The New Testament in the original Greek - Introduction and Appendix (1882)/Part III
PART III
APPLICATION OF PRINCIPLES OF CRITICISM TO THE TEXT OF THE NEW TESTAMENT
96. The principles of criticism explained in the foregoing section hold good for all ancient texts preserved in a plurality of documents. In dealing with the text of the New Testament no new principle whatever is needed or legitimate: but no other ancient text admits of so full and extensive application of all the various means of discriminating original from erroneous readings which have been suggested to scholars by study of the conditions of textual transmission. On the one hand the New Testament, as compared with the rest of ancient literature, needs peculiarly vigilant and patient handling on account of the intricacy of evidence due to the unexampled amount and antiquity of mixture of different texts, from which few even of the better documents are free. On the other it has unique advantages in the abundance, the antiquity, and above all in the variety of its documentary evidence, a characteristic specially favourable to the tracing of genealogical order.
CHAPTER I. PRELIMINARY CHRONOLOGICAL SURVEY OF DOCUMENTS
97—128
97. Before entering on the historical phenomena of the text itself, and the relations between its principal documents, we think it best to interpose a short general survey of the written evidence with which all criticism has to deal, presenting it in a form somewhat different from that of the detailed catalogues which it is the office of other books to supply. The entire body of documentary evidence, with inconsiderable exceptions, consists of three parts; extant Greek MSS, ancient translations or 'Versions' in different languages, and quotations from the New Testament made by ancient Christian writers or 'Fathers'.
A. 98—106. Greek MSS
98. The Greek MSS of the New Testament are divided into two classes, conventionally though somewhat incorrectly termed 'Uncials' and 'Cursives', according as they are written in capital or in minuscule characters. Since Wetstein's time (1751, 1752) it has been customary to distinguish Uncials by capital letters, and Cursives for the most part by arabic numerals. At the head of the list of Uncials stand four great MSS belonging to the fourth and fifth centuries. When complete, they all evidently contained the whole Greek Bible. At least three, and not improbably all four, had all the books of the New Testament that have been subsequently recognised as canonical, at least two containing other books in addition: as two are mutilated at the end, it is impossible to speak with greater precision. These four MSS are products of the earlier part of that second great period of Church history which begins with the reign of Constantine; the time when the various partial Canons of Scripture were brought together and as it were codified in various ways, the first step in the process being probably the catalogue of Eusebius in his Church History (of about 325), and the most decisive step, at least for the Greek churches, the catalogue of Athanasius in his 39th Paschal Epistle, of 367. About 332 Constantine directed Eusebius to have fifty easily legible copies of the complete Scriptures executed by skilful calligraphers for the use of the churches in his newly founded capital. We learn nothing of the texts or the contents of these "sumptuously prepared volumes" (Eus. Vit. Const. IV 37): but if the contained books corresponded with Eusebius's own list of a few years earlier (H. E. III 25), none of our present MSS can well have been of the number. The incident illustrates however a need which would arise on a smaller scale in many places, as new and splendid churches came to be built under the Christian Empire after the great persecution: and the four extant copies are doubtless casual examples of a numerous class of MSS, derived from various origins though brought into existence in the first instance by similar circumstances. These four are the Codex Vaticanus (B), containing the whole New Testament except the later chapters of Hebrews, the Pastoral Epistles, Philemon, and the Apocalypse; the Codex Sinaiticus (א), containing all the books entire; the Codex Alexandrinus (A), containing all, except about the first 24 chapters of St Matthew's and two leaves of St John's Gospel and three of 2 Corinthians; and the Codex Ephraemi (C), containing nearly three fifths of the whole (145 out of 238 leaves), dispersed over almost every book, one or more sheets having perished out of almost every quire of four sheets. The two former appear to belong to the middle part of the fourth century: the two latter are certainly of somewhat later date, and are assigned by the best judges to the fifth century.
99. The remaining uncial MSS are all of smaller though variable size. None of them shew signs of having formed part of a complete Bible, and it is even doubtful whether any of them belonged to a complete New Testament. Six alone (including one consisting of mere fragments) are known to have contained more than one of the groups of books, if we count the Acts and the Apocalypse as though they were each a group. The Gospels are contained in fair completeness in nineteen uncial MSS (including אABC), the Acts in nine, the Catholic Epistles in seven, the Pauline Epistles in nine (besides the transcripts E3 and F2), and the Apocalypse in five. The numbers given for the Gospels, Acts, and Pauline Epistles do not include some more or less considerable fragments: but the line is hard to draw, and much is lost of C and Γ, which are included in the list.
100. After the four great Bibles the chronological distribution becomes remarkable. The fifth century supplies (besides AC) only Q and T, both consisting of fragments of Luke and John: the sixth century supplies for the Gospels D (all four, but incomplete), Ν and Ρ (fragments of all four), Σ (Matthew and Mark, almost complete), R (fragments of Luke), and Ζ (fragments of Matthew); for the Acts D and E2 (both incomplete); and for the Pauline Epistles D2 (not quite complete): under each head some lesser fragments are not reckoned. The seventh century furnishes merely a few fragments; the eighth, besides lesser fragments, EL (Gospels), Ξ (large fragments of Luke), and B2 (Apocalypse). But the MSS of the ninth and tenth centuries are about as numerous as those of all preceding centuries together. The preceding assignation of uncials to this or that century is founded in most cases on no independent judgement, but on the published estimates of the best qualified palæographers. It is quite possible that some of the intermediate uncials may be placed a century too high or too low, for the absence of dated MSS before the ninth century renders palæographical determination of the absolute chronology as yet insecure. The approximate outlines of the relative or sequential chronology appear however to have been laid down with reasonable certainty; so that the total impression left by a chronological analysis of the list of uncials can hardly be affected by possible errors of detail.
101. The bilingual uncial MSS have a special interest. They are, in Greek and Latin, DΔ of the Gospels, DE2 of the Acts, and D2[E3F2]G3 of the Pauline Epistles; in Greek and Thebaic (the language of Upper Egypt), the fragmentary Τ of Luke and John, with some still smaller fragments of the same kind.
102. The Cursive MSS range from the ninth to the sixteenth centuries. Many of them contain two or more groups of books, and about 30 the whole New Testament. If each MS is counted as one, irrespectively of the books contained, the total number is between 900 and 1000.
103. An accessory class of Greek MSS is formed by Lectionaries or books of ecclesiastical lessons taken from the New Testament, of which above 400 have been catalogued. Above four fifths contain only Gospel lessons, most of the rest lessons from the Acts and Epistles, some few being mixed. About 70 are uncials, and the rest cursives. None however are believed to be older than the eighth or possibly the seventh century, and uncial writing continued in use for Lectionaries some time after it had become obsolete for complete copies of the New Testament or complete divisions of it.
104. Such is the nominal roll of Greek MSS. If however we confine our attention to those sufficiently known to be used regularly as direct evidence, a numerically large deduction has to be made, the amount of which, as distinguished from its value, cannot be estimated even in a rough manner. Comparatively few Lectionaries have as yet been collated. Some of these have been found to contain readings of sufficient value and interest to encourage further enquiry in what is as yet an almost unexplored region of textual history, but not to promise considerable assistance in the recovery of the apostolic text. Of the numerous cursive MSS of the New Testament and its parts hardly any have been printed in extenso. We have however complete and trustworthy collations of a select few from Tregelles, and of a large miscellaneous (English) array from Dr Scrivener, both most careful collators; and tolerably complete collations of other miscellaneous assemblages from Alter (Vienna) and Matthæi (chiefly Moscow and Dresden); with which other collations might probably be classed. On the customary mode of reckoning, by which the four traditional divisions of the New Testament (Acts and Catholic Epistles being counted as one) are taken separately, the full contents of about 150 cursives, besides Lectionaries, may be set down as practically known from these sources. A much larger number are known in various degrees of imperfection, some perhaps almost as well as those included in this first class, from the labours of a series of collators, of whom Mill, Wetstein, Griesbach, Birch, Scholz, and Muralt deserve special mention. Many others have been examined only in selected passages, by which rough presumptions, but hardly more, can be formed as to the general character of the text; and many others again are entirely unknown.
105. This large amount of present ignorance respecting the contents of cursives is much to be lamented. Valuable texts may lie hidden among them; many of them are doubtless sprinkled with relics of valuable texts now destroyed; and fresh collations always throw more or less light on the later history of the text generally, and sometimes on its earlier history. But enough is already known to enable us to judge with reasonable certainty as to the proportional amount of valuable evidence likely to be buried in the copies as yet uncollated. If we are to trust the analogy thus provided, which agrees with what might have been anticipated from the average results of continued transcription generally, nothing can well be less probable than the discovery of cursive evidence sufficiently important to affect present conclusions in more than a handful of passages, much less to alter present interpretations of the relations between the existing documents.
106. The nominal list of uncials needs hardly any appreciable deductions to make it a true representation of the uncial evidence completely available. With the exception of the lately discovered Σ, all the older and more important uncials, some fragments excepted, have now been published in continuous texts, and the various readings of the rest are included in the apparatus critici of Tischendorf and (with unimportant exceptions) of Tregelles.
B. 107—122. Versions
107. The second class of documents consists of Versions, that is, ancient translations of the whole or parts of the New Testament, made chiefly for the service of churches in which Greek was at least not habitually spoken. Besides some outlying Versions, there are three principal classes, the LATIN, the SYRIAC, and the EGYPTIAN. The history of all is still more or less obscure.
108. The LATIN MSS are usually classified under two heads, 'Old Latin' (sometimes miscalled 'Italic') and 'Vulgate'. For some purposes the distinction is convenient and almost necessary: but it disguises the fact that there is a wider difference between the earlier and the later stages of the 'Old Latin' (in this comprehensive sense of the term) than between the later stages and the Vulgate. The statements of Tertullian leave no doubt that when he wrote, near the beginning of the third century, a Latin translation of the New Testament was already current in North Africa. How much earlier it came into existence, and in what manner, cannot be ascertained; but it may be reasonably assumed to have originated in Africa. An exact and authentic transcript of portions of the African text is conveyed to us by the early Latin patristic quotations. The rich evidence supplied by Tertullian's works is indeed difficult to disentangle, because he was fond of using his knowledge of Greek by quoting Scripture in immediate and original renderings, the proportion of which to his quotations from the existing version is indeterminate but certainly large. This disturbing element is absent however from Cyprian's quotations, which are fortunately copious and carefully made, and thus afford trustworthy standards of African Old Latin in a very early though still not the earliest stage.
109. In the fourth century we find current in Western Europe, and especially in North Italy, a second type of text, the precise relation of which to the African text of the second and third centuries has not yet been clearly ascertained. These two Latin texts have very much in common, both in the underlying Greek text and in language; and many of the differences are fully compatible with the supposition that the African was the parent of the European text, having undergone revision when it travelled northwards, and been in some measure adapted to the needs of a more highly cultivated population. On the other hand, other differences, not so easily accounted for by this process, afford some justification for the alternative view that Italy had an indigenous version of her own, not less original than the African. The distinctively African renderings which occur not unfrequently in some of the best European documents may be explained in conformity with either view; as survivors from an earlier state, or as aliens introduced by mixture. Recent investigations have failed to solve this difficult problem, and it must be left for further examination: fortunately the value of the two early forms of the Latin text is not appreciably affected by the uncertainty. The name 'Old Latin', in its narrower and truer sense, may properly be retained for both, where there is no need of distinguishing them, and for the European text, where the African is not extant or never existed; the special designations 'African Latin' and 'European Latin' being employed where they bear a divided testimony.
110. After the middle of the fourth century we meet with Latin texts which must be referred to a third type. They are evidently due to various revisions of the European text, made partly to bring it into accord with such Greek MSS as chanced to be available, partly to give the Latinity a smoother and more customary aspect. In itself the process was analogous to that by which the European text must have been formed, on the supposition that it was of African parentage: but, as we shall see presently, the fundamental text now underwent more serious changes, owing to the character of the Greek MSS chiefly employed. The fact that the Latin text found in many of Augustine's writings is of this type has long been used with good reason to shew what he meant by the Itala which he names in a single laudatory notice (De doct. Chr. ii 15). Without doubt this name was intended to distinguish the version or text which he had in view from the 'African' version or text with which he was likewise familiar ('codices Afros' Retr. i 21 3). The only open question is whether he had definitely before his mind a special text due to a recent North Italian revision, as has been usually assumed by those who have interpreted rightly the general bearing of his words, or was merely thinking of the text of Italy in such a comprehensive sense as would include what we have called the European text. The former view was a necessary inference from the assumption that the best known Old Latin MSS of the Gospels had a strictly African text: but much of its probability is lost when it is seen how far removed they are from a Cyprianic standard. But whatever may be the precise force of the term as used by Augustine, such revised texts as those which he himself employed constitute an important stage in the history of the Latin New Testament: and it can hardly lead to misunderstanding if we continue to denote them by the convenient name 'Italian'.
111. The endless multiplicity of text in the Latin copies at length induced Jerome, about 383, to undertake a more thorough revision of the same kind. We learn from his own account nothing about his Greek MSS except that they were "old"; or about his mode of proceeding except that he made no alterations but such as were required by the sense, and that he kept specially in view the removal of the numerous interpolated clauses by which the Gospels were often brought into factitious similarity to each other in parallel passages. Internal evidence shews that the Latin MSS which he took as a basis for his corrections contained an already revised text, chiefly if not wholly 'Italian' in character. In the Gospels his changes seem to have been comparatively numerous; in the other books of the New Testament, which he left without any explanatory preface, but which he must have taken in hand as soon as the Gospels were finished, his changes were evidently much scantier and more perfunctory. It is worthy of notice that readings distinctly adopted in his own writings are not seldom at variance with the revised text which bears his name. These discrepancies may possibly be due to a change of view subsequent to the revision: but in any case it would be rash to assume that Jerome deliberately considered and approved every reading found in his text, even of the Gospels, and much more of the other books which passed through his hands. The name 'Vulgate' has long denoted exclusively the Latin Bible as revised by Jerome; and indeed in modern times no continuous text of any other form of the Latin version or versions was known before 1695.
112. Generations not a few had passed before the Hieronymic revision had even approximately displaced the chaos of unrevised and imperfectly revised Latin texts; and during the period of simultaneous use the Latin Vulgate, as we may now call it, suffered much in purity by the casual resumption of many readings expelled or refused by Jerome. Scribes accustomed to older forms of text corrupted by unwitting reminiscence the Vulgate which they were copying; so that an appreciable part of Jerome's work had been imperceptibly undone when the Vulgate attained its final triumph. Partly from this cause, partly from the ordinary results of transcription, the Vulgate text underwent progressive deterioration till long after the close of the Middle Ages, notwithstanding various partial attempts at correction. At length the authoritative 'Clementine' revision or recension of 1592 removed many corruptions. Many others however were left untouched, and no critically revised text of the Latin Vulgate New Testament founded systematically on more than one or two of the best MSS has yet been edited. The text of at least two of the best as yet known, and a very few others comparatively good, has however been printed at full length.
113. The existing MSS of the Old Latin Gospels, distinguished by small letters, belong for the most part to the fourth, fifth, and sixth centuries: one however (c), strange to say, was written as late as the eleventh century. Hardly any are quite complete, and those which contain more than inconsiderable fragments amount to about fourteen, of which on an average scarcely more than half are extant in any one passage: in this computation Ante-Hieronymic texts of all types are included. Among the few fragments not counted are two leaves which agree closely with one of the comparatively complete MSS: but with this exception all known MSS shew more or less textual individuality, and there are many traces of sporadic and casual mixture. Two of the MSS (e k) are substantially African, a large proportion of their texts being absolutely identical with that of Cyprian, where he differs from European MSS and Fathers; but each has also an admixture of other readings: both are unfortunately very imperfect, e having lost above two-fifths of its contents, chiefly in Matthew and Mark, and k above three-fourths, including the whole of Luke and John. Two other MSS (f q), and one or two fragments, must be classed as 'Italian'. The remaining ten, though African readings are found to a certain extent in some of them, and Italian readings in others, have all substantially European texts.
114. Various modifications of late revision and mixture are represented in some Latin, MSS of the Gospels, which do not properly fall under any one of the preceding heads. Four of them are usually marked as Old Latin (ff1 g1,2 l); but most of the number pass simply as copies of the Vulgate. With few exceptions their texts are as yet imperfectly known; and the relations of their texts to each other, and to the Hieronymic or any other late revisions, have still to be investigated. They are certainly however in most cases, and not improbably in all, monuments of the process described above (§ 112) by which Old Latin readings, chiefly European but in a few cases African, found their way into texts fundamentally Hieronymic. The chief worth of these Mixed Vulgate MSS for the criticism of the Greek text consists in the many valuable particles of Latin texts antecedent to the Vulgate which have thus escaped extinction by displacing Jerome's proper readings. Mixed texts of this class are not confined to the Gospels; but in the other books, so far as they are yet known, their Ante-Hieronymic elements contain a much smaller proportion of valuable materials.
115. The Gospels alone are extant in a series of tolerably complete Old Latin MSS. For most of the other books we have, strictly speaking, nothing but fragments, and those covering only a small proportion of verses. The delusive habit of quoting as Old Latin the Latin texts of bilingual MSS has obscured the real poverty of evidence. These MSS are in Acts Cod. Bezae (D, d; as in the Gospels) and Cod. Laudianus (E2, e), and in St Paul's Epistles Cod. Claromontanus (D2, d) and Cod. Boernerianus (G3, g; without Hebrews). The origin of the Latin text, as clearly revealed by internal evidence, is precisely similar in all four MSS. A genuine (independent) Old Latin text has been adopted as the basis, but altered throughout into verbal conformity with the Greek text by the side of which it was intended to stand. Here and there the assimilation has accidentally been incomplete, and the scattered discrepant readings thus left are the only direct Old Latin evidence for the Greek text of the New Testament which the bilingual MSS supply. A large proportion of the Latin texts of these MSS is indeed, beyond all reasonable doubt, unaltered Old Latin: but where they exactly correspond to the Greek, as they do habitually, it is impossible to tell how much of the accordance is original, and how much artificial; so that for the criticism of the Greek text the Latin reading has here no independent authority. The Latin texts of Δ of the Gospels and F2 of St Paul's Epistles are Vulgate, with a partial adaptation to the Greek. Besides the Græco-Latin MSS there are four Gothico-Latin leaves of Romans.
116. The relics of genuine Old Latin MSS of the books after the Gospels are as follows. For Acts: a few palimpsest leaves of an African text (h); a complete European copy (g), and also the story of Stephen from a Lectionary (g2), both agreeing closely with the quotations of Lucifer; and some palimpsest fragments of the later chapters (s), with a text of the same general type. For the Catholic Epistles: one (? European) MS of St James, and some fragments of the next three epistles in a later (? Italian) text (q): the palimpsest fragments of James and 1 Peter accompanying s of Acts are apparently Vulgate only. For the Pauline Epistles: considerable Italian fragments of eight epistles (r), with leaves from two other MSS having similar texts (r2 r3). For the Apocalypse: two palimpsest leaves of a purely African text (h), and a late European text of the whole book (g). Other portions of Ante-Hieronymic texts of different books are said to have been discovered in Italy; and doubtless others will in due time be brought to light.
117. This is the fitting place to speak of the quotations made by Latin Fathers, for they constitute a not less important province of Old Latin evidence than the extant MSS; not only furnishing landmarks for the investigation of the history of the version, but preserving numerous verses and passages in texts belonging to various ages and in various stages of modification. Even in the Gospels their aid is always welcome, often of the highest value; while in all other books they supply not only a much greater bulk of evidence than our fragmentary MSS, but also in not a few cases texts of greater antiquity. Some books and parts of books are of course much worse represented than others, more especially such books as formed no part of the original North African Canon. But in the Apocalypse Primasius, an African writer of the sixth century, has preserved to us an almost uninterrupted text, which is proved by its close similarity to the quotations of Cyprian to be African Latin of high purity. Thus, singularly enough, the Apocalypse possesses the unique advantage of having been preserved in a Latin text at once continuous and purely African. The quotations of other late African Fathers from various books exhibit an African text much altered by degeneracy and mixture, but preserving many ancient readings.
118. The SYRIAC versions are, strictly speaking, three in number. The principal is the great popular version commonly called the Peshito or Simple. External evidence as to its date and history is entirely wanting: but there is no reason to doubt that it is at least as old as the Latin version. Till recently it has been known only in the form which it finally received by an evidently authoritative revision, a Syriac 'Vulgate' answering to the Latin 'Vulgate'. The impossibility of treating this present form of the version as a true representation of its original text, without neglecting the clearest internal evidence, was perceived by Griesbach and Hug about the beginning of this century: it must, they saw, have undergone subsequent revision in conformity with Greek MSS. In other words, an Old Syriac must have existed as well as an Old Latin. Within the last few years the surmise has been verified. An imperfect Old Syriac copy of the Gospels, assigned to the fifth century, was found by Cureton among MSS brought to the British Museum from Egypt in 1842, and was published by him in 1858. The character of the fundamental text confirms the great antiquity of the version in its original form; while many readings suggest that, like the Latin version, it degenerated by transcription and perhaps also by irregular revision. The rapid variation which we know the Greek and Latin texts to have undergone in the earliest centuries could hardly be absent in Syria; so that a single MS cannot be expected to tell us more of the Old Syriac generally than we should learn from any one average Old Latin MS respecting Old Latin texts generally. But even this partially corrupted text is not only itself a valuable authority but renders the comparatively late and 'revised' character of the Syriac Vulgate a matter of certainty. The authoritative revision seems to have taken place either in the latter part of the third or in the fourth century. Hardly any indigenous Syriac theology older than the fourth century has been preserved, and even from that age not much available for textual criticism. Old Syriac readings have been observed as used by Ephraim and still more by Aphraates: but at present there are no means of supplying the lack of Old Syriac MSS to any appreciable extent from patristic quotations. Of the Old Syriac Acts and Epistles nothing as yet is known. The four minor Catholic Epistles and the Apocalypse, not being included in the Canon of the Syrian Churches, form no part of the true Syriac Vulgate, but are extant in supplementary versions. None of the editions of the Syriac Vulgate come up to the requirements of criticism: but considerable accessions to the evidence for the Greek text are hardly to be looked for from this source.
119. A second version, closely literal in its renderings, was made by Polycarpus for Philoxenus of Mabug in 508. Little is known of it in this its original condition. We possess a revision of it made by Thomas of Harkel in 616, containing all the New Testament except the Apocalypse. The margin contains various readings taken from Greek MSS, which must either have been ancient or have had ancient texts. A third version, written in a peculiar dialect, is found almost exclusively in Gospel Lesson-books, and is commonly called the Jerusalem Syriac. The text is of ancient character: but there is no other evidence to shew when the version was made. Besides one almost complete Lesson-book known for some time, a few considerable fragments have lately come to light. They include a few verses of the Acts. Various signs render it likely that both these versions were in some sense founded on one or other of the two forms of the Peshito. But the whole subject awaits fuller investigation.
120. The third great group of Versions is the EGYPTIAN. The Coptic or Egyptian versions proper are three, very unequally preserved. The Memphitic, the version of Lower Egypt, sometimes loosely designated as the Coptic, contains the whole New Testament, though it does not follow that all the books were translated at the same period, and the Apocalypse was apparently not treated as a canonical book. The greater part of the version cannot well be later than the second century. A very small number of the known MSS have been used in the existing editions, and that on no principle of selection. A cursory examination by Dr Lightfoot has recently shown much diversity of text among the MSS; and in Egypt, as elsewhere, corruption was doubtless progressive. The version of Upper Egypt, the Thebaic or Sahidic, was probably little if at all inferior in antiquity. It in like manner contained the whole New Testament, with the Apocalypse as an appendix. No one book is preserved complete, but the number of extant fragments, unfortunately not yet all published, is considerable. Of the third Egyptian version, the Bashmuric, about 330 verses from St John's Gospel and the Pauline Epistles alone survive. With the Egyptian versions proper it is at least convenient to associate the Æthiopic, the version of ancient Abyssinia, dating from the fourth or fifth century. Though written in a totally different language, it has strong affinities of text with its northern neighbours. The best judges maintain its direct derivation from a Greek original: but neither this question nor that of the relation of the Thebaic to the Memphitic version can be treated as definitively settled while so much of the evidence remains unpublished. The numerous MSS of the Æthiopic have been ascertained to vary considerably, and give evidence of revision: but the two editions yet printed are both unsatisfactory. No book of the New Testament is wanting.
121. Besides the three great groups two solitary versions are of considerable interest, the one from outlying Asia, the other from outlying Europe. These are the ARMENIAN and the GOTHIC. The ARMENIAN, which is complete, was made early in the fifth century. Some modern copies, followed by the first printed edition, contain corruptions from the Latin Vulgate: but the Armenian translators certainly followed Greek MSS, probably obtained from Cappadocia, the mother of Armenian Christianity. The GOTHIC version, the work of Ulfilas the great bishop of the Goths, dates from the middle of the fourth century. He received a Greek education from his Christian parents, originally Cappadocians: and Greek MSS unquestionably supplied the original for his version. We possess the Gospels and the Pauline Epistles (Hebrews excepted), with many gaps, admirably edited from MSS of about the sixth century.
122. The other versions are of comparatively late date, and of little direct value for the Greek text, though some of them, as the Slavonic, bear traces of ancient texts. Most of them are only secondary translations from other versions, chiefly the Latin and Syriac Vulgates. C. 123—126. Fathers
123. The third class of documentary evidence is supplied by the writings of the Fathers, which enable us with more or less certainty to discover the readings of the MS or MSS of the New Testament which they employed. The quotations naturally vary in form from verbal transcripts of passages, short or long, through loose citations down to slight allusions. Nay there are cases in which the absence of even an allusion allows the text read by an author to be inferred with tolerable certainty: but this negative evidence is admissible only with the utmost caution.
124. Besides the evidence as to the texts used by ancient writers which is supplied by their quotations, allusions, or silences, a few of them sometimes make direct assertions as to variations of reading within their knowledge. The form of assertion varies much, now appearing as a statement that, for instance, "some" or "many" or "the most accurate" "copies" contain this or that variant, now as an allegation that the true reading has been perversely depraved by rash or by heretical persons for some special end. This whole department of patristic evidence has a peculiar interest, as it brings vividly before the reader the actual presence of existing variations at a remote antiquity. Its true value is twofold: for the history of the whole text it certifies two or more alternative readings as simultaneously known at a definite time or locality; and for the settlement of the text in a given passage it usually enables the reading adopted by the writer to be known with a higher degree of certainty than is attainable in a majority of cases by means of ordinary quotations. But this superior certitude must not be confounded with higher authority: the relative excellence or the historical position of the text employed by a Father has nothing to do with the relative adequacy of our means of ascertaining what his text actually was. Moreover in the statements themselves the contemporary existence of the several variants mentioned is often all that can be safely accepted: reliance on what they tell us beyond this bare fact must depend on the estimate which we are able to form of the opportunities, critical care, and impartiality of the respective writers.
125. An enumeration of the Greek Fathers would be out of place here. The names most important in textual criticism will come before us presently, when we have to speak of the peculiar value of their evidence as enabling us to trace the outlines of the early history of the text. This is however the place for observing that the extent of patristic evidence still preserved is considerably less than might have been a priori anticipated. Numerous verses of the New Testament are rarely or never quoted by the Fathers: the gaps in the evidence are still more striking if we take the Ante-Nicene Fathers by themselves. A small portion of Origen's commentaries is virtually all that remains to us of the continuous commentaries on the New Testament belonging to this period: they include Matt. xiii 36—xxii 33 in the original Greek (perhaps in an abridged form), and Matt. xvi 13—xxvii 66 in a condensed Latin translation, preserving matter not found in the Greek now extant; some verses of St Luke (a much condensed Latin translation of Homilies on i—iv, not continuous, and on five later passages of St Luke being also extant); John i 1—7, 19—29; ii 12—25; iv 13—54; viii 19—25 and 37—53; xi 39—57; xiii 2—33 (little more than a sixth of the whole) in the full original text; Romans in the much condensed and much altered version of Rufinus; many verses of 1 Corinthians and Ephesians; and a few scattered verses of some of the other books. The extant commentaries and continuous series of homilies written before the middle of the fifth century are as follows:—Theodore of Mopsuestia on the minor Pauline Epistles in a Latin translation; Chrysostom's Homilies, which include St Matthew, St John, Acts (ill preserved), and all the Pauline Epistles; Theodoret on all the Pauline Epistles, his notes being chiefly founded on the works of Theodore of Mopsuestia and Chrysostom; and Cyril of Alexandria's Homilies on St Luke (many fragments in Greek and large portions in a Syriac translation) and Commentary on John i 1—x 17; xii 49—end, with fragments on the rest of the book and on the other Gospels and several of the Pauline Epistles; together with fragments by other writers preserved in Catenæ under various conditions, sometimes apparently in their original integrity, but much oftener in a condensed and partly altered shape.
126. It is on the whole best to class with patristic evidence a few collections of biblical extracts, with little or no intervening matter, selected and arranged for doctrinal or ethical purposes. The Ethica of Basil of Cæsarea (Cent. IV) and the Parallela Sacra of John of Damascus (Cent. VIII) are the best known Greek examples: parts of some of Cyril of Alexandria's dogmatic writings, especially the Thesaurus, have nearly the same character. A Latin collection of a similar kind, the Speculum which wrongly bears the name of Augustine, but is of unknown authorship, has usually been placed with Old Latin MSS under the signature m, and contains an interesting but not early Old Latin text. Of much the same structure are the three books of Testimonia by Cyprian, and indeed a large part of his little treatise De exhortatione martyrii addressed to Fortunatus.
127, 128. Documentary preparation for this edition
127. It is right that we should here explain to what extent we have thought it our duty to take part ourselves in the indispensable preparatory work of collecting documentary evidence. Great services have been rendered by scholars who have been content to explore and amass texts and readings for the use of others; or again who have discussed principles and studied documents without going on to edit a text. On the other hand an editor of the New Testament cannot completely absolve himself from either of these two preliminary tasks without injury to his own text: but the amount of personal participation required is widely different for the two cases. If he has not worked out at first hand the many and various principles and generalisations which are required for solving the successive problems presented by conflicts of evidence, the resulting text is foredoomed to insecurity: but the collection of evidence is in itself by no means an indispensable apprenticeship for the study of it.
128. We have accordingly made no attempt to follow the example of those editors who, besides publishing critical texts of the New Testament, have earned the gratitude of all who come after them by collation of MSS and accumulation of registered evidence in the form of an apparatus criticus. As we have never proposed to do more than edit a manual text, so we have no considerable private stores to add to the common stock. The fresh evidence which we have obtained for our own use has been chiefly patristic, derived in a great measure from writings or fragments of writings first published during the last hundred years, or now edited from better MSS than were formerly known. While in this and other respects the evidence already accessible to all students has been to a certain limited extent augmented, it has of course been frequently verified and re-examined, not only for the sake of clearing up ambiguities or doubts, but because the needful experience could hardly be otherwise acquired. The exigencies of our task demanded a personal acquaintance with the outward phenomena of MSS, with the continuous texts of individual MSS and versions, and with the varying conditions under which the New Testament is quoted and referred to by the Fathers; for no information at second hand can secure the conveyance of a correct and vivid impression of the true and complete facts by bare lists of authorities cited for a succession of detached and sharply defined various readings. But we have deliberately chosen on the whole to rely for documentary evidence on the stores accumulated by our predecessors, and to confine ourselves to our proper work of investigating and editing the text itself. Such a concentration of labour ought at least to favour an impartial survey of the entire field of evidence, and to give time and opportunity for prolonged consideration of the text and its history in various lights.
CHAPTER II. RESULTS OF GENEALOGICAL EVIDENCE PROPER
129–255
SECTION I. DETERMINATION OF THE GENEALOGICAL RELATIONS OF THE CHIEF ANCIENT TEXTS
129—168
129. After this short preliminary survey of the existing documents out of which the text of the New Testament has to be recovered, we have now to describe the chief facts respecting their ancestry and the character of their texts which have been learned by study of their contents or from any other sources, and which render it possible to deal securely with their numerous variations in accordance with the principles of criticism explained in the preceding section. We have already seen, first, that decision upon readings requires previous knowledge of documents, and secondly that the most valuable part of the knowledge of individual documents implies a previous knowledge of the genealogical history of the text as a whole. The first step therefore towards fixing the places of the existing documents relatively to each other is to employ them conjointly as evidence for discovering the more ancient ramifications of transmission; and for this purpose the whole mass of documents of all dates and all kinds must at the outset be taken into account.
A. 130, 131. Priority of all great variations to Cent. v
130. A glance at any tolerably complete apparatus criticus of the Acts or Pauline Epistles reveals the striking fact that an overwhelming proportion of the variants common to the great mass of cursive and late uncial Greek MSS are identical with the readings followed by Chrysostom (ob. 407) in the composition of his Homilies. The coincidence furnishes evidence as to place as well as time; for the whole of Chrysostom's life, the last ten years excepted, was spent at Antioch or in its neighbourhood. Little research is needed to shew that this is no isolated phenomenon: the same testimony, subject to minor qualifications unimportant for the present purpose, is borne by the scattered quotations from these and other books of the New Testament found in his voluminous works generally, and in the fragments of his fellow-pupil Theodorus of Antioch and Mopsuestia, and in those of their teacher Diodorus of Antioch and Tarsus. The fundamental text of late extant Greek MSS generally is beyond all question identical with the dominant Antiochian or Græco-Syrian text of the second half of the fourth century. The community of text implies on genealogical grounds a community of parentage: the Antiochian Fathers and the bulk of extant MSS written from about three or four to ten or eleven centuries later must have had in the greater number of extant variations a common original either contemporary with or older than our oldest extant MSS, which thus lose at once whatever presumption of exceptional purity they might have derived from their exceptional antiquity alone.
131. The application of analogous tests to other groups of documents leads to similar results. The requisite chronological criteria are to be found in the Greek patristic evidence of the second, third and fourth centuries; in the Latin patristic evidence of the third and fourth centuries; in the Old Latin version, as dated indirectly by the Latin patristic evidence; in the Vulgate Latin, the Gothic, and virtually the Armenian versions, as dated by external evidence; and the two (or possibly three) oldest extant Greek MSS, B, א, and A; the Armenian version and probably A being however a little over the line. To this list may safely be added the Old and Vulgate Syriac, as they have some sufficient if slight patristic attestation in the early part of the fourth century, although the evidence which completely establishes their antiquity, being inferential, would not entitle them to a place here; and also the two principal Egyptian versions, the early age of which, though destitute of the testimony which it would doubtless have received from the preservation of an early Coptic literature, is established by historical considerations independent of the character of the texts. The list, however limited, contains a sufficient variety of strictly or approximately direct historical evidence to enable us at once to refer to the fourth century at latest the original of nearly every considerable group of extant documents which frequently recurs in the apparatus criticus, and indeed to carry back some to the third, and others to the second century. In each case the genealogical process here employed can of course do no more than supply an inferior limit of age: a lost original thus proved to be as old as the fourth century may, for all that we have thus far seen, be in reality as old as the other lost originals which can be positively referred to earlier times. What we have gained is the limitation of enquiry by the knowledge that all the important ramifications of transmission preceded the fifth century.
B. 132—151. Posteriority of Syrian (δ) to 'Western' (β) and other (neutral, α) readings shown (1) by analysis of Conflate Readings
132. Within this comparatively restricted field we have next to investigate the genealogical relations of the principal groups of documents, or, what is virtually the same thing, of their respective lost originals, following partly, as before, external evidence, partly the indications of sequence obtained by Internal Evidence of the Groups as wholes. The presence of early and extensive mixture betrays itself at once in the number and intricacy of cross distributions of attestation (see § 60), and thus it becomes important to ascertain at the outset whether any whole groups have been affected by it; and if such can be found, to determine the contributory groups which are thereby proved not merely to be of earlier date, but to have been the actual parents of the groups of mixed origin.
133. The clearest evidence for this purpose, as we have already seen (§ 62), is furnished by conflate readings, where they exist; and in the case of some of the primary groupings of the textual documents of the New Testament they are fortunately not wanting. Before proceeding however to examine some examples of this kind, it may be well to notice a few illustrations of the phenomenon of 'conflation' in its simpler form, as exhibited by single documents. Here and always we shall use the ordinary notation, unless there is sufficient reason for departing from it: a list of special symbols and abbreviations employed is given in the Appendix. In Acts vi 8, where the two readings πλήρης χάριτος and πλήρης πίστεως are attested each by a plurality of documents, E2 alone combines them, by means of a conjunction, reading πλήρης χάριτος καὶ πίστεως. In Mark vi 56 the Latin MS a couples the readings ἐν ταῖς ἀγοραῖς and ἐν ταῖς πλατείαις by a conjunction, and slightly modifies them, reading in foro et in plateis. In John v 37 D makes ἐκεῖνος αὐτός out of ἐκεῖνος and αὐτός without a conjunction, and similarly John xiii 24 stands in one principal text as νεύει οὖν τούτῳ Σ. Π. καὶ λέγει αὐτῷ Εἰπὲ τίς ἐστιν περὶ οὗ λέγει in another as νεύει οὖν τούτῳ Σ. Π. πυθέσθαι τίς ἂν εἴη περὶ οὗ λέγει, while א adds one form to the other, merely changing a tense, and reads νεύει οὖν τούτῳ Σ. Π. πυθέσθαι τίς ἂν εἴη περὶ οὗ ἔλεγεν, καὶ λέγει αὐτῷ Εἰπὲ τίς ἐστιν περὶ οὗ λέγει. In 1 Cor. x 19 the readings τί οὖν φημί; ὅτι εἰδωλόθυτόν τί ἐστιν; ἢ ὅτι εἴδωλόν τί ἐστιν; and τί οὖν φημί; ὅτι εἰδωλόθυτόν ἐστίν τι· οὐχ ὅτι εἴδωλόν ἐστίν τι, or their Latin equivalents, are ingeniously interwoven by fuld. as quid ergo dico? quod idolis immolatum sit aliquid, aut quod idolum sit aliquid? non quod idolum sit aliquid. Luke xvi 30 illustrates another kind of combination, in which part of a longer reading is replaced by the whole of the shorter reading: for ἐάν τις ἐκ νεκρῶν πορευθῇ πρὸς αὐτούς or ἐάν τις ἐκ νεκρῶν ἀναστῇ (implied in the Latin reading si quis ex mortuis resurrexerit [v. 1. surrexerit]) א has ἐάν τις ἐκ νεκρῶν ἀναστῇ πρὸς αὐτούς, while two or three other documents retain both verbs. In 1 Cor. i 8 the Latin Vulgate effects the combination by making the one element dependent on the other, changing the Old Latin in adventu Domini nostri (ἐν τῇ παρουσιᾳ τοῦ κυρίου ἡμῶν) into in die adventus Domini nostri by incorporating the Greek reading ἐν τῇ ἡμέρᾳ τοῦ κυρίου ἡμῶν. Bold conflations, of various types, are peculiarly frequent in the Æthiopic version, at least in the extant MSS.
134. We now proceed to conflate readings involving important groups of documents, premising that we do not attempt to notice every petty variant in the passages cited, for fear of confusing the substantial evidence.
Mark vi 33 (following καὶ εἶδαν αὐτοὺς ὑπάγοντας καὶ [ἐπ]έγνωσαν πολλοί, καὶ πεζῇ ἀπὸ πάσων τῶν πολέων συνέδραμον ἐκεῖ)
| (α) | καὶ προῆλθον αὐτούς אΒ (LΔ 13) It (39) 49 lat.vg me arm (LΔ 13 It 39 have προσῆλθον) |
| καὶ προῆλθον αὐτὸν αὐτοῦ syr.vg | |
| (β) | καὶ συνῦλθον αὐτοῦ D 28 b |
| καὶ ἦλθον αὐτοῦ 81 ff i | |
| καὶ ἦλθον a | |
| om. cu3 (c) | |
| (δ) | καὶ προῆλθον αὐτοὺς καὶ συνῦλθον πρὸς αὐτόν AEFGHKMUVΓΠ cu.omn.exc.8 f q syr.hl aeth |
136. If now we compare the three readings with reference to Transcriptional Probability, it is evident that either δ is conflate from α and β, or α and β are independent simplifications of δ; for the similarity αὐτοῦ and αὐτούς, combined with the relative dissimilarity of both to πρὸς αὐτόν, shews that δ can hardly have been a passage from α to β or from β to α; and the independent derivation of β and δ from α, or of α and δ from β, would be still more incredible. There is nothing in the sense of δ that would tempt to alteration: all runs easily and smoothly, and there is neither contradiction nor manifest tautology. Accidental omission of one or other clause would doubtless be easy on account of the general similarity of appearance (και...ηλθον...αυτο...), and precedents are not wanting for the accidental omission of even both clauses in different documents or groups of documents. On the other hand the change from πρὸς αὐτόν of δ to αὐτοῦ of β is improbable in itself, and doubly improbable when ἐκεῖ has preceded. Supposing however α and β to have preceded δ, the combination of the two phrases, at once consistent and quite distinct in meaning, would be natural, more especially under the influence of an impulse to omit no recorded matter; and the change from αὐτοῦ to πρὸς αὐτόν (involving no change of historical statement, for the place denoted by αὐτοῦ was the place to which the Lord had gone) might commend itself by the awkwardness of αὐτοῦ (itself a rare adverb in the New Testament) after συνέδραμον ἐκεῖ, and by the seeming fitness of closing this portion of narrative with a reference to the Lord Himself, who is moreover mentioned in the opening words of the next verse.
137. As between α and β the transcriptional probabilities are obscure, Συνῆλθον αὐτοῦ is certainly otiose after συνέδραμον ἐκεῖ, and a sense of the tautology might lead to change; but the changes made by scribes hardly ever introduce such vivid touches as this of the arrival of the multitude before the apostles. On the other hand προῆλθον αὐτούς might be altered on account of the unfamiliarity of the construction or the unexpectedness of the sense, which harmonises with the earlier words εἶδον αὐτοὺς ὑπάγοντας but would hardly be suggested by them; and then συνέδραμον might suggest to the ear and perhaps to the mind συνῆλθον, after which αὐτούς would be inevitably read as αὐτοῦ, αὐτοῖς being in manifest contradiction to the contrast between ἐν τῷ πλοίῳ and πεζῇ: the tautology introduced might easily escape notice at first under the different phraseology, especially if συνῆλθον were taken to express the arrival subsequent to the running, though it was perceived afterwards, as we see by the omission of αὐτοῦ in a, and of the whole clause in c, where convenerunt stands for cognoverunt above.
138. As regards Intrinsic Probability, β may be dismissed at once, on grounds virtually given already. Had δ been the only extant reading, it would have roused no suspicion: but when it has to be compared with α, we cannot but notice the irrelevance of the repetition of σύν in composition with two different verbs not in immediate sequence, and the intrusiveness of καὶ προῆλθον αὐτούς between the local and the personal endings of the journey expressed by ἐκεῖ and πρὸς αὐτόν; the position of this clause can be justified only if συνέδραμον is inserted merely to account for the prior arrival, and in that case ἐκεῖ is out of place. Nor is St Mark's characteristic abundance of detail to the purpose here, for his multiplication of accessory facts is at least equalled by his economy of words. Had he wished to introduce the only fresh point in δ, that conveyed by πρὸς αὐτόν, the language natural to him would have been ἔδραμον καὶ (or better δραμόντες) προῆλθον αὐτούς καὶ συνῆλθον πρὸς αὐτόν. But the truth is that this fresh point simply spoils the point of ἐξελθών in v. 34; the multitude 'followed' (Matt, Luke) the Lord to the desert region (ἐκεῖ), but the actual arrival at His presence was due to His act, not theirs, for He 'came out' of His retirement in some sequestered nook to meet them. Thus, if we look below the surface, the additional phrase in δ is found to disarrange the diction and confuse rather than enrich the sense; while according to the clear and exact language of α the fact to which the whole sentence leads up stands emphatically at its close, and there is no premature intrusion of what properly belongs to the next part of the narrative.
139. Accordingly the balance of Internal Evidence of Readings, alike from Transcriptional and from Intrinsic Probability, is decidedly in favour of the derivation of δ from α and β rather than of α and β from δ; so that, as far as can be judged without the aid of other passages, the common original of the documents attesting α and the common original of the documents attesting β must both have been older than the common original of the documents attesting δ.
140. To examine other passages equally in detail would occupy too much space. For the following similar variations it will for the most part suffice to add but brief comments to the documentary attestation.
Mark viii 26 (following καὶ ἀπέστειλεν αὐτὸν εἰς οἶκον αὐτοῦ λέγων)
(α)Μηδὲ εἰς τὴν κώμην εἰσέλθῃς (א)BL i*-209 me
(β)Ὕπαγε εἰς τὸν οἶκόν σου καὶ μηδενὶ εἴπῃς εἰς τὴν κώμην D(q)
(β2) Ὕπαγε εἰς τὸν οἶκόν σου καὶ ἐὰν εἰς τὴν κώμην εἰσέλθῃς μηδενὶ εἴπῃς μηδὲ ἐν τῇ κώμῃ 13-69-346 28 61 81; also (omitting μηδέ) i, and (omitting μηδὲ ἐν τῇ κώμῃ b f ff g1.2 vg
Ὕπαγε εἰς τὸν οἶκόν σου καὶ μὴ εἰς τὴν κώμην εἰσέλθῃς μηδέ τινι εἴπῃς a
Μηδὲ εἰς τὴν κώμην εἰσέλθῃς ἀλλὰ ὔπαγε εἰς τὸν οἶκόν σου καὶ ἐὰν εἰς τὴν κώμην εἰσέλθῃς μηδὲ εἴπῃς τινὶ (or μηδενὶ εἴπῃς) [μηδὲ] ἐν τῇ κώμῃ arm; also apparently (omitting ἀλλὰ...σου) syr.hl.mg
Μηδενὶ εἴπῃς εἰς τὴν κώμην (or ἐν τῇ κώμῃ) (c) k
(δ) Μηδὲ εἰς τὴν κώμην εἰσέλθῃς μηδὲ εἴπῃς τινὶ ἐν τῇ κώμῃ ACNXΔEFGHKMSUVΓΠ cu.omn.exc.8 syr.vg-hl aeth go
Here a is simple and vigorous, and it is unique in the N. T.: the peculiar initial Μηδέ has the terse force of many sayings as given by St Mark, but the softening into Μή by א* shews that it might trouble scribes. In β we have α deprived of its novelty by the μηδενὶ εἴπῃς of Matt. ix 6 and its parallel, and of its abruptness by the previous insertion of Ὕπαγε εἰς τὸν οἶκόν σου from Matt. viii 4 and its parallels. Then follow several different but not all independent conflations of α and β. By the insertion of α, a little modified, in the midst of β the Greek form of β2 arises; and this, with the superfluous last words removed, is the prevalent Latin reading. In one MS, α, a fresh conflation supervenes, the middle clause of the Latin β2 being replaced by α, almost unaltered. Arm. (and apparently with one omission the margin of syr.hl) prefixes α to β2. The reading of (c) k is as short as α, and may be derived directly from it; but is more probably β delivered from its extraneous first clause by the influence of α. Lastly δ combines α with β by substituting it for the first clause of β; a less clumsy means of avoiding the contradiction latent in the probability that the 'house' would be in the 'village' than the introduction of ἐάν in β2. This neat combination retains Μηδέ without its abruptness by making it a conjunction, but involves a new contradiction unless τινὶ ἐν be taken as τινὶ τῶν ἐν by a laxity ill suited to the context. The documents attesting δ, it is to be observed, include the early uncials CN as well as A, and also Δ and the Syriac Vulgate.
141. Mark ix 38 (following Διδάσκαλε, εἴδαμεν τινα ἐν τῷ ὀνόματί σου ἐκβάλλοντα δαιμόνια,)
(α) καὶ ἐκωλύομεν αὐτόν, ὅτι οὐκ ἠκολούθει ἡμῖν א ΒΔ (?νν)
καὶ ἐκωλύομεν αὐτόν, ὅτι οὐκ
ἀκολουθεῖ μεθ' ἡμῶν L
καὶ ἐκωλύσαμεν αὐτόν, ὅτι οὐκ ἀκολουθεῖ ἡμῖν C cu3 ƒ
(syr.vg-hr me aeth)
(β) ὃς οὐκ ἀκολουθεῖ μεθ' ἡμῶν καὶ ἐκωλύομεν αὐτόν D
ὃς οὐκ ἀκολουθεῖ μεθ' ἡμῶν καὶ ἐκωλίσαμεν αὐτόν a k
ὃς οὐκ ἀκολουθεῖ μεθ' ἡμῖν καὶ ἐκωλύομεν αὐτόν 1-209
ὃς οὐκ ἀκολουθεῖ ἡμῶν καὶ ἐκωλύσαμεν αὐτόν
Χ 13-69-346 28 al4 b c ff i vg syr.hl.mg arm
(δ) ὃς οὐκ ἀκολουθεῖ ἡμῖν, καὶ ἐκωλύσαμεν αὐτόν, ὅτι οὐκ ἀκολουθεῖ ἡμῖν ANEFGHKMSUVΓΠ cu.omn.exc.20 syr.hl.txt go
(81 has ἠκολούθει and al1 μεθ' ἡμῶν in the first clause and al2 μεθ' ἡμῶν in the third: 33 is defective.)
Part of the confusion of readings is due to obvious causes, which throw little light on genealogy. From Luke ix 49 come ἀκολουθεῖ and μεθ' ἡμῶν; while in both Gospels a general proneness to alter imperfects and the influence of the preceding aorist have together produced ἐκωλύσαμεν. But in β, besides assimilation to St Luke, there is a bold transposition of the last clause bringing it into proximity to its subject, with a necessary change of ὅτι to ὅς (cf. Matt. ν 45 in similar documents); while in two modifications of β the aorist ἐκωλύσαμεν reappears, and one of them, β2, the most widely spread, has also ἡμῖν in conformity with α. The transposed clause is preserved in both places by δ with exact similarity of ending. Here again δ is supported by Ν as well as A, but not by any early version.
142. Mark ix 49
(α) πᾶς γὰρ πυρὶ ἁλισθήσεται (א)BLΔ 1-118-209 61 81 435 al9 me.codd the arm.codd
(β) πᾶσα γὰρ θυσία ἁλὶ ἁλισθήσεται D cu2 (a) b c ff i (k) tol holm gig (a c tol holm gig omit ἁλί: α omits γάρ: k has words apparently implying the Greek original πᾶσα δὲ (or γὰρ) οὐσία ἀναλωθήσεται, Ο being read for Θ, and ⲁⲛⲁⲗⲱ for ⲁⲗⲓⲁⲗⲓⲥ.)
(δ) πᾶς γὰρ πυρὶ ἁλισθήσεται, καὶ πᾶσα θυσία ἁλὶ ἁλισθήσεται ACNXEFGHKMSUVΓΠ cu.omn.exc.15 f q vg syr.vg-hl me.codd aeth arm.codd go Vict(cu10 vg.codd.opt omit ἁλί; X adds it after πυρί.)
A reminiscence of Lev. vii 13 (και πᾶν δῶρον θυσίας ὑμῶν ἁλὶ ἁλισθήσεται) has created β out of α, ⲡⲩⲣⲓⲁⲗⲓⲥⲑ being read as ⲑⲩⲥⲓⲁⲁⲗⲓⲁⲗⲓⲥⲑ with a natural reduplication, lost again in some Latin copies. The change would be aided by the words that follow here, καλὸν τὸ ἅλας κ.τ.λ. In δ the two incongruous alternatives are simply added together, γάρ being replaced by καί. Besides ACNX, δ has at least the Vulgate Syriac and the Italian and Vulgate Latin, as well as later versions.
143. Luke ix 10 (after καὶ παραλαβὼν αυτοὺς ὑπεχώρησεν κατ' ἰδίαν)
| (α) | εἰς πόλιν καλουμένην Βηθσαιδά (אca) BLXΞ 33 me the | |
| εἰς. κώμην καλουμένην. Βηθσαιδά. D | ||
| (β) | εἰς τόπον ἔρημον | א* et cb [? 13-346-] (69) 157 (syr.vt) |
| (cf. Tert) (εἰς ἔ.τ. 13-69-346 | ||
| (syr.vt) | ||
| εἰς τόπον ἔρημον | Βηθσαιδά c ff q vg syr.vg | |
| εἰς τόπον ἔρημον καλούμενον Βηθσαιδά a e f | ||
| (δ) | εἰς τόπον ἔρημον πόλεως καλούμένης Βηθσαιδά (A) C | |
| EGHKMSUVΓΔΛΠ cu.omn.exc. | ||
| 3(5) syr.hl aeth arm go | ||
| (A cu4 place ἔρημον before τόπον, 1-131-209 omit it) | ||
The change from α to β would be suggested by the occurrence of ἔρημος τόπος in the two parallels (Matt. xiv 13; Mark vi 31), by the words ὅτι ὧδε ἐν ἐρήμῳ τόπῳ ἐσμέν two verses later, and by the difficulty of associating the incident with a 'city'. Two forms of β, in taking up the name from α, still avoid this difficulty by refusing πόλιν. In δ the difficulty is ingeniously overridden by keeping both α and β, but making β dependent on α. For δ we find, with AC, the four latest but no early version. In this variation א* goes with β, and D virtually with α.
144. Luke xi 54 (after ἤρξαντο οἱ γραμματεῖς καὶ οἱ Φαρισαῖοι δεινῶς ἐνέχειν καὶ ἀποστοματίζειν αὐτὸν περὶ πλειόνων,)
(α) ἐνεδρεύοντες αὐτὸν θηρεῦσαί τι ἐκ τοῦ στόματος αὐτοῦ אBL me aeth Cyr.syr (om. αὐτόν א me Cyr.syr)
(β) ζητοῦντες ἀφορμήν τινα λαβεῖν αὐτοῦ ἵνα εὕρωσιν κατηγορῆσαι αὐτοῦ D syr.vt
ζητοῦντες ἀφορμήν τινα λαβεῖν αὐτοῦ ἵνα κατηγορήσωσιν αὐτοῦ lat.vt (om. αὐτοῦ I0 c e rhe)
(δ) ἐνεδρεύοντες αὐτόν, ζητοῦντες θηρεῦσαί τι ἐκ τοῦ στόματος αὐτοῦ, ἵνα κατηγορήσωσιν αὐτοῦ ACXEGHKMUVΓΔΛΠ cu.omn.exc.5 lat.vg syr.vg-hl (om. αὐτόν X 130 lat.vg: καὶ ζητοῦντες cu.mu lat.vg syr.hl arm: om. ἐνεδρεύοντες αὐτόν arm: om. ζητοῦντες 1-118-131-209 239)
interrogantes (? ἐπερωτῶντες) αὐτόν, ζητοῦντες θηρεῦσαί τι ἐκ τοῦ στόματος αὐτοῦ, ἵνα ἀφορμὴν εὕρωσιν κατηγορῆσαι αὐτοῦ f
The figurative language of α is replaced in β by a simply descriptive paraphrase, just as in the preceding sentence the chief documents that attest β change δεινῶς ἐνέχειν to δεινῶς ἔχειν and ἀποστοματίζειν αὐτόν to συνβάλλειν αὐτῷ: and in the second or Latin form of β εὕρωσιν κατηγορήσαι becomes κατηγορήσωσιν in conformity with Matt. xii 10; Mark iii 2. In δ both phrases are kept, the descriptive being used to explain the figurative: the now superfluous middle part of β however is dropped, and ζητοῦντες is transposed to ease the infinitive θηρεῦσαί. Again the documents of δ include ACX, both Vulgates, and a later version. Besides the readings of some good cursives and of the Armenian, in which the influence of α and of β respectively leads to some curtailment of δ, f presents an interesting secondary conflation, the last phrase of which is derived with a neat transposition from the earliest form of β, whereas the β used in δ is the second form, no longer separately extant in Greek.
145. Luke xii 18 (after καθελῶ μου τὰς ἀποθήκας καὶ μείζονας οἰκοδομήσω, καὶ σηνάξω ἐκεῖ πάντα)
| (α) τὸν σῖτον καὶ τὰ ἀγαθά μου (אac) BTL(X) 1-118-131-(209) (13-69-124) 157 (al) (syr.hr me the aeth) arm (the bracketed documents add μου to σῖτον) | ||
| (β) τὰ γενήματά μου א*D 435 al2(3) b ff i q rhe | ||
| (? Iren.lat) Amb | syr.vt | |
| τοὺς καρπούς μου lt 39 a c d e m } | ||
| (δ) τὰ γενήματά μου καὶ τὰ ἀγαθά μου AQEFGHKMSUVΓΔΛΠ cu.omn.exc.12 f vg syr.vg-hl Bas Cyr | ||
| τὸν σῖτον μου καὶ τὰ γενήματά μου 346 | ||
For the rather peculiar combination of τὸν σῖτον and τὰ ἀγαθά the single general term τὰ γενήματα, common in the LXX and Apocrypha, is substituted by β, the precise combination συνάγειν τὰ γενήματα being indeed found in Ex. xxiii 10; Lev. xxv 20; Jer. viii 13: some documents have the similar τοὺς καρπούς μου from v. 17. In δ the full double form of α is retained, but the plural τὰ γενήματα replaces τὸν σῖτον in accordance with the plural τὰ ἀγαθα. Another form of conflation of a and β appears in 346. Besides AQ and Cyril, δ has, as in Mark ix 49, the Vulgate Syriac and the Italian and Vulgate Latin in addition to the Harklean Syriac versions: both א* and D support β.
146. Luke xxiv 53 (after καὶ ἦσαν διαπαντὸς ἐν τῷ ἱερῷ)
(α) εὐλoγoῦvτες τὸν θεόν אBC*L me syr.hr
(β) αἰνoῦvτες τὸν θεόν D a b e ff vg.codd Aug
(δ) αἰνoῦvτες καὶ εὐλoγoῦvτες τὸν θεόν AC2XFHKMSUVΓΔΛΠ cu.omn c f q vg syr.vg-hl arm
εὐλoγoῦvτες καὶ αἰνoῦvτες τὸν θεόν aeth
This simple instance needs no explanation. The distribution of documents is fairly typical, δ having AC2X with the two Vulgates, the Italian Latin (and another MS containing a similar element), and two later versions; while the Æthiopic has an independent conflation in inverse order.
147. It is worth while to note at once the distribution of the chief MSS and versions with reference to the three classes of readings contained in these eight ternary variations. Only the first hand is taken into account, cursives differing from the main body are not noticed, and slightly aberrant readings are classed with those from which they deviate least. Several MSS and versions are too fragmentary to give more than faint indications of the origin of their texts within these narrow limits, and indeed for the rest of them the results can be only provisional.
| α | β | δ | Total | α | β | δ | Total | ||
| א | 6 | 2 | 0 | 8 | Lat.vt | 0 | 8 | 0 | 8 |
| A | 0 | 0 | 8 | 8 | it | 1 | 2 | 5 | 8 |
| B | 8 | 0 | 0 | 8 | vg | 1 | 3 | 4 | 8 |
| C | 2 | 0 | 4 | 6 | Syr.vt | 0 | 3 | 0 | 3 |
| D | 1 | 7 | 0 | 8 | vg | 2 | 1 | 5 | 8 |
| L | 8 | 0 | 0 | 8 | hl | 0 | 0 | 8 | 8 |
| N | 0 | 0 | 2 | 2 | hr | 3 | 0 | 0 | 3 |
| Q | 0 | 0 | 1 | 1 | Memph | 8 | 0 | (1 codd) | 8 |
| T | 1 | 0 | 0 | 1 | Theb | 3 | 0 | 0 | 3 |
| X | 2 | 1 | 4 | 7 | Aeth | 3 | 0 | 5 | 8 |
| Δ(Mc) | 3 | 0 | 1 | 4 | Arm | 3 (or 2) | 2 | 3 (or 4) | 8 |
| (Lc) | 0 | 0 | 4 | 4 | Goth | 0 | 0 | 4 | 4 |
| Ξ | 1 | 0 | 0 | 1 | |||||
| Late uncials and Cursives | 0 | 0 | 8 | 8 | |||||
148. Comparison of these eight variations strongly confirms the conclusion to which the independent evidence respecting each has provisionally led, that the longer readings marked δ are conflate each from two earlier readings. The fundamental grouping of documents also remains the same throughout, notwithstanding the partial fluctuation. The conflate readings marked δ are found in AC(N) of the earlier and in all later uncials except L, not invariably however in C, X, or Δ; as also in the great mass of cursives, and in the Gothic and Harklean Syriac, two versions known to be late. On the other hand no δ or conflate readings are found in אBDL lat.vt syr.vt me (the), these four versions being also the most ancient. The most constant witnesses for the readings marked β are D and most or all of the Old Latin MSS, though they do not always support the same modification of β: and in the three places in which it is extant the Old Syriac is with them. The most typical group attesting the readings marked α, which in these passages we have found reason to believe to be the original readings, consists of אBL and the Egyptian versions, with the Jerusalem Syriac in its three places; though א twice passes over to the ranks of β, even in Luke ix 10, where D is virtually with α. The five remaining comparatively late versions or forms of versions contain either readings of all three classes in different proportions, or (Æthiopic) both δ readings and α readings: and CX have a similar variable character.
149. Speaking roughly then we may assign the attestation of Greek MSS thus: to α a small handful of uncials, including the two oldest, and a few varying cursives, sometimes wanting; to β D and sometimes a few varying cursives, with the rare accession of א or another uncial; to δ nearly all the later uncials, with two or three of the older, especially A, and nearly all the cursives. The like rough distribution of the three great families of versions which date from early times will be as follows: to α the Egyptian, and to β the Old Latin and Old Syriac; while the later versions, dating from the fourth and following centuries (one perhaps a little earlier), with one limited exception include δ readings, and two here exhibit δ readings alone.
150. To the best of our belief the relations thus provisionally traced are never inverted. We do not know of any places where the α group of documents supports readings apparently conflate from the readings of the β and δ groups respectively, or where the β group of documents supports readings apparently conflate from the readings of the α and δ groups respectively. Hence it is certain not only that the δ readings were always posterior in date to the α and the β readings in variations illustrating the relation between these three groups by means of conflation, but also that the scribes or editors who originated these δ readings made use in one way or another of one or more documents containing these α readings, and one or more documents containing these β readings; that is, they either wrote with documents of both classes before them, or wrote from documents of one class which had readings from the other class written in the margin, or wrote from documents of one class while carrying in their own minds reminiscences from documents of the other class of which they had had knowledge at some previous time.
151. Now it is morally impossible that their use of documents of either or both classes should have been confined to those places in which conflation enables us to detect it in actual operation. The facts observed thus far do not forbid the hypothesis that the originators of the δ readings made use likewise of documents belonging to some additional class, conceivably purer than the documents which furnished them with α and with β readings respectively, and that these additional documents may have been followed by them in a greater or less part of the rest of their text. But the proved actual use of documents of the α and β classes in the conflate readings renders their use elsewhere a vera causa in the Newtonian sense. With every allowance for the provisional possibility of some use of other hypothetical documents, it may be safely taken for granted that those documents which we know to have been either literally or virtually in the hands of the δ scribes were freely employed by them in other parts of their text.
C. 152—162. Posteriority of 'Syrian' to 'Western' and other (neutral and 'Alexandrian') readings shown (2) by Ante-Nicene Patristic evidence
152. The next step accordingly is to discover whether traces of such employment can be found. The variations in the Gospels afford innumerable opportunities for recognising singly the three principal groups of documents, detached from the rest. Oppositions of each of the three groups in turn to all or nearly all the other extant documents abound everywhere, presenting a succession of Distinctive readings of each group, that is, readings having no other attestation: ternary variations in which each of the three groups approximately attests a different variant occur also, but much more rarely. The large field of documentary evidence over which we are now able to range enlarges at the same time our knowledge of the groups themselves. Other Greek MSS and other MSS of versions become available: but above all we obtain some valuable geographical and historical data from the patristic quotations which in many cases give clear additional attestation to the several groups.
153. It will be convenient from this point to designate two of the primary groups of documents no longer by Greek letters but by names. We shall call the β group 'Western', an appellation which has for more than a century been applied to its leading members. It was given at a time when the patristic evidence was very imperfectly known, and its bearing ill understood; and was suggested by the fact that the prominent representatives of the group were Græco-Latin MSS, certainly written in the West, and the Old Latin version, which throughout its range from Carthage to Britain is obviously Western. The fitness is more open to question since it has become evident that readings of this class were current in ancient times in the East as well as the West, and probably to a great extent originated there. On the whole we are disposed to suspect that the 'Western' text took its rise in North-western Syria or Asia Minor, and that it was soon carried to Rome, and thence spread in different directions to North Africa and most of the countries of Europe. From North-western Syria it would easily pass through Palestine and Egypt to Ethiopia. But this is at present hardly more than a speculation; nor do any critical results depend upon it. Whatever may have been the original home of the 'Western' text, a change of designation would now cause more confusion than it would remove, and it remains true that the only continuous and approximately pure monuments of the 'Western' texts now surviving have every right to the name. The δ group we propose to call 'Syrian', for reasons which have partly been noticed already, and which will appear more clearly further on. To these must here be added another group, which would be fitly marked γ, for, as we shall see, its originals must have preceded those of the Syrian group. The local relations of those of its habitual representatives which can be geographically fixed prescribe for it the name 'Alexandrian'.
154. We have hitherto spoken of the primary groups and the ancient texts attested by them with reference to the Gospels alone, where the evidence is at once most copious and most confused. For a full knowledge of their characteristics however it is necessary to pursue them through other books of the New Testament. St Paul's Epistles stand next to the Gospels in the instructiveness of their variations, and fortunately tolerably unmixed Western texts of them are preserved in two independent Greek uncials and in a large body of quotations from Latin Fathers. The Western attestation of the Acts is much less full, and suffers grievously in parts by the loss of leaves in the Codex Bezae (D); but still it can be fairly made out; while the Alexandrian text stands out in much prominence, far more so than in the Pauline Epistles. In the Catholic Epistles the Western text is much obscured by the want of the requisite documents, either Greek or Latin, and probably also by the limited distribution of some of the books in early times; so that it can rarely be relied on for the interpretation of evidence: on the other hand the Alexandrian text is as conspicuous as in the Acts. In the Apocalypse the difficulty of recognising the ancient texts is still greater, owing to the great relative paucity of documents, and especially the absence or loss of this book from the Vatican MS (B) which is available for nearly all the rest of the New Testament; and thus the power of using a directly genealogical method is much limited.
155. The variations here mentioned between different parts of the New Testament are, it will be noticed, of two kinds, being due partly to the varying amount and distribution of documentary evidence which happens to be extant at the present day, partly to the facts of ancient textual history disclosed by the evidence. It is important to observe that, wherever the evidence is copious and varied enough to allow the historical facts to be ascertained, the prevalent characteristics of the ancient texts, as regards both their readings and their documentary attestation, are identical or at least analogous throughout, the diversities which exist being almost wholly confined to proportion.
156. Patristic evidence, which we have now to examine for indications of the ancient texts, needs at all times to be handled with much circumspection, for it includes data of every degree of trustworthiness. The uncertainty which affects many apparent patristic attestations, that is, the difficulty of knowing how far they can safely be taken as conveying to us the readings of the MSS used by the Fathers, arises from two causes. First, what a Father actually wrote is very liable to be falsified by the proneness of both scribes and modern editors to alter the text before them into conformity with the written or printed text most familiar to themselves; and since a text substantially identical with that of δ was unquestionably the only text likely to be known to transcribers generally throughout the centuries to which existing Greek patristic MSS with the rarest exceptions belong, as also to the authors of nearly all the current editions of the Greek Fathers till quite lately, it is no wonder that those Greek corruptions which can on sufficient evidence be determined as such are almost invariably found to consist in the introduction, not in the removal, of δ readings; and nearly the same may be said as to Vulgate readings in the texts of Latin Fathers. This kind of corruption is hardly ever systematic or thorough, but it is common enough; it is usually abundant in those passages of Christian writers which owe their preservation to Catenæ, especially where, as frequently happens, they have been evidently condensed by the compiler. It may often be detected by recourse to better MSS, by comparison with other quotations of the same passage by the same writer, or, best of all, by close examination of the context: but in many cases a greater or less degree of doubt remains as to the words actually written by a Father.
157. The second possible cause of error in dealing with patristic evidence is laxity of quotation by the writers themselves, more especially when they quote indirectly or allusively. The laxity may arise either from conscious or semi-conscious modification for the sake of grammar or convenience, or from error of memory, a frequent cause of error being confusion with other similar passages. Here too there is a considerable residuum of more or less doubtful cases, though comparison with other quotations of the same passage and above all experience will remove many prima facie ambiguities. Allusive references are sometimes as decisive as full and direct quotations, and they have the advantage of being much less liable to corruption by scribes and editors. But whatever imperfections of verification of patristic evidence may cling to particular passages, they do not to any appreciable extent affect the generalisations as to the patristic attestation of particular groups of documents obtained by taking a large number of passages together. The broad facts come out clearly: where there is doubt, it for the most part relates to the presence or absence of rare exceptions.
158. When we examine the remains of the Ante-Nicene Christian literature with a view to collect evidence respecting the ancient texts which the groupings of the extant documents shew to have existed, we are for some time after the apostolic age hampered both by the paucity of the writings preserved and by the scantiness and comparative vagueness of the textual materials contained in them. The only period for which we have anything like a sufficiency of representative knowledge consists roughly of three quarters of a century from about 175 to 250: but the remains of four eminent Greek Fathers, which range through this period, cast a strong light on textual history backward and forward. They are Irenæus, of Asia Minor, Rome, and Lyons; his disciple Hippolytus, of Rome; Clement, of Athens and Alexandria; and his disciple Origen, of Alexandria and Palestine. To the same period belong the Latin representatives of North Africa, Tertullian and Cyprian, as also Cyprian's Roman contemporary Novatian. Towards the close of the third century we have somewhat considerable remains of Methodius, of Lycia and Tyre, an enemy of the Origenian school; and in the first third of the fourth century several writings of Eusebius of Cæsarea in Palestine, the most learned of its disciples. For the second half of the third century we have other fragments, but they are few in number.
159. The most striking phenomenon of the evidence belonging to the time before 250 is the number of places in which the quotations exhibit at least two series of readings, Western and what may be called Non-Western. The first clear evidence of any kind that we possess, that obtained from recorded readings of Marcion (Pontus and Rome) and from the writings of Justin Martyr (Samaria and Rome), is distinguished by readings undoubtedly Western, and thus shews that texts of this character were in existence before the middle of the second century. The same character of text is found in Irenæus and Hippolytus, and again in Methodius and predominantly in Eusebius. Thus the text used by all those Ante-Nicene Greek writers, not being connected with Alexandria, who have left considerable remains is substantially Western. Even in Clement of Alexandria and in Origen, especially in some of his writings, Western quotations hold a prominent place.
160. On the other hand the many Non-Western readings supplied by Clement of Alexandria prove that great divergencies were in existence at latest by the end of the second century. Any possible doubts on this head that could be suggested by his free mode of citation would be entirely swept away by what we find in Origen's extant writings. Many of the verses which he quotes in different places shew discrepancies of text that cannot be accounted for either by looseness of citation or by corruption of the MSS of his writings; and in most instances the discrepant readings are those of the primary extant groups, including the 'Alexandrian' group, of which we shall presently have to speak in detail. It is even possible, as Griesbach shewed long ago, to trace to a certain extent his use of different MSS when writing different treatises; and moreover he now and then refers in express words to variations between MSS, as indeed Irenæus had at least once done. Many of his readings in variations in which Western documents stand opposed to all other documents are distinctly Western, many more are distinctly Non-Western. On the other hand his quotations to the best of our belief exhibit no clear and tangible traces of the Syrian text.
161. That these characteristics, positive and negative, of the quotations found in Origen's writings are due to accident is in the highest degree improbable. A long and laborious life devoted chiefly to original biblical studies, combined with a special interest in texts, and the twofold opportunities supplied by the widely different circumstances of Alexandria and Palestine, to say nothing of varied intercourse with other lands, could hardly fail to acquaint him with all leading types of Greek text current in the Churches, and especially in the Eastern Churches: and as a matter of fact we find all other known great types of text represented in his writings except the one; that one moreover, had it then existed, being more likely to have come to the notice of a dweller in Palestine than any other.
162. Nor is the testimony that of a single Father, however well placed and well fitted for reflecting the lost testimony of all contemporary Churches on such a matter. The whole body of patristic evidence down to his death, or later, tells the same tale. Before the middle of the third century, at the very earliest, we have no historical signs of the existence of readings, conflate or other, that are marked as distinctively Syrian by the want of attestation from groups of documents which have preserved the other ancient forms of text. This is a fact of great significance, ascertained as it is exclusively by external evidence, and therefore supplying an absolutely independent verification and extension of the result already obtained by comparison of the internal character of readings as classified by conflation.
D. 163—168. Posteriority of Syrian to Western, Alexandrian, and other (neutral) readings shewn (3) by Internal Evidence of Syrian readings
163. The Syrian conflate readings have shown the Syrian text to be posterior to at least two ancient forms of text still extant, one of them being 'Western', and also to have been, at least in part, constructed out of both. Patristic evidence has shewn that these two ancient texts, and also a third, must have already existed early in the third century, and suggested very strong grounds for believing that in the middle of the century the Syrian text had not yet been formed. Another step is gained by a close examination of all readings distinctively Syrian in the sense explained above, comparing them on grounds of Internal Evidence, Transcriptional and Intrinsic, with the other readings of the same passages. The result is entirely unfavourable to the hypothesis which was mentioned as not excluded by the phenomena of the conflate readings, namely that in other cases, where the Syrian text differs from all other extant ancient texts, its authors may have copied some other equally ancient and perhaps purer text now otherwise lost. In themselves Syrian readings hardly ever offend at first. With rare exceptions they run smoothly and easily in form, and yield at once to even a careless reader a passable sense, free from surprises and seemingly transparent. But when distinctively Syrian readings are minutely compared one after the other with the rival variants, their claim to be regarded as the original readings is found gradually to diminish, and at last to disappear. Often either the transcriptional or the intrinsic evidence is neutral or divided, and occasionally the two kinds of evidence appear to be in conflict. But there are, we believe, no instances where both are clearly in favour of the Syrian reading, and innumerable where both are clearly adverse to it.
164. The testimony of the simpler variations in which the other ancient texts are united against the Syrian reading is remarkably confirmed by that of many of those variations in which they are divided among themselves. Here one of the readings has to approve itself on transcriptional grounds by its fitness to give rise not to one but to two or more other readings, that is either to each independently or to one which will in like manner account naturally for the third (or the rest); and the failure of the Syrian reading to fulfil this condition is usually manifest. The clearest cases are those in which the immediate parent of the Syrian reading is seen to be itself in turn derived from another, so that the two steps of the process illustrate each other: not a few distinctively Syrian readings are in reality Western or Alexandrian readings, somewhat trimmed and modified.
165. To state in few words the results of examination of the whole body of Syrian readings, distinctive and non-distinctive, the authors of the Syrian text had before them documents representing at least three earlier forms of text. Western, Alexandrian, and a third. Where they found variation, they followed different procedures in different places. Sometimes they transcribed unchanged the reading of one of the earlier texts, now of this, now of that. Sometimes they in like manner adopted exclusively one of the readings, but modified its form. Sometimes they combined the readings of more than one text in various ways, pruning or modifying them if necessary. Lastly, they introduced many changes of their own where, so far as appears, there was no previous variation. When the circumstances are fully considered, all these processes must be recognised as natural.
166. Thus not only do the relations disclosed by the conflate Syrian readings reappear conspicuously in the much larger field of distinctively Syrian readings generally, but no fresh phenomenon claims to be taken into account, unless it be the existence of the Alexandrian text, which has its own extant attestation apart from the Syrian text. Taking these facts in conjunction with the absence of distinctively Syrian readings from the patristic evidence of the Origenian and Ante-Origenian periods, while nevertheless distinctive readings of all the texts known to have been used in the production of distinctively Syrian readings abound in the Origenian period, as also, with the possible exception of distinctively Alexandrian readings, in the Ante-Origenian period, we are led to conclude that the hypothesis provisionally allowed must now be definitively rejected, and to regard the Syrian text as not only partly but wholly derived from the other known ancient texts. It follows that all distinctively Syrian readings may be set aside at once as certainly originating after the middle of the third century, and therefore, as far as transmission is concerned, corruptions of the apostolic text.
167. The same facts lead to another conclusion of equal or even greater importance respecting non-distinctive Syrian readings, which hold a conspicuous place by their number and often by their intrinsic interest. Since the Syrian text is only a modified eclectic combination of earlier texts independently attested, existing documents descended from it can attest nothing but itself: the only authority which they can give to readings having other documentary attestation, that is to readings Syrian but not distinctively Syrian, is the authority of the Syrian text itself, which resolves itself into that of a lost ancient MS of one or possibly more of those older texts from which the Syrian text was in any given variation derived. Accordingly a reading supported both by the documents belonging to the Syrian group and by those belonging to e.g. the Western group has no appreciably greater presumption in its favour than if it were supported by the Western group alone: the only accession is that of a lost Western MS not later in date than the time when the Syrian text was formed; and in almost all cases this fact would add nothing to our knowledge of the ancestry of the reading as furnished by the Non-Syrian documents attesting it.
168. If our documents were free from all mixture except that contained in the Syrian text, that is, if no document of later origin itself combined elements from different texts, the application of this principle would be always clear and certain. Since however most of the more important documents are as a matter of fact affected by later mixture, the origin of any given reading in them can only be determined by grouping; and since grouping is sometimes obscure, a greater or less degree of doubt about the antecedents of a non-distinctive Syrian reading may in such cases remain. Thus it may be clear that a reading was first Western and then Syrian, while yet there may be a doubt whether certain of the attesting documents derived it from a Syrian or from an earlier source. If from the former, the reading must be held to be in effect distinctively Western: if from the latter, the possibility or probability of its having existed not only in the Western but in a Non-Western Pre-Syrian text has to be taken into account. These occasional ambiguities of evidence do not however affect the force or the ordinary applicability of the principle itself: and in practice the doubt is in most cases removed by Internal Evidence of Groups.
SECTION II. CHARACTERISTICS OF THE CHIEF ANCIENT TEXTS
169—187
A. 170—176. Western characteristics
170. On all accounts the Western text claims our attention first. The earliest readings which can be fixed chronologically belong to it. As far as we can judge from extant evidence, it was the most widely spread text of Ante-Nicene times; and sooner or later every version directly or indirectly felt its influence. But any prepossessions in its favour that might be created by this imposing early ascendancy are for the most part soon dissipated by continuous study of its internal character. The eccentric Whiston's translation of the Gospels and Acts from the Codex Bezae, and of the Pauline Epistles from the Codex Claromontanus, and Bornemann's edition of the Acts, in which the Codex Bezae was taken as the standard authority, are probably the only attempts which have ever been made in modern times to set up an exclusively or even predominantly Western Greek text as the purest reproduction of what the apostles wrote. This all but universal rejection is doubtless partly owing to the persistent influence of a whimsical theory of the last century, which, ignoring all Non-Latin Western documentary evidence except the handful of extant bilingual uncials, maintained that the Western Greek text owed its peculiarities to translation from the Latin; partly to an imperfect apprehension of the antiquity and extension of the Western text as revealed by patristic quotations and by versions. Yet, even with the aid of a true perception of the facts of Ante-Nicene textual history, it would have been strange if this text as a whole had found much favour. A few scattered Western readings have long been approved by good textual critics on transcriptional and to a great extent insufficient grounds; and in Tischendorf's last edition their number has been augmented, owing to the misinterpreted accession of the Sinai MS to the attesting documents. To one small and peculiar class of Western readings, exclusively omissions, we shall ourselves have to call attention as having exceptional claims to adoption. But when the Western readings are confronted with their ancient rivals in order to obtain a broad comparative view of the two texts, few scholars could long hesitate to pronounce the Western not merely to be the less pure text, but also to owe its differences in a great measure to a perilous confusion between transcription and reproduction, and even between the preservation of a record and its supposed improvement; and the distrust thus generated is only increased by further acquaintance.
171. What has been here said is equally true whether we confine ourselves to Western readings having only a Western attestation or include with them those Western readings which, having been adopted into the Syrian text, have a combination of Western and Syrian attestation. When once the historical relations of the texts have been ascertained, it would be arbitrary to refuse the evidence of the latter class in studying the general character of Western readings apart from attestation, for the accident of their appropriation by the Syrian text when the other Western readings were neglected can have no bearing on the antecedent relations of the whole class to the apostolic originals. But as a matter of fact the general conclusions would be the same in either case: throughout both classes of Western readings there is no diversity of salient characteristics.
172. To what extent the earliest MSS of the distinctively Western ancestry already contained distinctive Western readings, cannot now be known. However they may have differed from the apostolic autographs, there was at all events no little subsequent and homogeneously progressive change. It is not uncommon to find one, two, or three of the most independent and most authentically Western documents in agreement with the best representatives of Non-Western Pre-Syrian texts against the bulk of Western authorities under circumstances which render it highly difficult to account for the concurrence by mixture: and in such cases these detached documents must attest a state of the Western text when some of its characteristic corruptions had not yet arisen, and others had. On the other hand it is probable that even the relatively latest Western readings found in distinct provinces of Western documents, for instance in different languages, were already in existence at a very early date of Church history, it may be before the end of the second century.
173. The chief and most constant characteristic of the Western readings is a love of paraphrase. Words, clauses, and even whole sentences were changed, omitted, and inserted with astonishing freedom, wherever it seemed that the meaning could be brought out with greater force and definiteness. They often exhibit a certain rapid vigour and fluency which can hardly be called a rebellion against the calm and reticent strength of the apostolic speech, for it is deeply influenced by it, but which, not less than a tamer spirit of textual correction, is apt to ignore pregnancy and balance of sense, and especially those meanings which are conveyed by exceptional choice or collocation of words. An extreme form of the paraphrastic tendency is shown in the interpolation of phrases extending by some kind of parallelism the language of the true text; as καὶ τῆς νύμφης after εἰς ὑπάντησιν τοῦ νυμφίον in Matt. xxv 1; γεννῶνται καὶ γεννῶσιν between οἱ υἱοὶ τοῦ αἰῶνος τούτου and γαμοῦσιν καὶ γαμίσκονται in Luke xx 34; and ἐκ τῆς σαρκὸς αὐτοῦ καὶ ἐκ τῶν ὀστέων αὐτοῦ after μέλη ἐσμὲν τοῦ σώματος αὐτοῦ in Eph. ν 30. Another equally important characteristic is a disposition to enrich the text at the cost of its purity by alterations or additions taken from traditional and perhaps from apocryphal or other non-biblical sources; as Σὺ εἶ ὁ υἱύς μου ὁ ἀγαπητός, ἐν σοὶ εὐδόκησα (originating of course in Ps. ii 7) given as the words spoken from heaven at the Baptism in Luke iii 22; and a long interpolation (printed in the Appendix) beginning Ὑμεῖς δὲ ζητεῖτε after Matt. xx 28. The two famous interpolations in John v and viii, which belong to this class, will need special notice in another place. Under the present head also should perhaps be placed some of the many curious Western interpolations in the Acts, a certain number of which, having been taken up capriciously by the Syrian text, are still current as part of the Received text: but these again will require separate mention.
174. Besides these two marked characteristics, the Western readings exhibit the ordinary tendencies of scribes whose changes are not limited to wholly or partially mechanical corruptions. We shall accordingly find these tendencies, some of them virtually incipient forms of paraphrase, in other texts of the New Testament: but in the Western text their action has been more powerful than elsewhere. As illustrations may be mentioned the insertion and multiplication of genitive pronouns, but occasionally their suppression where they appeared cumbrous; the insertion of objects, genitive, dative, or accusative, after verbs used absolutely; the insertion of conjunctions in sentences which had none, but occasionally their excision where their force was not perceived and the form of the sentence or context seemed to commend abruptness; free interchange of conjunctions; free interchange of the formulæ introductory to spoken words; free interchange of participle and finite verb with two finite verbs connected by a conjunction; substitution of compound verbs for simple as a rule, but conversely where the compound verb of the true text was difficult or unusual; and substitution of aorists for imperfects as a rule, but with a few examples of the converse, in which either a misunderstanding of the context or an outbreak of untimely vigour has introduced the imperfect. A bolder form of correction is the insertion of a negative particle, as in Matt. xxi 32 (οὐ being favoured, it is true, by the preceding τοῦ), Luke xi 48, and Rom. iv 19; or its omission, as in Rom. ν 14; Gal. ii 5; ν 8.
175. Another impulse of scribes abundantly exemplified in Western readings is the fondness for assimilation. In its most obvious form it is merely local, abolishing diversities of diction where the same subject matter recurs as part of two or more neighbouring clauses or verses, or correcting apparent defects of symmetry. But its most dangerous work is 'harmonistic' corruption, that is, the partial or total obliteration of differences in passages otherwise more or less resembling each other. Sometimes the assimilation is between single sentences that happen to have some matter in common; more usually however between parallel passages of greater length, such especially as have in some sense a common origin. To this head belong not only quotations from the Old Testament, but parts of Ephesians and Colossians, and again of Jude and 2 Peter, and, above all, the parallel records in the first three Gospels, and to a certain extent in all four. It is difficult to exaggerate the injury thus inflicted upon the resources for a right understanding of the Gospel history by the destruction of many of the most characteristic and instructive touches contributed by the several narratives, whether in the form of things otherwise said, or of additional things said, or of things left unsaid. A sense of the havoc wrought by harmonistic corruption in the Old Latin texts, in their origin Western texts, has been already mentioned as one of the primary motives alleged by Jerome for his revision; and though his effort had only a limited success, the Vulgate contrasts favourably with prior Latin texts of the Gospels in this respect. It should be observed that the harmonistic changes in the Western as in all other texts were irregular and unsystematic. Nor is it rare to find Western changes proceeding in an opposite direction; that is, to find paraphrastic or other impulses followed in the text of one Gospel in unconsciousness or disregard of the creation of new differences from the language of a parallel narrative.
176. It must not be supposed that the liberties taken by the authors of the Western readings, though far exceeding what we find appearing for the first time in other texts of the New Testament, are unknown in other literature transmitted under not unlike circumstances. Several books of the Apocrypha of the Old Testament exist in two forms of text, of which one is evidently an amplified and interpolated modification of the other. Analogous phenomena in various manners and degrees occur in the texts of some of the earliest post-apostolic Christian writings, as the Epistle of Barnabas and the Shepherd of Hermas; and even the interpolations of the Ignatian Epistles are to a certain extent of the same kind. In the Christian 'apocryphal' or legendary literature, some of which, in its elements if not in its present shape, is undoubtedly as old as the second century, much of the extraordinary diversity in different MSS can only be explained by a hardly credible laxity of idea and practice in the transmission of texts. Some at least of the writings here mentioned, if not all of them, had a large popular currency: and it is probably to similar conditions of use and multiplication, prevailing during the time of the slow process by which the books of the New Testament at last came to be placed on the same footing as those of the Old, that we must look for a natural explanation of the characteristics of their Western texts. In surveying a long succession of Western readings by the side of others, we seem to be in the presence of a vigorous and popular ecclesiastical life, little scrupulous as to the letter of venerated writings, or as to their permanent function in the future, in comparison with supposed fitness for immediate and obvious edification.
B. 177—180. The neutral text and its preservation
177. We now proceed to other Pre-Syrian texts. If it be true, as we have found reason to believe, first, that during that part of the Ante-Nicene period of which we have any direct knowledge 'Western' texts were at least dominant in most churches of both East and West, and secondly, that, whatever may be the merits of individual Western readings, the Western texts generally are due to a corruption of the apostolic texts, it is natural to ask where comparatively pure texts were preserved. The only extant patristic writings which to any considerable extent support extant Pre-Syrian readings at variance with Western readings are connected with Alexandria, that is, the remains of Clement and Origen, as mentioned above (§ 159), together with the fragments of Dionysius and Peter of Alexandria from the second half of the third century, and in a certain measure the works of Eusebius of Cæsarea, who was deeply versed in the theological literature of Alexandria. In like manner, of the three great versions or families of versions which must date from the earliest centuries, two in their Old or unrevised form must be classed as Western, the Latin clearly and almost entirely, the very imperfectly preserved Syriac more obscurely: but it is only the two versions of Lower and of Upper Egypt, and the latter, which is the further from Alexandria, less than the former, that can be pronounced extensively Non-Western. That a purer text should be preserved at Alexandria than in any other church would not in itself be surprising. There, if anywhere, it was to be anticipated that, owing to the proximity of an exact grammatical school, a more than usual watchfulness over the transcription of the writings of apostles and apostolic men would be suggested and kept alive. But the rapid total extinction of comparatively pure texts in all other places would undeniably be a riddle hard of solution.
178. No such enigmatic history however demands acceptance. The early traces of a text free from Western corruption in churches remote from Alexandria, though relatively few in number, are indubitable and significant. They are the same facts that were mentioned above (§ 172) in speaking of the progressiveness of Western changes, only seen from the other side. When we find that those very Western documents or witnesses which attest some of the most widely spread and therefore ancient Western corruptions attest likewise ancient Non-Western readings in opposition to most Western documents, we know that they must represent a text in process of transition from such a text as we find at Alexandria to a more highly developed Western text, and consequently presuppose a relatively pure Non-Western text. This early evidence is sometimes at once Greek, Latin, and Syriac, sometimes confined to one or two of the languages. It shews that at least in remote antiquity the Non-Western text was by no means confined to Alexandria.
179. As regards the other facts of the Ante-Nicene period, the negative evidence is not of a trustworthy kind. If we deduct from the extant Ante-Nicene Greek patristic quotations those of the Alexandrian Fathers, the remainder, though sufficient to shew the wide range of the Western text, is by no means sufficient by itself to disprove the existence of other texts. What we have urged in a former page (§ 162) respecting the absence of patristic evidence for the Syrian text before the middle of the third century at earliest was founded on the whole evidence, including that of Clement and Origen, Origen's evidence being in amount more than equal to all the rest put together, and in probable variety of sources and actual variety of texts exceptionally comprehensive: and moreover this negative argument was confirmed by the internal phenomena of the Syrian text itself. But further, much positive evidence for the persistence of Non-Western texts in various regions throughout the Ante-Nicene period is contained in the varied texts of Fathers and versions of the fourth and fifth centuries. It is true that the only considerable text of a Father or version of this later period which closely approximates to a Non-Western Pre-Syrian text, that of the younger Cyril, has again Alexandria for its locality. It is true also that it is not absolutely impossible for the large Non-Western Pre-Syrian elements which enter into many mixed texts of the later period to have all radiated from Alexandria in the third century. Nevertheless the preservation of early Non-Western texts in varying degrees of purity in different regions would account for the facts much more naturally than such a hypothesis. On the one hand there is no reason to think the prominence of Alexandria in the extant evidence accidental: nowhere probably was the perpetuation of an incorrupt text so much an object of conscious desire and care, and the local influence of Origen's school for some generations after his death was likely to establish a tradition of exceptional jealousy for the very words of Scripture. On the other hand our documentary evidence, taken as a whole, equally suggests, what historical probability would have led us to anticipate, that in various and perhaps many other places the primitive text in varying degrees of purity survived the early Western inundation which appeared to submerge it.
180. Such being the facts, we have not thought it advisable to designate Non-Western Pre-Syrian readings generally as 'Alexandrian', although this, or something like this, is the sense in which the term 'Alexandrian' is commonly used, when it is not extended to all ancient readings alike that are not found in the later Greek MSS. Not only were these readings not confined to Alexandria, but a local name suggests erroneous associations when applied to a text which owes its comparative isolation to the degeneracy of its neighbours. On the laxity with which existing MSS are themselves often called Alexandrian we shall have occasion to remark hereafter.
C. 181—184. Alexandrian characteristics
181. There is moreover, as we have already intimated, a class of ancient readings to which the name 'Alexandrian' of right belongs. They are brought to light by a considerable number of variations among those documents which have chiefly preserved a Non-Western Pre-Syrian text, and which are shown by the whole distribution of documentary evidence to have nothing to do with variations between Western and Non-Western texts. They enter largely, as we shall see presently, into the texts of various extant uncial MSS, and with the help thus afforded to the recognition of documentary grouping it is usually easy to see which variants in successive variations have the distinctively 'Alexandrian' attestation, and thus to arrive at a comparative view of the general internal characteristics of the two series of readings.
182. The differences of type are by no means so salient here as in the previous comparison of Western with Non-Western texts; but on due consideration the case becomes clear. On grounds of Intrinsic and Transcriptional Probability alike, the readings which we call Alexandrian are certainly as a rule derived from the other Non-Western Pre-Syrian readings, and not vice versa. The only documentary authorities attesting them with any approach to constancy, and capable of being assigned to a definite locality, are quotations by Origen, Cyril of Alexandria, and occasionally other Alexandrian Fathers, and the two principal Egyptian Versions, especially that of Lower Egypt. These facts, taken together, shew that the readings in question belong to a partially degenerate form of the Non-Western Pre-Syrian text, apparently limited in its early range, and apparently originating in Alexandria. It cannot be later in date than the opening years of the third century, and may possibly be much earlier. Some of its readings at one time attracted the attention of critics, owing to certain peculiarities in their secondary attestation: but the greater number have been confused with other Non-Western readings, doubtless owing to the accidental loss of all Greek MSS having an approximately unmixed Alexandrian text. Had D of the Gospels and Acts and D2E3F2G3 of the Pauline Epistles all in like manner perished, it would have been in like manner far harder than now to form a clear conception of the Western text, and consequently of early textual history.
183. The more startling characteristics of Western corruption are almost wholly absent from the Alexandrian readings. There is no incorporation of matter extraneous to the canonical texts of the Bible, and no habitual or extreme licence of paraphrase; though a certain amount of paraphrase and what may be called inventive interpolation finds place in the less read books, that is, the Acts and Catholic Epistles (especially 1 Peter), and probably the Apocalypse. The changes made have usually more to do with language than matter, and are marked by an effort after correctness of phrase. They are evidently the work of careful and leisurely hands, and not seldom display a delicate philological tact which unavoidably lends them at first sight a deceptive appearance of originality. Some of the modes of change described above as belonging to incipient paraphrase occur as distinctly here as in the Western texts, though as a rule much more sparingly; and the various forms of assimilation, especially harmonistic alteration and interpolation in the Gospels, recur likewise, and at times are carried out in a very skilful manner.
184. Alexandrian changes sometimes occur in places where Western changes exist likewise, sometimes where they do not; and again the Syrian text sometimes follows one, sometimes another, of the three antecedent texts in the former case, of the two in the latter. Considerable variety of distribution, irrespective of Non-Syrian mixture, accordingly arises in the documentary attestation. We often find the Alexandrian group opposed to all other documents, often the Alexandrian and Syrian groups combined in opposition to the others, implying an adoption of an Alexandrian reading by the Syrian text. But the most instructive distributions, as exhibiting distinctly the residual Pre-Syrian text which is neither Western nor Alexandrian, are those produced by the simultaneous aberration of the Western and Alexandrian texts, especially when they severally exhibit independent modes of easing an apparent difficulty in the text antecedent to both.
D. 185—187. Syrian characteristics
185. The Syrian text, to which the order of time now brings us back, is the chief monument of a new period of textual history. Whatever petty and local mixture may have previously taken place within limited areas, the great lines of transmission had been to all appearance exclusively divergent. Now however the three great lines were brought together, and made to contribute to the formation of a new text different from all. As we have seen, the reading now of one, now of another was adopted, such adoption being sometimes a mere transcription but often accompanied by a varying amount of modification not rarely resulting in an entirely new reading. Occasionally also the readings of two of the antecedent texts were combined by simple or complex adaptations. The total process to which these operations belonged was essentially different from the preceding processes of change. In itself the mixture of independent texts might easily be, and perhaps usually was, fortuitous or even unconscious. But the complexity of the Syrian text as derived from three distinct sources simultaneously, the elaborate manner in which they are laid under contribution, and the interfusion of adjustments of existing materials with a distinctly innovative process, shown partly in verbal transformation of adopted readings, partly in assimilative or other interpolations of fresh matter, belong to a manner of change differing as widely from change of either the Western or the Alexandrian type as even Western change from ordinary careless transcription. The Syrian text must in fact be the result of a 'recension' in the proper sense of the word, a work of attempted criticism, performed deliberately by editors and not merely by scribes.
186. The guiding motives of their criticism are transparently displayed in its effects. It was probably initiated by the distracting and inconvenient currency of at least three conflicting texts in the same region. The alternate borrowing from all implies that no selection of one was made,—indeed it is difficult to see how under the circumstances it could have been made,—as entitled to supremacy by manifest superiority of pedigree. Each text may perhaps have found a patron in some leading personage or see, and thus have seemed to call for a conciliation of rival claims: but at all events, if a new measure was to be adopted for promoting unity of text, no course was so natural and convenient as the acceptance of the traditional authority of each text already accredited by honour and use, at least in an age when any really critical perception of the problem involved in the revision of a written text would have been an anachronism. It would have been no less an anachronism at each variation to find reasons for the preference to be given to this or that text in specialities of documentary attestation or again in consideration of Transcriptional Probability. The only grounds of selection, affording any true means of advancing towards textual purity, that could find place in the conditions of the time, or that can now be discerned in the resulting text, depend on a rough and superficial kind of Intrinsic Probability. But the governing impulses, just as in the case of nearly all licentious as distinguished from inaccurate transcription, unquestionably arose from a very natural failure to distinguish between the purity of a text and its present acceptability or usefulness.
187. The qualities which the authors of the Syrian text seem to have most desired to impress on it are lucidity and completeness. They were evidently anxious to remove all stumbling-blocks out of the way of the ordinary reader, so far as this could be done without recourse to violent measures. They were apparently equally desirous that he should have the benefit of instructive matter contained in all the existing texts, provided it did not confuse the context or introduce seeming contradictions. New omissions accordingly are rare, and where they occur are usually found to contribute to apparent simplicity. New interpolations on the other hand are abundant, most of them being due to harmonistic or other assimilation, fortunately capricious and incomplete. Both in matter and in diction the Syrian text is conspicuously a full text. It delights in pronouns, conjunctions, and expletives and supplied links of all kinds, as well as in more considerable additions. As distinguished from the bold vigour of the 'Western' scribes, and the refined scholarship of the Alexandrians, the spirit of its own corrections is at once sensible and feeble. Entirely blameless on either literary or religious grounds as regards vulgarised or unworthy diction, yet shewing no marks of either critical or spiritual insight, it presents the New Testament in a form smooth and attractive, but appreciably impoverished in sense and force, more fitted for cursory perusal or recitation than for repeated and diligent study.
SECTION III. SKETCH OF POST-NICENE TEXTUAL HISTORY
188—198
A. 188—190. The two stages of the Syrian text
188. We have thus far found it conducive to clearness to speak of the Syrian text in the singular number. Two stages of it however can be traced, which may have been separated by an interval of some length. At an early period of modern textual criticism it was perceived that the Vulgate Syriac version differed from early versions generally, and from other important early documentary authorities, in the support which it frequently gave to the common late Greek text: and as the version enjoyed a great traditional reputation of venerable antiquity, the coincidence attracted much interest. Eventually, as has been already noticed (§ 118), it was pointed out that the only way of explaining the whole body of facts was to suppose that the Syriac version, like the Latin version, underwent revision long after its origin, and that our ordinary Syriac MSS represented not the primitive but the altered Syriac text: and this explanation has been signally confirmed in our own day by the discovery of part of a copy of the Gospels in which the national version is preserved approximately in its Old or unrevised state. Two facts render it highly probable that the Syriac revision was instituted or sanctioned by high authority, personal or ecclesiastical; the almost total extinction of Old Syriac MSS, contrasted with the great number of extant Vulgate Syriac MSS; and the narrow range of variation found in Vulgate Syriac MSS, so far as they have yet been examined. Historical antecedents render it tolerably certain that the locality of such an authoritative revision, accepted by Syriac Christendom, would be either Edessa or Nisibis, great centres of life and culture to the churches whose language was Syriac, but intimately connected with Antioch, or else Antioch itself, which, though properly Greek, was the acknowledged capital of the whole Syrian population of both tongues. When therefore we find large and peculiar coincidences between the revised Syriac text and the text of the Antiochian Fathers of the latter part of the fourth century, and strong indications that the revision was deliberate and in some way authoritative in both cases, it becomes natural to suppose that the two operations had some historical connexion.
189. Nevertheless the two texts are not identical. In a considerable number of variations the Vulgate Syriac sides with one or other of the Pre-Syrian texts against the Antiochian Fathers and the late Greek text, or else, as we have already found (§§ 134, 143), has a transitional reading, which has often, though not always, some Greek documentary attestation. These lesser irregularities shew that the Greek Syrian revision in its ultimate form, the only form adequately known to us, and the Syriac revision, though closely connected in origin, cannot both be due to a single critical process performed once for all. The facts would, we believe, be explained by the supposition, natural enough in itself, that (1) the growing diversity and confusion of Greek texts led to an authoritative revision at Antioch, which (2) was then taken as a standard for a similar authoritative revision of the Syriac text, and (3) was itself at a later time subjected to a second authoritative revision, carrying out more completely the purposes of the first; but that the Vulgate Syriac text did not undergo any corresponding second revision. The revision apparently embodied in the Harklean Syriac will be noticed further on.
190. The final process was apparently completed by 350 or thereabouts. At what date between 250 and 350 the first process took place, it is impossible to say with confidence; and even for conjecture the materials are scanty. There can be little doubt that during the long respite from persecution enjoyed by the Church in the latter half of the third century multiplication of copies would be promoted by the increase of converts and new security of religious use, and confusion of texts by more frequent intercourse of churches. Such a state of things would at least render textual revision desirable; and a desire for it might easily arise in a place where a critical spirit was alive. The harmony between the characteristics of the Syrian revision and the well known temper of the Antiochian school of critical theology in the fourth century, at least on its weaker side, is obvious; and Lucianus the reputed founder of the school, himself educated at Edessa, lived in the latter part of the third century, and suffered martyrdom in 312. Of known names his has a better claim than any other to be associated with the early Syrian revision; and the conjecture derives some little support from a passage of Jerome, which is not itself discredited by the precariousness of modern theories which have been suggested by it. When he says in his preface to the Gospels "Praetermitto eos codices quos a Luciano et Hesychio nuncupatos paucorum hominum adserit perversa contentio", he must have had in view some definite text or texts of the Gospels or the New Testament generally, appealed to by some definite set or sets of men as deriving authority from names honoured by them. Jerome's antagonism to Antiochian theology would readily explain his language, if some Antiochian Father had quoted in controversy a passage of the New Testament according to the text familiar to him, had been accused of falsifying Scripture, and had then claimed for his text the sanction of Lucianus. Whether however Lucianus took a leading part in the earlier stage of the Syrian revision or not, it may be assigned with more probability either to his generation or to that which immediately followed than to any other; and no critical results are affected by the presence or absence of his name.
B. 191—193. Mixture in the fourth century
191. Two successive external events which mark the opening years of the fourth century, the terrible persecution under Diocletian and his colleagues and the reaction under Constantine, doubtless affected the text not less powerfully than the Canon of the New Testament. The long and serious effort of the imperial government to annihilate the Scriptures could not be otherwise than unequally successful in different places; and thus while throughout whole regions all or nearly all existing MSS would perish without leaving their text transmitted through fresh copies, the vacant places would presently be filled, and more than filled, by transcripts which would import the texts current in more fortunate lands. Thus whatever irregularities in the geographical distribution of texts had grown up in the earlier centuries would be suddenly and variously multiplied. Moreover the tendency of the changes brought about in that century of rapid innovation by the new relations between the Church and the empire, and by the overwhelming influence of theological controversies, was unfavourable to the preservation of local peculiarities of any kind. It is therefore no wonder that the ancient types of text now lose themselves in a general medley, not indeed vanishing entirely from view, but discernible only in fragments intermingled with other texts. Whatever may be the causes, mixture prevails everywhere in the fourth century: almost all its texts, so far as they can be seen through the quotations of the Fathers, are more or less chaotic.
192. The confusion was naturally most extensive in the Greek texts; but the versions did not altogether escape it. Enough is already known of the Latin texts to enable us to see what kind of processes were at work. Along with the old Western licence as to diction, in which Latin scribes must have long continued to indulge, we find not only indigenous mixture, the combination of diverging or possibly of independent Latin types, but also mixture with Greek texts. Combinations of this latter kind were in fact more or less rude revisions, not differing in essential character from the Hieronymic revision to which the Vulgate is due. As in that better known case, they proceeded from a true feeling that a Greek MS as such was more authentic than a Latin MS as such, uncontrolled by any adequate sense of the difference between one Greek MS and another. As was to be expected, the new Greek elements of these revised Latin MSS came from various sources, now Pre-Syrian with or without the specially Alexandrian corruptions, now distinctly Syrian, Greek readings of this last type being however almost confined to the Italian and Hieronymic revisions. How far the mixture perceptible in Egyptian texts should be referred to this time, it is not as yet possible to say.
193. Exact knowledge of the patristic texts of the fourth century is much impeded by the uncritical manner in which the works of most of the Greek Fathers have been edited. But wherever firm ground can be reached, we find essentially the same characteristics; almost total absence of all the ancient texts in approximate integrity, and infinitely varying combinations of them, together with an increasing infusion of the later Syrian readings. The most remarkable fact, standing out in striking contrast to the previous state of things, is the sudden collapse of the Western text after Eusebius: a few writers offer rare traces of the expiring tradition in occasional purely Western readings which subsequently vanish; but even this slight and sporadic testimony is exceptional. On the other hand elements derived from Western texts entered largely into most of the mixtures which encounter us on every side. A similar diffusion of large elements derived from the Alexandrian text, discernible in the patristic evidence, is still better attested by versions or revisions of versions in this and the next following period, and apparently by the phenomena of subsequent Greek MSS. At Alexandria itself the Alexandrian tradition lives on through the fourth century, more or less disguised with foreign accretions, and then in the early part of the fifth century reappears comparatively pure in Cyril. On the growing influence of the Syrian texts throughout this time enough has already been said.
C. 194, 195. Final supremacy of the Syrian text
194. The history of the text of the New Testament in the following centuries is obscure in details; but the facts which stand out clearly are sufficient for the purposes of criticism. The multiplicity of texts bequeathed by the fourth century was of long continuance. If, passing over the four great early Bibles אBAC, and also the Græco-Latin and Græco-Egyptian MSS, we fix our attention on what remains to us of purely Greek MSS down to the seventh or eighth century, we cannot but be struck by the considerable though unequal and on the whole decreasing proportion in which Pre-Syrian readings of all types are mingled with Syrian. On the other hand before the close of the fourth century, as we have said, a Greek text not materially differing from the almost universal text of the ninth century and the Middle Ages was dominant, probably by authority, at Antioch, and exercised much influence elsewhere. It follows that, however great and long continued may have been the blending of texts, the text which finally emerged triumphant in the East was not a result of any such process, in which the Antiochian text would have been but one factor, however considerable. With one memorable exception, that of the Story of the Woman taken in Adultery, there is evidence of but few and unimportant modifications of the Antiochian text by the influence of other ancient texts before it became the current text of the East generally.
195. Two classes of causes were at work to produce this singular result. On the one hand Greek Christendom became more and more contracted in extent. The West became exclusively Latin, as well as estranged from the East: with local exceptions, interesting in themselves and valuable to us but devoid of all extensive influence, the use and knowledge of the Greek language died out in Western Europe. Destruction of books, which had played so considerable a part in textual history at the threshold of the Constantinian age, was repeated again and again on a larger scale, with the important difference that now no reaction followed. The ravages of the barbarians and Mahometans annihilated the MSS of vast regions, and narrowly limited the area within which transcription was carried on. Thus an immense number of the MSS representing texts furthest removed in locality from Antiochian (or Constantinopolitan) influence perished entirely, leaving no successors to contribute readings to other living texts or to transmit their own texts to the present day. On the other hand Greek Christendom became centralised, and the centre, looked up to increasingly as such while time went on, was Constantinople. Now Antioch is the true ecclesiastical parent of Constantinople; so that it is no wonder that the traditional Constantinopolitan text, whether formally official or not, was the Antiochian text of the fourth century. It was equally natural that the text recognised at Constantinople should eventually become in practice the standard New Testament of the East.
D. 196, 197. Relics of Ρre-Syrian texts in cursives
196. We have hitherto treated the Greek text of the Middle Ages as a single text. This mode of representation, strictly true in itself, does not convey the whole truth. An overwhelming proportion of the text in all known cursive MSS except a few is as a matter of fact identical, more especially in the Gospels and Pauline Epistles, however we may account for the identity. Further, the identity of readings implies identity of origin; the evidence already given has shown many of the characteristic readings to have originated about 250—350, assigning them at the same time a definite single origin, for we need not here distinguish stages in the Syrian revision; and there are no reasons whatever for assigning a different origin to the rest. If an editor were for any purpose to make it his aim to restore by itself as completely as possible the New Testament of Antioch in 350, he could not help taking the approximate consent of the cursives as equivalent to a primary documentary witness; and he would not be the less justified in so doing for being unable to say precisely by what historical agencies the one Antiochian original was multiplied into the cursive hosts of the later ages. But it is no less true that the consent is only approximate. Although numerous important variations between the Antiochian and other more ancient texts have left no trace in known cursive texts, hardly a verse is free from deviations from the presumed Constantinopolitan standard, sometimes found in a few cursives or one, sometimes even in a large array; and there are not wanting cursives which suggest a doubt whether such a standard forms any part of their ancestry. These diversities of cursive texts, perceptible enough even in Mill's pages, and brought into clearer relief by the collations made or employed by Griesbach and Scholz, can now be studied as to all their characteristic phenomena by means of Dr Scrivener's exhaustive collations.
197. Variations of cursives from the prevalent late text are of two kinds, differing in origin, though not always capable of being distinguished. They are due either to mixture with other texts, or to ordinary degeneracy of transmission. In the latter case they must of course have originated in an age which deprives them at once of all critical value and of all but the most subordinate historical interest: in the former case they not only often supply important documentary evidence for the restoration of the apostolic text, in which light we shall have to consider them presently, but form a remarkable link historically between the ninth and following centuries and the preceding periods, being in fact analogous to the Old Latin readings often preserved in Vulgate Latin MSS. They are virtually copies of minute fragments of lost MSS, belonging doubtless in most instances to the middle or late uncial times, but sometimes of an earlier date, and in either case derived directly or indirectly, wholly or partially, from ancient texts. They shew that the final victory of the Antiochian text did not carry with it a total suppression of MSS of other texts; while the fact that the cursives with distinctly mixed texts are not only proportionally but absolutely much more numerous in the tenth and eleventh than in the twelfth and later centuries shews equally that the MSS of other texts fell more and more into neglect. The cursives mentioned above as probably or possibly independent of any Constantinopolitan origin are doubtless on this supposition copies, more or less pure, of MSS similar to those which, immediately or remotely, furnished detached ancient readings to the mixed cursives. They might be compared to the Old Latin c, written several centuries not only after the formation of the Latin Vulgate, but even after its general adoption.
E. 198. Recapitulation of history of text
198. The continuity, it will be seen, is complete. Early in the second century we find the Western text already wandering into greater and greater adulteration of the apostolic text, which, while doubtless holding its ground in different places, has its securest refuge at Alexandria; but there in turn it suffers from another but slighter series of changes: and all this before the middle of the third century. At no long time after we find an attempt made, apparently at Antioch, to remedy the growing confusion of texts by the editing of an eclectic text combining readings from the three principal texts, itself further revised on like principles, and in that form used by great Antiochian theologians not long after the middle of the fourth century. From that date, and indeed earlier, we find a chaos of varying mixed texts, in which as time advances the elder texts recede, and the Antiochian text now established at Constantinople increasingly prevails. Then even the later types with mixed base disappear, and with the rarest exceptions the Constantinopolitan text alone is copied, often at first with relics of its vanquished rivals included, till at last these too dwindle, and in the copies written shortly before the invention of printing its victory is all but complete. At each stage there are irregularities and obscurities: but we believe the above to be a true sketch of the leading incidents in the history of the text of the New Testament; and, if it be true, its significance as a key to the complexities of documentary evidence is patent without explanation.
SECTION IV. RELATIONS OF THE PRINCIPAL EXTANT DOCUMENTS TO THE CHIEF ANCIENT TEXTS
199—223
A. 199, 200. Nature of the process of determination
199. In the preceding pages we have been tracing the history of ancient lines of transmission, divergent and convergent, by means of evidence chiefly furnished by the existing documents. In order to use the knowledge thus obtained for the restoration of the text, we have next to follow the converse process, and ascertain which ancient text or texts are represented by each important document or set of documents. Up to a certain point this exploration of the ancestry of documents has been performed already at an earlier stage of the investigation, for we could have made little progress if we had not been able to recognise certain more or less defined groups of documents as habitually attesting analogous ancient readings, and thus as being comparatively faithful representatives of particular ancient texts. But we are now enabled both to verify with increased exactness the earlier classifications, and to extend them to other documents the texts of which were too ambiguous at first sight to allow them to be classified without the aid of standards external to themselves.
200. The evidence is supplied by the numerous variations in which each variant can at once be assigned with moral certainty to some one of the ancient texts, to the exclusion of those variations in which the grouping of documents is at this stage obscure. At each variation we observe which ancient text is attested by the document under examination. The sum of these observations contains the required result. Neglecting petty exceptions as probably due to some unnoticed ambiguity, unless they happen to be of special clearness, we find that the document habitually follows some one ancient text; or that it sometimes follows one, sometimes another, but has no characteristic readings of the rest; or again that it follows all in turn. Thus we learn that it has transmitted one ancient type of text in approximate purity; or that it is directly or indirectly derived by mixture from two originals of different defined types; or that it has arisen from a more comprehensive mixture. The mixture may of course have taken place in any proportions, and the same observations which bring to light the various elements will supply also a fair estimate of the proportions between them: most commonly there is no difficulty in recognising one text as the base on which readings of one or more other types have been inserted in greater or less number. From the component elements of the text of a document as thus empirically ascertained to be present in the illustrative variations taken into account, and also, more roughly, from their proportions, the component elements of its text generally, and their proportions, become approximately known. This knowledge supplies a key to other less simple variations, by shewing either to which ancient text a given reading must be referred, so far as its attestation by each such document is concerned, or at least to which ancient text or texts each such document gives little or no warrant for referring it. The uses of the information thus obtained, and their limitation, will appear in due time.
B. 201—212. Texts found in Greek MSS
201. We have next to give a brief account of the relations of the principal extant documents to ancient texts as ascertained in the manner described above. Greek Uncial MSS are arranged here in the order that seems most convenient for exhibiting their textual composition, without reference to any supposed order of excellence. Some repetitions have been found unavoidable.
202. Western texts virtually unmixed survive exclusively in Græco-Latin MSS written in Western Europe. They are well represented in the Gospels and Acts by D, some leaves in different places and some whole chapters at the end of Acts being however lost. Though the MS was written in Cent. VI, the text gives no clear signs of having undergone recent degeneracy: it is, to the best of our belief, substantially a Western text of Cent. II, with occasional readings probably due to Cent. IV. Much more numerous are readings belonging to a very early stage of the Western text, free as yet from corruptions early enough to be found in the European or even in the African form of the Old Latin version, and indeed elsewhere. In spite of the prodigious amount of error which D contains, these readings, in which it sustains and is sustained by other documents derived from very ancient texts of other types, render it often invaluable for the secure recovery of the true text: and, apart from this direct applicability, no other single source of evidence except the quotations of Origen surpasses it in value on the equally important ground of historical or indirect instructiveness. To what extent its unique readings are due to licence on the part of the scribe rather than to faithful reproduction of an antecedent text now otherwise lost, it is impossible to say: but it is remarkable how frequently the discovery of fresh evidence, especially Old Latin evidence, supplies a second authority for readings in which D had hitherto stood alone. At all events, when every allowance has been made for possible individual licence, the text of D presents a truer image of the form in which the Gospels and Acts were most widely read in the third and probably a great part of the second century than any other extant Greek MS.
203. Western texts of the Pauline Epistles are preserved in two independent uncials, D2 and G3, in G3 to the exclusion of Hebrews. What has been said of D of the Gospels may be applied with little deduction to the Pauline D2, allowance being made for the inferior interest of all Western texts of St Paul. The text of G3, to a great extent coincident, apparently represents a later type, but still probably not later than Cent. IV. It is to be observed that though many readings of D2 in opposition to G3 are supported by other very ancient texts, others receive no such confirmation, and are shown by Latin evidence to be no less Western than those of G3. But this is merely an example of the variety of Western texts. Since G3 was apparently written late in Cent. IX, probably at St Gallen by an Irish scribe (though it may possibly have been brought to St Gallen from Ireland), the nature of its text may be due either to the preservative power of the seclusion of Greek learning in the West or to direct transcription from a very much older copy. The text of the Gospels in what was originally part of the same MS is, we shall see, entirely different. Two of the uncial Græco-Latin copies of the Pauline Epistles, E3 and F2, cannot count as independent sources of evidence: E3 has long been recognised as a transcript of D2, and we believe F2 to be as certainly in its Greek text a transcript of G3; if not, it is an inferior copy of the same immediate exemplar. Not a single Greek MS of any age, as we have already (§ 171) had occasion to notice, has transmitted to us an Alexandrian text of any part of the New Testament free from large mixture with other texts.
204. Tried by the same tests as those just applied, Β is found to hold a unique position. Its text is throughout Pre-Syrian, perhaps purely Pre-Syrian, at all events with hardly any, if any, quite clear exceptions, of which the least doubtful is the curious interpolation in Rom. xi 6. From distinctively Western readings it seems to be all but entirely free in the Gospels, Acts, and Catholic Epistles: in the Pauline Epistles there is an unquestionable intermingling of readings derived from a Western text nearly related to that of G3; and the facility with which they can generally be here recognised throws into clearer relief the almost total absence of definite Western influence in the other books. Here and there indeed may be found readings which are perhaps in some sense Western, having some slight Old Latin or similar attestation: but they are few and not clearly marked, so that their existence does not sensibly render less significant the absence of distinctively Western readings manifestly such. Respecting Alexandrian readings negative statements as to a document containing a Non-Western Pre-Syrian text can never be made without hesitation, on account of the narrow limitation of the difference of documentary attestation characteristic of the two forms of this text respectively. But we have not been able to recognise as Alexandrian any readings of Β in any book of the New Testament which it contains; so that, with the exceptions already noticed, to the best of our belief neither of the early streams of innovation has touched it to any appreciable extent. This peculiar character is exhibited to the eye in the documentary evidence of those variations in which both a Western and an Alexandrian corruption is present, and one of these corruptions is adopted in the Syrian text, Β being then conspicuous in the usually slender array supporting the reading from which both have diverged. It must not of course be assumed to follow that Β has remained unaffected by sporadic corruption independent of the three great lines, Western, Alexandrian, and Syrian. In the Gospel of St Matthew for instance it has occasionally admitted widely spread readings of very doubtful genuineness. But the influence of these three lines upon almost all extant documents has been so enormous that the highest interest must already be seen to belong to a document of which thus far we know only that its text is not only Pre-Syrian but substantially free from Western and Alexandrian adulteration.
205. The relations to ancient texts which disclose themselves on analysis of the text of א are peculiarly interesting. As in its contemporary B, the text seems to be entirely, or all but entirely, Pre-Syrian: and further a very large part of the text is in like manner free from Western or Alexandrian elements. On the other hand this fundamental text has undergone extensive mixture either with another text itself already mixed or, more probably, with two separate texts, one Western, one Alexandrian. Thus, widely different as is א from the Syrian text, as well as independent of it, it is analogous in composition, except that it shews no trace of deliberate adjustment and critical modification. The mixture is unequally distributed, being most abundant in the Gospels and apparently in the Apocalypse, and least abundant in the Pauline Epistles; but it is never absent for many verses together. The Western readings are specially numerous in St John's Gospel, and in parts of St Luke's: they belong to an early and important type, though apparently not quite so early as the fundamental text of D, and some of them are the only Greek authority for Western readings which, previous to the discovery of א, had been known only from versions.
206. Every other known Greek MS has either a mixed or a Syrian text, mixture becoming rarer as we approach the time when the Syrian text no longer reigned supreme, but virtually reigned alone. Moreover every known Greek MS except those already mentioned contains a Syrian element, which is in almost all cases large, but is very variable. The differences in respect of mixture fall under three chief heads;—difference in the proportion of Syrian to Pre-Syrian readings; difference in the proportion of Pre-Syrian readings neither Western nor Alexandrian to those of both these classes; and difference in the proportion of Western to Alexandrian readings. It is to be observed that the Non-Syrian element of these mixed Greek MSS is hardly ever, if ever, exclusively Western or exclusively Alexandrian. Sometimes the one type predominates, sometimes the other, but neither appears quite alone. This state of things would naturally arise if, as was to be anticipated from the phenomena of the fourth century, the Pre-Syrian texts in their purer forms quickly died out, and were replaced by a multitude of mixed texts. In like manner it is no wonder that the Pre-Syrian text neither Western nor Alexandrian, which already by the fourth century was apparently less popular than that of either the Western or the Alexandrian type, is afterwards found less conspicuously represented in mixed texts than its rivals.
207. The text of A stands in broad contrast to those of either Β or א, though the interval of years is probably small. The contrast is greatest in the Gospels, where A has a fundamentally Syrian text, mixed occasionally with Pre-Syrian readings, chiefly Western. In the other books the Syrian base disappears, though a Syrian occurs among the other elements. In the Acts and Epistles the Alexandrian outnumber the Western readings. All books except the Gospels, and especially the Apocalypse, have many Pre-Syrian readings not belonging to either of the aberrant types: in the Gospels these readings are of rare occurrence. By a curious and apparently unnoticed coincidence the text of A in several books agrees with the Latin Vulgate in so many peculiar readings devoid of Old Latin attestation as to leave little doubt that a Greek MS largely employed by Jerome in his revision of the Latin version must have had to a great extent a common original with A. Apart from this individual affinity, A both in the Gospels and elsewhere may serve as a fair example of the MSS that, to judge by patristic quotations, were commonest in the fourth century. Even the difference of text in the Gospels, though very possibly due only to accidental use of different exemplars for different groups of books, corresponds to a difference existing on a larger scale; for the Syrian text of the Gospels appears to have become popular before that of the rest of the New Testament.
208. In C the Syrian and all three forms of Pre-Syrian text are combined in varying proportions; distinctively Syrian readings and such distinctively Western readings as were not much adopted into eclectic texts being however comparatively infrequent.
209. With respect to the texts of extant uncial MSS of the Gospels later than the four great Bibles, a few words on some of the more important must suffice. The Greek text of the Græco-Thebaic fragments of St Luke and St John (T, Cent. v) is entirely Pre-Syrian and almost entirely Non-Western. That of the considerable fragments of St Luke called Ξ has a similar foundation, with a larger share of Alexandrian corrections, and also a sprinkling of Western and Syrian readings: this character is the more remarkable as the date seems to be Cent. viii. Of greater general importance is L of about the same date, which contains the Gospels in approximate completeness. The foundation of the text is Non-Western Pre-Syrian. No extant MS has preserved so many Alexandrian readings in the Gospels, but the early readings neither Western nor Alexandrian are also very numerous. On the other hand the fundamental text has been largely mixed with late Western and with Syrian elements. The composition, it will be seen, has analogies with that of א, though the actual texts are entirely independent, and the much smaller proportion of Alexandrian corrections in א, the great dissimilarity of its Western element, and the absence of a Syrian element, constitute important differences. In three Gospels the St Gallen MS Δ (see above on G3 of the Pauline Epistles, § 203) has an ordinary Syrian text sprinkled thinly with Alexandrian and a few Western readings. But in St Mark this fundamental text is for the most part displaced by mixture with a Non-Western Pre-Syrian text of the same type as the fundamental text of L and Ξ, and thus full of Alexandrian corrections as well as other early Non-Western readings: traces of the process remain in conflate or intermediate readings. The numerous fragments of PQRZ of the Gospels (see § 100) are variously mixed, but all have a large proportion of Pre-Syrian readings; in such MSS as ΝΧΓ(?Σ), and still more as KM, Pre-Syrian readings are very much fewer. The smaller fragments we must pass over, with one exception: too few lines of Wd (St Mark) survive to enable us to form a trustworthy conception of its text generally; but it includes a large Western element of a very curious type.
210. The Codex Laudianus (E2) of Acts is interesting on more accounts than one. It was apparently the identical Greek MS used by Bede. As it is Græco-Latin in form, its text might be expected to be Western. A Western text it does contain, very distinctly such, though evidently later than that of D; but mixed on apparently equal terms, though in varying proportions, with a no less distinctly Alexandrian text: there are also Syrian readings, but they are fewer in number. P2 is all but purely Syrian in the Acts and 1 Peter, while in the other Epistles and the Apocalypse a similar base is variously mixed with another text predominantly but not exclusively Alexandrian, often agreeing with A where A has readings of this class. The Pauline fragments M2 and H3 have mixed texts, that of M2 being of more ancient character and more interesting. The historical antecedents of B2, and indeed of all MSS of the Apocalypse, are still obscure.
211. A few words must suffice here on Greek Cursives. By far the most free from Syrian readings is 61 of the Acts, which contains a very ancient text, often Alexandrian, rarely Western, with a trifling Syrian element, probably of late introduction. The cursive which comes nearest to 61 of Acts in antiquity of text, though at a long interval, is 33 of the Gospels; which has indeed a very large Syrian element, but has also an unusual proportion of Pre-Syrian readings, chiefly Non-Western of both kinds though also Western: the same type of text runs through the whole MS, which is called 13 in the Acts and Catholic Epistles, and 17 in the Pauline Epistles. Most cursives of the Gospels which contain many ancient readings owe more to Western than to Alexandrian sources. Among these may be named four, 13, 69, 124, and 346, which have recently been shown by Professors Ferrar and T. K. Abbott to be variously descended from a single not very remote original, probably uncial: its Non-Syrian readings belong to very ancient types, but their proportion to the fundamentally Syrian text as a whole is not great. Nearly the same may be said of 1 and 209 of the Gospels, which contain a large common element of ancient origin, partly shared by 118, as also by 131. The most valuable cursive for the preservation of Western readings in the Gospels is 81, a St Petersburg MS called 2pe by Tischendorf as standing second in a list of documents collated by Muralt. It has a large ancient element, in great measure Western, and in St Mark its ancient readings are numerous enough to be of real importance. Another more than usually interesting text, somewhat of the same type but much more largely Syrian, is that of lt 39, the British Museum Gospel Lectionary called y by its collator Dr Scrivener. In 157 of the Gospels we have the best example of the few cursives which more nearly resemble 33 in the composition of their Pre-Syrian element, though not connected with 33 by any near affinity.
212. The proportion of cursives of the Acts and Catholic Epistles containing an appreciable amount of Pre-Syrian readings is much larger than in the Gospels or even in the Pauline Epistles, and the Alexandrian readings thus attested are greatly in excess of the Western, without taking into account 61 or 13. Fortunately however Western texts are not altogether ill represented, though only by scattered readings, chiefly in 137, 180, and 44, this last being a MS belonging to the Baroness Burdett-Coutts (iii 37), for the loan of a collation of which we have to thank Dr Scrivener's kindness; and to these MSS should be added 31 (the Leicester MS called 69 in the Gospels), which has many Non-Alexandrian Pre-Syrian readings of both kinds. The chief characteristics of the ancient elements in the cursive texts of St Paul are the extreme irregularity with which they appear in different parts of his epistles, and the small proportion of Western readings to others. Certain corrections in the margin of 67 (66 of the Acts and Catholic Epistles) stand apart by their inclusion of a relatively large number of very ancient readings, which have no other cursive attestation, some distinctively Western, others not so: these marginal readings must have been derived from a MS having a text nearly akin to that of the fragmentary MS called M2, though not from M2 itself. Besides 17, mentioned above, no other MSS of St Paul require special notice. Much ancient evidence is assuredly preserved in not a few cursive texts of the Apocalypse: but they have not as yet been traced with any clearness to their sources.
C. 213—219. Texts found in Versions
213. Analogous phenomena of mixture to those observed in most Greek MSS recur in the later Versions and states of versions: but the want of adequate knowledge of individual MSS of all versions except the Old Latin leaves much uncertain that will doubtless hereafter be cleared up. The African and European Latin, as has been already intimated, represent Western texts of different antiquity: but most of the aberrant readings found in single MSS are probably due to independent mixture with other Greek texts. In the Italian and Vulgate revisions mixture with Greek texts of various types played a large part: in the Italian Latin the Syrian contingent is especially conspicuous. We have already spoken of the various forms of Latin mixture which are perceptible in 'Mixed Vulgate' MSS (§ 114): it is likewise possible that some of their Non-Western readings may have come directly from Greek MSS.
214. The textual character of the Old state of the national or Peshito Syriac version is to a certain extent ambiguous, as being known only through a solitary and imperfect MS. We cannot always distinguish original readings of the version, antecedent to the bulk of Western readings, from readings in no sense Western introduced into it by mixture in the later generations before our MS was written. In many cases however the discrimination is rendered morally certain by the grouping of documents: and at all events the widest examination of all classes of documents only confirms the general conclusions on the history of the Syriac version set forth above (§ 118) as suggested by the prima facie relations of early grouping. In its origin the version was at least predominantly Western of an early type, such few Alexandrian readings as occur having probably come in at a later though still early time. At the revision, whether independent or conforming to a Greek Syrian revision, changes having the Syrian characteristics already described were introduced into the fundamental text. The revised or Vulgate Syriac text differs from the final form of the Greek Syrian text chiefly in retaining many Non-Western readings (some few of them apparently Alexandrian) which afterwards gave way to Western or to new (distinctively Syrian) readings.
215. The Harklean Syriac, which the thorough recasting of diction constitutes rather a new version founded on the Vulgate Syriac than a revision of it in the ordinary sense, receives its predominant character from the multitudes of ordinary Antiochian readings introduced; but readings of more ancient Greek types likewise make their appearance. Taken altogether, this is one of the most confused texts preserved: but it may be rendered more intelligible by fresh collations and better editing, even if they should fail to distinguish the work of Thomas of Harkel from that of his predecessor Polycarpus. It would not be surprising to find that Polycarpus simply converted the Vulgate Syriac into an exact imitation of the Greek Antiochian text, and that the more ancient readings were introduced by Thomas from the "three (v. l. two) approved and accurate Greek copies in the Enaton of the great city of Alexandria, in the holy monastery of the Enatonians", with which he states that he carefully compared his predecessor's version. In this case the readings noted in the margin might well be those which he did not see fit to adopt, but thought it best to place on record in a secondary place. The Non-Antiochian readings in the text, with or without an asterisk, have the same general character as the marginal readings, and can mostly claim a very high antiquity: many of them are distinctively Western, and they include a large proportion of the peculiar Western variations and interpolations in the Acts. In the Catholic Epistles the readings of the Harklean Syriac have a more mixed character than in the other books.
216. The Jerusalem Syriac Lectionary has an entirely different text, probably not altogether unaffected by the Syriac Vulgate, but more closely related to the Old Syriac. Mixture with one or more Greek texts containing elements of every great type, but especially the more ancient, has however given the whole a strikingly composite character. Variations occur to a certain extent between repetitions of the same passages in different parts of the Lectionary, and also between the several MSS in the few places where the new fragments contain the same portions with each other or with the principal MS. These differences are probably caused by mixture with late Greek MSS; which is indeed likely to have affected this Syriac text in all the extant copies: but for the most part the same peculiar text presents itself throughout.
217. The Egyptian versions are substantially true to their prima facie character. The main body of both versions is founded on a very ancient Non-Western text, sometimes affected by the Alexandrian corrections, sometimes free from them. Neither of them however has escaped mixture. Syrian readings are rare, even in the printed editions, and it is probable that they belong only to a late and degenerate state of the versions: the variation which Dr Lightfoot has found as to the presence or absence of some conspicuous interpolations, Syrian by either origin or adoption, in different Memphitic MSS, and the appearance of a series of them in the margins but not the text of the leading Oxford MS, suggest that this element may have been wholly wanting in the first few centuries. The Western influence is more deeply seated, but is probably of two kinds. The Memphitic no less than the Thebaic has Western readings, but they are with comparatively few exceptions, readings much current in the fourth century, and possibly owe their place to comparatively late mixture. The Thebaic on the other hand has a large proportion of distinctively Western readings of an older type. Whatever may be the real origin of the Æthiopic, it is on the one hand strongly Syrian, on the other in strong affinity with its Egyptian neighbours, and especially its nearer neighbour the Thebaic: both ancient Western and ancient Non-Western readings, Alexandrian and other, are conspicuous in its unsettled but certainly composite text.
218. The two solitary outlying versions bear marks of their late date, but not less of the valuable texts which were still current when they were made. The Armenian includes at least three large elements, Syrian, early Western, and early Non-Western, including some Alexandrian modifications. The coincidence of many of the Western readings in the Armenian with the Latin Vulgate, in conjunction with the real adulteration of the first printed edition from the Latin Vulgate, as mentioned above (§ 121), has brought this version under a vague suspicion of having been at some period subjected to Latinising corruption. The coincidences however with the Old Latin in peculiar readings against the Vulgate Latin are likewise numerous, and can only be explained by descent from a Greek Western original. The Gothic has very much the same combination as the Italian revision of the Old Latin, being largely Syrian and largely Western, with a small admixture of ancient Non-Western readings. Whether the copies which furnished the Western element were obtained by Ulfilas in Europe or brought by his parents from Cappadocia, cannot be determined: in either case they were Greek, not Latin.
219. It will be seen that, extensive and intricate as have been the results of mixture upon Versions, the broad historical relations of their texts correspond to the relations found among other documentary authorities. The only readings, belonging to distinctive types, that can with any certainty claim the authority of either of the three great independent families of versions originating in the earliest period are either Western or Alexandrian. Apparent exceptions to this statement may be found in occasional Syrian readings, or what appear to be such, attested by the Old Syriac or the Memphitic: but the evident presence of a late or extraneous element in the solitary MS of the one and in the printed editions, founded on late MSS, of the other, together with the prevailing character of both texts, renders it highly improbable that these exceptions existed in the versions in their earlier days. The Revised Syriac is the first version to betray clearly the existence of the Greek Syrian revision, exhibiting a large proportion of the characteristically Syrian new readings and combinations of old readings. Various Latin revised texts follow, with analogous but different combinations, two alone deriving a very large share of their complexion from the Syrian text. The Egyptian texts, and especially the Memphitic, likewise sooner or later became adulterated, as we have said, with extraneous elements; but at what dates is uncertain. The only versions, besides the Italian and Vulgate Latin, in which the completed Syrian text is clearly and widely represented are definitely known to be of the fourth or later centuries, that is, the Gothic, Æthiopic, Armenian, and Harklean Syriac: the date of the Jerusalem Syriac is unknown.
D. 220—223. Texts found in Greek Fathers
220. Enough has already been said (§§ 158—162) on the texts which can be recognised in the extant remains of the several Ante-Nicene Greek Fathers. A few supplementary remarks must however be inserted here on the peculiar nature of the textual evidence furnished by Greek works preserved, wholly or in great part, only in ancient translations. In the quotations found in these works the texts of Versions and Fathers are variously blended together, so that their testimony needs to be examined with special care, while it is often too valuable to be neglected. Irenæus furnishes the most prominent example. Of his great treatise against heresies, which is extant in a Latin translation, no Greek MS is known to exist. Epiphanius however, writing about 375, has transcribed into his own principal work the greater part of the first of the five books. Other Greek writers and compilers, from Eusebius onwards, have preserved many short fragments, a few being likewise extant in a Syriac or Armenian dress. Secure knowledge of the character of the text of the New Testament used by Irenæus himself can of course be obtained only from the Greek extracts and from such readings extant only in Latin as are distinctly fixed by the context; and it is solely from these materials that we have described his text as definitely Western. In the use of the Greek extracts the age and other circumstances of the several sources from which they are derived have to be considered. The Greek transmission is independent of the Latin transmission, but not always purer. Greek corruptions absent from the Latin version, due either to the use of degenerate MSS of Irenæus by late writers or to degenerate transmission of the works of these writers themselves, can often be detected in the language of Irenæus himself, and might therefore be anticipated in his quotations. But these individual ambiguities do not disturb the general results. The passages subject to no reasonable doubt render it certain that the translator largely modified biblical quotations in conformity with an Old Latin text familiar to him, but perhaps unconsciously, certainly irregularly and very imperfectly. We thus learn what antecedents to the Latin readings we have to take into account as possible where the Greek has perished, aided by the fact that passages quoted several times exhibit a text sometimes identical, sometimes modified in various degrees. Occasionally, with the help afforded by the other Old Latin evidence, we can arrive at moral certainty that the translator has faithfully reproduced his author's reading: but more commonly the two alternatives have to be regarded as equally possible. Both texts are Western; and the evidence is valuable, whether it be that of Irenæus or virtually of a fresh Old Latin MS, though in the former case it is much more valuable. Were indeed Massuet's commonly accepted theory true, that the Latin version of Irenæus was used by Tertullian, the biblical text followed by the translator would take precedence of all other Old Latin texts in age. We are convinced however, not only by the internal character of this biblical text but by comparison of all the passages of Irenæus borrowed in substance by Tertullian, that the Greek text alone of Irenæus was known to him, and that the true date of the translation is the fourth century. The inferior limit is fixed by the quotations made from it by Augustine about 421.
221. Several important works of Origen are likewise, wholly or in part, extant only in Latin, and need similar allowance for two alternatives in the employment of their evidence as to biblical texts. Caution is especially needed where Rufinus is the translator, as in the early treatise De Principiis, the commentaries on Canticles and Romans, and the Homilies on several early books of the Old Testament and on three Psalms: for his well known licence in manipulating Origen's own language undoubtedly extended to the quotations; and at least in the commentaries the depravation of text has apparently been increased by the condensation of the voluminous original. Yet even here numerous readings can be determined with certainty as Origen's. More reliance can be placed, though still with some reserve, on Jerome's translations, that is, those of the Homilies on St Luke, (Isaiah?), Jeremiah (mostly also extant in Greek), and Ezekiel, and of two on Canticles. For part of the commentary on St Matthew we have an interesting anonymous translation, the portion for xvii 34—xxvii 66 being preserved in no other shape. For xvi 13—xxii 33 it overlaps an extant section of the Greek text; and comparison suggests that they are both independent condensations of a fuller original, so that neither can be safely neglected, though the Latin has the disadvantages of Old Latin modification as well as greater brevity. It has however occasionally preserved matter omitted altogether by the Greek abbreviator. Other Greek patristic writings extant in Latin may be passed over.
222. The Syriac MSS brought to England within the present century have contributed some valuable patristic texts. The Theophania of Eusebius, edited and translated by Dr Lee, presents phenomena analogous to those of the Latin Irenæus. Some of the readings are undoubtedly of Old Syriac parentage, and introduced by the translator; others as certainly belong to Eusebius; and many may have either origin. Moreover the predominant colour of both texts is Western, though the influence of a Non-Western text over Eusebius is also perceptible. The help of Greek fragments is available both here and in the other Syriac patristic translation most useful to the textual critic, that of a large part of the younger Cyril's Homilies on St Luke, edited and translated by Dr Payne Smith. In this instance the disturbing element is the Vulgate Syriac: but the great bulk of the text of the biblical quotations is unaffected by it, and takes high rank as a documentary authority for a Non-Western Pre-Syrian text of the verses which it covers.
223. Respecting Post-Nicene Greek patristic writings generally it will suffice here to refer to what has been said already (§ 193) on the extremely mixed character of their texts, shewing a growing preponderance of Syrian readings even where the text of Antioch was not adopted almost or altogether without modification. With the works of Cyril of Alexandria may be named an obscure exposition of faith (Κατὰ μέρος πίστις), formerly called a work of Gregory of Neocæsarea (Cent. III), and now attributed with much probability to Apollinaris, which has a remarkable Pre-Syrian and chiefly Non-Western text. A more than average proportion of similar elements presents itself in the quotations of Epiphanius; and even so late a writer as John of Damascus (Cent. VIII) makes considerable use of an ancient text.
SECTION V. IDENTIFICATION AND ESTIMATION OF READINGS AS BELONGING TO THE CHIEF ANCIENT TEXTS
224—243
A. 224. Nature of the process of identification
B. 225, 226. Identification and rejection of Syrian readings
225. The first point to decide with respect to each reading is whether it is Pre-Syrian or not. If it is attested by the bulk of the later Greek MSS, but not by any of the uncials אBCDLPQRTZ (Δ in St Mark) Ξ (also 33) in the Gospels (the smaller fragments we pass over here), אABCDE2 (also 13 61) in Acts, אABC (also 13) in the Catholic Epistles, or אABCD2G3 (also 17 67**) in the Pauline Epistles, and not by any Latin authority (except the latest forms of Old Latin), the Old or the Jerusalem Syriac, or either Egyptian version, and not by any certain quotation of a Father earlier than 250, there is the strongest possible presumption that it is distinctively Syrian, and therefore, on the grounds already explained (§ 158), to be rejected at once as proved to have a relatively late origin. It is true that many documents not included in these privileged lists contain Pre-Syrian elements; but only in such small proportion that the chance of a Pre-Syrian reading finding attestation in these late relics of vanishing or vanished texts, and none in the extant documents wholly or mainly of Pre-Syrian ancestry, is infinitesimal; and, when this hypothetical possibility is set against the vera causa supplied by the Syrian revision, becomes yet more shadowy. The special need of strictly limiting early patristic authority for the present purpose to what is 'certain' will be explained further on.
226. The Syrian or Post-Syrian origin of a reading is not much less certain if one or two of the above Greek MSS, as CLPQR 33 in the Gospels, AC[E2] 13 in the Acts and Catholic Epistles, and AC 17 in the Pauline Epistles, are found on the side of the later MSS, or even if similar testimony is prima facie borne by such a version as the Memphitic, the MSS of which have not yet been subjected to a critical sifting. It would be useless to attempt to lay down absolute rules of discrimination; the essential prerequisites for striking the balance are familiarity with the documents, and a habit of observing their various groupings: but the fundamental materials of judgement must be such facts and combination of facts, slightly sketched in the preceding pages, as are implied in the rough arrangement of documents just given. The doubt that must sometimes remain is not often whether a given reading is Syrian, but whether it is distinctively Syrian, that is, whether it originated with the Syrian revision, or was an older reading, of whatever type, adopted by the Syrian revisers. In the final decision, as will be seen, this doubt is very rarely of practical moment.
C. 227—232. Identification of Western and of Alexandrian readings
227. Distinctively Syrian and Post-Syrian readings being set aside, there remain only such readings as the nature of their documentary attestations marks out, often with certainty, often with high probability, as older than 250. Such readings may with substantial truth be called 'Ante-Nicene'; but the term 'Pre-Syrian', if less familiar, is not less convenient, and certainly more correct. The account which we have already given of the early history of the text must have dispelled any anticipation that textual criticism, in reaching back to the middle of the third century, would have nearly fulfilled its task. In truth not only the harder but the larger part remains. We have to begin with simply endeavouring to range under the three principal types or lines of text all readings evidently worthy of attention as possibly right, at the same time making full use of the instruction to be gained by observing the attestations of all Pre-Syrian readings whatever, whether they have any appearance of being possibly right or not. Of the variations in which the endeavour is baffled we shall speak presently. Multitudes of variations present no difficulty at all, and as many need only a little consideration to interpret them.
228. Such Western readings as have acquired no accessory attestation by adoption into the Syrian or other mixed texts catch the eye at once in books or parts of books in which we have one or more Greek MSS with a tolerably unmixed Western text and in which Old Latin evidence is not wanting. In the Gospels such readings are attested by D, the chief Old Latin MSS and Fathers, the Old Syriac, and the Greek Ante-Nicene Fathers, those of Alexandria partially excepted. They are not materially less conspicuous if in the Gospels they are likewise supported by a stray uncial as א or X or Γ, or by a few cursives, as 81 (especially), or 1 and its kindred, 13 and its kindred, 22, 28, 157, &c., or by the Latin or Syriac Vulgate (indeed any Syrian text), or the Thebaic, Æthiopic, Armenian, or Gothic. In Acts D and the Old Latin fragments and Fathers, with the Greek patristic evidence as above, are the primary attestation: א, E2, 31, 44, 61, 137, 180, &c., or any of the above versions except the Gothic, especially the Harklean Syriac or Thebaic, may be the secondary; the numerous quotations by Irenæus taking a prominent place. In the Pauline Epistles the primary documents are D2G3, (E3 and F2 need no further mention), the Old Latin fragments and Fathers, and Greek patristic quotations as above: in the second place may stand א or B, 31, 37, 46, 80, 137, 221, &c., or any of the above versions, the Gothic in particular. The secondary documents here named are only those whose sporadic attestation of Western readings not afterwards Syrian is most frequent: from readings of this class few if any uncials having a large Pre-Syrian element are entirely free.
229. The analogous Alexandrian readings need more attention to detect them. Since it has so happened that every MS containing an approximately unmixed Alexandrian text has perished, the Alexandrian readings can have no strictly primary attestation among extant documents, and are therefore known only through documents containing large other elements. In the Gospels they are chiefly marked by the combination אCLX 33, and also Ζ in St Matthew, Δ in St Mark, Ξ and sometimes R in St Luke, with one or both of the Egyptian versions, and sometimes another version or two, especially the Armenian or the Vulgate or another revised Latin text; and of course Alexandrian Fathers. The least inconstant members of this group are CL and the Memphitic. In the Acts the chief representatives are אACE2 13, 61, and other cursives, as 27 29 36 40 68 69 102 110 112; and the same in the Catholic Epistles, with the loss of E2 and 61, and the partial accession of P2; and in the Pauline Epistles אACP2 5 6 17 23 39 47 73 137 &c.; with the same versions, so far as they are extant, and Fathers as in the Gospels. As however all these documents abound in neutral readings, and most of them in Western readings, the identification of Alexandrian readings can be effected only by careful observation and comparison of contrasted groupings in successive variations. The process is a delicate one, and cannot be reduced to rule: but, though many cases must remain doubtful, we believe that the identification can usually be made with safety.
230. In each of the two classes of variations just noticed the array opposed to the group representing the aberrant text, that is, the Western or the Alexandrian text, as the case may be, owes much of its apparent variety, and more of its apparent numbers, to the presence of the irrelevant Syrian contingent. Two other classes of variations, differing from these in nothing but in the transposition of the habitually Syrian documents to the aberrant side, must evidently be interpreted in precisely the same way. Readings having only characteristic Western and characteristic Syrian attestation must have belonged to the Western text: readings having only characteristic Alexandrian and characteristic Syrian attestation must have belonged to the Alexandrian text.
231. On the other hand the rival readings cannot be exactly described except in negative terms. Against a Western stands a Non-Western Pre-Syrian reading: against an Alexandrian stands a Non-Alexandrian Pre-Syrian reading. The attestation of these readings is simply residual; that is, each of them must have been the reading of all extant Pre-Syrian texts, whatever they may be, except the Western in the one case, the Alexandrian in the other. It follows that, unless reason has been found for believing that all attestation of texts neither Western nor Alexandrian has perished, it must be presumed that the rival reading to a Western reading is not exclusively Alexandrian, and that the rival reading to an Alexandrian reading is not exclusively Western.
232. A large proportion of variations still remains in which the assignation of the readings to different types of ancient text is in various degrees difficult or uncertain. The difficulty arises chiefly from two causes, the mixed composition of some of the principal extant documents, especially Greek uncials, and the not infrequent opposition of documents habitually agreeing as witnesses for one of the aberrant types, resulting in apparent cross distribution. Owing to the former cause Western readings, for instance, which were saved from the extinction which befel their parent texts in the Greek East in the fourth century by their reception into eclectic texts of that period, must naturally be often found attested by documents lying outside the properly Western group. Almost all our better uncials occur singly in their turn as supporters of very distinctly Western readings, and therefore it would be surprising if two or three of them were never to hold the same position together; so that a reading which two or three of them concur in supporting may quite possibly have had a Western origin. But where there is no clear inequality of number and also of predominant character in the attestation which documents of this kind give to the two rival readings of a variation, it may be difficult or impossible to say whether the opposition is between a Western and a Non-Western, or between a Non-Alexandrian and an Alexandrian reading. The cases of apparent cross distribution, of which the Old Latin evidence furnishes the most conspicuous examples, are of course equally due to mixture, and especially to the mixture produced by revision of versions after Greek MSS. Latin MSS known to contain revised texts may naturally be taken to follow a Non-Western source where they stand in opposition to MSS of purer Old Latin pedigree; and in many similar instances a complete survey of the documentary evidence suffices to bring to light the essential features of the grouping in spite of partial confusion. But among these cases likewise there remain ambiguities which can be cleared up only by other kinds of evidence, or which cannot be cleared up at all.
D. 233—235. Identification of neutral readings
233. Besides all the various classes of binary variations examined in the preceding paragraphs, and besides those ternary variations in which the third variant is distinctively Syrian, there are, as we have already seen (§ 184), many other ternary variations in which one reading has a characteristic Western attestation, another has a characteristic Alexandrian attestation, the Syrian evidence being in support of either the first or the second, while the third is attested by documents ascertained to be of wholly or chiefly Pre-Syrian origin: in other words, both the principal aberrant texts stand clearly side by side, each clearly distinguished from a third text. Such third reading may doubtless be, and often manifestly is, nothing but a secondary modification of one of the other readings; for, as has been already intimated, it is not unusual to find together less and more developed Western readings, or less and more developed Alexandrian readings, or both together: nor are mixtures of the two lines unknown. But there are many other third readings which cannot without great difficulty be assigned on either external or internal grounds to such an origin, and which must stand on at least an equal rank with the other two, as having to all appearance an independent ancestry.
234. If then a Pre-Syrian text exists which is neutral, that is, neither Western nor Alexandrian, the phenomena of attestation provide two resources for learning in what documents we may expect to find such a text preserved, comparison of the two fundamental types of binary variations, and direct inspection of the ternary or yet more complex variations last mentioned. In order to avoid needless repetition, the information thus obtained has been to a certain extent employed already in the account of the constituent elements of different documents (§§ 199—223): but, strictly speaking, it is only at the present stage of the investigation that the large body of evidence supplied by the binary variations becomes available. By comparison of binary variations we find what documents recur oftenest in the attestations of Non-Western and the attestations of Non-Alexandrian readings, taken together; in other words, what documents are oftenest found joining others in opposition to either of the aberrant texts singly. By inspection of ternary variations we find what documents oftenest stand out in clear detachment from all others by patent opposition to a Western and an Alexandrian text simultaneously.
235. As might be expected, the results of both processes are accordant as to the documents which they designate as most free at once from Western and from Alexandrian peculiarities. We learn first that, notwithstanding the lateness of our earliest Greek MSS as compared with some of the versions, and the high absolute antiquity of the fundamental texts which the older versions represent, the constituent texts of our better Greek MSS must be in the main of at least equal antiquity, and that the best of them are, even as they stand, more free from Western and Alexandrian peculiarities than any version in its present state. We learn next that Β very far exceeds all other documents in neutrality of text as measured by the above tests, being in fact always or nearly always neutral, with the exception of the Western element already mentioned (§ 204) as virtually confined to the Pauline Epistles. At a long interval after B, but hardly a less interval before all other MSS, stands א. Then come, approximately in the following order, smaller fragments being neglected, Τ of St Luke and St John, Ξ of St Luke, L, 33, Δ (in St Mark), C, Ζ of St Matthew, R of St Luke, Q, and P. It may be said in general terms that those documents, Β and א excepted, which have most Alexandrian readings have usually also most neutral readings. Thus among versions by far the largest amount of attestation comes from the Memphitic and Thebaic; but much also from the Old and the Jerusalem Syriac, and from the African Latin; and more or less from every version. After the Gospels the number of documents shrinks greatly; but there is no marked change in the relations of the leading uncials to the neutral text, except that A now stands throughout near C. In Acts 61 comes not far below א, 13 being also prominent, though in a much less degree, here and in the Catholic Epistles. The considerable Pre-Syrian element already noticed (§ 212) as distinguishing a proportionally large number of cursives in this group of books includes many neutral readings: for examples of these cursives it will suffice to refer to the two lists given above (§§ 228, 229), which include the more important MSS. In some of the Catholic Epistles, as also in the subsequent books, an appreciable but varying element of the text of P2 has the same character. For the Pauline Epistles there is little that can be definitely added to אBAC except 17 and P2: the best marked neutral readings are due to the second hand of 67.
E. 236—239. Suspiciousness of Western and of Alexandrian readings
236. Nearly all that has been said in the preceding pages respecting the documentary attestation of the three leading types of Pre-Syrian text remains equally true whatever be the historical relation of these types to each other. On the other hand, it was necessary at an earlier stage (§§ 173 ff., 183), in describing the characteristics of the Western and Alexandrian texts, to state at once the general conclusions on this head to which we are irresistibly led by Internal Evidence of Texts, alike on that more restricted study of Western and Alexandrian readings which is limited to variations in which their characteristic attestation is least disguised by extraneous evidence, and on the more comprehensive study of all readings that can be ultimately recognised as Western or Alexandrian. In a vast majority of instances the result is identical: in binary variations the Non-Western reading approves itself more original than the Western, the Non-Alexandrian than the Alexandrian: in ternary variations the neutral reading, if supported by such documents as stand most frequently on the Non-Western and Non-Alexandrian sides in binary variations, approves itself more original than the Western and also more original than the Alexandrian. The Western and Alexandrian texts as wholes are therefore in the strictest sense, as we have called them partly by anticipation, aberrant texts.
237. It does not follow however that none of their distinctive readings are original. If it could be shown with reasonable certainty that the three lines diverged simultaneously from the apostolic autographs, or from a common original derived almost immediately from the autographs, the chance that one line alone has preserved true readings where the two others agree, that is, that two transcribers have independently made the same changes, would be infinitesimal (see § 75), except as regards changes of a very obvious and tempting kind. No such presupposition is however imposed by the actual evidence: we have no right to affirm that the two great divergences were simultaneous, not successive. Both are indeed of such extreme antiquity that a strong presumption must always lie against an exclusively Western or exclusively Alexandrian reading; since, apart from accidental coincidence, its genuineness would presuppose as a necessary condition, not only that the two divergences were not simultaneous, but that the rival reading came into existence either at the first divergence or between the first and the second.
238. Of the unfavourable presumptions arising out of the internal character of distinctive Western and distinctive Alexandrian readings generally we have said enough already (§§ 170 ff., 181 ff.). A certain number might on purely internal grounds be received or rejected with equally or almost equally good reason: it is however, we believe, quite safe to dismiss them along with their much more numerous associates that are condemned by individual internal evidence no less than by the prevailing character of the text to which they belong: it may be added that they are seldom intrinsically of much interest. Others remain which by strong internal probability of some kind plead against summary rejection. The plea can never with prudence be set entirely aside: but the number of such readings which eventually make good a claim to a possible place in the apostolic text is, in our judgement, exceedingly small.
239. There are indeed some Western readings in the Gospels, and perhaps in the Acts, which cannot be explained by accidental error of transcription, or by any of the ordinary causes of textual corruption, such as paraphrase, or assimilation to other passages of the New or Old Testament; and in such cases an incautious student may be easily tempted by the freshness of the matter to assume that it must have come from the hand of the writer of the book before him. The assumption would be legitimate enough were the Western texts of late origin: but it loses all its force when we remember (see § 173) that in the second century oral traditions of the apostolic age were still alive; that at least one written Gospel closely related to one or more of the four primary Gospels, together with various forms of legendary Christian literature concerning our Lord and the Apostles, was then current in some churches; and that neither definition of the Canon of the New Testament nor veneration for the letter as distinguished from the substance of its sacred records had advanced far enough to forbid what might well seem their temperate enrichment from such sources as these. Transcriptional probability is likewise of no little weight here: the absence of Western readings of this kind from the NonWestern texts is inexplicable on the supposition that they formed part of the apostolic text.
F. 240—242. Exceptional Western non-interpolations
240. On the other hand there remain, as has been before intimated (§ 170), a few other Western readings of similar form, which we cannot doubt to be genuine in spite of the exclusively Western character of their attestation. They are all omissions, or, to speak more correctly, non-interpolations, of various length: that is to say, the original record has here, to the best of our belief, suffered interpolation in all the extant Non-Western texts. The almost universal tendency of transcribers to make their text as full as possible, and to eschew omissions, is amply exemplified in the New Testament. Omissions of genuine words and clauses in the Alexandrian and Syrian texts are very rare, and always easy to explain. In the Western text, with which we are here concerned, they are bolder and more numerous, but still almost always capable of being traced to a desire of giving a clearer and more vigorous presentation of the sense. But hardly any of the omissions now in question can be so explained, none in a satisfactory manner. On the other hand the doubtful words are superfluous, and in some cases intrinsically suspicious, to say the least; while the motive for their insertion is usually obvious. With a single peculiar exception (Matt. xxvii 49), in which the extraneous words are omitted by the Syrian as well as by the Western text, the Western non-interpolations are confined to the last three chapters of St Luke. In various parts of the Gospels other Western omissions are to be found, which it would be rash to condemn absolutely, the attestations being precisely similar to those of the non-interpolations which we accept, and the internal evidence, intrinsic and transcriptional, being open to some doubt; in other words, an intermediate class of Western omissions that may perhaps be non-interpolations must be admitted. Examples will be found in Matt. (vi 15, 25;) ix 34; (xiii 33;) xxi 44; (xxiii 26;) Mark ii 22; (x 2;) xiv 39; Luke v 39; x 41 f.; xii 19, 21, 39; xxii 62; (xxiv 9;) John iii 32; iv 9. With the difficult question of notation here involved we are not for the moment concerned: it is enough here to repeat that we find ourselves wholly unable to believe some of the clauses and sentences omitted by Western documents to be genuine, while in other not obviously dissimilar cases our judgement remains suspended.
241. These exceptional instances of the preservation of the original text in exclusively Western readings are likely to have had an exceptional origin. They are easily reconciled with the other phenomena if we suppose, first, that the text which became fixed at Alexandria, and in due time was partially adulterated by Alexandrian corruptions, was an offshoot from the text which we have called the neutral text, and which had parted company from the earliest special ancestry of the Western text at a yet earlier date; and secondly, that the interpolations which give rise to the appearance of Western omissions took place in the interval, if not at the actual divergence, and thus stand in all Non-Western texts, whether derived through Alexandria or not. These interpolations are for the most part quite unlike Alexandrian interpolations, and have much more of a 'Western' character; so that the hypothesis which might at first sight suggest itself, of their having originated at Alexandria, and thence spread by mixture to Non-Western texts elsewhere, is set aside by internal evidence as well as by the want of other corroborative instances. The purely documentary phenomena are compatible with the supposition that the Western and the Non-Western texts started respectively from a first and a second edition of the Gospels, both conceivably apostolic: but internally none of the Non-Western interpolations certainly justify this claim to a true though a secondary kind of originality, and some of them, it is not too much to say, shew a misunderstanding which renders it impossible to assign to them any worthier origin than to ordinary Western interpolations.
242. Nothing analogous to the Western non-interpolations presents itself among distinctively Alexandrian readings of any form, omissions, additions, or substitutions. Now and then, though fortunately but rarely, the attestation of what seems to be an Alexandrian reading, unusually well attested, approaches too near the attestation of some neutral readings to exclude doubt as to the true origin, while internal evidence is likewise indecisive. But this occasional ambiguity of external evidence is not to be confounded with incongruities of internal character in readings of clearly defined external type. No variations are known to us in which a distinctively Alexandrian reading, indubitably such, approves itself as genuine against Western and neutral texts combined, or even against the neutral text alone. Of the numerous variations which at first sight appear to involve conflicts between the neutral text and the Western and Alexandrian texts combined it will be more opportune to speak further on. G. 243. Recapitulation of genealogical evidence proper
243. To sum up what has been said on the results of genealogical evidence proper, as affecting the text of the New Testament, we regard the following propositions as absolutely certain. (I) The great ancient texts did actually exist as we have described them in Sections II and III. The main line of neutral and comparatively pure text was from an early time surrounded and overshadowed by two powerful lines containing much aberration, the 'Western' being by far the most licentious and the most widely spread, and the Alexandrian being formed by skilful but mostly petty corrections which left the neutral text untouched, at all events in the Gospels and Pauline Epistles, except in a very small proportion of its words. Late in the third century, or soon after, MSS came to be written in which the three main texts were mixed in various proportions, and the process went forward on a large scale in the following century, when all the unmixed texts began to die out. The Western, hitherto the most influential of all texts, now disappeared rapidly, lingering however, it would seem, in the West. One of the mixed texts was formed in Syria with care and contrivance, modifying as well as combining the earlier texts, and by the middle of the fourth century was established in influence. For some centuries after the fourth there was in the East a joint currency of the Syrian and other texts, nearly all mixed, but at last the Syrian text, the text of Constantinople, almost wholly displaced the rest. (II) In the Gospels and Pauline Epistles, and to a less extent in the Acts, all the four principal forms of text are fairly represented in extant documents; in other books the representation of one or more of the texts is seriously incomplete or doubtful. (III) The extant documents contain no readings (unless the peculiar Western non-interpolations noticed above are counted as exceptions), which suggest the existence of important textual events unknown to us, a knowledge of which could materially alter the interpretation of evidence as determined by the above history. (IV) In a large proportion of variations the assignation of the several readings to the several ancient texts by means of extant documents is clear and certain, and thus affords a sure clue to the original reading. (V) In many other ancient variations the distribution of documentary evidence must as a matter of fact be due to ancient distribution among the several texts, with or without subsequent mixture, although the extant documentary evidence is too scanty or too confused to allow confident decision between two or more possible views of the historical antecedents of the several readings. This last proposition implies that we have to do with many variations in which the tests supplied by the general history of the text of the New Testament are not available for direct use, and other critical resources are needed. To these we must presently turn.
SECTION VI. REVIEW OF PREVIOUS CRITICISM WITH REFERENCE TO ANCIENT TEXTS
244—255
A. 244—246. Foundation of historical criticism by Mill, Bentley, and Bengel
244. Before however we pass from the great ancient texts, it will be right to interpose a few words of comment on previous criticism dealing with the same subject. Although the series of editions which can be said to approximate to a true text of the New Testament begins in 1831, the preliminary studies of the eighteenth century, unduly neglected since the earlier part of the present century, form the necessary introduction to all secure progress hereafter. It will be sufficient to mark the most salient points in the progress of criticism.
245. Mill led the way in 1707 not only by his ample collection of documentary evidence but by his comprehensive examination of individual documents, seldom rising above the wilderness of multitudinous details, yet full of sagacious observations. He incidentally noticed the value of the concurrence of Latin evidence with A, the most conspicuous and the only complete representative of an ancient Non-Western Greek text then sufficiently known; and this glimpse of genealogical method was not lost upon Bentley, who with clear and deliberate purpose made Greek and Latin consent the guiding principle of his own project for a restoration of the text. The actual project fell to the ground until it was revived and carried out in Lachmann's edition of 1831, the starting point of the later period; in which however it assumed a somewhat different shape through the substitution of the Old Latin for the Vulgate Latin, and the ranging of the Greek Western uncials on the Latin or, as it was more properly called, the 'Western' side. But the principle itself was received at once into fruitful soil, and contributed more than any other antecedent to the criticism of the intervening period.
246. How deeply the value of the principle, as set forth in Bentley's Proposals of 1720, impressed Bengel, although he accepted it only in part, is evident from many pages of his Introduction of 1734. Bengel himself pointed out the deceptiveness of numerical superiority detached from variety of origin, prepared for sifting the confused mass of Greek MSS by casting upon it, as he said, the Versions and Fathers as an additional heap, and endeavoured to classify the documents known to him according to their presumed derivation from ancient texts. He divided them into two great 'nations' or 'families', the 'Asiatic' and the 'African', answering roughly to what we have called Syrian and Pre-Syrian; and further, less distinctly, subdivided the latter into two subordinate 'nations' or 'families', represented typically by A and by the Old Latin. At the same time he laid great stress on internal evidence, in this as in other respects making large use of materials scattered through Mill's notes; and it is chiefly to his earnest if somewhat crude advocacy that Transcriptional Probabilities under the name of 'the harder reading' owe their subsequent full recognition.
B. 247—249. Development of historical criticism by Griesbach, in contrast with Hug's theory of recensions
247. Bengel was succeeded in Germany by Semler, and under his influence by a group of acute and diligent textual critics, stimulated to fresh researches both by Bengel's writings and by the rich accession of new materials from Wetstein's edition of 1751-2, and from the various explorations and collations which were vigorously carried on in the later years of the century. What Bengel had sketched tentatively was verified and worked out with admirable patience, sagacity, and candour by Griesbach, who was equally great in independent investigation and in his power of estimating the results arrived at by others. Bengel's 'Asiatic' text he called 'Constantinopolitan': the two more ancient texts, which he clearly defined, he called 'Western' and 'Alexandrian'. Unfortunately he often followed Semler in designating the ancient texts by the term 'recension', and thus gave occasion to a not yet extinct confusion between his historical analysis of the text of existing documents and the conjectural theory of his contemporary Hug, a biblical scholar of considerable merit, but wanting in sobriety of judgement.
248. Hug started from what was in itself on the whole a true conception of the Western text and its manifold licence. He called it the κοινὴ ἔκδοσις, or 'Vulgate Edition', taking the name from the text of the LXX as it was in its confusion before the reform attempted by Origen in his Hexapla. But further he conjectured that the disorderly state of this popular text led to its being formally revised in three different lands, the product of each revision being a 'recension' in the strict sense of the word. The alleged evidence consists in two well known passages of Jerome. In the first he speaks of the diversity of copies of the LXX in different regions; Alexandria and Egypt appeal, he says, to the authority of Hesychius; Constantinople and Antioch approve of the copies of Lucian the Martyr; the intermediate provinces read the Palestinian volumes, wrought out by Origen and published by Eusebius and Pamphilus; and the whole world is set at discord by this threefold difference. In the second passage, already cited (§ 190), he is stating vaguely to what Greek sources he proposes to have recourse in correcting the Latin Gospels. "I pass by", he says, "those volumes which bear the names of Lucianus and Hesychius, and are upheld by the perverse contentiousness of a few men": he adds in obscure language that 'they had neither been allowed to make corrections (emendare) after the Seventy in the Old Testament, nor profited by making corrections in the New Testament'. The latter quotation, enigmatic as it is, distinctly implies the existence of copies of the New Testament or the Gospels bearing in some way the names of Lucianus and Hesychius, and supposed to have in some way undergone correction; and likewise associates the same names with some analogous treatment of the LXX. As they appear in company with Origen's name in a similar connexion in the first quotation, Hug supposed that Hesychius had made a recension of both Testaments for Alexandria, Lucianus for Antioch, and Origen for Palestine. He had next to discover descendants of the supposed recensions in existing groups of documents, and had no difficulty in assigning the Constantinopolitan text to Lucianus: but since Hesychius plausibly claimed the 'Alexandrian' text, he could find no better representation of Origen's supposed work than an ill defined and for the most part obscure assemblage headed by AKM.
249. Origen's quotations prove conclusively that no such text as these documents present can ever have proceeded from him: and it is hardly less certain, as Griesbach shewed by the implicit testimony of various passages, that he never made anything like a recension of the New Testament. It does not follow that the same can be said of Lucianus and Hesychius. As we have already observed (§§ 185, 190), the Syrian text must have been due to a revision which was in fact a recension, and which may with fair probability be assigned to the time when Lucianus taught at Antioch. Of the Alexandrian corrections more than one stage can certainly be traced: whether the primary corrections were due to a distinct revision cannot, we think, be determined, and it would be little gain to know. That Hesychius had no hand in any revision which can have produced them is proved by the occurrence of many of them in Origen's writings, at a much earlier date. But it is quite conceivable that Hesychius made or adopted some eclectic text too short-lived to have left recognisable traces of itself in extant evidence, though it may be a hidden factor in the process of mixture to which some of our texts are partly due. Thus much it is but just to Hug to say, though the point is of no practical consequence. But neither the deserved discredit into which Hug's theory of recensions as a whole has fallen, nor the uncertainty as to the precise nature of the facts referred to in Jerome's second passage, create any doubt as to the soundness of Griesbach's fundamental classification of texts, which rests entirely on the independent base furnished by the observed phenomena of existing documents.
C. 250—253. Defects of Griesbach's criticism
250. There are indeed some defects in Griesbach's view which he could hardly have failed to correct if all the evidence now accessible had been in his hands. Perhaps the most important of these is a confusion between the classification of ancient texts and the classification of documents derived from them. He was aware indeed that no existing MS preserves any 'recension' or leading ancient text in absolute purity, and that one source of corruption was the intrusion of readings out of another 'recension' (Preface to Gospels of 1796, p. lxxviii; cf. Meletemata, pp. xxxviii f.). But still in effect he treated our documents as capable of being each on the whole identified with some one ancient text. In other words, he failed to apprehend in its true magnitude the part played by mixture in the history of the text during the fourth and following centuries, or to appreciate the value of the observation of groupings as a critical instrument by which a composite text can be to a great extent analysed into its constituent elements.
251. Hardly if at all less important was his confusion of Alexandrian readings with readings preserved wholly or chiefly at Alexandria. His discrimination of the internal character of Western and Alexandrian corrections (ib. p. lxxvii) is excellent as far as it goes, and may supply useful guidance in some cases of obscure attestation. But his mode of using the two great texts can be justified only on the impossible assumption that the Alexandrian text, with its bulk of pure readings and its distinctive corruptions alike, was, so to speak, full-blown from the beginning. The very fact that these corruptions originated at Alexandria implies that MSS free from them, as well as from Western corruptions, existed previously at Alexandria; and there is no apparent reason why this earlier form of text should not have been propagated in greater or less purity at Alexandria by the side of the altered text or texts. If it was, and if any existing documents represent it, their text, whatever its value may be, has not the defects of a distinctive Alexandrian text. But further there is no apparent reason why documents should not exist derived from sister MSS to those which originally came to Alexandria, and which thus were the parents of later MSS current at Alexandria, including those in which the Alexandrian corrections originated; and if so, no ordinary internal evidence can enable us to decide whether the ancestry of any given existing documents having this character of text was altogether independent of Alexandria, or had its home at Alexandria but was unaffected by any distinctive Alexandrian corruption. Griesbach seems however to have tacitly assumed both that Alexandria had but one Non-Western text, and that no early Non-Western text survived except at Alexandria; and accordingly in most variations the critical problem which virtually presented itself to him was merely whether it was more likely on internal grounds that the (assumed) Western reading was a corruption of the (assumed) Alexandrian or the Alexandrian of the Western, the characteristics of each 'recension' and the special probabilities of the immediate context being considered together.
252. Thus owing to an imperfect conception of the process of transmission, leading to a misinterpretation of quite the most important evidence, unchecked by attention to grouping, Griesbach was driven to give a dangerously disproportionate weight to internal evidence, and especially to transcriptional probability, on which indeed for its own sake he placed excessive reliance: and this, not his wise anxiety to discriminate the ancient sources of readings before counting or weighing authorities, is the chief cause of the inferiority of his own text of the New Testament, which stands in singular contrast to the high qualities of his criticism. The other great cause of its insufficiency we have already mentioned (§§ 16, 17), his use of the Received Text as a basis for correction. To have taken as his basis those ancient texts in which he himself placed most confidence would have increased the difficulties of his task as an editor, since they frequently did not offer him the same reading; but, as Lachmann triumphantly shewed, in no other way was it possible to avoid the errors that must often find acceptance when numberless variations are approached from the wrong side.
253. The limitations of view in Griesbach and his predecessors were the natural result of the slenderness of their materials. Bentley and Bengel wrote when A was for practical purposes the one ancient purely Greek uncial; and the peculiarities of its text, used as a standard, coloured their criticism, and to a certain extent even that of Griesbach. He learned much from his study of C and L: but the very large distinctively Alexandrian element which they contain had probably a considerable share in leading him implicitly to assume that any extant ancient text not Western must be Alexandrian, and that in the most exclusive sense. A later generation has less excuse for overlooking the preservation of a neutral text, in approximate integrity in B, and in greater or less proportions in many other documents; or for questioning the vast increase of certainty introduced by its recognition in weighing the claims of rival Pre-Syrian readings.
D. 254, 255. Permanent value of Griesbach's criticism
254. In dwelling on Griesbach's errors at some length, notwithstanding the neglect into which his writings have unhappily fallen, we should be grieved even to seem regardless of a name which we venerate above that of every other textual critic of the New Testament. It was essential to our purpose to explain clearly in what sense it is true, and in what sense it is not true, that we are attempting to revive a theory which is popularly supposed to have been long since exploded. No valid objection can, we believe, be brought against the greater part of Griesbach's historical view. It is commonly met by vague sceptical assertions which make no attempt to deal with the actual phenomena. Criticisms which merely shewed that he had been led into too broad and unqualified assertions as to this or that document have left untouched or even unawares strengthened his main positions. The most plausible allegation, that his latest discoveries as to Origen's readings compelled him to abandon his attempt to distinguish between his 'Western' and his 'Alexandrian' readings, and thus destroyed the basis of what is called his theory, depends on a double misconception. The recognition of the fact that Origen sometimes used a MS either 'Western' or containing a large 'Western' element did indeed render it impossible to affirm that a reading found in Origen must needs be 'Alexandrian', that is, it prescribed special care in the interpretation of one single source of evidence; but it made no change in other respects: and the Meletemata of 1811, in which the recognition is conveyed, reiterate Griesbach's familiar statements in precise language, while they shew a growing perception of mixture which might have led him to further results if he had not died in the following spring.
255. It is not necessary to our purpose to pass under review the principles and texts of Griesbach's three great successors, all of whom have published texts of a substantially ancient type, and from each of whom, from Tregelles in particular, we have learned much. But we are bound to express our conviction that the virtual abandonment of Griesbach's endeavour to obtain for the text of the New Testament a secure historical foundation in the genealogical relations of the whole extant documentary evidence has rendered the work of all appreciably more imperfect in itself, and less defensible on rational grounds. Such corrections of Griesbach's leading results as have been indicated above (§§ 250—252) would have removed the difficulties which have unquestionably been felt by dispassionate judges, though they have also been distorted and exaggerated by partisans. In taking up his investigations afresh, we have, we trust, found a way not only to make a somewhat nearer approximation to the apostolic text than our immediate predecessors, but also to strengthen the critical bases on which their own texts are for the most part founded.
CHAPTER III. RESULTS OF INTERNAL EVIDENCE OF GROUPS AND DOCUMENTS
256—355
SECTION I. DOCUMENTARY GROUPS AS LIMITED BY REFERENCE TO PRIMARY GREEK MSS GENERALLY
256—280
A. 256—260. General considerations on Documentary Groups
256. In attempting to give an account of the manner in which the historical relations of the great ancient texts of the New Testament can be safely used for decision between rival readings, we have of necessity (see § 72) transgressed the limits of purely genealogical evidence, in so far as we have dwelt on the general internal character of the Western and Alexandrian texts as a ground for distrusting readings apparently Western only, or Western and Syrian only, or Alexandrian only, or Alexandrian and Syrian only. The evidence which has been thus appealed to is in effect Internal Evidence of Groups (§§ 77, 78), in principle identical with Internal Evidence of Documents in virtue of the genealogical axiom that, accidental coincidences apart, identity of reading implies ultimate identity of origin. Thus, to take the simplest case, finding a frequent recurrence of D, the Old Latin, and the Old Syriac in isolated combination, we knew that in each such reading they must be all lineally descended from a single common ancestor. Having found reason to think that readings attested by this particular group of documents are of great antiquity, we examined them successively in order to ascertain their prevailing internal character by means of variations in which the internal evidence is morally free from doubt.
257. Now a moment's consideration shews that the essentials of this process are independent of the historical adjuncts here attached to it, and remain the same for every possible combination of documents; and that therefore its power of employing easy variations as a key to difficult variations is of universal range. So applied, it is essentially a particular mode of using Internal Evidence of Documents; only not continuous extant documents but, as it were, fragmentary lost documents. Whenever a particular detached combination of documents is of sufficiently frequent occurrence to give room for generalisations, and those of its readings which admit of being provisionally accepted or rejected on Internal Evidence of Readings, Intrinsic and Transcriptional, are found to be all or nearly all apparently right, we are justified in anticipating that its other readings, as to which our judgement has thus far been suspended, or even on the whole adverse, are right too, and in requiring on re-examination very strong local internal evidence to rebut the favourable presumption. A similar recurrence of numerous apparently wrong readings will throw suspicion on the other or doubtful readings of the same group, provided that it remains in all cases literally or practically detached: we say practically, because the accession of a group containing no document outside the habitual attestation of such a text as the Syrian violates detachment in appearance alone. Either the favourable or the unfavourable presumption may also be further defined according to particular classes of readings.
258. Since in all cases tire inference depends on assumed homogeneousness of text, its basis may appear to be subject to uncertainty; for homogeneousness is interrupted by the intrusion of mixture, and it is theoretically possible that lost originals of groups might be mixed, as well as extant MSS. But the originals from which most groups which it is in practice worth while to keep in mind must have diverged can with difficulty be referred to so late a date as the times of general mixture, and no clear evidence of antecedent mixture has come to our own notice. The homogeneousness of the fundamental texts of all important groups may therefore, we believe, be safely trusted.
259. The limitation, more or less strict, to detached combination is necessary because otherwise the characteristics of the special common ancestor will be mixed up with the characteristics of a remoter and for present purposes less important ancestor. In all places where there is no variation D and the two associated versions are likewise found in combination, not the less truly because all other documents have the same reading; and this combination points with equal certainty to a single common ancestor: but here the single common ancestor was the apostolic autograph, followed perhaps by an indefinite number of immediate descendants; whereas what we want to know is the character of the special ancestor, as displayed either in departure from the original text or in fidelity shewn to it where others have departed from it. Similarly, where we find D and its associates agreeing with, for instance, אBCL and the Memphitic against all other documents, if we have ascertained that this second group often stands in opposition to the first, we know that the reading must have existed in a common ancestor of the two special ancestors, and that therefore it can tell us nothing about the special characteristics of either.
260. The most delicate and difficult part of the use of groupings in criticism consists in judging how far a group loses its virtual identity by slight losses or slight accessions of constituent members. The least important losses and accessions from this point of view are evidently those which accompany fragmentariness of text, so that the change is not, for instance, from concurrence to opposition, but from concurrence to total absence, or vice versa: in such cases much depends on the number and variety of the remaining members. Others again, which look as if they ought to be important, are found in experience to be of little or no account: that is, if we treat separately the groupings with and without the varying member, the characteristics are found to be identical; so that the same results would have been reached by treating both forms of combination as a single group. An excellent example is supplied by many of the Alexandrian corrections in St Mark, where we have every binary and ternary combination of אCLΔ besides the full quaternion. But the accession or loss of any primary document should always be treated as constituting a new group until observation has shown that no real difference can be detected in the results. How easily readings having the same origin might come to have an attestation perpetually varying within certain limits may be readily understood, for instance in such an example as that just cited, as soon as we apprehend clearly the manner in which ordinary casual mixture came to pass. Whether two or more MSS were deliberately compared for simultaneous use, or variations were noted in a margin and then at the next stage taken up into the text, or reminiscences of a text formerly heard or read became intermingled with the immediate impressions of eye and ear in transcription,—in all these cases a transcriber was making a conscious or unconscious selection of readings to insert into his fundamental text; and no two transcribers would make exactly the same selection. However great may be the superficial complexities of existing attestation, the primitive relations of text from which they are derived must have been simple; as otherwise each variation must have exhibited a much greater number of variants: and thus it is no wonder that after a while we find ourselves enabled to ascribe practical identity to groups not identical as to all their members.
B. 261—264. Progressive limitation of Groups with reference to Primary Greek MSS
261. It might perhaps be imagined that the possible combinations of our numerous documents would constitute an intractable multitude of groups: but no such difficulty exists in practice. Genealogical possibilities make up the merest fraction of arithmetical possibilities; and of the combinations that actually occur only a small proportion deserve more than momentary attention. The Syrian text as a whole must, we believe, be condemned by Internal Evidence of Groups almost as surely as by the evidence connected with the history of texts; and texts supported by only a portion of the Syrian phalanx have still less claim to consideration. Greek manuscripts containing a large amount of Pre-Syrian text, early Versions, and early Fathers are not numerous, and to a great extent are fragmentary or discontinuous; and combinations into which none of them enter may evidently in most cases be safely neglected. A student soon becomes aware that the groupings which can by any possibility affect his judgement in doubtful variations are sure to contain one or more of a very small number of primary documents. If at any time in the examination of a specially difficult case his attention is attracted by a reading supported by a group hitherto neglected by him, he will naturally take fresh opportunities of observing its characteristics. But the whole operation is simpler than it seems on paper.
262. No one, we believe, who agrees explicitly or implicitly with the account which we have given of the Syrian text and its attestation would hesitate, after studying the Internal Evidence of Groups, to take אBCDL, 33 in the Gospels, אABCDE2, 13 61 in Acts, אABC 13 in the Catholic Epistles, and אABCD2G3 17 in the Pauline Epistles, as the primary documents in the sense just mentioned. This is of course entirely consistent with the assignation of substantial weight to numerous other documents in different degrees in the decision between rival readings. What is meant is that all groups containing none of these primary documents are found so habitually to support the obviously wrong variants where internal evidence is tolerably clear, that they must lie under the strongest suspicion in doubtful variations. Some few other Greek MSS, mostly fragmentary, might to a certain extent claim to be placed in the same class (see § 225): but it is safer to keep to these conspicuously preeminent and approximately complete copies. In strictness the African and European Latin, the Old Syriac, the Egyptian versions, and the Ante-Nicene Fathers should be added to the list: we venture however to omit them here for the sake of simplicity, the practical effect of omitting them being extremely small, as will be explained further on.
263. Now if each of the Greek MSS singled out as primary is individually entitled to this exceptional distinction as a representative of Pre-Syrian texts, we should naturally expect the complete combinations of them to attest a specially pure text; the text thus attested being certified by the concurrence of all the great lines of transmission known to have existed in the earliest times, since undoubtedly all known Pre-Syrian forms of text are sufficiently represented among the primary MSS except the Western texts of the Catholic Epistles (in so far as they have a Western text) and of part of the Acts, and these exceptions are shown by the analogies of other books to affect little beyond degrees of certainty. And this is precisely what we do find: the groups formed by the complete combinations of these primary documents attest clearly the purity of their ancestry by the prevailing internal excellence of their readings. The number of their readings which can with any show of reason be pronounced to be apparently corruptions of other existing readings is exceedingly small; and in our opinion the claim is in all these cases unfounded.
264. When these groups lose their most distinctively Western members, D in the Gospels and Acts and D2G3 in the Pauline Epistles, and with them, as usually happens, one or more of the predominantly Western versions, totally different because less comprehensive groups come into view, אBCL 33 in the Gospels, אABC and the one or two cursives in the other books; but these also, when tried by internal evidence, are found not less constantly to bear the marks of incorrupt transmission. Thus far we have been dealing with essentially the same distributions as in former pages, though from a different point of view: the last result is nearly equivalent to the former conclusion that, certain peculiar omissions excepted, the Western text is probably always corrupt as compared with the Non-Western text.
C. 265—267. Relation of Primary Greek MSS to other documentary evidence
265. Before we proceed to examine the character of the more narrowly limited groups, it is necessary to consider in some little detail the bearing of the evidence of Greek MSS not singled out for primary authority, and of all versions and patristic quotations. Texts in all the languages supply a greater or less amount of various Pre-Syrian evidence having a strong prima facie claim to authority, the true force of which manifestly cannot be left undetermined. It is needless to discuss variations in which the secondary Pre-Syrian evidence (the Syrian evidence may be passed over here and elsewhere) is predominantly on the side of the primary group, or in which it divides itself with anything like equality: the apparent difficulty begins with the numerous cases in which the reduced band of primary MSS is sustained by only a small proportion of the secondary evidence; and then the question arises whether any and if so what amount or weight of secondary evidence, in conjunction with outlying primary MSS, ought to balance or outweigh the strong antecedent authority of the primary band of primary MSS. The question here is not, as it was above (§ 262), whether this or that document should be included among primary documents, but whether the documents accepted as primary, whichever they may be, can safely be allowed an absolutely paramount authority. Taking for granted that all the documentary evidence contributes, more or less appreciably, to the formation of a right judgement as to the merits of all rival readings, and further that in many variations documents not classed as primary contribute materially to a right decision, either directly or as aiding the interpretation of the whole evidence, we have still to ask how far primary documents can be implicitly trusted where they have little or no support from other documents. The doubt presents itself most strongly in readings attested by a very small number of primary MSS exceptionally commended by Internal Evidence of Groups and Documents: but the principle is not affected by the number.
266. The strongest presumption against the legitimacy of any such separate authority of the primary MSS is derived from the prima facie superiority of composite to homogeneous attestation (see § 75); while on the other hand (see § 76) it is checked by the contingency, varying in probability according to the ascertained elements of the secondary documents that may be in question, that apparent compositeness of attestation may really be due to mixture and therefore delusive. A satisfactory answer to the question can however be obtained from two sources only, Internal Evidence of such groups as consist wholly or almost wholly of primary MSS, and consideration of the nature of the texts of the secondary documents as bearing on the point at issue. On the Internal Evidence of the more important groups of this class enough will be said in the following sections. We are for the present concerned with the preliminary enquiry whether any class of secondary documents has such a textual character that their total or almost total absence from the attestation of a reading otherwise sufficiently attested by primary MSS should throw doubt on its genuineness.
267. To conduct the enquiry with due circumspection, it is necessary to pay special attention to those variations in which the extant evidence includes important secondary documents preserved only in fragments, and especially documents which would merit a place on the primary list but for their imperfect preservation. If in such cases the result were often unfavourable to the primary MSS, it would evidently in variations where they are absent be requisite to take into account the twofold contingency of their hypothetical presence on this or on that side. If however, on careful consideration of every kind of evidence, their actual presence is not found to justify doubts as to the antecedent authority of the primary MSS, we can with the more confidence trust the primary MSS in those more numerous variations where, with perhaps no accession to the number of their allies, they are confronted by a less imposing array.
D. 268. Absence of Secondary Greek MSS from Groups containing Primary Greek MSS
268. The first class of secondary documents, according to the usual order, is formed by the secondary Greek MSS; in which we do not include those whose texts are wholly or almost wholly of Syrian origin. Nothing can be clearer than the mixed character of all these MSS; so that, in supposing them to have derived a given reading from, for instance, a Western origin, ultimate or immediate, we are not contradicting the known fact that they have numerous ancient Non-Western readings, when it is equally known that they contain numerous Western readings. If in some places their aggregation in opposition to the primary MSS appears too great to be explained by accidental coincidence of several separate mixtures with Western or other sources, we have to remember, first, that none or almost none of them are without a large Syrian element, and secondly, that there is no reason to suppose the Syrian to have been the only eclectic text which had a wide influence about the fourth century.
E. 269—273. Absence of Versions from Groups containing Primary Greek MSS
269. Respecting Versions, it is to be observed at the outset that the large extent to which they have either from the first or at some later time participated in Western corruption must lead us to expect from them but scanty support to the true reading in a large proportion of Pre-Syrian variations. Of the versions more ancient than the times of general mixture, the Old Latin being wholly Western, and the Old Syriac, as now extant for not quite half of the Gospels and for no other books, being almost wholly Western, there remain only the two closely related Egyptian versions, of which the Thebaic, itself preserved only in fragments, contains so large a Western element that earlier critics reckoned it as wholly Western. It is certain, on evidence already given (§§ 120, 217), that the original Memphitic version became ultimately corrupted from common Greek sources, and the printed editions to a great extent represent this debased form of Memphitic text; so that till the best MSS have been completely collated, we have no security that Memphitic readings at variance with the general character of the version belong to its primitive state. Moreover, as we have seen, even in its earlier days it was probably touched by the Western influence. There remain the later versions and the revised forms of the Latin and Syriac versions; and though they all contain Non-Western Pre-Syrian elements in various proportions, and accordingly have all a certain number of readings in common with the primary Greek MSS against most versions, we have no right to regard their predominant or even concordant opposition as outweighing an otherwise trustworthy attestation.
270. This distribution of Western and Non-Western texts among versions is reflected in the range of support which the primary Greek MSS (in opposition to D in the Gospels and Acts, D2G3 in the Pauline Epistles) most usually receive from the several versions. Their most constant allies are, as we should expect, one or both of the Egyptian versions. Next to them probably come documents essentially Western, but preserving much of the earlier state of text which existed when many of the Western readings had not yet arisen, such as the Old Syriac and the African Latin. But, as we have said, the primary Greek MSS likewise receive in turn the support of every other version, sometimes of several at once, not seldom even where all or nearly all other Greek MSS stand in opposition.
271. On the other hand the support of versions is sometimes wholly wanting. Before however this distribution can be rightly judged, a very large majority of the variations prima facie belonging to it must be cleared away. The causes of the irrelevance fall under two principal heads, inability to express Greek distinctions, and freedom of rendering. Where the variation lies between two approximately synonymous words, it is often impossible to say which it was that the author of a given version had before him. Such version cannot therefore be cited for either variant, and the necessary absence of a version from the side of the primary Greek MSS in an apparatus criticus leaves it undecided whether the Greek original of the version had or had not their reading. A similar uncertainty attends grammatical forms partially identical in meaning, such as the aorist and perfect of verbs; and also, though not in all cases, the presence or absence of the article. The ambiguity caused by freedom of rendering is sometimes not essentially different from the preceding cases, namely, where the genius of the translator's language would have rendered literal translation of one of the Greek readings unendurably stiff, or even impossible, and the most obvious rendering of it coincides with what would be a literal representation of the other Greek reading.
272. But, apart from this involuntary licence, most translators are liable to deviate from their original by slight verbal paraphrase in just the same way as transcribers of the fundamental text: in other words, many associations of versions with Greek evidence in support of changes of diction are due to accidental coincidence. Every paraphrastic impulse which affects a transcriber is not less likely to affect a translator, who has a strong additional temptation to indulge the impulse in the fact that he is creating a new set of words, not copying words set one after another before him. One of the commonest forms of paraphrase is a change of order; and a large proportion of the readings in which the primary Greek MSS stand alone differ from the rival readings in order only. How little reliance can be placed on the adverse testimony of versions in such a matter is indeed proved by the absence of Greek or any other authority for numberless scattered inversions of order, to be found in MSS of so literal a version as the Old Latin. Other changes of a paraphrastic kind, in which versions may have the appearance of supplying attestation in another language to similar Greek readings, but which doubtless were often in fact made by the translators and the Greek scribes independently, are the insertion of expletives, more especially pronouns (very liberally added as suffixes by Syriac translators), καί after οὕτως, and the like; the resolution or introduction of participial constructions; and permutations of conjunctions, and introductory language generally. In some of these cases a peculiarity of form in one Greek reading renders it probable that versions which attest it are faithfully reproducing their original, while it remains uncertain which original underlies any or all of the versions on the opposite side: in other cases either Greek reading might so easily be paraphrased by the other, either in Greek or in any other language, that no single version can be safely taken to represent exactly its original; though it is usually probable that some only of the versions have disguised their fundamental reading.
273. But, when allowance has been made for all these cases in which the apparent isolation of the primary Greek MSS is possibly or probably delusive, a certain number of variations remain in which the isolation must in the present state of our evidence be counted as unambiguous. For the reasons given above, the supposition that readings thus unattested by any version may yet be original is consistent with the known facts of transmission; and continuous examination of the readings attested by the primary Greek MSS without a version fails to detect any difference of internal character between them and readings in which the primary Greek MSS are sustained by versions. While therefore so narrow a range of attestation renders special caution imperative with respect to these readings, and some of them cannot be held certain enough to render all recognition of their rivals superfluous, we have found no sufficient reasons either for distrusting them generally or for rejecting any of them absolutely.
F. 274—279. Absence of Fathers from Groups containing Primary Greek MSS
274. The presence or absence of Fathers as allies of the primary Greek MSS is evidently to a great extent fortuitous, depending as it does so much on the nature of the passage, as causing it to be quoted often, seldom, or not at all. Except therefore in the comparatively few cases in which it is morally certain that a passage must have been quoted by one or more given Fathers in given contexts, had it stood with a particular reading in the text used by him or them, negative patristic evidence is of no force at all.
275. This universal rule is completely applicable to the variations which we are now considering, where neither variant is attested by any Father who does not habitually follow a Syrian text: it is applicable in principle, but subject to more or less qualification, where the reading opposed to that of the primary Greek MSS has patristic attestation not obviously Syrian, and their reading has none. The extent of its applicability must be affected by the usual character of the text of the Fathers who cite the passage. Almost all Greek Fathers after Eusebius have texts so deeply affected by mixture that their dissent, however clearly established, cannot at most count for more than the dissent of so many secondary Greek uncial MSS, inferior in most cases to the better sort of secondary uncial MSS now existing. The patristic evidence which can appreciably come into account must thus be limited to that of Ante-Nicene Fathers, and those very few later Fathers who used approximately Ante-Nicene texts.
276. But further, the apparent patristic evidence literally or virtually Ante-Nicene requires in its turn critical sifting. All the possible sources of error explained in former pages (§§ 156, 157) have to be kept constantly in mind; with the additional consideration that here we are dealing with detached variations, in which, except in the way of observation of analogies, we can obtain no corrective help from other variations. Positive grounds for distrusting the faithful transmission of a patristic attestation concordant with the Syrian text may very often be found, for instance in a recorded variation of MSS or in the clear implication of the context. Where this is the case, there is nothing arbitrary in ignoring the printed testimony, or even, if the evidence is strong enough, in reckoning it as favourable to the rival reading. Wherever a transcriber of a patristic treatise was copying a quotation differing from the text to which he was accustomed, he had virtually two originals before him, one present to his eyes, the other to his mind; and, if the difference struck him, he was not unlikely to treat the written exemplar as having blundered. But since the text familiar to nearly all transcribers after the earlier ages, to say nothing of editors, was assuredly the Syrian text, this doubleness of original could arise only where the true patristic reading was Non-Syrian. For the converse supposition there is no similar justification: for the only known causes that can be assigned for the appearance of a Non-Syrian reading in a patristic quotation are faithful transmission and accidental error; and where the reading is independently known to be of high antiquity, the chance of accidental coincidence in error is in an immense preponderance of cases too minute to come into account.
277. Even where there is no obvious positive internal ground for doubting whether the words written by a Father have been faithfully preserved, some slight uncertainty must always rest on a patristic attestation of a variant adopted by the Syrian text, since the supposed doubleness of original remains equally possible, and equally likely, whether the circumstances of the individual quotation do or do not happen to contain suspicious indications. This uncertainty ceases to be slight when the apparent position of the patristic testimony creates a grouping unlike any of the groupings into which it habitually enters, and when if transferred to the other side it would find itself in accustomed company.
278. Again, there is often reason to doubt whether what a Father wrote was identical with what he read: positive grounds may be found for distrusting a free quotation as faithfully representing the biblical text used, provided that the difference between one variant and another is such as might readily be reproduced accidentally by the free manner or the special purpose of the citation. Patristic quotations in short, like versions, may easily seem to make up a composite attestation, when it is really nothing more than an accidental coincidence. Such deceptive attestations might conceivably arise in either direction: but in a large majority of cases they would be due to a paraphrastic impulse such as that which we find working in scribes; that is, for either process the original peculiarities of order or diction which tempt to modification would be the same. In like manner the intermingling of unconscious reminiscences of parallel or similar passages, a specially fruitful cause of corruption in patristic quotations, may easily result in readings identical with readings due in MSS to harmonistic or other assimilation, and thus produce a deceptive semblance of joint attestation. Accordingly quotations apparently opposed to the primary Greek MSS are oftener found to be for these reasons questionable representatives of the texts used by the patristic writers than those which seem to support the primary Greek MSS. Suspicions as to fidelity of quotation, unsustained by other evidence, by the nature of the case can never transpose attestation from one side to the other; they can only create uncertainty: but uncertainty suffices to destroy the force of the prima facie contrast between the presence of patristic attestation on the one side and its absence on the other.
279. Lastly, even the presence of tried and verified Pre-Syrian patristic evidence in opposition to the primary Greek MSS, in conjunction with its absence from their side, loses much of the weight to which it would otherwise be entitled, when the actual texts employed in the extant writings of the Ante-Nicene Fathers are taken into consideration. Western readings, it will be remembered, are abundant in Clement and Origen, much more in Eusebius; and these are the only Ante-Nicene Fathers, represented to us by more than petty fragments, whose texts are not approximately Western. Now the readings of primary Greek MSS with which we are here concerned have opposed to them D in the Gospels and Acts, D2G3 in the Pauline Epistles and almost always other Western documents as well, making up a clear Western element in the attestation, whether the origin be 'Western' or not. If therefore even Clement or Origen swell the array, the source of their readings in these passages, as in many others where no doubt is possible, may be Western; and if so, they contribute nothing towards shewing that these readings were only preserved by the Western text, not originated by it. Nevertheless, since the greater part of the texts of the Alexandrian Fathers is Non-Western (see § 159), their certified opposition to a reading of the primary Greek MSS ought to forbid its unqualified acceptance except after the fullest consideration.
G. 280. Absence of Versions and Fathers from Groups containing Primary Greek MSS
280. We have spoken separately of the absence of Versions and of Fathers from the company of the primary Greek MSS: it remains to consider the rare and extreme cases in which Versions and Fathers are absent together. Independently of the special utility of versions and patristic quotations in supplying the landmarks of textual history their certified testimony has a high corroborative worth. The unknown Greek MSS from which they all derive their authority preceded our earliest extant MSS in several cases by long periods eventful in textual history, and thus at least rescue any reading of our MSS which they undoubtedly attest from the suspicion of having come into existence at any recent stage of transcription, in the century, we may say, preceding 350. This ancillary aid of Versions and Fathers in individual variations is invaluable, notwithstanding their unfitness to supply a primary and continuous standard of text as compared with our best Greek MSS. But, though the security of verification is withdrawn where Versions and Fathers are both absent, it by no means follows that a positive insecurity takes its place. Every version, so far as it is at present known to us, contains so many readings which it is morally impossible to believe to be right, and a certain proportion of these readings are scattered in such apparent irregularity, that we have no right to assume either that the deficiencies of one version, as the Memphitic, would in every case be made up by some other version, or that deficiencies of all versions and deficiencies of all extant patristic evidence would never happen to coincide. Moreover the transition to total absence of Versions and Fathers is bridged over by the many places in which a secondary version, as the Æthiopic or Armenian, supplies the only accessory authority. The whole number of cases where the primary Greek MSS stand alone is extremely small, when the deceptive variations mentioned above (§§ 271, 272), have been set aside: and neither in their internal character nor in their external relations to other documents have we found reason to deny to such readings the favourable presumption which their attestation by the better of the extant Greek MSS would confer.
SECTION II. DOCUMENTARY GROUPS AS LIMITED BY REFERENCE TO THE BEST PRIMARY GREEK MSS
281—355
A. 281—283. Relation of variations between Primary Greek MSS to the chief ancient texts
281. After this examination of the relation of the evidence of Versions and Fathers to that of the primary Greek MSS in respect of the final process of determining the text, we must now resume the consideration of the numerous variations in which the primary Greek MSS differ widely among themselves. Here, in investigating Internal Evidence of Groups for each individual group or class of groups, we lose clear and obvious parallelism with the great ancient texts. But the distribution of attestation for most of the groups must as a matter of fact have in most cases been determined by the great ancient texts, with or without subsequent mixture, whether it be in our power to assign each document to a definite text or not (see § 243 V); and therefore that cannot well be the right reading which would render the documentary distribution incompatible with known genealogies. It is not indeed requisite that we should be able to decide between two or more possible histories of a variation; but an important confirmation is wanting when we are unable to suggest at least one such history consistent alike with the composition of documents as known through the simpler and more normal distributions of attestation, and with the genuineness of the reading commended by Internal Evidence of Groups and other considerations. Before therefore we proceed to enquire into the character of special groups in detail, it will be right to examine a little more closely the probable relation of the primary ancient lines of transmission to many important variations now to be considered.
282. The principal difficulty with which we have to deal arises from an apparent combination of Western and Alexandrian attestations in opposition to a group of documents which bears no clear and obvious marks of compositeness of attestation, but which is commended by Internal Evidence of Groups; so that the preference accorded to this group seems to involve the paradox of a preference of a single line of descent to two concordant lines of descent. Given the independence of the Western and Alexandrian texts, the supposed preference is genealogically untenable as regards readings which could not owe their place in both texts to accidental coincidence in error. Now, though no contradiction is involved in the hypothesis of the adoption of early Alexandrian readings into a late Western text or of early Western readings into a late Alexandrian text, the actual evidence contains comparatively few traces of any such relation of dependence; while the definite original parallelism of the two texts is evinced by the many places in which they smooth away difficulties of language by entirely different devices. Either therefore (1) the readings of which we are now speaking as found only in the better of the primary Greek MSS must be of Alexandrian origin; or (2) they must have originated in some indeterminate equally aberrant text, assignation of them to a Western origin being in most cases clearly impossible; or (3) the opposed attestation cannot rightly be said to combine the two primary aberrant texts. 283. The two former suppositions stand in so flagrant opposition to the suggestions of internal evidence, howsoever obtained, and harmonise so ill with the results furnished by other groupings, that nothing but the proved inadmissibility of the third supposition could justify their acceptance. The third supposition is however natural enough, as soon as we recognise on the one hand the wide and early prevalence of Western readings, and on the other the mixed composition of the Greek MSS which are the chief extant representatives of the Alexandrian text (compare § 269). The Alexandrian text of the Gospels for instance would have been hopelessly obscure but for the very large Alexandrian elements which אCL(Δ) 33 contain in various places and proportions: yet the presence of a Western element in these MSS is equally indubitable, and it furnishes what must be in most cases the true key to the paradox. The readings attested by the best of the primary Greek MSS are as a rule simply Non-Western readings which are extant in an exceptionally small number of existing documents because the Western corruptions of them obtained an exceptionally early and wide popularity in one or other of the eclectic texts of the third and fourth centuries. That one of these eclectic texts arose at Alexandria, the text of Hesychius (see § 249) being indeed probably of this character, is likely enough; and, if so, it might be called a late Alexandrian text: but such a fact would only serve to illustrate the conclusion just stated. This conclusion harmonises in every respect with all known facts; and we are unable to think of any other interpretation which can be consistently applied without startling incongruities alike of external and of internal evidence. Β. 284—286. General relations of Β and א to other documents
284. When the various subordinate groupings which arise by the defection of one or another member of the leading groups of primary Greek MSS described as mainly Non-Western are tested by the prevalent character of their readings, the results thus obtained are for most of them as well marked as in the cases where the primary Greek MSS agree together. Two striking facts here successively come out with especial clearness. Every group containing both א and Β is found, where Internal Evidence is tolerably unambiguous, to have an apparently more original text than every opposed group containing neither; and every group containing B, with the exception of such Western groups as include Β in the Pauline Epistles, is found in a large preponderance of cases, though by no means universally, to have an apparently more original text than every opposed group containing א.
285. Thus Internal Evidence of Groups conducts us to conclusions respecting these two MSS analogous to, and confirmatory of, the conclusions obtained independently by ascertaining to what extent the principal extant documents severally represent the several ancient lines of text. We found א and Β to stand alone in their almost complete immunity from distinctive Syrian readings; א to stand far above all documents except Β in the proportion which the part of its text neither Western nor Alexandrian bears to the rest; and Β to stand far above א in its apparent freedom from either Western or Alexandrian readings with the partial exception in the Pauline Epistles already mentioned more than once (§§ 204 ff.).
286. The two processes deal with distinct classes of phenomena, the one with distributions of external attestation, the other with internal characteristics. The former simply registers in what company a given document is or is not found, with reference to certain well marked assemblages constantly recurring and having a conspicuously ancient origin: the latter deduces from those variations which on internal grounds afford clear presumptions the quality of the texts attested by the various groups into which a given document enters, and thus ultimately the quality of the document itself as a whole. The results of the former process are brought into comparison with those of the latter by a similar but independent deduction of the texts of the observed assemblages of documents. To a certain limited extent the materials in this case are identical with those employed in the latter process, for the various Syrian, Western, and Alexandrian assemblages are included among the numerous groups. But this partial coincidence does not materially impair the independence of the two processes, at least as regards any mixed or any approximately neutral document; for among the variations from which the character of, let us say, the Western text is deduced there will be found many in which each of the mixed documents now in question stands in opposition to the Western reading; and again many groupings, which by the ascertained quality of their texts go to shew the quality of a given document included in them all, are of too ambiguous composition to be used as evidence of the character of the Western or other assemblages. Thus the correspondence between the results of the two modes of investigating the groups containing א and B, and again those containing Β without א, is not created, as might be incautiously surmised, by a twofold presentation of inferences essentially the same, but amounts to a real verification. On the other hand the ascertainment of the quality of any single document by bringing together the ascertained qualities of the texts of the different groups of which it is a member is not essentially different from the direct ascertainment of its quality on internal grounds without intermediate reference to groups, except in its omission to take into account those variations in which the document stands absolutely alone.
C. 287—304. Relation of B to א and characteristics of Groups containing both Β and א
287. It now becomes necessary to scrutinise more closely the trustworthiness of the propositions laid down above respecting the preeminent excellence of the Vatican and Sinaitic MSS, which happen likewise to be the oldest extant Greek MSS of the New Testament. It is at the outset essential to distinguish carefully the readings and the groups of documents in which they stand side by side from those in which one of them stands alone. Following the gradual narrowing of groups, we come first to the combination אB, which is, as we have intimated, wherever it occurs, the constant element of those variable groups that are found to have habitually the best readings. The statement remains true, we believe, not less when the groups dwindle so as to leave אB comparatively or absolutely alone than when they are of larger compass. The cases in which אΒ have no support of Greek MSS, or no support at all, are connected by every gradation with the cases in which they stand at the head of a considerable group; and the principle is not affected by the size of the groups. But when the number of members is nearly or quite reduced to two, it is of consequence to find out what can be known respecting the antecedents of each, and especially respecting their mutual relations.
288. The first point that arises for examination is the independence of their testimony. The numerous readings in which they stand alone against all or nearly all extant Greek MSS suggests at once the enquiry whether they had separate ancestries or were, to a greater or less extent, copies of a single exemplar. The enquiry is the more necessary because the two MSS are really brought together as to their transcription in a singular manner by the fact observed by Tischendorf, that six leaves of the New Testament in א, together with the opening verses of the Apocalypse, besides corrections, headings, and in two cases subscriptions, to other parts, are from the hand of the same scribe that wrote the New Testament in B. The fact appears to be sufficiently established by concurrent peculiarities in the form of one letter, punctuation, avoidance of contractions, and some points of orthography. As the six leaves are found on computation to form three pairs of conjugate leaves, holding different places in three distant quires, it seems probable that they are new or clean copies of corresponding leaves executed by the scribe who wrote the rest of the New Testament, but so disfigured, either by an unusual number of corrections of clerical errors or from some unknown cause, that they appeared unworthy to be retained, and were therefore cancelled and transcribed by the 'corrector'. However this may be, their internal character of text differs in no respect from that of their neighbours. The fact that the scribe of Β was a 'corrector' of א shews that the two MSS were written in the same generation, probably in the same place: but as regards the text it has no independent force, though it would have to be taken into account if the internal evidence were to point to the use of a common exemplar. On the other hand a strong presumption to the contrary is created by remarkable differences in the order of the books, the divisions into sections, and other externals.
289. Turning then to the internal evidence afforded by the texts themselves, we are at once confronted by the question,—How can we know that any two MSS are both derived from a common parent or near ancestor? Certainly not, as is often assumed, from the bare fact that they have many readings in common, with or without the support of other documents. What is absolutely certain in these cases is that those readings have some common ancestor, coincidences in independent error being always excepted; and it is morally certain that the same ancestor supplied more or less of the rest of the text. But this ancestor may have been at any distance from the MSS, near or remote, back to the autograph itself inclusive. That this is no exaggeration will be seen at once by following the course of transmission downwards instead of upwards. Whenever an original reading has disappeared from all representatives of all originally independent lines of transmission except two, and each of these two lines has either but a single extant representative or has itself lost the true reading in all its extant representatives but one, the resulting distribution is precisely as supposed, two MSS against the rest: and this is a common case in many texts. To what stage in the transmission the common ancestor implied by the identical readings belonged, can in fact, so far as it can be determined at all, be determined only by the internal character of these readings, and by the genealogical relationships to other documents disclosed by these and the other readings.
290. As soon as the test furnished by the most elementary analysis of attestations, and consequently of genealogies, is applied, the supposition that the texts of א and Β as wholes are in any one book or chapter of the Testament derived from a single near ancestor falls to the ground. It is negatived at the first glance by the multitude of variations in which they are divided, while each is associated with a variety of attestation. Apart from the associated attestations the diversities of reading would be inconclusive: they might have been produced by the independent carelessness or licence of two transcribers of the same exemplar. But where each discrepant reading has other witnesses, and there is no room for accidental coincidence, the discrepancies in two transcripts of the same exemplar can have no other origin than mixture; that is, at least one of the transcripts must be virtually a transcript of two different originals. In this restricted sense alone is the hypothesis of a proximate common origin of א and Β worthy of being seriously examined; that is, in the sense that a single proximate original has supplied a large common element in their texts.
291. To examine the hypothesis in this shape, we must put out of sight all the elements of each MS which it owes to undoubted mixture with texts capable of being recognised through a long succession of variations, and which may therefore easily have come in together; that is, every clearly Western and every clearly Alexandrian reading of א in such books as are preserved in B, and every clearly Western reading of Β in the Pauline Epistles. The residue would then approximately represent each text reduced to the form which it must have had just before the great final independent mixture, upon the hypothesis that antecedent to this mixture the two texts had a common proximate origin. To make comparison clearer, we may further leave out of account every reading of either MS singly which has no other attestation whatever.
292. The resulting text however would still entirely fail to shew the imagined agreement. Multitudes of discrepancies between א and Β would remain, in which each MS would have some very early documentary evidence supporting it. Doubtless the hypothesis might still be rendered possible by supposing all the readings in which א and Β differ to have been taken simultaneously in one of these MSS from a single accessory original, or each MS to have its own accessory original. But the same conjectural mode of composition might be imagined with equal propriety for any other pair of MSS having at least an equal number of coincidences peculiar to themselves and no greater number of discrepancies. It is only one among an almost infinite number of at least equally probable contingencies, and has therefore no a priori probability of its own, though it would have no inherent improbability if other textual phenomena pointed to it. The problem cannot possibly be solved on the ground of attestation alone: but, so far as the phenomena of attestation contribute to its solution, they do not suggest a near common origin for even the residuary portions of א and B.
293. We now come to the indications furnished by the internal character of identical readings. If some of the identical readings are manifestly wrong, and if they further are of such a nature that accidental coincidence will not naturally account for their having the double attestation, they must have had a common original later than the autograph; and it becomes probable that some at least of those other identical readings which afford no clear internal evidence of the intrinsic kind had likewise only that later MS than the autograph for their common original. But this negative fact is all that we learn; and it is compatible with even the extreme supposition that the common source of the identical readings was the original of all extant documents, though itself but imperfectly representing the autograph, and thus that these readings, wrong though they be, were the ancestors of all other existing variants of the same variations (see §§ 86, 87). If on the other hand some of the wrong identical readings are manifestly derived from other existing readings, the common original must of course have been later than the common original of the other readings; but the question of its remoteness or proximateness to the two extant MSS remains undecided.
294. The only quite trustworthy evidence from internal character for derivation from a common proximate original consists in the presence of such erroneous identical readings as are evidently due to mere carelessness or caprice of individual scribes, and could not easily have escaped correction in passing through two or three transcriptions. To carry weight, they must of course be too many to be naturally accounted for by accidental coincidence of error in two independent scribes. Now, to the best of our belief, א and Β have in common but one such reading, if we set aside the itacisms, or permutations of vowels, current in uncial times, as between ο and ω, η and ει; including the confusion between ἡμεῖς and ὑμεῖς. This solitary blunder is παραλλαγὴ ἢ τροπῆς ἀποσκιάσματος for π. ἢ τ. ἀποσκίασμα in James i 17. The final -ατος might possibly be derived from an αὐτός which stands at the head of the next verse in a good cursive (40) and in two Syriac texts, and which has much intrinsic force: on this supposition the reading of א and B, though erroneous, would be nearer to the true reading than the common reading. But the evidence as a whole does not point to so deeply seated a corruption; and it may be fairly assumed that the reading -ατος is due either to thoughtless assimilation to the preceding genitive or to a mental separation of ἀπό from σκίασμα and consequent correction of the supposed solecism. But, though a series of such coincidences would imply community of proximate origin, a single instance does not, nor would two or three. Our extant MSS afford examples of more startling coincidences, unquestionably accidental, as σειροῖς ζόφοις (אA) for σειροῖς ζόφου in 2 Pet. ii 4, φθορᾶς φθαρτῆς (אAC) for σπορᾶς φθαρτῆς in 1 Pet. i 23, and ἐξίσταντο (א*C*D*) for ἐξίστατο, followed by Ἀκούσαντες δὲ οἱ ἀπόστολοι, in Acts viii 13, the subject of the verb being ὁ Σίμων. The coincident readings of א and Β likewise include one or two peculiar spellings having a somewhat problematical appearance: they occur however in peculiar words, in which it is difficult to find a trustworthy criterion of intrinsic certainty or even probability. They include likewise a few substantive readings which are capable of being accounted for as blunders, but which may as reasonably be admitted as genuine, and in most cases are sustained by internal evidence.
295. Thus far we have obtained only negative results. We have found readings that are explicable by the supposition of a common proximate original: we have found none that it is difficult to explain without it. We must now turn to such positive indications of the relative antiquity of the common original as can be obtained by taking genealogical relations into account. These are of two kinds, arising from comparisons in which the two MSS are taken together, and from those in which they are taken separately.
296. Under the former head we have to compare the readings in which א and Β together stand unsupported with those in which they have the concurrence of one or two important MSS or of ancient versions and quotations without extant MSS. Here we are merely reconsidering from a special point of view the evidence from which the enquiry started (§ 287), the Internal Evidence of Groups. Having found אB the constant element in various groups of every size, distinguished by internal excellence of readings, we found no less excellence in the readings in which they concur without other attestations of Greek MSS, or even of Versions or Fathers. The two sets of groupings, containing no reading in common, illustrate and confirm each other. The general character of the readings of both is the same, so that there is no internal evidence against the natural presumption that they come from the same source. But the readings of אB in which they are associated with other and various witnesses for very early texts cannot by the nature of the case have originated with the scribe of a proximate common source; so that, if the common source was proximate, they must have been received and transmitted from an earlier source: and accordingly there is no reason, in the absence of constraint from internal evidence, to imagine a different origin for those readings of אB which have no other attestation. It might indeed be suggested that both sets of readings were obtained from a single proximate common source, but that the one set originated there, while the other was transmitted. But against this contingent possibility must be set the comparative inconstancy of the members of the smaller groups containing אB, and the consequent probability that occasionally they would all be found ranged against readings having the same parentage as those which they elsewhere concur with אΒ in supporting (see § 280).
297. These considerations shew that the common original of אΒ for by far the greater part of their identical readings, whatever may have been its own date, had a very ancient and very pure text, and that there is no sufficient reason for surmising that the rest of their identical readings came from any other source. They prove that one of three alternatives must be true: either the respective ancestries of א and Β must have diverged from a common parent extremely near the apostolic autographs; or, if their concordant readings were really derived from a single not remote MS, that MS must itself have been of the very highest antiquity; or, lastly, such single not remote MS must have inherited its text from an ancestry which at each of its stages had enjoyed a singular immunity from corruption. For practical purposes it is of little moment which alternative is true. The second and third alternatives would leave open the possibility that single readings of אB, otherwise unsupported, may have originated with the common proximate source here implied: but there is no difference between the three alternatives as regards the general character and date of the readings taken together, and the consequent presumption in favour of any one of them.
298. When however we go on, secondly, to compare the identical readings of אB with the readings of א unsupported by Β and of Β unsupported by א, the first alternative obtains so much positive corroboration that the second and third may be safely dismissed. For the present purpose we must neglect the numerous readings in which א or Β forms part of a large group, and attend to those readings only in which they stand respectively in opposition to all or almost all other Greek MSS, but with some other support: with the places where they stand absolutely alone we are not for the present concerned. It is then seen that a large proportion of the small groups containing one or other of the two MSS contain also other documents (versions or quotations) attesting a high antiquity of text. Many of the readings of Β having this accessory attestation are doubtless wrong, and, as we shall see presently, a much greater number of the readings of א: what we are now concerned with however is not genuineness but antiquity. Each of the two MSS is proved by these readings to be at least in part derived from an original preserving an extremely ancient text, for the most part not represented by our other extant MSS: and these two texts are by the nature of the case different from each other.
299. The distinct existence of these two independent texts is further illustrated by places where they emerge into view simultaneously; that is, in a certain number of those ternary or yet more composite variations in which the readings of א and of Β are different from each other, but are closely connected together in opposition to the reading or readings of the great bulk of documents, and in which each of the two MSS is supported by a small number of documents having a largely Pre-Syrian text. In these cases, allowance being made for the possibility of an occasional accidental coincidence, the reading of neither א nor Β can have originated in the process of transcription from a proximate common source, and the two MSS confront each other with exclusively early texts of different ancestry.
300. It follows from the binary and the ternary variations alike that the hypothesis of a proximate common original for the identical readings of אB involves the necessity of postulating at least three independent sources of exceptionally ancient character of text for the two MSS, independently of sources akin to documents still largely extant. It is at once obvious that the same phenomena are accounted for with much greater probability by the simple explanation that the identical readings do not represent a third and proximate common original, containing a single pure text preserved with extraordinary fidelity, but are merely those portions of text in which two primitive and entirely separate lines of transmission had not come to differ from each other through independent corruption in the one or the other.
301. The importance of this conclusion is so great that we venture to repeat in other and fewer words the principal steps which lead to it. Whatever be the mutual relation of א and B, each of them separately, א in the Apocalypse excepted, is found on comparison of its characteristic readings with those of other documentary authorities of approximately determinate date to have a text more ancient by a long interval than that of any other extant Non-Western MS containing more than a few verses; to be in fact essentially a text of the second or early third century. This fact, which is independent of coincidences of אΒ, so that it would remain true of א if Β were unknown, and of Β if א were unknown, suggests the most natural explanation of their coincidences. They are due, that is, to the extreme and as it were primordial antiquity of the common original from which the ancestries of the two MSS have diverged, the date of which cannot be later than the early part of the second century, and may well be yet earlier. So high an antiquity would of course be impossible if it were necessary to suppose that the 'common original' was a single archetypal MS comprising all the books as they now stand in either existing MS. But, as has been noticed elsewhere (§ 14: see also § 352), there is reason to suspect that the great MSS of the Christian empire were directly or indirectly transcribed from smaller exemplars which contained only portions of the New Testament; so that the general term 'common original', which we have used for the sake of simplicity, must in strictness be understood to denote the several common originals of the different books or groups of books. There is however no clear difference of character in the fundamental text common to Β and א in any part of the New Testament in which Β is not defective. The textual phenomena which we find when we compare them singly and jointly with other documents are throughout precisely those which would present themselves in representatives of two separate lines diverging from a point near the autographs, and not coming into contact subsequently. Other relations of pedigree are doubtless theoretically possible, but involve improbable combinations.
302. An answer, in our opinion a true and sufficient answer, is thus found to the question how far the testimonies of א and Β are independent of each other. Their independence can be carried back so far that their concordant testimony may be treated as equivalent to that of a MS older than א and Β themselves by at least two centuries, probably by a generation or two more. Here, as always, high relative and absolute antiquity supplies a strong presumption of purity, but cannot guarantee it: on the one hand the writings of the New Testament were liable to textual change in the earliest generations of their existence as well as a little later; on the other the close approach to the time of the autographs raises the presumption of purity to an unusual strength. It must be remembered however that part of the evidence with which we have been dealing relates to quality as well as to antiquity: Internal Evidence of Groups, independently of the aid which it gives towards ascertaining the proximity or distance of the common original of א and B, retains its own direct value. As was pointed out above (§ 296), even if it were credible that they were divided from their common ancestor by no more than two or three transcriptions, we should have on this ground to ascribe to the ancestry of the common ancestor an extraordinary freedom from corruption.
303. That absolute purity cannot be ascribed to all readings attested by אB is implied in the existence of the Western non-interpolations (§ 240). We shall presently have to notice the possibility of a concurrence of א and Β in support of wrong Western readings in St Paul's Epistles, implying a departure in the ancestries of both from their common fundamental text; and this is perhaps the most natural explanation of the attestation of the unquestionably wrong reading ἦλθεν for ἦλθον by אBD2G3 cu2 Orig in Gal. ii 12. Account must likewise be taken of the places in which, without difference of reading between א and B, the true text appears to be lost in all existing documents, or in all but one or two of a subsidiary character. Besides these clear or possible errors in אB there are some few variations in which their joint reading, though supported by some other testimony, is subject to more or less of doubt. But we have not found reason to make any further deduction from their united authority. In this as in all similar cases no account of course can be taken of coincidences that might be easily due to the independent origination of the same error by two different scribes. Under this head preeminently fall identical changes of an itacistic kind, as the confusion between imperatives in -ε and infinitives in -αι, and also between ἡμεῖς and ὑμεῖς: it seldom happens that both MSS go unquestionably astray together in such points, for their laxity is but comparative, but examples do occur. When these indecisive coincidences have been set aside, no readings of אB remain which we could venture to pronounce certainly or probably wrong as against other existing readings. This general immunity from substantive errors that can without room for doubt be recognised as errors in the common original of אB, in conjunction with its very high antiquity, provides in a multitude of places a safe criterion of genuineness, not to be distrusted except on very clear internal evidence. Accordingly, with the exceptions mentioned above, it is our belief (1) that readings of אB should be accepted as the true readings until strong internal evidence is found to the contrary, and (2) that no readings of אB can safely be rejected absolutely, though it is sometimes right to place them only on an alternative footing, especially where they receive no support from Versions or Fathers. 304. Sufficient examples of important or interesting readings attested by אB, but lost from the texts of all other extant uncials, will be found in the Appendix, as in the notes on Matt. ν 22; x 3; xi 19; xvi 21; xvii 20; xxviii 6; Mark ix 29; xvi 9—20; Acts xx. 5, 28; 1 Pet. ν 2; Eph. i 1. Two or three additional places may be noticed here, in which there is reason to think that the bearing of the internal evidence is liable to be misunderstood.
Mark iv 8 καὶ ἄλλα ἔπεσεν κ. τ. λ., καὶ ἐδίδου καρπὸν ἀναβαίνοντα καὶ αὐξανόμενα אΒ (αὐξανόμενον ADLΔ cu1, αὐξάνοντα C and most documents). Here the true force of the parable requires that not the fruit, but the plants into which the seeds have expanded, be said to mount up and grow. The temptations to corruption were peculiarly strong; ἀναβαίνοντα, immediately following καρπόν, had an ambiguous termination readily assumed to belong to the masculine accusative, and thus drew after it the other participle, one text adopting the middle form, which involved least change, the other the neuter form, which coincided with ἀναβαίνοντα: an additional motive for alteration would be the apparent paradox of seeds being said to 'mount up', a paradox which St Mark apparently intended to soften by means of the order of words. Finally the Western and Syrian texts completed the corruption by changing ἄλλα to the ἄλλο of vv. 5, 7.
John iv 15 ἵνα μὴ διψῶ μηδὲ διέρχωμαι (or -ομαι) ἐνθάδε ἀντλεῖν א*Β Orig5 (ἔρχωμαι most documents). Διέρχομαι is here used in its idiomatic sense 'come all the way', which expresses the woman's sense of her often repeated toil. Being commonly used in other senses, the word was easily misunderstood and assumed to be inappropriate; and the change would be helped by the facility with which one of two similar consecutive syllables drops out.
Acts xxviii 13 καταχθέντες εἰς Συρακούσας ἐπεμείναμεν ἡμέρας τρεῖς ὅθεν περιελόντες κατηντήσαμεν εἰς Ῥήγιον א*B g (tulimus et [='weighed anchor,' as vg cum sustulissent de Asso for ἄραντες ἆσσον in xxvii 13]) memph ('going forth'); where most documents have περιελθόντες. Περιελόντες here is explained by the use of the same verb in xxvii 40, καὶ τὰς ἀγκύρας περιελόντες εἴων εἰς τὴν θάλασσαν, where it clearly means the casting loose (literally 'stripping off') of the anchors (with their cables) in order to set the vessel free to drive, though it is otherwise unknown as a nautical term. By analogy it must here mean the casting loose of the cables which attached the vessel to the shore in harbour (called in ampler phrase τὰ ἀπόγεια λύσασθαι, λῦσαι, ἀποκόψαι &c.), the elliptic employment of transitive verbs being common in Greek nautical language as in English (compare ἄραντες in xxvii 13, cited above). The general sense then is merely 'and loosing from thence', that is, from Syracuse, where there had been a stay of three days. On the other hand the run from Syracuse to Rhegium could never be described as circuitous (περιελθόντες), unless the ship were thrown out of her course by contrary winds, a circumstance not likely to be noticed by means of an obscure implication (cf. xxvii 4, 7, 8); while scribes, to whom this geographical difficulty was not likely to suggest itself, would be tempted by the superficial smoothness of περιελθόντες.
D. 305—307. Binary uncial combinations containing Β and א respectively
305. We come next to the variations in which א and Β stand on different sides. The first step towards dealing successfully with the problems which here arise is to examine the internal character of the readings attested by the two series of binary groups formed by א and by Β combined with each other primary Greek MS. Now every such binary group containing Β is found by this process to offer a large proportion of readings which on the closest scrutiny have the ring of genuineness, while it is difficult to find any readings so attested which look suspicious after full consideration. Such groups are in the Gospels BL, BC, BT, ΒΞ, BD, AB, BZ, Β 33, in St Mark ΒΔ; in the Acts AB, BC, BD, BE2, Β 61; in the Catholic Epistles AB, BC, BP2; in the Pauline Epistles AB, BC, BM2, (BP2) Β 17, Β 67**. These readings are in fact for most of the groups, especially those belonging to the Gospels, hardly of less uniformly good character than the readings of אB. Once more, their character is not found appreciably different whether they do or do not receive the support of Versions or Fathers.
306. One binary group containing Β requires separate mention, namely BD2 of the Pauline Epistles. From what has been already said (§§ 204, 228) on the Western element of Β in these Epistles it will be evident that the combinations BD2G3 and BG3, when they are unsustained by clear Non-Western Pre-Syrian attestation, may be taken to imply a Western reading. The question thus arises whether the same is to be said of BD2. On the one hand D represents on the whole an earlier and purer form of the Western text than G3, so that, were not Β known to contain a Western element in these epistles; the combination BD2 would, like the BD of the Gospels and Acts, have a strong presumption in its favour; and the presumption, though weakened, is by no means destroyed by the contingency which has thus to be taken into account. On the other hand D2 has some clearly Western corruptions from which G3 is free; and the analogy of BD2G3 and BG3 preclude any assumption that BD2 could not have this character. The decision must accordingly rest with Internal Evidence, which is on the whole definitely favourable to the BD2 readings, while some of them are not free from doubt. They cannot as a class be condemned with the readings of BD2G3 and BG3; but neither is it certain that none of them are of the same origin and quality. Since the inferior quality of BG3 and the ambiguity as to BD2 are explained by the exceptional intrusion of an alien element into the Pauline text of B, the existence of which alien element is ascertained independently of the quality of its readings, the character of the fundamental text of B, as shown by the other binary combinations, evidently remains unaffected.
307. When א is tested in like manner, the results are quite different. None of its binary combinations, if their readings are examined consecutively, are found to be habitually of good character, though here and there readings occur which are not to be hastily dismissed. The readings of אD in the Gospels and Acts are often interesting, but they are shown by the Versions and Fathers which usually support them to be simply Western: the character of אD with the Old Latin, of א with the Old Latin, and of D with the Old Latin is identical. Except in the peculiar Western non-interpolations we have never found reason to trust אD. It is worth mention here that much the most considerable deduction to be made from the superiority of text in Tischendorf's editio octava to his earlier editions is due to the indiscriminate vagueness of his estimate of א: a large proportion of those readings adopted by him which we have been obliged to reject are ordinary Western readings which are attested by א in consequence of the Western element which it contains. With אD of the Gospels may be classed אG3 of the Pauline Epistles; while the rarer combination אD2 of the Pauline Epistles contains both bad and good readings, the latter being apparently confined to the parts where Β is defective, and elsewhere to those variations in which the reading of Β is that of its Western element peculiar to these books, so that in the absence of this element we might have expected אBD2 in place of אD2. Trial by Internal Evidence is likewise unfavourable to such groups as in the Gospels אL, אC, אT, אΞ, אZ, א 33, in St Mark אΔ; in the Acts אA, אC, אE2, א 61; in the Catholic Epistles אA, אC, אP2; in the Pauline Epistles אA, אC, (אP2,) (א 17; though they contain a few readings which may perhaps be genuine. Their pedigree is usually, we believe, perhaps almost always, Alexandrian. The character is here, as elsewhere, ascertained independently of the origin: but it is instructive to see how completely the results of the comparison of binary groups containing א and Β respectively are explained by the presence of large Western and Alexandrian elements in א. The character of what remains of the text of א after their subtraction must be largely excellent, as the character of אB shews; an estimate of the degree of excellence cannot however be formed till we have taken another step.
E. 308—325. Singular and subsingular readings of B
308. The readings of Β and of א respectively have now to be compared in those variations in which they stand unsustained by any other Greek uncial MS. Such readings are of two kinds, 'singular readings', as they are usually called, which have no other direct attestation whatever, and what may be called 'subsingular readings', which have only secondary support, namely, that of inferior Greek MSS, of Versions, or of Fathers, or of combinations of documentary authorities of these kinds. Subsingular readings of B, which are in fact the readings of a particular class of groups containing B, will require consideration presently. What we have to say on the singular readings of Β may be made clearer by a few remarks on singular readings generally.
309. The attention prima facie due to singular readings of any one document is evidently variable, according to the number and genealogical relations of the whole body of extant documents. If a text is preserved in but two documents, every reading of each where they differ is a singular reading, one or other of which must be right; unless indeed both are wrong, and the true reading has perished. If the documents are more numerous, the singular readings of one document have no less prima facie authority than the rival readings found in all other documents alike, provided that the other documents have had a common original (see § 52), making the readings common to them to be virtually, though not in appearance, as 'singular' as the others. The same principle holds good whatever be the total number of documents, unless they have all only one common ancestor; that is, the prima facie authority of the singular readings of any document cannot be estimated by the bare numerical relation (see §§ 54—57), but varies partly with the independence of ancestry of the one document in relation to all the rest, partly with the affinities of ancestry among the rest. Where the whole pedigree is very complex, as in the New Testament, any documents which frequently stand in very small groups attesting evidently genuine readings, against the bulk of documents of various ages, must evidently contain so large elements having an independent ancestry that the a priori presumption against their singular readings cannot be much greater than against singular readings at their best, that is, in texts preserved in two documents only.
310. On the other hand (see §§ 56, 58) the singular readings of a document may always be due either to inheritance from a more or less remote ancestry, which may be of any degree of purity, or to quite recent corruption, or, which is much the commonest case, partly to the one, partly to the other. Whatever a document has inherited of the autograph text is of necessity included in its proper or ancestral text; and in order to ascertain the character of those of its singular readings which belong to its ancestral text, we must sift away as far as possible those other singular readings which are mere individualisms, so to speak, originating with the scribe or one of his immediate predecessors. Complete discrimination is of course impossible in the absence of the exemplar or exemplars; but every approximation to it is a gain. Except by conjecture, which does not concern us here, no scribe can make a text better than he found it; his highest merit is to leave it no worse. The inherited text of a document must therefore have been usually better, never worse, than the text which it actually presents to the eye; and the character of the inherited text is inevitably disguised for the worse by every 'individualism' which remains undetected.
311. Individualisms may obviously belong to various types, from purely clerical errors to alterations of purely mental origin. Sufficient clerical errors betray themselves, beyond the possibility of doubt, to enable us with a little care to form an estimate of the degree of general accuracy attained by the scribe of a given document, and also of the kinds of mistakes to which he was prone (see § 45). The mere subtraction of a large number of irrelevant readings from the gross list of singular readings gives, as we have seen, greater exactness to the appreciation of the character of the ancestral text. But moreover the further knowledge gained respecting the habits of the scribe becomes of use both positively and negatively in dealing at a later stage with individual variations. Singular readings which make good sense and therefore need imply no clerical error, but which might also be easily explained as due to a kind of clerical error already fixed upon the scribe by undoubted examples, are rendered by the presence of possible clerical error as a vera causa more doubtful than they would otherwise be. Singular readings which make good sense, and which cannot be explained by clerical error except such as lies outside the known proclivities of the scribe, acquire a better title to consideration. Again, those singular readings which are evidently errors, but are not clerical errors, can likewise be classified, and the results of classification used in the same manner: for instance, in the New Testament an appreciable number of the singular readings of A consist in the permutation of synonyms, and it can hardly be doubted that these readings are true individualisms. Whether however such singular readings are individualisms or of older date, is often not easy to tell: but it is always useful to remember that the text of a document as it stands is partly ancestral, partly due to transcriptional error in the last stage or stages of transmission, though definite indications of the one or the other origin may be wanting for each individual variation.
312. When the singular readings of Β are examined for the purpose here explained, it is found that on the one hand the scribe reached by no means a high standard of accuracy, and on the other his slips are not proportionally numerous or bad. Like most transcribers, he occasionally omits necessary portions of text because his eye returned to the exemplar at the wrong place. As the longer portions of text so omitted consist usually either of 12 to 14 letters or of multiples of the same, his exemplar was doubtless written in lines of this length. Often, but not always, an obvious cause of omission may be found in homoeoteleuton, the beginning or ending of consecutive portions of text with the same combinations of letters or of words. Reduplications due to the same cause likewise occur, but more rarely. More characteristic than these commonest of lapses is a tendency to double a single short word, syllable, or letter, or to drop one of two similar consecutive short words, syllables, or letters. The following are examples: Mark ix 25 ⲉⲅⲱⲉⲅⲱⲉⲡⲓⲧⲁⲥⲥⲱ for ⲉⲅⲱⲉⲡⲓⲧⲁⲥⲥⲱ; Acts xviii 17 ⲧⲟⲩⲧⲱⲛⲧⲱⲛⲧⲱ for ⲧⲟⲩⲧⲱⲛⲧⲱ; Mark xiii 13 ⲉⲓⲥⲥⲧⲉⲗⲟⲥ for ⲉⲓⲥⲧⲉⲗⲟⲥ; John xiv 10 ⲁⲉⲅⲱ for ⲁⲉⲅⲱⲗⲉⲅⲱ; Luke vii 24 ⲥⲁⲗⲉⲩⲟⲙⲉⲛ for ⲥⲁⲗⲉⲩⲟⲙⲉⲛⲟⲛ; Mark iii 5 ⲗⲉⲓ for ⲗⲉⲅⲉⲓ; vi 22 ⲉⲓⲉⲗⲑⲟⲩⲥⲏⲥ for ⲉⲓⲥⲉⲗⲑⲟⲩⲥⲏⲥ; vii 21 ⲇⲓⲗⲟⲅⲓⲥⲙⲟⲓ for ⲇⲓⲁⲗⲟⲅⲓⲥⲙⲟⲓ; also without similarity of form, Mark vi 1 ⲉⲝⲏⲑⲉⲛ for ⲉⲝⲏⲗⲑⲉⲛ; vii 18 ⲁⲥⲩⲛⲧⲟⲓ for ⲁⲥⲩⲛⲉⲧⲟⲓ. Occasionally we find assimilations of ending, as Mark ν 38 αλαλαζοντας πολλας (for πολλα); Rom. xiv 18 δοκιμοις τοις ανθρωτποις (for δοκιμος); or even, but very rarely, such verbal assimilations as κήρυγμα ὃ ἐκήρυξεν in Acts x 37 for βάπτισμα ὃ ἐκήρυξεν.
313. The singular readings of Β which cannot strictly be called clerical errors, and yet which appear to be individualisms of the scribe, are confined within still narrower limits. A current supposition, to which frequent repetition has given a kind of authority, that the scribe of Β was peculiarly addicted to arbitrary omissions, we believe to be entirely unfounded, except possibly in the very limited sense explained below, while the facts which have given it plausibility are everywhere conspicuous. In the New Testament, as in almost all prose writings which have been much copied, corruptions by interpolation are many times more numerous than corruptions by omission. When therefore a text of late and degenerate type, such as is the Received Text of the New Testament, is consciously or unconsciously taken as a standard, any document belonging to a purer stage of the text must by the nature of the case have the appearance of being guilty of omissions; and the nearer the document stands to the autograph, the more numerous must be the omissions laid to its charge. If Β is preeminently free from interpolations, Western, Alexandrian, or Syrian, it cannot but be preeminently full of what may relatively to the Received Text be called omissions. Strictly speaking, these facts have no bearing on either the merits or the demerits of the scribe of B, except as regards the absolutely singular readings of B, together with those nearly singular readings in which the other attestation may easily be due to accidental coincidence: multitudes of the so called omissions of Β are found in other good documents, few or many, and therefore, if not genuine, must at least have originated at a point in the line of transmission antecedent to B. It has seemed best however to speak of the supposed omissions of Β here once for all, both those which concern the character of Β individually and those which concern the character of the older text or texts from which it was derived.
314. The great mass of omissions, or rather for the most part non-interpolations, which Β shares with other primary documents being set aside as irrelevant, it remains to be considered whether its singular readings, which alone are relevant, include such and so many omissions as to indicate a characteristic habit of the scribe. It is a conceivable hypothesis that the scribe of B, besides inheriting a text unusually free from interpolations, was one of the very few transcribers addicted to curtailment, and thus corrupted the inherited text in a direction opposite to the usual course of transcription: the question is whether such a hypothesis is borne out by a comprehensive examination of the facts. What has been said above (§ 312) as to omissions due to purely clerical error need not be repeated. The only readings of Β which can with any plausibility be urged on behalf of the hypothesis are the instances in which it omits slight and apparently non-essential words found in all other documents, such as pronouns and articles. It is on the one hand to be remembered that such words are peculiarly liable to be inserted, especially in Versions and quotations by Fathers; and still more that we find numerous similar omissions in good groups containing B, with every gradation in the amount of support which it receives, so that these omissions in Β alone might be taken as genuine non-interpolations without incongruity as to the attestation, as well as consistently with the general character of the text of B. In our opinion this is the most probable account of the matter in some cases, and possibly in all: but it is on the whole safer for the present to allow for a proneness on the part of the scribe of Β to drop petty words not evidently required by the sense, and therefore to neglect this class of omissions in Β alone, where good confirmatory external or internal evidence is wanting. If however a like scrutiny is applied to important words or clauses, such as are sometimes dropped in the Western texts for the sake of apparent directness or simplicity, we find no traces whatever of a similar tendency in B. Omissions due to clerical error, and especially to homoeoteleuton, naturally take place sometimes without destruction of sense: and all the analogies suggest that this is the real cause of the very few substantial omissions in Β which could possibly be referred to a love of abbreviation. As far as readings of any interest are concerned, we believe the text of Β to be as free from curtailment as that of any other important document.
315. The chief feature of the few remaining individualisms of B, so far as they can be recognised with fair certainty as such, is their simple and inartificial character. Nearly all of them are due to easy assimilation, chiefly between neighbouring clauses or verses, occasionally between parallel passages. Consecutive words are perhaps occasionally transposed: but here on the other hand account has to be taken of the peculiar habitual purity of the text of Β in respect of the order of words; a purity which is specially exhibited in numerous ternary or more composite variations, in which Β is the sole or almost the sole authority for the one collocation which will account for the other variants. Of paraphrastic change there is little or nothing. The final impression produced by a review of all the trustworthy signs is of a patient and rather dull or mechanical type of transcription, subject now and then to the ordinary lapses which come from flagging watchfulness, but happily guiltless of ingenuity or other untimely activity of brain, and indeed unaffected by mental influences except of the most limited and unconscious kind.
316. This examination of the tolerably certain individualisms of B, of all kinds, prepares the way for an examination of the character of its remaining singular readings. We must first however consider the readings of a set of groups intermediate between those last considered (§§ 281—304) and B, that is, what we have called the subsingular readings of B. When the groups formed by Β with one or more secondary Greek MSS and with one or more Versions or Fathers are tried by Internal Evidence, the proportional number of readings which are to all appearance genuine is very large indeed. Readings so attested cannot in fact be well distinguished in character from readings of אB. When Β stands supported by only a single version, the results are by no means so uniform. When it is followed only by the Old Latin, or one or more Old Latin MSS or Fathers, the readings seldom commend themselves as worthy of unreserved confidence, though it is no less true that they are seldom manifestly wrong (see § 204): they may as a rule be strictly called doubtful readings. On the other hand when the associated version is the Memphitic, Thebaic, or Old Syriac, the presumption of genuineness raised by the habitual character of the readings is much greater, and not a few of them are almost certainly right. With other versions the combinations are various in quality, as might be expected from the mixed origin of the versions themselves and their present condition as edited.
317. These diminutions of attestation lead us continuously to the singular readings proper. Here too so many readings of Β by itself commend themselves on their own merits that it would be rash to reject any hastily, though undoubtedly not a few have to be rejected at last. Occasionally too some stray quotation of a Father shews that readings of Β which might have been thought to be individualisms were really at least several generations older than the age when Β was written. Thus in 1 Cor. xiii 5 it has τὸ μὴ ἑαυτῆς with Clem. Paed. 252 for τὰ ἑαυτῆς, retained by Clem. Strom. 956; both readings being shown by the respective contexts to have been actually used by Clement, and both making excellent sense. But, wherever there is no such accessory authority, clear internal evidence is needed to justify the acceptance of singular readings of B, since the possibility that they are no more than individualisms is constantly present.
318. The special excellence of Β displays itself best perhaps in ternary or more than ternary variations. This has been already noticed (§ 315) in reference to collocations of words; but the statement is equally true as regards readings of all kinds. Where the documents fall into more than two arrays, the readings of Β are usually found to be such as will account for the rival readings, and such as cannot easily be derived from any one of them, or any combination of them. Not the least instructive are what may be termed composite ternary variations, which easily escape notice in the cursory use of an ordinary apparatus criticus. They arise when two independent aberrant texts have removed a stumbling-block due to the original form of a phrase or sentence by altering different parts of the phrase, not by altering the whole or the same part in a different manner. If, as is usual, the evidence affecting each alteration is presented separately, we have in form not a single ternary variation but two or more successive binary variations. Now in such cases it is of frequent occurrence to find Β nearly or even quite alone in supporting what is evidently the genuine variant in each binary variation, while most of the other documents, representing ancient as well as later texts, divide themselves into those which are right in one place and those which are right in another.
319. If it is suggested that these phenomena might be due to a skilful selection and combination of readings from two sources by the scribe of B, the hypothesis is decisively negatived by several considerations. If it were true for composite variations, it should fit also the ternary variations of the more obvious type, in which Β similarly supports the neutral reading; whereas in most of them it would be peculiarly difficult to derive the neutral reading from any kind of coalescence of the aberrant readings. Secondly, the process hypothetically attributed to the scribe of Β is incongruous with all that is known of his manner of transcription and capacity of criticism. Thirdly, the ternary variations in which Β stands absolutely alone are not separable in character from those in which its readings are 'subsingular', having the support of, for instance, one or two early versions; and thus the operation would have to be attributed to one or more scribes of the first or early second century, while it would demand a degree of skill of which we have no example in extant records. Fourthly, the hypothesis is distinctly condemned by transcriptional evidence, which has an exceptional force in ternary variations (see § 29).
320. It should be noticed that some few variations in the Pauline Epistles, in which the local Western element of Β has affected the text, present a deceptive appearance of exceptions to what has been stated. Thus the accessory Western text, which makes itself felt in simple conflations (Col. i 12 καλέσαντι καὶ ἱκανώσαντι Β from ἱκανώσαντι and the Western καλέσαντι, 2 Thess. iii 4 καὶ ἐποιήσατε καὶ ποιεῖτε καὶ ποιήσετε Β from [καὶ] ποιεῖτε καὶ ποιήσετε and the Western καὶ ἐποιήσατε καὶ ποιεῖτε), is but partially followed in the composite ternary variation of Rom. x 5. Here the scribe of Β adopted two out of three closely connected Western (and subsequently Syrian) changes, the transposition of ὅτι and the insertion of αὐτά after ποιήσας, but in the third place negligently left αὐτῇ untouched, doubtless the reading of his primary exemplar, and thus produced an impossible combination. Combinations like these imply imperfect workmanship, not skilful choice. Nor is it material to know whether the scribe of Β himself took the Western readings from a second exemplar, or, as seems more likely, merely copied a single exemplar with marginal or interlinear corrections which he incorporated into the text (see §§ 335 ff.): the essential nature of the process is not changed by its being carried a single step back. Except in so far as even the slightest mixture may be said to involve some kind of selection, we hold it to be certain that the readings of Β are never the result of any eclectic process. Its occasional individual aberrations of course sometimes take place where there is variation already, and therefore sometimes go to make up ternary variations. But it remains true that the readings of Β in ternary variations, simple or composite, are habitually those of the original text, and the readings of the other texts divergent attempts to amend it.
321. What has been said on the excellence usually shown by the readings of Β in ternary variations will be made more intelligible by two or three examples of different types.
James v 7 ἰδοὺ ὁ γεωργὸς ἐκδέχεται τὸν τίμιον καρπὸν τῆς γῆς, μακροθυμών ἐπ’ αὐτῷ ἕως λάβῃ πρόϊμον καὶ ὄψιμον Β (? 31) lat.vg the (? aeth) arm. One text supplies the concluding adjectives with καρπόν (from the first clause) as a substantive (א 9 f me syr.hl.mg pp, with slight variations), another, the Syrian, with ὑeτόv (AK2L2P2 cupl syr. vg-hl.txt ppser). Here the elliptic expression has manifestly given rise to two different corrections; and Β is the only certain Greek authority for the true text. This is an example of the simplest and most fundamental form of ternary readings, with the neutral text clearly exhibited.
322. Mark vi 43 καὶ ἦραν κλάσματα δώδεκα κοφίνων πληρώματα Β. The easier κλασμάτων of viii 20 (πόσων σφυρίδων πληρώματα κλασμάτων ἤρατε, where the necessary order enforces the genitive) is adopted by א 13-69-124-346 209 (1 omits). The Western (and Syrian) text, starting from this last reading, borrows κοφίνους πλήρεις, to replace the last two words, from viii 19; Matt. xiv 20 (AD unc11 cupl latt syrr me); most Latins, with 33 and some secondary Greek MSS, introducing further assimilations to Matt. There are also two remarkable conflations: LΔ vary from Β only by adopting κοφίνους from the Western reading (or the antecedent parallel passages); 28, which has many relics of a very ancient text hereabouts, retains the κλάσματα of Β, but for the rest follows the Western and Syrian text. Here the choice clearly lies between three readings, those of B, of א and the lost early originals of two texts now partially preserved in cursives, and of LΔ; and the difficulty of accounting for the well attested κλάσματα is unfavourable to the second. The reading of LΔ, κλάσματα δώδεκα κοφίνους πληρώματα, which has no intrinsic probability, may be due to accidental mixture (in v. 31 they, and they alone, have the impossible εὐκαίρου): the reading of B, which has much intrinsic probability, was likely to be changed on account of the double accusative, even apart from the influence of parallel passages, and might easily give rise to all the other variants with the help of harmonistic assimilation. If we take the three parts of the composite variation separately, a good group is found supporting each of the three readings of Β; κλάσματα being attested by ΒLΔ 28, κοφίνων by אB 1-209 13-69-124-346, and πληρώματα by אBLΔ 1-209 13-69-124-346. This last specially certain attestation marks the virtual authority for the entire fundamental text from which the Western correction departed, the peculiar word πληρώματα being the turning-point of change; and evidently the common ancestor of א &c. altered one of the three preceding words, and the common ancestor of LΔ another, while Β alone held fast the true text throughout.
323. Once more, the unique character of Β in a series of separate but mutually related variations, making up as it were an extended composite variation, is illustrated by St Mark's account of the denials of St Peter. Alone of the evangelists St Mark notices two crowings of a cock. According to the true text he follows the same lines as St Matthew and St Luke, while he makes the requisite additions in three places: that is, he inserts the word 'twice' (δίς) in both the prediction (xiv 30) and St Peter's recollection of the prediction (xiv 72 b), and the phrase 'a second time' (ἐκ δευτέρου) in the statement that 'a cock crew' immediately after the third denial (xiv 72 a). Thus all the points are tersely but sufficiently given. The text however, as it thus stood, presented more than one temptation to correction. At the first of the four places (v. 30) the direct harmonistic influence from the other Gospels was naturally strong and unchecked, and thus the first δίς is largely omitted (by אC* aeth arm as well as the Westerns, D cu2 lat.afr-eur). When v. 72 a was reached, ἐκ δευτέρου was as naturally a stumbling-block for a different reason, because there had been no mention of a previous cock-crowing. The supposed difficulty was met in two ways: a text now represented by a small group (אL c vg.cod), doubtless Alexandrian, assimilated v. 72 to v. 68 and the parallel narratives by striking out ἐκ δευτέρου; while the Western text boldly adapted v. 68 to v. 72 by inserting καὶ ἀλέκτωρ ἐφώνησεν after προαύλιον. Lastly v. 72 b was affected by the various texts both of the preceding words and of the original prediction (v. 30), here expressly repeated and thereby brought into strict parallelism, and accordingly δίς is omitted by more documents than ἐκ δευτέρου. The Syrian text makes the whole uniformly symmetrical and complete by accepting the Western interpolation in v. 68, while it retains δίς in both places. The confusion of attestation introduced by these several cross currents of change is so great that of the seven principal MSS אABCDLΔ no two have the same text in all four places. Neither of the two extreme arrangements, the Syrian (with A), which recognises the double cock-crowing in all four places, and that of א c, which recognises it nowhere but simply follows the other Gospels, could have given rise to the other readings. The chief cause of disturbance is manifestly the attempt to supply an explicit record of the first cock-crowing; and the original absence of καὶ ἀλέκτωρ ἐφώνησεν in v. 68 is sufficiently attested by אBL lt 17 c me. Half however of this group, as we have seen, followed the alternative expedient of omitting ἐκ δευτέρου, two of the number going on to omit the following δίς: and thus it appears that the only consistent authorities for the true text in this series of variations are B, a lectionary, and the Memphitic.
324. Such being the results of an examination of ternary variations, it is no wonder that binary variations likewise supply us with multitudes of readings of B, slenderly supported or even alone, which have every appearance of being genuine, and thus exemplify the peculiar habitual purity of its text. Readings like these are striking illustrations of the danger of trusting absolutely to even an overwhelming plurality of early and good authorities (see § 282 f.), and the need of bearing in mind the distorting effects of mixture. For instance it is morally certain that in Gal. vi 15 B, with two good cursives and some Versions and Fathers, is right in reading οὔτε γάρ for ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε which is borrowed from v 6; and yet the array sustaining the interpolation includes אACD2G3P2 with Versions and Fathers. Such a distribution could never have arisen except by a wide early adoption of a yet earlier aberration of some influential text, which here was evidently Western. On the other hand there are many subsingular readings of Β that cannot claim more than the secondary rank of alternative readings which may possibly be genuine, and there are many others that may be safely rejected. The claims of absolutely singular readings of Β in binary variations are naturally found to be usually of no great strength, though some among them appear to be very possibly genuine, and their genuineness would not be out of harmony with the known textual relations of B.
325. The existence of numerous genuine subsingular readings of Β in binary variations gives the key to the origin of another class of variations, fundamentally the same but different in appearance, which, though rare in the Gospels, are not uncommon in the other books preserved in B. The peculiarity of these variations consists in the agreement of Β with the Syrian text against the great mass of documents representing the more ancient texts. How is this distribution to be explained? Are these readings of Β corruptions of its fundamental text from a Syrian source, or do they belong to its fundamental text, so that they must have stood in the purest of the texts out of which the Syrian text was constructed? Internal evidence is decisively favourable to the second answer for at least the larger number of passages, and thus affords a strong presumption for the rest. Perhaps the most striking example is the well known variation in 1 Cor. xv 51, where there can be no doubt that the peculiar form of St Paul's words, together with forgetfulness of the language of the apostolic age (1 Thess. iv 15, 17), led to a transposition of the negative from the first clause to the second, and the introduction of a seemingly easy but fallacious antithesis. Here the wrong position of the negative is supported by א(A)CG3 17 with some Versions and Fathers, and also, with a verbal change, which probably formed part of the corruption in its earliest shape, by D2 with other Versions and Fathers. Thus Β alone of primary uncials, sustained however by the Memphitic and apparently by Origen and other good Fathers, as also by lost MSS mentioned by Fathers, upholds the true position in company with the Syrian text. The only difference of distribution between such cases and those noticed in the last paragraph is the shifting of the Syrian documents from the one side to the other; and such a shifting is the natural result of the eclecticism of the Syrian revisers ( see §§ 185 f.). Two causes have doubtless contributed to the unequal occurrence of the readings here described, genuine readings attested by Β almost alone in addition to the Syrian documents, so that if the Syrian attestation were removed they would be subsingular readings of Β; their greater abundance in the Acts and Epistles than in the Gospels being partly due to the more rapid and more widely current corruption of the Gospels, and partly to the relative paucity of extant uncials containing the Acts and Epistles. The former cause belongs to the actual history of the text; the latter is a mere accident in the preservation of documents to this day.
F. 326—329. Singular and subsingular readings of א and other MSS
326. Turning from Β to א, we find ourselves dealing with the handiwork of a scribe of different character. The omissions and repetitions of small groups of letters are rarely to be seen; but on the other hand all the ordinary lapses due to rapid and careless transcription are more numerous, including substitutions of one word for another, as when γινώσκει αὐτούς replaces σκηνώσει ἐπ’ αὐτούς in Apoc. vii 15. Some of these substitutions have a kind of sense of their own which is out of all relation to the context, as εἰς τὴν Ἀντιπατρίδα (from Acts xxiii 31) for εἰς τὴν πατρίδα in Matt. xiii 54; and ἀγαπήσας τοὺς Ἰουδaίους (for ἰδίους) τοὺς ἐν τῷ κόσμῳ in John xiii 1. The singular readings are very numerous, especially in the Apocalypse, and scarcely ever commend themselves on internal grounds. It can hardly be doubted that many of them are individualisms of the scribe himself, when his bold and rough manner of transcription is considered; but some doubtless are older. Little encouragement however to look favourably upon them is given by an examination of the subsingular readings. Many of these, as has been already noticed (§ 205), are clearly Western corruptions, of which οἶνον οὐκ εἶχον ὅτι συνετελέσθη ὁ οἶνος τοῦ γάμου in John ii 3 is an example; and many others are probably of Alexandrian origin: but, whatever may be the sources, the prevalent internal character where it can be known is such as to raise a strong presumptive suspicion where it is obscure. There are however a few subsingular readings of א which recall the predominant character of subsingular readings of B, and are possibly or even probably genuine. Such are the omission of υἱοῦ θεοῦ in Mark i 1, and of ἡ πύλη in Matt. vii 13; the insertion of Ἠσαίου in Matt. xiii 35; μηδένα (for μηδὲv) ἀπελπίζοντες in Luke vi 35; ᾔτησαν τὸν (for ᾐτήσαντο) Πειλᾶτον in Acts xiii 28; ἔδωκα for ἔδωκαν in Matt. xxvii 10. The fact that Origen's name occasionally stands among the accessory authorities is a warning against hasty rejection; and though subsingular readings of א attested by Origen are doubtless often only Alexandrian, this is probably not always the case.
327. These various characteristics of the singular and subsingular readings of א are easily explained in connexion with the relation between the texts of Β and of א described above, and at the same time enable this relation to be ascertained with somewhat greater precision. The ancestries of both MSS having started from a common source not much later than the autographs, they came respectively under different sets of influences, and each in the course of time lost more or less of its original purity. With certain limited exceptions already noticed, the concordance of Β and א marks that residual portion of the text of their primitive archetype in which neither of the two ancestries had at any point adopted or originated a wrong reading. Where their readings differ, at least one of the ancestries must have departed from the archetypal text. The possibility that both have gone astray in different ways must remain open, for it would be only natural that there should be an occasional coincidence of place between corruptions admitted into the one line of transmission and corruptions admitted into the other; and as a matter of fact there are a few passages where it is difficult to think that either Β or א has preserved the reading of the common original. But these coincidences are likely to be only exceptional; and all that has been observed up to this point respecting the character of our two MSS justifies a strong initial presumption in each particular case that the text of their archetype is preserved in one or other of them.
328. It follows that any subsingular, or even singular, reading of either Β or א may owe the limitation of its attestation to either of two totally different sets of antecedents. A subsingular reading of Β (or א) may be, first, equivalent to a subsingular reading of אB combined, which has lost part of its attestation by the accidental defection of א (or B); it may be, secondly, an early corruption limited in range of acceptance. Both explanations being in all cases possible, the antecedent probabilities differ widely according as the one or the other MS is in question. The ancestry of Β posterior to the common archetype was probably a chain of very few links indeed; certainly the various transcribers who had a hand in making it must either have been in a position which kept them ignorant of the great popular textual corruptions of the second and third centuries or must have for the most part preferred to follow their own inherited exemplars. It was not so in all cases, as is shown by such examples as those which have been cited above (§ 326); and an exceptional adulteration of the fundamental text of Β must be recognised as having occasionally left א alone where אB ought, so to speak, to have stood together. On the other hand the certainty that the ancestry of א posterior to the common archetype must, at one or more points in its history, have been exposed to contact with at least two early aberrant texts, since it accepted a considerable number of their readings (§ 205), enables us to account at once for the good internal character of most subsingular readings of B, and for the questionable internal character of most subsingular readings of א. Where the corrupt readings adopted by the ancestors of א happened to be widely adopted in current texts likewise, Β would be left with little or no support from Greek MSS ; that is, the true text of the common archetype would be preserved in subsingular readings of B. Where the corrupt readings adopted by the ancestors of א happened to find little or no reception in eclectic texts, Β and mixed Greek texts generally would be found alike attesting the true text of the common archetype, and subsingular readings of א would be nothing more than examples of early aberration early extinguished. The erroneous subsingular readings of B, proportionally as well as absolutely much less numerous than those of א, may be described in the same general terms with respect to their genealogical character, subject to the difference that the sources of corruption in Β are for the most part of a sporadic and indeterminate character (§ 204). Finally, the absence of any external criterion for referring the various singular and subsingular readings of either MS to one or other of the two possible origins, combined with the exceptional antiquity and purity of the fundamental text which they both preserve intact in very large though unequal proportions, demands a specially vigilant consideration for every such reading of both before it is definitely rejected.
329. It may be added explicitly here that, except for the Apocalypse, and the peculiar Western non-interpolations of the Gospels, a similar examination of the singular and subsingular readings of every extant MS except Β and א leads to entirely unfavourable results. There are a few, a very few, cases in which the genuineness of such a singular or subsingular reading must be admitted as possible: but all such readings occur, we believe, in ternary or more composite variations, and differ from the readings of Β or א merely by the absence of some slight erroneous modification. The same general statement may likewise be made respecting the trial of individual MSS by means of binary combinations into which א and Β do not enter (as in the Gospels CD, CL, CZ, CΔ, DL, DZ, LΔ, LΞ, AC, AD &c.), or indeed respecting any other application of Internal Evidence of Groups to the testing of their internal character.
G. 330—339. Determination of text where Β and א differ
330. It will be evident from the foregoing pages that Β must be regarded as having preserved not only a very ancient text, but a very pure line of very ancient text, and that with comparatively small depravation either by scattered ancient corruptions otherwise attested or by individualisms of the scribe himself. On the other hand to take it as the sole authority except where it contains self-betraying errors, as some have done, is an unwarrantable abandonment of criticism, and in our opinion inevitably leads to erroneous results. A text so formed would be incomparably nearer the truth than a text similarly taken from any other Greek MS or other single document: but it would contain many errors by no means obvious, which could with more or less certainty have been avoided by the free use of all existing evidence.
331. Enough has already been said on the determination of the text where Β is supported by א. A few words must be added here on the mode of dealing with the numerous variations in which these two preeminent MSS differ from each other. Setting aside ternary variations, most of the distributions in which the conflict of א and Β requires notice belong to one or other of the three following types: (1) Β with a small group against the rest; (2) א and Β each with a large group dividing the array; and (3), much less important, א with a small group against the rest. The characteristics and twofold genealogical antecedents of the first and third have been already considered (§§ 324, 326 ff.). In the first two cases, and also to a limited extent in the third, Genealogy and Internal Evidence of Groups have brought us to the point of having two readings before us, with so real a conflict of authority that, notwithstanding the habitually greater integrity of text in Β than in א, the normal relations between the different kinds of evidence are to a certain extent disturbed. Two classes of evidence rise into unusual importance here, Secondary documentary evidence and Internal evidence. The effects of both under these circumstances are the same; first to rescue a slenderly attested reading from being entirely set aside, and next, if the two classes of evidence sustain each other, or either is of exceptional strength, to render superfluous the retention of the other reading as an alternative. The bearing of Internal evidence, which here can be only Internal Evidence of Readings, requires no special comment. The change in the relative importance of Secondary documentary evidence will need a little explanation.
332. All Secondary documentary evidence has its value for these variations, in so far as it shews a given reading attested by a primary MS not to be an individualism; provided of course that the coincidence is such as cannot well be accidental. By supplying diversity of attestation, it has at the least the effect of proving that the reading had some sort of pedigree; and, considering the absence of very close and immediate relations of affinity between most extant documents, the pedigree must usually have been of some length. Little would be gained by this were the uncial itself secondary: but if its readings are habitually good in an exceptional proportion, the relative probability of the given reading is at once much increased.
333. There is however a much greater increase of authority when the secondary evidence is that of a peculiarly good element in a mixed document, being then equivalent to fragments of a document which if continuously preserved would have been of primary or not much lower rank. Such elements are found, for instance, in some Mixed Latin MSS, and also in some cursive Greek MSS. If a given cursive is observed to concur several times with the very best documents against not only all or almost all other cursives but almost all uncials in favour of a manifestly right reading, we know that it must contain an element of exceptional purity, and reasonably infer that the same element is the parent of other less certain readings in supporting which it joins with perhaps a single primary uncial only. Under these conditions the uncial may receive weighty documentary support from an apparently insignificant document.
334. On a superficial view it might seem arbitrary to assign a given cursive or other mixed document high authority in those variations which differ from the common text, and refuse it any authority where it agrees with the common text. As however has been implicitly shown in former pages (§ 197), this view derives its plausibility from neglect of the conditions on which criticism allows authority to a document on the ground that it is 'good', that is, gives it relative confidence in doubtful cases because it has been found on the right side in clear cases in which most documents are on the wrong side. If the homogeneousness of a cursive text is found to be broken by sporadic ancient readings, we know that we have virtually two distinct texts to deal with under the same name; that is, the readings discrepant from the common text proclaim themselves as derived from a second ancestor which had an ancient text. It can never indeed be positively affirmed that all the readings agreeing with the common text came distinctively from the principal or Syrian ancestor of the supposed cursive, for in regard of any one such reading it is always speculatively possible that it may have had a place in the virtually Pre-Syrian as well as in the Syrian ancestor: but in the face of the certainty that it must have existed in the Syrian ancestor this speculative possibility has no appreciable force for the purposes of criticism.
335. It so happens that the relation between two extant uncial MSS of St Paul's Epistles illustrates vividly the composite origin of many texts, including the texts of some at least of such cursives as have been noticed above. The St Germain MS E3, apparently written in Cent. X or late in Cent. IX, has long been recognised as a copy of the Clermont MS D2, executed after D2 had suffered much revision by correcting hands: all possible doubt as to the direct derivation of the one from the other is taken away by the senseless readings which the scribe of E3 has constructed out of a combination of what was written by the original scribe of D2 and what was written by its correctors;—an interesting illustration, it may be observed in passing, of the manner in which the strange Βεωορσόρ of א* in 2 Pet. ii 15 must have resulted from a fusion of the two readings Βεώρ and Βοσόρ. D2, it will be remembered (§§ 100 f., 203), was written in Cent. VI, and has a Western text. The readings introduced by the two chief correctors, referred to Cent. VII (D2a) and Cent. IX (D2b) respectively, and especially the readings due to the later of the two, are for the most part Syrian: on the other hand, while the later corrector alters many Pre-Syrian readings which his predecessor had passed over, he fails to make his own assimilative revision complete.
336. A short passage from D2 (Rom. xv 31—33) will sufficiently exhibit the chief phenomena of the corrections and transcription, the readings of the correctors being set between the lines: ἵνa ῥυσθῶ ἀπὸ τῶν ἀπιθούντων ἐv τῇ
Ἰουδαίᾳ ἵνα ἡ διακονία μου ἡ εἰs
Ἰουδαίᾳ καὶ ἡ δωροφορία μου ἡ ἐν Ιη̄μ̄ εὐπρόσδεκτος γένηται
τοῖς ἁγίοις, ἵνα ἐv χαρᾷ ἔλθω πρὸς ὑμᾶς διὰ θελήματος θῡ
τοῖς ἁγίοις, ἵνα ἐv χαρᾷ ἔλθω πρὸς ὑμᾶς διὰ θελήματος Χῡ Ἰῡ
καὶ ἀναψύξω μεθ’ ὑμῶν· ὁ δὲ θεὸς τῆς εἰρήνης dots
καὶ ἀναψύξω μεθ’ ὑμῶν· ὁ δὲ θεὸς τῆς εἰρήνης ἤτω μετὰ
πάντων ὑμῶν· ἀμήν. This passage contains five distinctively Western readings, of which the first four, ἡ δωροφορία, ἐν (before Ἰερουσαλήμ), Χρίστοῦ Ἰησοῦ, and the interpolation of ἤτω, are brought by the correctors into conformity with the true and the Syrian texts alike; the fifth, ἀναψύξω μεθ’ ὑμῶν for συνaνaπaύσωμaι ὑμῖν, remains untouched. The two Western readings which are also Syrian, γένηται τοῖς ἁγίοις for τ. ἁ. γ. and ἔλθω...καί for ἐλθών, are likewise left as they were. Lastly, the second ἵνα, omitted by all Pre-Syrian authorities, is inserted in agreement with the Syrian text. Of the five changes here made E3 adopts the first three, substituting them for the original readings of D2. The last two it neglects, retaining the original readings: the correctors' omission of ἤτω was apparently expressed by cancelling dots, which might easily escape the eye; the disregard of θεοῦ is probably due merely to carelessness, of which the scribe gives abundant signs. It will be seen at once that, if both the later corrector of D2 and the scribe of E3 had done effectually that which they evidently proposed to do, E3 would in this place have simply represented the Syrian text; and that the combined negligence was the cause of the survival of three Western readings.
337. These instructive phenomena naturally receive little consideration now, because the exact knowledge that we possess of the original D2 renders attention to the copy E3 superfluous. Supposing however that D2 had been lost, the complex antecedents of the text of E3 would have been unknown: it would have presented itself merely as a Syrian document sprinkled with Western readings. When then we find other late MSS having a Syrian text sprinkled with Western or other Pre-Syrian readings, we may reasonably take D2 and E3, as exhibiting the manner in which the mixture has probably arisen, and indirectly illustrating other possible modes of mixture. Evidently the textual value of E3 is virtually confined to the fragments which it preserved of the original writing of D2, while in the absence of D2 there would be no way of distinguishing these fragments from the rest of the text except by their discrepance from the Syrian text: and in like manner discrepance from the Syrian text is the only safe test for the readings of the ancient element in any late mixed document, because in late times the texts which would be virtually taken as standards for assimilative correction were naturally Syrian, no others being current.
338. It is true that by attending to the discrepant readings alone we should be neglecting some readings which as a matter of fact were in the original writing of D2, namely the Western readings that became Syrian (in the passage cited these are the change of order and the resolved construction): but if D2 had been lost there would have been no means of knowing this. Two courses alone would have been open; to attend exclusively to the readings discrepant from the Syrian text, as being almost certainly derived from the Non-Syrian element in the ancestry of E3; or to allow to all the readings of E3 whatever authority the discrepant readings might claim. In the former case there would be a negative disadvantage; a necessary loss of evidence, but no falsification of it: the composite text of E3 would be virtually ignored outside the definite limits, but the risk of attributing to the better element of its ancestry readings due in fact to the worse would be avoided. In the latter case there would be a certainty of extensive positive error, since E3 obviously abounds in purely Syrian readings, and yet, for want of a discriminative test, they would be included with the rest in the general attribution of the authority belonging properly to the more ancient element alone. Here again D2 and E3 elucidate the necessity of limiting the separate authority of cursives containing ancient elements of text to their Non-Syrian readings (see the end of § 334).
339. Some weight might doubtless be consistently given to the cumulative negative evidence against a reading supplied by the absence of any cursive attestation whatever; because it might be anticipated that the fortuitous irregularity with which the ancient readings are scattered over any one mixed text would be neutralised by the juxtaposition of all mixed texts, so that a genuine reading would be likely to obtain attestation from at least one or other of the number. But the anticipation is not verified by experience, for numerous absolutely certain readings have no cursive or other similar attestation; and this fact has to be taken into account in doubtful cases. Here, as in all cases where textual character is in question, what is said of cursives applies equally to late uncials: the outward and formal difference between the two classes of MSS involves no corresponding difference of texts.
H. 340—346. Determination of text where B is absent
340. The comparative certainty afforded by the peculiar character of Β is felt at once when we pass to parts of the text where it is wanting. As regards the ancient texts, we lose the one approximately constant Greek neutral document: as regards Internal Evidence of Groups, we lose all the groups into which Β enters. This state of evidence occurs under three different conditions; first, in detached variations in the Pauline Epistles, where the Western element of Β has displaced its fundamental or neutral element, the absence of which is virtually equivalent to the absence of Β; secondly, in those parts of the Pauline Epistles which were contained in the lost leaves of B, but in which the relations of the other documents are to a considerable extent illustrated by facts of grouping observed in those parts of the same series of books for which Β is extant; and thirdly, in the Apocalypse, where analogies of grouping are to say the least imperfect, and the few important documents common to the rest of the New Testament present themselves in novel relations.
341. First both in order of books and in gradation come the isolated Western readings of Β in the Pauline Epistles. Where BD2G3 or BG3 with other chiefly Western documents stand alone among Pre-Syrian documents, there is no difficulty. Distinctively Western substitutions or additions attested by Β are with a few doubtful exceptions, as κημώσεις 1 Cor. ix 9, ἑρμηνευτής xiv 28, ἐνδεικνύμενοι 2 Cor. viii 24, ὑμεῖς...ἐστέ Gal. iv 28, which it is prudent to retain as alternatives, of no better character than similar distinctively Western readings not supported by B. Such readings therefore as πληροφορήσαι for πληρώσαι Rom. xv 13 (cf. v. 29 v.l.), φιλοτιμοῦμαι xv 20, δωροφορία for διακονία xv 31, 'Αριστοβόλον xvi 10, οὐδὲ ἀπῆλθον Gal. i 17, and the transposition of τῇ οὔσῃ ἐν Κορίνθῳ and ἡγιασμένοις ἐν Χριστῷ Ἰησοῦ (ancient lines) in 1 Cor. i 2 we have had no hesitation in rejecting. The internal evidence is not so clear with respect to distinctively Western omissions, and for the present at least it is safest to indicate doubt about words omitted by this group. But where other documents not clearly Western form part of the attestation, interpretation of the evidence is often difficult, if the rival reading is well attested. We can have no security in these cases that Β derived its reading from its neutral element: and, if it derived it from its Western element, then two alternatives are possible: either the accessory documents are really Non-Western, in which case the rival reading is often Alexandrian; or they are mixed (usually Syrian) and have adopted a Western reading, in which case the rival reading is more likely to be simply Non-Western, although its attestation is consistent with its being Alexandrian. In these cases we have exactly the state of things, as far as regards extant attestation, which Griesbach assumed to have from early times existed everywhere (see § 251), an attestation which might easily be only Western opposed to an attestation which might easily be only Alexandrian. If however these variations are examined together. Internal Evidence is generally favourable to the apparently Non-Western readings: but in not a few cases the other reading must be retained as an alternative, or even appears to be the more probable of the two.
342. Since in the Pauline Epistles Β (as well as א, A, and C) sometimes supports distinctively Western readings, so that they gain, for instance, the attestation BD2G3 as well as אD2G3, AD2G3, and (more rarely) CD2G3 and even ACD2G3 and occasionally אACD2G3, it might be asked what security we have that אBD2G3, or even the same group with other uncials added, do not make a Western combination. As a matter of attestation the contingency contains no improbability; and the recognition of it prescribes special watchfulness where there is no sufficient accessory Non-Western attestation, this being in fact another of the cases in which secondary documentary evidence of the better sort acquires a high interpretative value. But Internal Evidence is so favourable to the group אBD2G3 that except in a very few cases, as οὗ Rom. iv 8, αἰχμαλωτίζοντά με ἐν τῷ νόμῳ vii 23, ἡ omitted after τοῦ θεοῦ 1 Cor. xv 10, ἁγίοις omitted 1 Thes. ν 27, and καὶ τῆς ἁγνότητος added 2 Cor. xi 3, we have not found reason to treat their readings as doubtful.
343. We come next to the analogous difficulties which arise where Β totally fails us as regards direct evidence, but still affords some indirect aid in the interpretation of groupings, namely in the latter part (ix 14—end) of the Epistle to the Hebrews, in the Pastoral Epistles, and in the Epistle to Philemon. Here too the main distinctive problem is how to distinguish oppositions of Western and Non-Western from oppositions of Non-Alexandrian and Alexandrian readings; and it has to be dealt with in the same manner as in the former case. Another uncertainty is suggested by a recollection of the excellence of subsingular readings of Β in those parts of the Pauline Epistles which are preserved in it, and of the similar excellence of readings differing in attestation from these by the mere addition of the Syrian documents (§§ 324 f). Evidently the only resource here is to allow an alternative place to readings slenderly supported, or supported chiefly by Syrian documents, provided that the attestation includes such documents as are often associated with Β in its subsingular readings, and that the local internal evidence is favourable. It would be convenient to an editor in this part of the New Testament to assign to א such an authority as a consideration of the whole evidence has up to this point constrained us to assign to B. But the absolute excellence of א is neither lessened nor increased by the loss of a purer MS: the comparative excellence of its fundamental text and the deterioration of that text by mixture alike remain unchanged, while the discrimination of the different elements through grouping is deprived of one important resource. Such being the case, the text of these eighteen or nineteen chapters of the Pauline Epistles is undeniably less certain than that of the rest, though, as far as we can judge, the uncertainty is small in amount and of no real moment.
344. When at last we reach the Apocalypse, new and troublesome conditions of evidence are encountered. Not only is Β absent, but historical landmarks are obscure, and familiar documents assume a new position. Probable traces of a Western and perhaps an Alexandrian text may be discerned, with analogous relations to the extant uncials which contain other books: but they are not distinct enough to give much help, and for the most part Internal Evidence of Groups is the highest available guide of criticism. As before, א has a large neutral element; but in addition to mixture, probably Western and Alexandrian, evident individualisms of the scribe, or of one of his immediate predecessors, come forth in much greater luxuriance than before, as also they do in the Epistle of Barnabas which follows the Apocalypse in the same handwriting; this less scrupulous treatment of the text being perhaps connected with the ambiguous authority of the Apocalypse in the canonical lists of Cent. IV. Nor is internal evidence as a rule here favourable to א unsupported by other uncials: indeed a large proportion of the readings of the binary combinations אΑ, אC, אP2 are questionable or clearly wrong. C preserves nearly the same character as in the Acts and Epistles. The elements of A apparently remain unchanged; but the ancient or neutral element is larger. Both these MSS however acquire a high relative eminence through the want of compeers, or documents approximately such. Their consent is well supported by internal evidence, even where it has no documentary confirmation; and A stands quite alone, or unsustained by any other Greek MS, in some manifestly right readings, such as κατήγωρ in xii 10, and εἴ τις εἰς αἰχμαλωσίαν εἰς αἰχμαλωσίαν ὑπάγει in xiii 10. On the other hand the absolute proportion of wrong readings is great in each of them singly. As in most of the Epistles, P2 contains, in the midst of a somewhat degenerate text, so many good readings that it is entitled to an appreciable authority in doubtful cases; while the comparatively few readings of B2 which rise above its generally low level of character are such as imply a source of no distinctive value. Cursives containing not a few ancient readings are fairly numerous, and yield valuable help; as do the Latin versions, and in a less degree the rest, which seem to be all of comparatively late date, and certainly have texts of an extremely mixed character. Careful study of grouping goes far towards shewing which readings may safely be neglected; and Internal Evidence of Readings is often sufficiently decisive in this book to allow a clear decision to be made between those that remain. Yet the state of the documentary evidence renders it necessary to leave a considerable number of alternative readings. With the fullest allowance for the peculiarities of the rough Palestinian Greek, which indeed for the most part may be classified under a very small number of grammatical heads, several places remain where no document seems to have preserved the true text, and it is quite possible that the discovery of new and better documents might bring to light other unsuspected corruptions. Nothing however in the extant evidence suggests the probability that they would be of any importance.
345. We are by no means sure that we have done all for the text of the Apocalypse that might be done with existing materials. But we are convinced that the only way to remove such relative insecurity as belongs to it would be by a more minute and complete examination of the genealogical relations of the documents than we have been able to accomplish, nor have we reason to suspect that the result would make any considerable change.
346. The relation of the 'Received Text' to the ancient texts in the Apocalypse requires separate notice. In all other books it follows with rare exceptions the text of the great bulk of cursives. In all the books in which there was an undoubted Syrian text the text of the great bulk of cursives is essentially Syrian, with a certain number of later ('Constantinoplitan') modifications; in other books the text is, if not Syrian, at least such as must have been associated with the original Syrian books at Constantinople. The exceptional readings of the 'Received Text', in which it abandons the majority of the cursives, are hardly ever distinctively Alexandrian; in almost all cases they are Western readings, sometimes very slenderly attested, which evidently owe their place to coincidence with the Latin Vulgate, having been adopted by Erasmus in the first instance, and never afterwards removed. The foundation of the 'Received Text' of the Apocalypse on the other hand was a transcript of the single cursive numbered 1: Erasmus had in his earlier editions no other Greek MS to follow, though eventually he introduced almost at random a certain number of corrections from the Complutensian text. Now 1 is by no means an average cursive of the common sort. On the one hand it has many individualisms and readings with small and evidently unimportant attestation: on the other it has a large and good ancient element, chiefly it would seem of Western origin, and ought certainly (with the somewhat similar 38) to stand high among secondary documents. While therefore the text of 1 differs very widely from the true text by its Western readings, its individualisms, and the large late or Constantinopolitan element which it possesses in common with other cursives, a text formed in the way that the 'Received Text' is formed in other books would probably have differed from the true text on the whole much more. Thus the 'Received Text' of the Apocalypse has a curiously anomalous position. Besides containing a small portion of text which, like some single words in other books with less excuse, was fabricated from the Latin by Erasmus without any Greek authority to supply a defect in his one MS, it abounds in readings which cannot be justified on any possible view of documentary evidence, and are as a matter of fact abandoned by all textual critics: and yet the proportion of cases in which it has adopted the readings most current in the degenerate popular Greek texts of the Middle Ages, though large, is probably smaller than in any other book of the New Testament. I. 347—355. Supplementary details on the birthplace and the composition of leading MSS
347. In all that we have hitherto said we have taken no account of the supposed locality in which MSS were written, except in certain definite cases. The reason is because we do not believe anything certain to be as yet known. Up to a certain point the bilingual MSS (Græco-Latin and Græco-Thebaic) tell their own tale: about no other important early MS is it as yet possible to make any geographical assertion with confidence. It is indeed usually taken for granted that the chief uncials of the New Testament were written at Alexandria. This floating impression appears to be founded on vague associations derived from two undoubted facts; (1) that the translations of the Old Testament which form the LXX were made at Alexandria, while the chief uncials of the New Testament agree in some prominent points of orthography and grammatical form (by no means in all) with the chief uncials of the LXX, the four oldest being moreover parts of the same manuscript Bibles, and (2) that A was at some unknown time, not necessarily earlier than the eleventh century, preserved at Alexandria, and is hence called the Codex Alexandrinus. The supposition cannot be pronounced incredible; but it is at present hardly more than a blind and on the whole improbable conjecture. An Alexandrian origin, much more an exclusively Alexandrian or Egyptian use, cannot be reasonably maintained for most of the unclassical orthographies and grammatical forms found in MSS of the New Testament, as we shall have to explain more at length in Part IV. The character of the substantive texts affords only the most uncertain indications; for (1) there is no reason to suppose that more than a small fraction of the readings often called Alexandrian had any special connexion with Alexandria, and (2) the clearest phenomena of Versions of the fourth and fifth centuries shew how widely spread at that time were Greek MSS containing a large proportion of those readings which did really originate at Alexandria.
348. Possibly hereafter some of the external accompaniments of the text may be found to contain trustworthy evidence. At present we know of almost nothing to appeal to except such orthographies as are shown by their isolated distribution to be due to scribes, not to the autographs. This evidence at best points only to the home or school of the scribe himself, and cannot take account of migration on his part. Such as it is, it suggests that A and C were connected with Alexandria. Orthographies apparently Alexandrian occur also in א, but chiefly or wholly in words for which A or C have them likewise. On the other hand some Western or Latin influence is very clearly marked in the usual or occasional spelling of some proper names, such as Ισακ and Ιστραηλ[ειτης] or Ισδραηλ[ειτης]. In Β the Alexandrian indications are to the best of our belief wholly wanting. Western indications are fainter than in א, but not absent. The superfluous euphonic τ is sometimes inserted in Ισραηλ[ειτης] but only in Acts, apparently implying the presence of Western or Latin influence in the scribe of that manuscript of Acts which was copied by the scribe of B. The substitution of Χριστὸς Ίησοῦς for Ίησοῦς Χριστὸς in places where it is almost certainly not right is mainly confined to Western documents, and it is also in St Paul's Epistles a favourite individualism of B.
349. Again it is remarkable that the principal Latin system of divisions of the Acts, found in the Codex Amiatinus and, slightly modified, in other Vulgate MSS, is indicated by Greek numerals both in א (with large irregular omissions) and in B, but is otherwise unknown in Greek MSS and literature. The numerals were apparently inserted in both MSS, certainly in א, by very ancient scribes, though not by the writers of the text itself, Β indeed having antecedently a wholly different set of numerals. The differences in detail are sufficient to shew that the two scribes followed different originals: the differences of both from the existing Latin arrangement are still greater, but too slight to allow any doubt as to identity of ultimate origin. The coincidence suggests a presumption that the early home, and therefore not improbably the birthplace, of both MSS was in the West.
350. The other systems of divisions marked in Β and א have not hitherto yielded any trustworthy indications; and, what is more surprising, the same must be said of the structure and contents of the MSS themselves. It might have been anticipated that in order to ascertain the regions in which they were written it would suffice to observe what books they do or do not include, and in what manner the books are arranged, account being taken of the Old as well as the New Testament. But the attempt is baffled by the scantiness of our information. Comparison with the few extant catalogues and other evidence of local use in the fourth century leads only to ambiguous results; and the difficulty of decision is increased by the wide differences of structure and arrangement between Β and א, and again between both and A.
351. Taking all kinds of indications together, we are inclined to surmise that Β and א were both written in the West, probably at Rome; that the ancestors of Β were wholly Western (in the geographical, not the textual sense) up to a very early time indeed; and that the ancestors of א were in great part Alexandrian, again in the geographical, not the textual sense. We do not forget such facts as the protracted unwillingness of the Roman church to accept the Epistle to the Hebrews, commended though it was by the large use made of it in the Epistle of Clement to the Corinthians: but the complex life of Christian Rome in the fourth century cannot safely be measured by its official usage; and it would be strange if the widely current History of Eusebius led no Roman readers to welcome the full Eusebian Canon, with the natural addition of the Apocalypse, a book always accepted in the West. The supposition here made would account for all ascertained facts and contradict none. Yet we are well aware that other suppositions may be possibly true; and we must repeat that the view which we have here ventured to put forward as best explaining the sum total of the phenomena is only a surmise, on which we build nothing.
352. The fundamental similarity of text throughout the whole of B, and again throughout the whole of א with the exception of the Apocalypse, deserves special notice, because it is more probable that the exemplars from which they were taken contained each only a single book or group of books than that they were large enough to contain the whole series of books (see §§ 14, 301). Even among cursives it is not uncommon to find one or more groups of books written in a different age from the rest, with which they are bound up; so that a transcript of the whole volume would really represent two different exemplars (see § 46): and for a different reason a similar diversity of sources must often have been disguised by transcription in the fourth and fifth centuries. The transition from small portable MSS of limited contents is strikingly illustrated by a fortunate accident in the transcription of one of the four great comprehensive MSS which are the earliest now extant. In the MS of the Apocalypse from which C was taken some leaves had been displaced, and the scribe of C did not discover the displacement. It thus becomes easy to compute that each leaf of the exemplar contained only about as much as 10 lines of the text of the present edition; so that this one book must have made up nearly 120 small leaves of parchment, and accordingly formed a volume either to itself or without considerable additions. The distinctive character of text exhibited by A in the Gospels, by Δ in St Mark, and by Β in the Pauline Epistles, as also the orthography of Β (Ιστρ.) peculiar to the Acts, are instances of indications which equally shew the precariousness of assuming with respect to any one MS of the New Testament that all the books in it were copied from a single volume. In some cases, as we have suggested above (§ 320) with reference to Β in the Pauline Epistles, the discrepant character of text in particular books or groups of books was doubtless introduced not by the immediate exemplar but by previous interlinear or marginal corrections made in its predecessor: but in most cases the range of the corrections would be limited by the contents of the accessory copy which furnished them; so that the cause of the discrepancy of text would be ultimately the same. It is indeed quite uncertain to what extent the whole New Testament was ever included in a single volume in Ante-Nicene times. On the other hand the average conditions to which different volumes of the sacred writings would be exposed in the same place were not likely to differ much, in so far as they were likely to affect the text. It is therefore not surprising that we find great fundamental similarity of text throughout MSS which probably derived different groups of books from different exemplars, and that definite evidence of separate origins is sometimes present, sometimes wanting.
353. A word may be added here respecting the different 'hands' of MSS. It sometimes happened that the original scribe ('first hand') of a MS discovered that he had begun to transcribe wrongly, and accordingly corrected himself before going further: in such cases what he first wrote may have been either a mere blunder or the unconsciously remembered reading of another copy. After the completion of a MS it was often revised by a 'corrector' with a view to the removal of clerical errors. The thoroughness with which this laborious process was carried out must however have varied to a singular extent: and moreover the revision appears sometimes to have included the occasional introduction of readings from a different exemplar. Changes made by a hand apparently contemporary with the original hand may usually be set down to the 'corrector'. Additional changes might be made subsequently at any date on account of observed difference of reading from another MS simultaneously read or another current text. Sometimes these changes were confined to a small portion of text, or were sprinkled very thinly over the whole, sometimes they were comparatively systematic: but it is hardly ever safe to assume that a reading left unchanged is to be taken as ratified by the copy or text from which neighbouring changes were derived. Since corrections in previously written MSS, as distinguished from corrections made in the process of transcription, are not likely to be conjectures, they may be treated as virtually particles of other lost MSS at least as early as the time of correction: the textual value of the lost MSS can of course be ascertained only by successive examination of their successive particles, and therefore often but imperfectly.
354. For some six centuries after it was written Β appears to have undergone no changes in its text except from the hand of the 'corrector', the 'second hand'. Among his corrections of clerical errors are scattered some textual changes, clearly marked as such by the existence of very early authority for both readings: the readings which he thus introduces imply the use of a second exemplar, having a text less pure than that of the primary exemplar, but free from clear traces of Syrian influence. The occurrence of these definite diversities of text renders it unsafe to assume that all singular readings which he alters were individualisms of the first hand, though doubtless many of them had no other origin. The scale of alteration was however very limited: hardly any of the corrections affect more than two or three letters, except the insertions of rightly or wrongly omitted words. Some few of the early corrections perceptible in the MS appear to have been made by the original scribe himself; and to his hand Tischendorf refers seven alternative readings placed in the margin of Matt. xiii 52; xiv 5; xvi 4; xxii 10; xxvii 4; Luke iii i (bis). In the tenth or eleventh century, according to Tischendorf's apparently well founded judgement, the faded characters of the fourth century were retraced in darker ink. The readings adopted for renewal were almost always those of the second hand; and words or longer portions of text wrongly repeated by the original scribe were left untouched. There was no systematic attempt to correct the text itself, except as regards the orthography, which was for the most part assimilated to the common literary standard; but Syrian readings were introduced here and there, though rarely, if ever, in cases where there would be more than a trifling difference in the space occupied by the old and the new readings respectively. We have passed over the readings of this third hand of Β in the Appendix because they not only were inserted at a very late period, but exhibit no distinctive internal character. Confusion between the second and third hands of Β has led to much error; and it is only of late that the true history of the changes undergone by the MS has been fully understood.
355. The original writing of א has escaped retracement, but it has been altered much at different times. The three principal hands alone need mention here. The 'corrector' proper (אa) made use of an excellent exemplar, and the readings which he occasionally introduces take high rank as authorities. Those of another hand (אb) of somewhat similar appearance but ill determined date (? Cent. vi) are likewise for the most part distinctly ancient, but include many of later origin. The much more numerous readings introduced by אc (? Cent. vii) are for the most part Syrian; but scattered among them are readings handed down from a high antiquity: the exemplar employed by this writer had apparently some such mixed character as we find in X of the Gospels. These examples will suffice to illustrate the phenomena of correction generally. The manner in which it produces mixture of texts in transcripts from corrected MSS has been already explained by the example of D2 and E3 (§§ 335—339). In some instances, as often in A and C, an erasure preceding correction has completely obliterated the original writing: but, as the amount of space which it occupied can almost always be ascertained, a comparison of the lengths of the existing variants is usually sufficient to determine the reading with tolerable certainty.
CHAPTER IV. SUBSTANTIAL INTEGRITY OF THE PUREST TRANSMITTED TEXT
356—374
356. Having now described the nature of the evidence available for settling the text of the New Testament, and explained the modes of applying it which leave least room for error, it is right that we should give some answer to the reasonable enquiry whether there is good ground for confidence that the purest text transmitted by existing documents is strictly or at least substantially identical with the text of the autographs. This enquiry will however be best approached through another, which is closely connected with the subject of the preceding chapter; namely, whether there is or is not reason to think that, notwithstanding the peculiar authority conferred on the best uncials by the clear results of Genealogical Evidence proper and of Internal Evidence of Groups, the true reading is sometimes one that is attested by inferior documents alone. This antecedent enquiry is complementary to a question discussed in another place (§§ 265—283), how far Primary Greek MSS may safely be trusted where accessory attestation is more or less completely wanting. From the nature of the case there is no room for absolute and unqualified answers: but we trust that the following considerations, taken along with what has been said already, will meet all such doubts as can be raised with a fair show of reason.
357—360. Approximate non-existence of genuine readings unattested by any of the best Greek uncials
357. The vague but necessary term 'inferior documents' covers two classes of evidence which demand attention on wholly different grounds; first, Greek uncials which in external character, as in conventional designation, have no generic difference from the best Greek uncials, and secondly, the earlier Versions and Fathers. First then it may be asked,—Given the relative supremacy which we have been led to ascribe under normal conditions to Β and א in most books, and to some extent to A and C in the Apocalypse, is there or is there not good ground to expect that the true reading should sometimes exist not in them but in less good or in secondary Greek uncials? There is no theoretical improbability in the supposition here made. This is obviously true in cases where א and Β are at variance, that is, where the positive evidence afforded by the coincidence of two extremely ancient independent lines is absent: for, where they differ from each other, the true reading may differ from that of either, and may have survived in an independent line to a somewhat later time, and so have found its way into other uncials. But the theoretical possibility holds good likewise where Β and א agree, though reduced within much narrower limits. Near as the divergence of the respective ancestries of Β and א must have been to the autographs, there must have been an appreciable interval of transcription (§§ 241, 301 ff.); and it is a priori conceivable that relics of a line of transmission starting from a yet earlier point should find their way into one or another uncial of the fifth or following centuries, and further that such relics should include genuine readings which disappeared in the writing of an intermediate ancestor of Β and א.
358. When however the readings of secondary or even primary uncials in opposition to Β and א are consecutively examined, they present no such phenomena, whether of accessory attestation or of internal character, as might have been expected were the supposition true. The singular readings with rare and unimportant exceptions have all the appearance of being individualisms. The scanty subsingular readings having some attestation by early Versions or Fathers will be noticed under the next head. The readings attested by two or more of these uncials, which make up by far the greater part of the whole number of these readings, can be recognised at once as distinctively Syrian or Alexandrian or Western, or as obvious modifications of extant readings having one or other such attestation and character. Among all the endless varieties of mixture there is a striking sameness in the elements mixed. The immediate sources of all our uncials not purely Syrian, except Β and א, were evidently for the most part the popular eclectic texts of about the fourth century, Syrian or other, and not the various earlier and simpler Ante-Nicene texts from which the eclectic texts were compounded, and which the eclectic texts soon drove out of currency. Lastly, the verdict of internal evidence is almost always unfavourable where it is not neutral.
359. Passing backwards to Ante-Nicene times, we have to deal with the second question,—May we or may we not reasonably expect to find true readings in very limited but very ancient groups of documents in opposition to Β and א? There are many Pre-Syrian readings the antiquity of which is vouched for by Versions or Fathers, but which nevertheless are supported by no Greek MS but a stray uncial or two, or only by a few cursives, (such cursives naturally as are otherwise known to contain ancient elements of text,) or even in many cases by no Greek MS at all. The attestation of these readings, or at least of the second and third classes of them, resembles the accessory attestation of the subsingular readings of B, which we have already learned to judge on the whole favourably: it resembles also the accessory attestation of the subsingular readings of א, which we have rarely found to have the stamp of genuineness. All such readings shew how plentiful a crop of variation existed in the early centuries and was swept out of sight by the eclectic texts.
360. Readings thus attested by Versions and Fathers almost without support from existing Greek MSS have as yet received from critics no attention proportionate to their historical interest. The accident of their neglect by the Greek editors of the fourth century, and their consequent approximate or complete extinction in Greek copies of the New Testament, can have no bearing on the character of their pedigree in the earlier ages. It is therefore but right to enquire whether the accidental preservation of Β and א does or does not give their texts an undeserved preeminence, which they would have lost had continuous uncials existed containing such texts as these stray readings represent. A scrutiny of the readings themselves dispels the suspicion. We have for our own part been quite prepared to find among these relics of ancient variation many readings highly commended by Internal Evidence: but experience has not justified any such anticipation. A very few readings absent from all existing Greek MSS we have thought it safest to retain as alternative readings; for instance in Matt. iv 17 Ἤγγικεν (for Μετανοεῖτε, ἤγγικεν γάρ), attested by syr.vt Orig(as represented by schol Procop.Es.144 Hier.Es.128) Vict.ant.Mc.273(expressly); and in 1 John iv 3 λύει (for μὴ ὁμολογεῖ), attested by 'ancient copies' mentioned by Socrates, and also by lat.vg Iren.lat(with context) Orig.Mt.lat:(?schol) Tert Lucif Aug Fulg. There are a few others supported by yet slighter authority, which have an appearance of intrinsic probability in places where the better attested readings seem to be specially difficult; and these we have not attempted to separate from purely conjectural readings. Readings belonging to either of these classes are however in the highest degree exceptional, and do not disturb the general impression produced by examination of the whole number. Most indeed of the readings of great antiquity which stand in no extant Greek uncial are seen at a glance to be ordinary Western readings; so that doubtless the reason why those of them which occur in the Gospels and Acts are deprived of the support of D is simply the comparative purity of its early Western text. While then it cannot be confidently affirmed that no relics of lines of transmission independent of the ancestries of Β and א now exist in one or more secondary documents of one kind or another (compare § 357), the utmost number of such relics is too petty, even with the inclusion of doubtful instances, to affect appreciably the conclusions already obtained. It is of course only with such evidence as actually exists that the primary uncials can be brought into comparison: but the fullest comparison does but increase the conviction that their preeminent relative purity is likewise approximately absolute, a true approximate reproduction of the text of the autographs, not an accidental and deceptive result of the loss of better Greek MSS.
361—370. Approximate sufficiency of existing documents for the recovery of the genuine text, notwithstanding the existence of some primitive corruptions
361. The way has now been cleared for the final question,—Is it or is it not reasonable to expect that in any considerable number of cases the true reading has now perished? Have we a right to assume that the true reading always exists somewhere among existing documents? The question is often foreclosed on one or both of two grounds which in our judgement are quite irrelevant. First, some think it incredible that any true words of Scripture should have perished. In reply it is a sufficient argumentum ad hominem to point to the existence of various readings, forming part of various texts accepted for long ages, and the frequent difficulty of deciding between them, even though we say nothing of difficulties of interpretation: on any view many important churches for long ages have had only an approximately pure New Testament, so that we have no right to treat it as antecedently incredible that only an approximately pure New Testament should be attainable now, or even in all future time. For ourselves we dare not introduce considerations which could not reasonably be applied to other ancient texts, supposing them to have documentary attestation of equal amount, variety, and antiquity. Secondly, the folly and frivolity of once popular conjectures have led to a wholesome reaction against looking beyond documentary tradition. Some of them are attempts to deal textually with what are really difficulties of interpretation only; the authors of others, though they propose remedies which cannot possibly avail, are not thereby shown to have been wrong in the supposition that remedies were needed; and a few have been perhaps too quickly forgotten. Though it cannot be said that recent attempts in Holland to revive conjectural criticism for the New Testament have shown much felicity of suggestion, they cannot be justly condemned on the ground of principle. The caution imposed by the numerous failures of the earlier critics has on the whole worked well; but it has no bearing on the question at issue.
362. On the other hand a strong presumption in favour of the immunity of the text of the New Testament from errors antecedent to existing documents is afforded by the facts mentioned under the last head (§§ 357—360). If among the very ancient evidence now extant, collected from various quarters, so little can be found that approves itself as true in opposition both to Β and א, there is good reason at the outset to doubt whether any better readings have perished with the multitudes of documents that have been lost.
363. The question however needs more careful consideration on account of the apparent ease and simplicity with which many ancient texts are edited, which might be thought, on a hasty view, to imply that the New Testament cannot be restored with equal certainty. But this ease and simplicity is in fact the mark of evidence too scanty to be tested; whereas in the variety and fullness of the evidence on which it rests the text of the New Testament stands absolutely and unapproachably alone among prose writings. For all other works of antiquity, the Old Testament (in translations) and some of the Latin poets excepted, MSS earlier than the ninth or even tenth century are of extreme rarity. Many are preserved to us in a single MS or hardly more; and so there is little chance of detecting corruption wherever the sense is good. Those only which are extant in many copies of different ages present so much as a distant analogy with the New Testament: and, if through the multitude of various readings, and the consequent diversities of printed editions, they lose the fallacious uniformity of text which is the usual result of extreme paucity of documents, there is always a nearer approximation to perfect restoration. Doubtful points are out of sight even in critical editions of classical authors merely because in ordinary literature it is seldom worth while to trouble the clearness of a page. The one disadvantage on the side of the New Testament, the early mixture of independent lines of transmission, is more than neutralised, as soon as it is distinctly perceived, by the antiquity and variety of the evidence; and the expression of doubt wherever doubt is really felt is owing to the paramount necessity for fidelity as to the exact words of Scripture.
364. But it will be seen from the preceding pages that we possess evidence much more precisely certified than by the simple and general titles of antiquity, excellence, and variety. Two or three of our best documents might have been lost, and yet those titles might still be justly claimed; while without those documents both the history of the text and its application would be so imperfectly understood that the results in that case would be both different and more uncertain. It is the minute study of the whole evidence in relation to the best documents which brings out their absolute and not merely their relative excellence. The external evidence is therefore such that on the one hand perfect purity is not a priori improbable, and a singularly high degree of purity is highly probable; and yet the conditions are not such—it is difficult to see how they could ever be such—as to exclude the possibility of textual errors.
365. These general probabilities however are but preparatory to the definite question,—Are there as a matter of fact places in which we are constrained by overwhelming evidence to recognise the existence of textual error in all extant documents? To this question we have no hesitation in replying in the affirmative. For instance in 2 Pet. iii 10 אBK2P2 with three of the best cursives and two Versions read στοιχεῖα δὲ καυσούμενα λυθήσεται καὶ γῆ καὶ τὰ ἐν αὐτῇ ἔργα εὑρεθήσεται. Before εὑρεθήσεται two other Versions insert a negative. C replaces εὑρεθήσεται by ἀφανισθήσονται, for which we find κατακαήσεται in AL2 and most cursives and several Versions and Fathers; while one representative of the Old Latin omits it altogether. External evidence is here strongly favourable to εὑρεθήσεται, as must be felt even by those who do not see any special significance in the concordance of א and B. Internal evidence of transcription is absolutely certain on the same side, for εὑρεθήσεται fully accounts for all four other readings, two of them being conjectural substitutes, two less audacious manipulations; while no other reading will account for the rest. Yet it is hardly less certain by intrinsic probability that εὑρεθήσεται cannot be right: in other words, it is the most original of recorded readings, the parent of the rest, and yet itself corrupt. Conditions of reading essentially the same, in a less striking form, occur here and there in other places.
366. But there is no adequate justification for assuming that primitive corruption must be confined to passages where it was obvious enough to catch the eye of ancient scribes, and would naturally thus lead to variation. Especially where the grammar runs with deceptive smoothness, and a wrong construction yields a sense plausible enough to cause no misgivings to an ordinary reader, there is nothing surprising if the kind of scrutiny required for deliberate criticism detects impossible readings accepted without suspicion by all transcribers. On the various kinds of primitive errors, and the nature of the evidence on which in each case their existence can be affirmed, we have said enough in the Second Part (§§ 85—92).
367. Little is gained by speculating as to the precise point at which such corruptions came in. They may be due to the original writer, or to his amanuensis if he wrote from dictation, or they may be due to one of the earliest transcribers. Except from extraneous sources, which here have no existence, it is never possible to know how many transcriptions intervened between the autograph and the latest common ancestor of all the elements in all extant documents; and a corruption affecting them all may evidently have originated at any link of that initial chain. Moreover the line of demarcation between primitive and other corruptions is less easy to draw than might be supposed. As was intimated above (§ 360), account has to be taken of a few places in which what appears to be the true reading is found exclusively in one or two secondary or hardly even secondary documents; perhaps transmitted from the autograph, and preserved by some rare accident of mixture notwithstanding the otherwise complete extinction of the line of transmission by which it had been conveyed, perhaps due only to a casual and unconscious emendation of an erroneous current reading. But these gradations of primitiveness in corruption have no practical moment. The only fact that really concerns us is that certain places have to be recognised and marked as insecure.
368. The number of such places which we have been able to recognise with sufficient confidence to justify the definite expression of doubt is not great. If we exclude books in which the documentary attestation of text is manifestly incomplete, as the Apocalypse, some of the Catholic Epistles, and the latter part of Hebrews, it is relatively extremely small. There may be and probably are other places containing corruption which we have failed to discover: but judging by analogy we should expect the differences to be of no real interest. We cannot too strongly express our disbelief in the existence of undetected interpolations of any moment. This is of course, strictly speaking, a speculative opinion, not a result of criticism. But we venture to think that the processes of criticism which it has been our duty to consider and work out have given us some qualifications for forming an opinion as to the probabilities of the matter. There are, it ought to be said, a few passages of St Matthew's Gospel (xii 40; [xiii 35;] xxiii 35; xxvii 9) in which it is difficult to believe that all the words as they stand have apostolic authority: the second part of xxvii 49 would have to be added to the list, if sufficient reasons should be found for accepting the possible but doubtful view that it is not a Non-Western interpolation, but an original reading omitted without authority by the Western text. But the question which these passages raise is rather literary than textual, for we see no reason to doubt that, as regards the extant form or edition of the first Gospel, their text as it stood in the autograph has been exactly preserved.
369. It will not be out of place to add here a distinct expression of our belief that even among the numerous unquestionably spurious readings of the New Testament there are no signs of deliberate falsification of the text for dogmatic purposes. The licence of paraphrase occasionally assumes the appearance of wilful corruption, where scribes allowed themselves to change language which they thought capable of dangerous misconstruction; or attempted to correct apparent errors which they doubtless assumed to be due to previous transcription; or embodied in explicit words a meaning which they supposed to be implied. But readings answering to this description cannot be judged rightly without taking into account the general characteristics of other readings exhibited by the same or allied documents. The comparison leaves little room for doubt that they merely belong to an extreme type of paraphrastic alteration, and are not essentially different from readings which betray an equally lax conception of transcription, and yet are transparently guiltless of any fraudulent intention. In a word, they bear witness to rashness, not to bad faith.
370. It is true that dogmatic preferences to a great extent determined theologians, and probably scribes, in their choice between rival readings already in existence: scientific criticism was virtually unknown, and in its absence the temptation was strong to believe and assert that a reading used by theological opponents had also been invented by them. Accusations of wilful tampering with the text are accordingly not unfrequent in Christian antiquity: but, with a single exception, wherever they can be verified they prove to be groundless, being in fact hasty and unjust inferences from mere diversities of inherited text. The one known exception is in the case of Marcion's dogmatic mutilation of the books accepted by him: and this was, strictly speaking, an adaptation for the use of his followers; nor had it apparently any influence outside the sect. Other readings of his, which he was equally accused of introducing, belonged manifestly to the texts of the copies which came into his hands, and had no exceptional character or origin. The evidence which has recently come to light as to his disciple Tatian's Diatessaron has shown that Tatian habitually abridged the language of the passages which he combined; so that the very few known omissions which might be referred to a dogmatic purpose can as easily receive another explanation. The absence of perceptible fraud in the origination of any of the various readings now extant may, we believe, be maintained with equal confidence for the text antecedent to the earliest extant variations, in other words, for the purest transmitted text, though here internal evidence is the only available criterion; and, as we have intimated above, any undetected discrepancies from the autographs which it may contain, due to other or ordinary causes, may safely on the same evidence be treated as insignificant. The books of the New Testament as preserved in extant documents assuredly speak to us in every important respect in language identical with that in which they spoke to those for whom they were originally written.
C. 371—374. Conditions of further improvement of the text
371. The text of this edition of course makes no pretension to be more than an approximation to the purest text that might be formed from existing materials. Much, we doubt not, remains to be done for the perfecting of the results obtained thus far. Even in respect of the discovery of new documents, and fuller acquaintance with the contents of some that have in a manner been long known, useful contributions to the better understanding of obscure variations may fairly be expected. It is difficult to relinquish the hope that even yet Lagarde may be able to accomplish at least a part of his long projected edition of the testimonies of the oriental versions, so that the New Testament may be allowed to enjoy some considerable fruits of his rare gifts and acquirements: a complete and critically sifted exhibition of the evidence of the Egyptian versions would be a specially acceptable boon. But it would be an illusion to anticipate important changes of text from any acquisition of new evidence. Greater possibilities of improvement lie in a more exact study of the relations between the documents that we already possess. The effect of future criticism, as of future discovery, we suspect, will not be to import many fresh readings; but there is reason to hope that the doubts between alternative readings will be greatly reduced.
372. We must not hesitate however to express the conviction that no trustworthy improvement can be effected except in accordance with the leading principles of method which we have endeavoured to explain, and on the basis of the primary applications of them which have been here made to the interpretation of the documentary phenomena of the New Testament. It is impossible to entertain an equal degree of confidence in the numerous decisions which we have felt ourselves justified in making in comparatively obscure or difficult variations; because in these cases a greater liability to error was involved in the proportionally larger part inevitably played by individual personal judgements. Even where a text is certain enough to make the exhibition of alternative readings superfluous, gradation of certainty is a necessary consequence of the manifold gradations of evidence. But, while we dare not implicitly trust our own judgement in details, the principles of criticism here followed rest on an incomparably broader foundation, and in an overwhelming proportion of cases their application is free from difficulty. As was said at the outset, the best textual criticism is that which takes account of every class of textual facts, and assigns to the subordinate method corresponding to each class of textual facts its proper use and rank. All that has been said in the intervening pages has been an attempt to translate into language the experience which we have gradually gained in endeavouring to fulfil that aim.
373. There is no royal road to the ascertainment of the true texts of ancient writings. Investigation of the history and character of documentary ancestries would indeed be out of place for the text of the New Testament if the documentary evidence were so hopelessly chaotic that no difference of authority could carry much weight as between readings all having some clearly ancient attestation. The consequent necessity of always judging chiefly by Internal Evidence of Readings would undeniably save much labour. But it would introduce a corresponding amount of latent uncertainty. The summary decisions inspired by an unhesitating instinct as to what an author must needs have written, or dictated by the supposed authority of 'canons of criticism' as to what transcribers must needs have introduced, are in reality in a large proportion of cases attempts to dispense with the solution of problems that depend on genealogical data. Nor would there be a material increase of security by the assignment of some substantial weight to documentary evidence, so long as it were found or thought necessary to deal with each passage separately, and to estimate the balance of documentary evidence by some modification of numerical authority, without regard either to genealogical affinities as governing the distribution of attestation or to the standard of purity which this or that document or group of documents habitually attains. Under all these circumstances the absence or neglect of the most essential kinds of textual evidence would leave a real precariousness of text which could be avoided only by an enormously increased exhibition of alternative readings.
374. For scepticism as to the possibility of obtaining a trustworthy genealogical interpretation of documentary phenomena in the New Testament there is, we are persuaded, no justification either in antecedent probability or in experience; and, if this be so, the range of uncertainty is brought at once within narrow limits. When it is clearly understood that coincidence of reading infallibly implies identity of ancestry wherever accidental coincidence is out of the question, all documents assume their proper character as sources of historical evidence, first respecting the antecedent lines of textual transmission, and then respecting the relation of each reading to these antecedent texts. Nearly a century and a half ago the more important ancient texts were clearly recognised, and the great subsequent accession of materials has but added certainty to this first generalisation, while it has opened the way for further generalisations of the same kind. Again, when it is seen that the variations in which decision is free from difficulty supply a trustworthy basis for ascertaining the prevalent character of documents and groups of documents, and thus for estimating rightly the value of their testimony in other places, little room is left for difference of estimate. Whatever may be the ambiguity of the whole evidence in particular passages, the general course of future criticism must be shaped by the happy circumstance that the fourth century has bequeathed to us two MSS of which even the less incorrupt must have been of exceptional purity among its own contemporaries, and which rise into greater preeminence of character the better the early history of the text becomes known.