature was Theogrimh. )

Chatelain (Revue de Philologie, vol. XIV. ^ Ed. Piper, Mon. Germaniae.) noticed that the end of the work of Gys- larius corresponded, even to a syllable, with the end of qua- • ternion IX of P ; that of Aldo, with the end of quaternion XVIII ; that of Fredeg, with quaternion XXVII ; that of Nauto and Theogriiim together, with quaternion XXXVI; that of Theodegri, with quaternion XLV ; that of Ansoaldus, with quaternion LIV ; and that the writing of the last page of the work of each scribe was spread out or condensed so as to co- incide with the end of the quaternion of P. He concluded, therefore, that the old fifth century uncial manuscript had been taken apart, and equal portions, of nine quaternions each, had been given out to the scribes to be copied simultaneously. It has further been pointed out by Traube (SitzungBherichte der Munchener Akademie^ 1891, Heft 3, p. 425), that the names of these scribes are to be found in the Libri Confraternitatum Sancti Q-alli [Gislarius (no. 3) Aldo (no. 10) Fredegaudus (no. 37) Nauto (no. 36) Teutcrinius (no. 26) Ansoaldus (no. 4) Landemarus (no. 24)] in the list of the monks of Tours, and that they all occur in a definite place upon the list, namely, in the second of the seven columns devoted to the monks of St. Martin's of Tours. From this he infers that in the monastery of St. Martin there was a definite class of monks who performed the duty of scribes, and were regularly engaged as such in the scriptorium of the monastery. The names corresponding to the signatures are as follows :

Signatures in R List in Col. 14, Lib. Confr. S. Galli Gyslaij Aldo Fredeg Nauto Theogrimn Ansoaldus Landemarus

It will be seen that one name is missing, namely, the one corresponding to the signature Theodegri, but it is just possible that Theogrimn and Theodegri were one and the same person. Nauto did not finish his full quota of nine quaternions, but stopped abruptly two-thirds of the way down fol. 112 recto^ after having completed a little over two quaternions. It is possible that Theodegri, after finishing his own portion, completed that of Nauto, using a different abbreviation in his signature. The abbot at the head of the list of monks of St. Martin's in the lAbri Confraternitatum is Fridegisus, the successor of Alcuin, who held the abbacy from 804-834.

( Chatelain, in his Paleographie des Classiques Latins (9® livraison), gives one the impression that the scribe whose signature was Fredeg, and Fridegisus the abbot, were one and the same person. In the hope that this might prove to be the case, and that I should find in the copy of the third of the scribes the work of a great Carolingian scholar, I studied that part of the manuscript with particular care. There was nothing, however, in the work of the scribe Fredeg to distinguish it from that of the other scribes. It contained even more than the average number of careless or ignorant blunders, and this portion of the copy could hardly have been made by a man who had a reputation for learning. Other considerations also add to the probability that the abbot was not the copyist. It is hardly likely that the chancellor of Louis le.D^bonnaire could find time to copy manuscripts with the monks in the scriptorium ; and, even if that were probable, he would have chosen the firet part of the work in preference to the third. I agree, therefore, with Traube in identifying the scribe whose signature is Fredeg with Fredegaudus, whose name is number 37 in column 14 of the Libri Confraternitatum. )

Traube is consequently justified in placing the date of R between those years. The fact that all these scribes were monks of Tours makes it certain that the work of transcribing was done at Tours ; for it is not likely that so many monks would be sent to Corbie to copy the Puteanus which at this time belonged to the monastery of that town. (Chatelain suggested this probability before Traube discovered that the scribes of the Beginensis were to be identified with the monks of the monastery of St. Martin at Tours.)

It is much more probable that the uncial manuscript was borrowed for copying, and this supposition would account for the haste shown in putting so many scribes to work upon making the copy, — the concentration, apparently, of the energies of the entire scriptorium upon this one task.

In R we have, therefore, an example of the developed calligraphy of Tours, produced within thirty years of the death of Alcuin. This monastery was, under his abbacy (from 796 to 804, the year of his death), the centre of the new activity in the production of manuscripts and of the reform in writing which spread over almost the whole of western Europe (Exceptions to this statement are England and Ireland and the monasteries of southern Italy).

The manuscript should therefore be thoroughly characteristic of the new movement, not merely in the style of the writing, but also in respect to the fitness of the ninth century monks for the task of copying the texts of the old Latin writers. In the handwriting of these eight scribes there is little variation. It is almost impossible in the case of several of them to distinguish at first sight the hand of one from that of another, which goes to show that, in this one scriptorium at least, the Caroline minuscule had been brought as nearly as possible to uniformity. On the other hand, it would seem that greater attention was given to uniformity in handwriting than in other features of the work of the copyist. In these there is considerable disparity. For instance, the signature of Theogrifi is attached to the quaternion which ends with fol. 119vo; but by means of the character of the errors, aside from the writing, one can see that the whole quaternion, and nearly a folio in addition, is not the work of the monk whose name is signed to it, but that of Nauto. The three quaternions copied by this scribe are relatively free from errors, while the quaternions signed with the names of Theogrin, Theodegri, Ansoaldus, and Landemarus are full of absurd blunders. These last-mentioned scribes are each prone to errors which are peculiar to themselves, a fact which, together with variations in orthography which are constant with certain scribes, would seem to indicate that more than one nationality was represented. The majority of the errors found in the manuscript are, however, common to all. They are due, for the most part, to carelessness, to a defective knowledge of Latin that was not suificient to enable the scribe to understand, except in a more or less random way, the meaning of the text he was copying, and to diificulties arising from lack of familiarity with the continuously written uncial script. Their work shows almost no intentional alteration, and the emendations are of the most superficial nature ; indeed, the majority of the scribes did not suificiently understand the meaning of the text to have been equal to any deliberate emendation of consequence. The errors, though numerous, are in themselves unimportant, but in the hands of scribes of a later age they would undoubtedly have become magnified into serious corruptions.

Each quaternion of the Meffinensis was corrected, as soon as it was completed, by some person or persons who supervised the work of the scriptorium. This is shown by the similarity between the ink of the correctors and that used by the scribes, and the non-recurrence, in the second quaternion of each scribe's work, of errors which were common in the first. These corrections, which will be treated in a subsequent chapter, are usually of a superficial nature and, in point of scholarship, are not much above the level of those made by the scribes themselves.


The comparative freedom of Capital and Uncial manuscripts from serious corruptions is due in large measure to the fact that the words of the text were usually not divided. (In the poem on the battle of Actium, found at Herculaneum, and in some early Virgil manuscripts, the words, though not spaced, are divided by points. This however was exceptional. )

The letters of the text were written one after the other, with no interruption except an occasional break to indicate the paragraph ; consequently it was never absolutely necessary for the scribe in making his copy to follow the sense of what he was copying. He might evade all difficulties arising from his own ignorance or from corruptions in his original, by simply writing the letters one by one without puzzling over the words they formed. He might indeed, as he copied, make wrong mental divisions of the words; but, if he did not in forming such wrong mental divisions also add, omit, or change a letter, his error could not in any way affect the text of his copy.

With the closing years of the eighth century, however, as a result of the Caroline reform, it became the regular practice to write each word separately. The Carolingian scribe, when set to make a copy of a continuously written majuscule manuscript, was forced to write his copy not letter by letter, but word by word. To do this rightly demanded a knowledge of the context, and the ability to read and understand Latin, — in which, as examples will clearly show, the eight scribes of R were anything but proficient. Their work is consequently full of wrong divisions of words, both where the sense of the passage was perfectly plain, and where corruptions in the original made the division really difficult. When in doubt, the scribes occasionally left the words undivided ; but, as a rule, they resorted more or less blindly to some random grouping of the letters.

Errors of this nature form by far the largest class of the mistakes made by the scribes of R, and many of the illustrations given under other headings can be indirectly traced to this source. These errors due to wrong division of words are in themselves comparatively insignificant, and, were the Puteanus lost, the emendation of this class of corruptions in the Meginensis would present little or no difficulty to a modern scholar. But the serious aspect of such errors is that they form the starting-point of further and more formidable corruptions in eleventh, twelfth, and thirteenth century copies of ninth century manuscripts. The scribes of a later age had education enough to recognize that there were errors, but not sufficient knowledge or care to rectify them ; and, in the superficial attempts which they made at restoring sense to the passages, all clues by which they might be emended by more careful scholars were frequently lost.


Repetition of Letters/Words

Dittography, the inadvertent repetition of a word, a syllable, or a letter, is a species of lapsus calami too common in our own writing to need further definition. It is a form of error to which the mediaeval scribe was perhaps a little more prone than we are for the reason that his task of copying manuscripts was almost entirely mechanical. In the case of the scribes of R, who were copying a manuscript in which there was no division of words, the opportunities for errors of this nature were greatly increased. Their mistakes were often due as much to the erring of the eye as to that of the hand, and for that reason were more likely to lead to corruption.

(1) Dittography of a Letter. — The largest number of examples in R of the dittography of a letter are due directly or indirectly to the confusion of the eye of the scribe caused by the continuously written text of the uncial manuscript. In glancing back to the page of his model after having written a word or syllable, inasmuch as there was nothing in the line to indicate the place where he had taken his eye from the page, the scribe sometimes unconsciously allowed the last letter of the word just written to arrange itself with the next group of letters. The letter was thus written twice, once at the end of one word, and again at the beginning of the next. Such doubling of letters is sometimes a cause, and sometimes a result, of wrong word-division.

Examples : XXIIII, 3, 9 arce satis . . . tuta P, arces satis . . . tuta R. — XXIIII, 8, 20 lacus thrasumennus et cannae^ tristia . . . exempla P, canna et tristia R. — XXIIII, 14, 7 seruili supplicio P, seruilis supplitio R. — XXIIII, 22, 15 quoniam eum P, quoniam meum R. — XXIIII, 33, 3 iouis it (= id) templum est P, iouis sit templum est R. — XXVII, 43, 10 haec senatu scripta P, haec senatus scripta R. — XXVIII, 35, 6 animo speciem P, animos speciem R. — XXVIIII, 24, 9 item circum Oram maritimam P, circum moram R. — XXII, 19, 10 anchoralia* (= ancoralia) P, ancora alia R. — XXIII, 44, 7 animaduertit P, anima aduertit R. — XXVIIII, 3, 1 tradenda deditionemque P, tradenda adeditionemque R.

Sometimes the repetition of a letter is a mere accident of writing, the scribe unconsciously writing it a second time. These repetitions, which have nothing to do with the division of words and do not usually affect the sense of the passage, are comparatively unimportant and do not usually lead to corruptions.

Examples are : XXVIIII, 3, 14 defectione P, defec- tio|one R. — XXVIIII, 36, 10 etruriam P, ettruriam R. — XXIIII, 18, 12 manu emiserat P, manuemisserat R. — XXIIII, 23, 2 comitia . . . hahita, creatuB . . . andranodorus P, habitaacreatus R.

(2) Dittography of Syllables and of Words, — The repetition of syllables or words is not nearly so common as the repetition of a letter. Examples of the dittography of a word are comparatively few in R, and none would be likely to lead to further corruption unless the following :

XXV, 41, 13 pisoni iurisdictio urbana pupio sicilia . . . evenit P. Here the scribe of R has written, pisoni iurisdictio urbana pupio urbana sicilia . . . evenit. XXIIII, 38, 7 aut uis aut fraus timeri possit P, aut ut uisa ut fraus R.

An interesting repetition of a syllable is seen in

XXVI III, 3, 5 frumentum sex mensum imperatum sagaque et togae exercitui P, frumentum sex mensum imperatum sagafw/nque et togae exercitui R.


The work of the eight scribes is full of omissions. These are for the most part of a trifling nature. The loss is confined chiefly to individual letters, syllables, and small unimportant words. It is rarely that a word of more than one syllable or a group of words has dropped out of the text. These omissions are occasionally the result of intentional emendation, but the majority of them are due to an oversight on the part of the scribe which resulted naturally from the lack of word-division in P. Thus the cause of errors of omission is practically the same as that of the errors illustrated in the two preceding chapters. In the confusion to the eye arising from the unbroken array of letters, it was easy enough for the omission of letters, syllables, and even words to pass unnoticed. The scribes of R never more than half grasped the meaning of the sentence which they were copying by reason of a far from perfect knowledge of Latin combined with the added difficulty caused by the lack of word-division in the uncial model. Consequently the failure of the eye to catch a letter was rarely checked by any feeling for the demands of the sense of the passage.

(1) Ordinary Haplography, — We have seen, in the last chapter, that there was an unconscious tendency on the part of the scribes of R to regard a letter which stands at the end of a word, or at the beginning of the next, as going with both (Dittography).

The opposite tendency is that of Haplography, the omission of one of two like letters, syllables, or words standing side by side. Omissions of this nature are, in the main, unconscious. Sometimes, however, they are the result of conscious emendation. The scribe imagines the repetition to be due to an error of dittography in the original, and omits one of the pair of letters, syllables, or words which he finds written twice.

(a) Examples of the omission of one of two identical letters standing side by side :

XXIII, 37, 2 quia muros satis per se altos P, muro satis per se altos R. Muros is here incorrect in P. The ablative is required. But, that there was no intelligence in the change of muros to muro in R, is shown by the fact that altos is left unchanged. — XXVI, 49, 10 eis praesentihus suos restituit P, praesentihus uos R. — XXIII, 48, 3 praetorem . . . misit tuerique . . . iussit P, misit uerique R. — XXVII, 41, 9 quos ubi . . . consul tiidet trihuno militum . . . imperat P, cos uide ir, mil. R. — XXVI, 38, 9 quo audacior res erat P, quo audatiores e|rat R. — XXIIII, 38, 8 ceteri superi infemique di P, ceteri super infemique di R. — A double example of this form of error is to be found in the following : XXIIII, 34, 5 ex ceteris nauibus sagittari (= sagittarii) funditoresque P, ex ceteris nauibus agit ari funditoresque R. — Examples of haplography such as praerant (= praeerant), uium (= uiuum), demisi (= demissi), clasis (= classis), nolent (= noUent), adferent (= adferrent), misise (= misisse), pasim (= passim), are quite common in R.

But it is difficult to say in any given case whether the omission of one of the double letters within a word is an oversight, or the result of a peculiarity of spelling on the part of a particular scribe.

(b) The same principle operates in the case of the repetition of a syllable, or where two syllables, made up in part of the same letters, stand side by side. The scribe is likely to omit one of them, either through oversight, or intentionally, in the belief that the repetition in P is the result of a dittography for which the pen of some previous scribe had been responsible. Examples are :

XXVIII, 33, 9 acies esset P, aciesset R. — XXVIII, 26, 2 gererent P, gerent R. — XXVI, 25, 3 m pela- goniam eadem celeritate uertit iter P, uertiter R. — XXVI, 28, 2 legionem inde deduct posse P, inde duci R. — XXVI, 6, 2 transitum hostibus dediL ibi, etc., . . . P, detibi R. — XXVI, 83, 12 qui se dediderunt P, qui se dederunt R.

Repetitions of a word such as are to be found in Plautus and Terence are rare in Livy ; consequently there are no examples of the omission of a word through haplography.

(2) Homoeoteleuton / HomoeoArcton

The recurrence of the same letter or series of letters in the same line was frequently the cause of the omission of all that was written between them. On glancing back to his uncial original after copying the text as far as the first of the two letters or combination of letters, the scribe's eye often caught sight of the second letter or syllable, and, imagining that this was what he had just written, he went on with his copy from that point, omitting all between. The name usually applied to this variety of omission is Corruptio ex Homoeotelevto, Tyrrell (Correspondence of Cicero^ vol. II, p. 64) suggests Para- blepsy as a more convenient name.

(a) Omission of one of two adjacent syllables containing the same vowel; e,g. XXIII, 43, 13 potiturwm P, potitum R. The scribe wrote potitu and upon glancing back to the page of P his eye caught sight of the second w, which he imagined was the one he had just written. [N.B. In the examples the part omitted in R is given in italics in citing the reading of P].

XXIIII, 10, 6 reli^iosi P, reliosi R. — XXV, 18, 4 creue- rat consuetudo quod aeger romae . . . P, creuerat consuetudo deger romae R. Here the scribe has omitted the o of quod^ and the letters between it and the last o of

consuetudo. — XXVI, 26, 6 scire se /requentes P, scire sequentis R. — XXVI, 39, 10 nawali P, nali R. — XXVI, 40, 14 ad qua^iraginta P, ad quaginta R. — XXVIII, 30, 7 sequer^tur P, sequetur R. — XXVIII, 31, 3 reJellione P, rellione R. — XXVIIII, 1, 24 carthagim'enses P, carthagienses P.

In the two following examples it is the first syllable which is omitted instead of the second, XXIII, 48, 4 scipionibus P, spionibus R. — XXVIII, 35, 4 ma«missam P, manissam R.

(b) Sometimes the omission of a whole word in R is caused by the fact that it ended with the same letter or syllable as the preceding one.

XXIII, 48, 2 claudiana castra super sessulam P^; castra is omitted in R. — XXVI, 27, 8 comprehensi ipsi familiaeque eorum P, compraehensi familiae quae eorum R. Here the omission of ipsi is due to the passing of the scribe's eye from the si of comprehensi to that of ipsi. — XXVIIII, 7, 2 cum primum aestu fraetum inclinatum est P^, cum primum aestu fraetum est R, the omission being due to the similar ending turn. — XXVIII, 32, 8 rerum suarum gestarum P, rerum suarum R, gestarum being omitted.

(c) In the two examples which follow, this same tendency is responsible for the omission of several words. In XXVIII, 29, 7 itaque quod ad uniuersos uos attinet, si err oris paenitet^ satis superque poenarum habeo P. The similarity of ending in attinet and paenitet has caused the omission of si erroris paenitet altogether. Also in XXVIIII, 21, 5 si quis miles aut in urbe restitisset aut secum extulisset P, the words aut secum extulisset have been omitted from the same cause.

(3) Particles

The largest class of omissions in R is made up of small unimportant words, such as prepositions, conjunctions, pronouns, and the various forms of the verb esse. Lindsay in his Introduction to Latin Textual Emendation has pointed out the importance of recognizing the tendency to omit these monosyllabic words in emending the texts of the Latin poets, where it is often necessary, in order to reduce a line to metrical regularity, to insert some small word, such as those described above. The reasons for the omission of these small words are various. Sometimes the scribes have, perhaps purposely, left them out in the belief that they were unnecessary, sometimes because they did not understand the context ; but in the majority of cases these little words, most of which consist of but two letters, must have been entirely overlooked by reason of the lack of word-division in P.

[For brevity, in citing examples of this variety of omissions I shall give simply the reading of P, printing in italics the words omitted in R.]

(a) Prepositions, — The omission of ab (a) and in is especially common.

XXII, 30, 9 famam a patribus accepissent. — XXIIII, 39, 2 productus ad populum a magistra- tibus. — XXVIIII, 1, 20 ceteris ab hannibale interfectis. — XXVI, 39, 8 ut ad componenda armamenta . . . satis temporis esset. — XXII, 41, 5, duas prope partes tironum militum in exercitu esse. — XXIIII, 5, 9 coniuratio in tyranni caput facta. — XXV, 19, 15 ut in nulla pari re. — XXVI, 1, 10 ne in oppidis hibernarent. — XXVI, 27, 4 qui in publicum redempta ac manu missi sunt. — XXVI, 28, 4 urbanae duae superioris anni in etruriam . . . mitterentur. — XXVI, 44, 5 trepidatio uero non in proelio maior quam tota urbe fuit. Here the cause for the omission seems to have been that in was not written with tota urbe. — XXVIII, 26, 14 qualem ne in acie quidem aiebant meminisse. — XXVIIII, 34, 12 sustinere ultra nequiere, the scribe apparently not understanding the adverbial use of the preposition.

(b) The conjunctions et^ ac^ aut^ and the enclitic qu^ are often omitted, especially when they go in pairs, the scribes regarding one of them as superfluous.

XXIII, 46, 9 muni- mentisquae firmatis et praesidio. The et is probably omitted on account of the presence of que, — XXIIII, 20, 5 blanda et apulorum aece (= Aecae) oppugnatae. — XXVI, 50, 9 pu- dore et gaudeo. — XXVIII, 24, 6 motae . . . eorum mentes sunt non tum primum . . . sed iam ante licentia ex diutino, ut fit, otio conlecta et non nihil quod in hostico laxius rapto suetis uiuere artiores in pace res erant. — XXVIIII, 5, 9 exercitum bolonum (= uolonum) ex etrxxTia, in galliam traducit, is an interesting example of the same tendency. Here the scribe of R supposed the et of Etruria to be the conjunction, and has left it out, writing bolonum ex ruria,

An example like the preceding is to be seen in XXVI, 50, 2. Here P has inter cetera ac|cepit, and the scribe of R, supposing ac to be the con- junction, and not seeing any possibility of its being used as such, has left it out and written inter cetera cepit, — XXIII I, 36, 5 legionem romanam quae exposita panormi est. Here the quae is omitted probably because it was taken for que. — XXIII, 48, 5 nee aliter aut exercitum aut prouinciam atteneri (P2) posse.

(c) Examples of the omission of pronouns :

XXVIIII, 6, 5 qui . . . regium se contulerant P, qui . . . regium contulerat R. The scribe seems to have regarded regium as subject of contulerat, and has altered the number in addition to the omis- sion of se. — XXIIII, 9, 9 praesenti fabio atque ipso comitia habente consulatus continuatus. The word ipso is here omitted apparently because the scribe did not understand its mean- ing. — XXIIII, 14, 8 liberatis auctorem eis non se fore solum. — XXII, 25, 19 seruili eius artis. — XXIIII, 16, 19 quam pater eiu8 in aventino . . . curauit.

(d) Omissions of the parts of the verb esse^ especially of the form est^ are very common; Examples would probably be superfluous here.

(4) Difficult or Illegible Passages

Another cause of omissions was the tendency of the scribes to leave out elements which they did not understand. Where a passage was corrupt in P, not knowing what to do with the corrupt word or words, they sometimes left them out altogether.

e.g. XXIIII, 18, 2 quae uelut diutinis morbis aegra corpora ex sese gignunt, aea nata bello erant. Here the scribe of R has omitted the corrupt aea. — XXIIII, 22, 14 set strictis simul (corruption of semel} gladis (= gladiis) P. — XXVIII, 27, 10 nee me uita iuuaret is the reading of Lucks; P has nee multa iuuaret ; R has omitted the corrupt multa alto- gether. — XXII, 13, 6 ab suo itinere Lucks; ad sue itinere P, a itinere R.

Sometimes the scribes of R, coming upon an unusual or unfamiliar word, or upon something which in their opinion interfered with the construction of the sentence, left out a syllable or a word so as to reduce the unknown element to the known.

e.g, XXIII, 16, 2 primo antesignam i)oenorum, dein signa perturbata P ; the scribe of R, not being familiar with the word antesignani^ omitted the wi, writing primo ante signa poenorum^ etc. — XXIIII, 37, 9 praesidio decedere aput romanos capita? esse P ; the same scribe (Fredeg) in R has left off the ?, and written apud romanos capita esse, — XXVI, 36, 3 pro wivvli parte P, pro uiri parte R ; the scribe in dividing the words first saw uiri and parte and not knowing what to do with the two additional letters left them out. — XXVIIII, 4, 1 animos rursus terror instans revocauit P, aninios rursus terror instare uocauit R ; this error seems to be due in the first place to a mental word-division of instansre uocauit^ and the scribe, supposing instansre was meant for the infinitive, omitted the ns and wrote instare, — XXV, 15, 6 itaque metapontini P, itaque etapontini R ; the omission of the m was apparently due to a wrong mental word-division ; itaquem not being intelligible, it was amended to itaque.

Other Omissions

Two varieties of errors of omission are treated in other chapters to which they more properly belong : Substitutions such as aetatis for aestatis^ temporum for templorum^ etc., are given in Chapter VII, and the omission of abbreviations in Chapter IX.

It is impossible, of course, to account for all of the omissions which occur in R ; most of them admit of being grouped under the heads given above, but there still remain omissions which had no apparent reason or starting-point, and which must be set down as accidents pure and simple.

R exhibits no omissions of any great length. The longest consist, at the most, of not more than three words.

Return to Index

Return to Index