Errors Of Scribes And Their Correction
( Originally Published 1893 )
THE object of most of the work bestowed at the present day on manuscripts is to discover and edit new literary or historical treasures, or to restore the actual words as written by an ancient author, by the exercise of a trained intellect on the more or less faulty materials which have survived to the present time. When we consider how liable a copyist is to errors of transcription, and how, when once an error has been made and has escaped correction, it cannot but be repeated by the next copyist, who also introduces his own new ones, and that in the course of centuries this process of deterioration is theoretically accelerated, and complicated by irresponsible at-tempts at conjectural emendation, we cease to wonder at the importance and honour accorded to Textual Criticism, the system by which the sources of error are classified, and an attempt is made to reverse the course of depravation and undo the accumulated perversions of many generations.
The textual critic then keeps chiefly before him the mind and the hand of the scribe, to which alone can be due variations from the original authentic text ; and in this spirit he considers the manuscripts of his author to which he has access, trying to separate them into classes, according to the peculiarities which they display. He may prove that several of them point back to a single older manuscript, from which they were all derived, and which in relation to these later copies is called their archetype. By this conclusion he has mounted one step nearer to the author's original text ; and if he can discern several archetypes of existing groups of MSS., he may compare even these phantom archetypal MSS., and so penetrate farther and farther into the mist of ages. This is the principle of Genealogy, and may be illustrated by a few simple examples. (I) The Discourses of Arrian, based on the teaching of his master Epictetus, called the were written in the second century after Christ, and exhibit the highest point which Stoic philosophy ever reached. But no MS. of them exists which is older than the twelfth century. In book I. 18. 9 of the Discourses there is a passage of which all the printed editions and all the MSS. make a hash ; some omit the passage, some print words which make no sense, and some indicate by blanks that they cannot deal with the difficulty. But in one MS. alone, the oldest, there is at this very passage a curious oval smear in the middle of the page, which on investigation accounts for all the vagaries of other MSS. and editions. We cannot possibly doubt that this MS. is the archetype of all other existing copies of the Discourses. It is seldom, however, that so clear and decisive a proof can be found. (2) In a MS. of mediæval letters there is one which begins, ` Frater AE. pauperum Christi' (brother AE[thelredus], one of the poor of Christ). This is quite straightforward ; but in one MS. nearly contemporary with the writer himself, the letter begins, ` Frater de pauperum Christi,' which cannot be translated. A comparison of several MSS. shows that the de is a mistake by an ignorant scribe, who mistook a peculiar form of YE for de. Now, whenever a MS. is found with de, we may feel pretty sure that such a peculiar blunder would not twice be made, and that any second later occurrence of the word is due to a copyist who had the first blunderer's own copy before him, and was unable or (we may hope) unwilling to attempt to restore the text by any conjecture of his own. (3) A master-mind like Traube's could do even more than this. As Professor W. M. Lindsay says, ` At one touch of Traube's magic wand the Berne MS. [of Valerius Maximus] has become one of the most precious monuments of mediæval learning.' Traube established the fact that it had been transcribed for Servatus Lupus, the great literary light of ninth-century France, that it was copied from an exemplar written probably at Fulda by an English or Irish scribe, and that that scribe had before him a codex written in capitals !
But when errors are not so easily traced, as in the second example, the textual critic is allowed to consider, not only the mind and the hand of the scribe, but also (and, so to speak, on his own responsibility) the mind of the author also. It is, however, a very slippery matter when one argues from general style, or from similar passages found elsewhere in the work, that the author must have written such and such words ; and the tendency of modern criticism is to confine the use of parallel passages to illustration or corroboration, and to deprecate that attractive exercise of ingenuity which suggests readings not found in any existing manuscript ( Conjectural Emendation ') ,— except within narrow limits, such as when a great author is only preserved to us in a few manuscripts (perhaps one only), as is the case with Catullus and part of the Annals of Tacitus.
It is a remarkable fact that instead of the corruptions and variations increasing in number in proportion to the distance of MSS. from the author's time, that number after a time seems not merely not to increase indefinitely, as might be expected, but actually to diminish ; partly from the correction of blunders by too intelligent scribes, partly from what is called ` mixture ' of MSS., one copy being used to correct and remove the faults of another, so that eccentricities of a single copyist are gradually eliminated by his successors through a comparison of other codexes.
We will now give a classification of sources of error in transcription, and a list of the chief principles on which these errors are corrected.
I. SOURCES OF ERROR IN TRANSCRIPTION
A. UNINTELLIGENT.—I. Errors of sight or hearing.
These occur when the eye of the scribe (or, in the less usual case of dictation, the ear) fails to grasp correctly what has to be copied. A common error is for a whole line of the MS. which is being copied to drop out ; and usually the cause of it is that two lines end with the same word or termination, and the scribe's eye has slipped over the first to the second one. Suppose part of the Lord's Prayer written in old style, with no stops or capitals, thus: bethynamethy kingdomcomethy willbedonein, a copyist might easily omit altogether the second line, because of the Homoioteleuton, as it is called, which might cause the eye to slip from the first thy to the second. Probably no Greek MS. of the New Testament is free from an example of this.
2. Errors of memory.—These occur when, in the interval between seeing and writing, some unconscious cerebration takes place in the copyist's head, and he puts down something wrong. Parallel pas-sages are a very common source of mistake, when the copyist remembers another set of words similar, but not identical, with that before him, and blunders. Thus, in Virgil's Æn. vii. 324, the copyist of an early MS. probably had ab sede dearum before him, but was misled by a recollection of Æn. vii. 454 (where MSS. do not differ) to put down ab sede sororum, which is now found in many MSS., and almost no editions. The point is, that it is so unlikely that a copyist would introduce dearum out of his own head, whereas he might easily introduce sororum wrongly, if he was well acquainted with his Virgil.
3. Errors of intellect.—In Latin, the contraction stands for both mater and martyr, and stands for both miseria and misericordia. It might easily happen that a copyist would unintelligently expand some such contraction wrongly.
B. INTELLIGENT.-I. Incorporation of marginal glosses or various readings.—Often when a word or passage is difficult, it is glossed by the scribe or by the reader ; that is to say, something is written just over it, or in the margin, to make it clearer. In Shakespeare's time the word ` owe ' had a meaning to ` own ' ; we can therefore imagine a copy of Shakespeare in which ` to him that owes it ' might be glossed by ` owns.' Some later copyist, say a foreigner, could easily be conceived as writing, ` to him that owes owns it,' thus incorporating the gloss, because he did not understand it, and thought it to be an addition accidentally omitted, and by all means to be inserted. So, in old days, when one MS. was compared with another, it was a custom to write a variation on the margin of one of the two, and this might be similarly incorporated.
2. Correction of apparent difficulties, such as un-usual forms and expressions, seeming contradictions, or incomplete quotations. Unfortunately, the tendency to make a text read well by removing difficulties led scribes in uncritical times to substitute, with no deliberate desire to mislead, easy words for archaic, and plain for obscure.
3. Deliberate falsification, such as a change of a theological text for dogmatic reasons. In ordinary MSS., even where there might be temptation, this fault is quite rare ; but a curious example may probably be found in Virgil, see p. 79.
II. PRINCIPLES ON WHICH TRANSCRIPTIONAL ERRORS ARE CORRECTED
In the following list the principles are in italics, followed by a brief explanation, if necessary, and references (1) to the foregoing sources of error, and (2) to passages in Virgil which may be taken as examples, if recourse be had to a critical edition such as Conington's or Ribbeck's ; but let it be under-stood that the passages are only adduced to show how a single principle operates in practice, for it is not suggested that that single principle in each case leads to the best text, since it may be overborne by still weightier considerations of another kind.
I. A short reading is to be preferred to one more verbose. This is due to the tendency of scribes to incorporate glosses, to expand hard phrases by easier ones which are often longer, and often to a conscientious desire to put into the text all there is in the MS. copied, whether glosses or various readings or corrections (B. i, B. 2).
Nec to comitem hinc asportare Creüsam Fas.comitem asportare
(Nor is it right for you to bear hence with you Creusa.)
Here the hinc, which is found both before and after comitem, and is in some MSS. omitted, excites suspicion that it is a gloss which has crept into the text ; and when we see how apposite as a gloss the word would be (to show that asportare, ` to carry off,' is quite distinct from apportare, ` to carry to '), our suspicion is increased. In fact, we can hardly doubt that the third line above is the right text, and that the shorter reading is to be preferred. But how soon the hinc found its way into the text may be judged from the fact that Servius, the early commentator on Virgil, declares that the common reading in his time (the fifth century ?) could not be scanned ; so clearly his text was No. 2 in the parallel readings at the head of this note. Another example of this principle will be found in Æn. vii. 464, 465. The whole principle has been forcibly challenged by Professor A. C. Clark, who has found that accidental omissions are more numerous and more important than have been suspected, but his criticism affects sentences and complete lines rather than words or expressions.
2. A difficult or obscure reading is better than one which is, from the point of view of the copyist, fuller and easier (B. 2).
Iam nosces, ventosa ferat cui gloria laudem. poenam.
(Quickly you shall know to which side wind-blown fame will bring defeat.)
This is a good example of a difficult reading (difficult to the mind of the scribe) being preferable. Frans, which seems at first sight impossible to trans-late, since there is no idea of deceit in the passage, had in early Latin the sense of injury, and here means harm, defeat. But copyists who did not know this old meaning substituted laudem, ` victory,' and some who did know it glossed it by pænam, which subsequent copyists took, not as a gloss, but as a better reading. Obviously we should prefer fraudem. Æn. i. 636 is in some points a similar case.
3. A less emphatic reading is nearer to the original text (B. 2) . No author is always at his best, and few writings cannot be improved by persons far inferior to the writer. The common quotation, Sic volo, sic jubeo : stet pro ratione voluntas, a forcible line, ought to be given in the weaker but more correct form, Hoc volo, sic jubeo : sit . . . (Juvenal).
Quique sui memores aliquos fecere merendo. alios
(And who by their good deeds made some to remember them.)
Aliquos is undoubtedly weak ; so weak that Conington declares alios (others) to be ` infinitely preferable,' since it introduces some little antithesis between ` themselves ' and ` others.' But aliquos has decidedly the better testimony, and Virgil probably let a careless line escape him.
4. Readings which owe their origin to simple carelessness on the part of the scribe are rightly rejected (A. 1-3).
Apollo ! Mortales medio .1 aspectu sermone relinquit. aspectus
(Apollo vanishes from mortal sight before his words are ended.)
The extraordinary circumstance about this line is that all the leading MSS. (three uncials and the chief cursive) read aspectu, which makes nonsense of the passage (` Apollo leaves mortals, while looking at them, with a remark ! '). In fact, a reading which has by far the best support is entirely given up by editors, because due to an early scribal blunder. The excuse for its existence at all is, of course, that medio precedes, and a word beginning with s follows, so that MEDIOASPECTUSSERMONE was, as it were, a trap for the inattentive.
5. Out of several readings, that one is best which lies apparently midmost among the others. It is very instructive, when a passage is beset with variations in the MSS., to attempt to reconstruct the Scala Vitiorum, and make a probable genealogy of the readings, whether blunders or corrections. The one which will best account for all the others, and with which the others can be most easily causally connected, is probably the right one (A. and B.).
Traicit." I, verbis virtutem illude superbis." ' Transiit.
(The arrow flies and pierces the hollow temples [of Remulus]
with iron barb. ` Go to, now mock my valour with thy vaunts.')
The first four readings above are those of the three uncials and the chief cursive MS. which contain the passage. The fifth is the corrected reading of the same cursive, and is found also in inferior MSS. Now, what is the pedigree of corruption, and which the form from which the error first sprang ? The ` I,' if not recognized as a complete word by itself, would naturally cause some confusion (TRAICITIVERBIS), and also the form traicio instead of the longer trans-icio. The probable genealogy would seem on the whole to be one which selects traicit i as the ` midmost ' reading from which the rest have most naturally sprung :—
transadigit transiit. I
But there must have been considerable variation even before our existing MSS. were written.
6. Omissions may be suspected when the passages wanting are repugnant to natural feeling or orthodox belief (B. 3).
This is a well-known passage, in which the hero of the Æneid discusses with himself whether he shall slay Helen in cold blood. The omission of the passage, whether by Tucca and Varius (see next page) or in some very early MS., is an argument for their genuineness, when we consider how shocking the idea must have seemed even to Roman minds. Most editors enclose them in brackets as doubtful, but they would appear to be genuine.
We may continue this chapter with two specimens of the literary history of famous books, selected because the amount of testimony to them, both in the number and importance of their MSS., is greater than of any other ancient authors whatever—Virgil and the Four Gospels.
1. Virgil (Publius Vergilius Mayo)
It is recorded of Virgil, as of Tennyson, that he wrote much more than he published, and that he was occupied for much of his time in cutting down and reducing to their best form passages thrown off in the heat of composition. Even at the end of his life he was so diffident of the merits of the Æneid that he wished to burn it. It appears to be certain that Virgil himself wrote out his poems in their best form ; for Aulus Gellius, in the second century of our era, distinctly records that he saw the autograph original manuscript of the Georgics. After Virgil's death, in B.C. 19, Tucca and Varius published, according to their discretion, the AEneid ; and even at this time may have removed from the text the lines in the second book (567-588), in which, as has just been said, the hero of the work expresses his deliberate desire to put Helen to death. Servius preserves the lines, and they occur in the text of only a few late MSS. Publication would, of course, mean that scribes were allowed to copy the auto-graph of Virgil, and that these transcripts were themselves copied, and the process continued till the invention of printing. Our oldest existing MSS. of the Æneid are the following : The Palatine (P), in the Vatican at Rome, of the third or fourth century ; the Vatican (R = Romanus), perhaps of the same date, and adorned with nineteen remark-able pictures illuminated in a classical style ; the Medicean (known as M), in the Laurentian Library at Florence, of the fifth century ; and three sets of fragments, in the Vatican (F, perhaps of the third century, still containing no less than fifty miniatures of the greatest interest and value), at St. Gall (G), and at Verona (V, a palimpsest 1). All these are in capitals, chiefly rustic capitals, and are followed by a host of minuscule MSS., from the ninth century onwards, which have never been enumerated at length, but must amount to two or three hundred at least.
2. The Four Gospels
It is generally agreed that after the death of Christ the history of His life and doctrine was carried on by oral tradition only, with no written record. But as soon as persecution began to disperse the Christians, it was inevitable that for fear of unconscious distortion, or even of simple forgetfulness of the facts, some record should be made which could be put in the hands of disciples departing to distant lands ; it was the only possible way of preventing hopeless disagreement, or incomplete exposition of their common doctrine. In this way, no doubt, originated the four histories now known as the Gospels. The earliest testimony we have to the text is contained in the writings of the early Fathers, where they quote passages from, or otherwise describe, the history of a MS. from which the old writing has been as far as possible scraped to allow of the parchment being again used for later writing. In the British Museum (MS. Add. 17,212) is a rare example of a double palimpsest. Lowest is an uncial MS. of the 5th cent., containing a fragment of the Annales of C. Granius Licinianus ; next, a grammatical Latin treatise, in cursive minuscule ; and on the top a Syriac translation of Chrysostom's Homilies, itself not later than the ninth or tenth century !
Christ ; and it is in accordance with the course of events described above that we find in the earliest notices a real but not a verbal agreement with our present text. Soon, however, the agreement be-comes close, and we can see that the reference is to standard written accounts. The next earliest class of witness to the text is, strange to say, that of versions or translations. There is evidence, for instance, of a Syriac version of the Gospels in the second century ; of an Egyptian, two forms of which are quite possibly also of the second century ; a Latin in the third century ; a Gothic, by Bishop Ulfilas, certainly of the fourth century ; and AEthiopic and Armenian versions before A.D. 600.
Lastly, we come to the age of our existing MSS. of the Greek Text. Two are ascribed to the fourth century ; one, the Codex Sinaiticus (known as Aleph), the romantic discovery of which is related on p. i i r, a MS. of the whole Bible in Greek, now imperfect, but still containing the whole of the New Testament, written (and it is unique in this respect) with four columns on a page ; the other, the Codex Vaticanus (B), also when perfect a MS. of the whole Bible, written in three columns, one of the chief treasures of the Vatican at Rome. Next in point of age ranks the Codex Alexandrinus (A), probably written in the fifth century, a complete Bible, presented to Charles I. in 1628 by Cyril Lucar, Patriarch of Constantinople, he having obtained it from Alexandria, of which he had been Patriarch, and where it had rested for many centuries. George II. placed it in the British Museum, with his library, in 1757, and it has ever since been its chief manuscript treasure. To the fifth century also belongs the Codex rescriptus Ephraemi (C), in the National Library at Paris, once a Bible, now a collection of fragments containing about two-thirds of its original contents. It is a palimpsest, the ancient writing, after being scraped, bearing above it some Greek works of Ephrem Syrus, written in the twelfth century. Codex Bezae (D), containing the Gospels and Acts (nearly complete), and originally the Catholic Epistles, in Greek and Latin, is the glory of the Cambridge University Library, and is generally ascribed to the sixth century. It was presented to the University in 1581 by Theodorus Beza, who obtained it from the monastery of St. Irenæus at Lyons, after it was sacked in 1562. It is celebrated for the extraordinary variations from and additions to the received text, but the most recent critics attach a high value to its readings.
All the five above-mentioned MSS. are written on fine vellum, in quarto form, with uncial letters. In all, there are twenty-one uncial Greek MSS. of the Gospels, or, counting fragments, sixty-six, each of which is known among scholars by a capital letter, or, in the case of fragments, by a capital letter together with a small letter : the majority belong to the sixth, seventh, eighth, and ninth centuries. This list is succeeded by a multitude of later MSS. in minuscule writing, the number of which, for the Gospels alone, is at least 1300, ranging from the ninth to the seventeenth century.
The number of witnesses to the text of Virgil and the Four Gospels is so great, that conjectural emendation is, in fact, not allowable for either book. And this superabundance has also its use when we have to consider less fortunate authors like Lucan or Propertius ; for we are able, by surveying the body of New Testament MSS., to estimate fairly the relation of late to early copies in point of accuracy and general worth, and to confirm our general confidence in a text even when no early testimony to it has survived.
Curiosities of Palaeography
At this point it may be allowable to lighten the subject of scribal errors by some examples of unusual or humorous results of their mistakes. The first place may be taken by Ghost Words—words entirely due to blunders. Every museum contains some ` cells,' or flint implements, a word which first appears about A.D. 1700. It appears that there is no Celtic connexion in the word, but that it arose from a misunderstanding of the Vulgate of Job xix. 23-24. ` Quis mihi det ut (sermones mei) exarentur in libro . . . vel celte sculpantur in silice ? ' Celte was taken as an ablative from a supposed celtis, an engraving instrument, but it is really a phonetic variant of the adverb ` certe ' (certainly, ` permanently,' as the Hebrew and the Septuagint) !
The resemblance of the strokes or pot-hooks which make up m, n, u and i in mediæval writing (and even in early printing), especially when i's had no dot and v was written u, is responsible for much. Not only can eight Latin words be made out of fifteen such strokes (immuniui, innuimini, inunimini, minimum, munimini, numanum,numrum, and uiminium), but the confusion has invaded our own T e Deum, where ` Make them to be numbered with Thy Saints in glory everlasting ' is due to misreading the Latin munerari (` Make them to be awarded with Thy Saints Thy glory everlasting ') as numerari. Till the last few years every loyal member of the Church of England was bidden to commemorate on September 7 a certain bishop Enurchus. No such bishop occurs in any Martyrology, and it turns out that the word ` Euurci, (the genitive of Evurtius in mediæval writing) was misprinted Enurchi in the Preces Privatæ of 1564, and so caused the error. September 7 was Queen Elizabeth's birthday, and in 1604, when the ghostly saint first appears, no better known saint than Evurtius bishop of Orleans could be found on whom to base a commemoration of that day. That is the plausible suggestion made in the Church Times of September 1, 1905, by the Rev. Vernon Staley. It is perhaps ungallant to call attention to the fact that the lady who wrote the Antiquities of Langharne (1871) after some years' residence in the town which she describes, found it necessary to head the list of Errata with the following notice :—` For Langharne wherever it occurs, read Laugharne ' ! ! In the second edition (188o) all is put right. Among other ghost-words are Grampian, Hebrides, St. Parasceve, Siatoutanda (see p. 147), Dedalricus (in German histories), and the commonest of all Roman praenomens, Caius !
The following may be taken as examples of remarkable blunders. The old way of writing MSS. without spaces between words caused even Virgil to make some odd mistakes. In Ecl. viii. 58, as a climax of impossibility he writes, ` Omnia vel medium fiant mare ' (` Let everything become mid ocean '), but the expression is peculiar and not obviously apt. He no doubt took the Greek expression, (` May everything be different '), and read EN'AÀAA as ENAAIA (` sea things ') ! That he was capable of this may be seen from Georg. i. 277 (Orcus), and AEn. ix. 716 (Inarime). All English Bibles of the Authorized Version of 1611 have perpetuated up to the present day the expression, ` Woe to the idol shepherd ! ' in Zech. xi. 17—a mere misprint for idle : and they all still present a hopeless error in Heb. x. 23, where the word ` faith ' (` Let us hold fast the profession of our faith ') is given as a translation of a ris, which is ` hope ' ! The name of Jervaulx Abbey (usually pronounced ` Jarvis'), on the Ure, is not obviously connected with Uredale, but the Latin name of the latter, Jorevallis, supplies the connexion. The signature of the Bishop of Salisbury (`Sarum ') is based on an error. The scribe of the Middle Ages abbreviated all common words, and in a Salisbury roll or document would write Sar' (for Sarisburia or Sarisburiensis), just as Oxon' would imply Oxonia (Oxford) or Oxoniensis. Earlier, it would be written Sar/, or more commonly San, using an old shape of the Y. But this shape of Y plus the bar of Suspension (/) together make up the symbol No. vii. on p. 37, and the word was therefore (quite wrongly) re-expanded as Sarum !