Since the earliest days that differences between N.T. manuscripts began to be observed,
critics of the text have been divisible into two main classes according to the aim that they set before themselves in dealing with
it. The one class is concerned to produce unity. An eclectic method is employed in order to preserve all that is felt to be best
in the multiplicity of texts. The other class aims at discovering, at all costs, the text that is the nearest possible to the
original. 'The almost universal tendency of transcribers to make their text as full as possible, and to eschew omissions' (Hort),
is reflected in some early editors, but not in all. The two methods are distinguished in an instructive passage of Eusebius, On
the Discrepance of the Gospels, quoted by Hort: (The New Testament in Greek,
1896, Appendix, p.31.)
For at this point [i.e. at xvi.8, ἐφοβοῦντο γάρ
- ephobounto gar] the end of the Gospel according to Mark is determined in nearly
all the copies of the Gospel according to Mark; whereas what follows, being but scantily current, in some but not in all copies,
will be redundant [i.e. as such should be discarded], and especially if it should contain a contradiction to the testimony of
the other evangelists... [That is the view of one class of critic.] While another,
not daring to reject anything whatever that is in any way current in the Scripture of the Gospels, will say that the reading is
double, as in many other cases, and that each [reading] must be received; on the ground that this finds no more acceptance than
that, nor that than this, with faithful and discreet persons.
The inclusive tendency, which is on a par with the harmonizing of commentators, is here sharply contrasted with what we think of as the critical method, the latter of which was chiefly characteristic of Alexandria, though there are manuscripts outside Alexandria which express a spirit of keen criticism. What were considered to be interpolations were carefully noted (both in pagan and Christian writings), and often obelized, or bracketed and dotted for deletion. The only course, however, open to an editor or commentator who desired as pure a text as possible was to employ the earliest manuscripts he could find. This was doubtless what was done by the scholars Origen and Jerome, who fill so important a place in the history of the text. The latter claims to have used ancient Greek codices rather than the emended ones containing the revisions attributed to Hesychius and Lucian. (Whether these attributions are correct, which Jerome makes in his Epistula ad Damasum, is very doubtful; cf. Sir F. G. Kenyon, The Text of the Greek Bible, pp.183, 209.) This, however, was not the bent of mind of the Christian world in general. The opposite tendency, which Eusebius evidently preferred, was followed by Lucian, who deliberately enriched his text with material drawn from manuscripts of the main types that existed in his day. His revision was the earliest form of what became the standard text represented by the mass of manuscripts that have come down to us, and in conformity with which nearly all Greek manuscripts have, to a greater or less extent, been corrected. It was a text 'smooth and readable in structure, and competently exact for all practical purposes' (Warfield). And it continued to be the text of Christendom, untouched by critical hands, till the Renaissance.
In 1514 was undertaken the first printed edition of the New Testament, that of Stunica in Cardinal Ximenes's Complutensian (Alca'la) Polyglot. But it was eight years before it saw the light.
In the meantime Erasmus, at Basle, prepared an edition in great haste to outstrip the other, and published it in 1516. The humanist worked ad maiorem Erasmi gloriam. He admitted himself that it was precipitatum verius quam editum. He used late medieval manuscripts of the Byzantine text, its rival—when it appeared seven or eight years later—being based on earlier manuscripts of the same text. Other editions followed, which were little more than reprints of these, especially of the Erasmian.
In 1550 appeared the magnificent edition of Stephanus, printed in Paris, which was almost entirely Erasmian. And an unsatisfactory revision of this, in five successive editions, by the reformer Beza was printed in the beautiful Elzevir 24mo editions issued at Leyden in 1624, 1633, &c. In that of 1633 it was stated that it contained 'the text now received by all'; and so the Stephanus-Elzevir text came to be known as the Textus Receptus, the Received Text, and is cited by the Greek symbol ς (= st for Stephanus). And reprints of the Stephanus or the Elzevir are the traditional text of the New Testament.
No Greek text intended to reproduce exactly that, which underlay our AV., has ever been printed.
Beza's fifth and last text, of 1598, was more likely than any other to be in the translators' hands, but they sometimes departed from it in retaining language inherited from Tyndale and his successors, which had been founded on the text of other Greek editions. They also adopted some readings, which Beza had mentioned in his notes, and others, perhaps, on independent grounds. (The RV in 1881 was translated from a text produced by a compromise between 'the text presumed to underlie the Authorized Version' and the text of Westcott and Hort.)
An impetus seems to have been given to the critical study of the text by the presentation to King Charles I of the codex Alexandrinus (A) in 1628. Nevertheless for two centuries no attempt was made to produce an edition independent of the Textus Receptus, and based upon the best manuscripts, because the material for it was as yet almost non-existent. One edition after another appeared in which the editors tried to revise the T.R. with the help of the manuscripts that were coming to light. This stage of textual criticism lasted from 1657 to 1831, the chief editions being those of Walton's Polyglot (1657), Fell (1695), Mill (1707), Wells (1709-19), all English scholars; and then a succession of German ones: Bengel (1734), Wetstein (1751, 1752), Griesbach (1775-1827), Matthaei (1782-8), and Scholtz (1830-6). The most important of these were those of Bengel and Griesbach, the latter being helped by suggestions of Semler, and by the need of reinforcing his position in 1811 against theories of Hug.
But in 1831 Lachmann began to work upon scientific principles laid down more than a century before by the genius of Bentley, whose intention of producing an edition had not found fulfilment. He constructed a text directly from ancient documents without regard to printed editions, and issued better editions in 1842-50.
He was immediately followed by Tischendorf (1840-72), whose eighth edition, published in parts in 1864 and 1872, is still the chief storehouse of variant readings, with which G. W. Homer's apparatus criticus to his Coptic editions may be compared.
The last important edition before that of Westcott and Hort was that of Tregelles, in parts from 1857 to 1879.
The advance in the science of criticism was made possible by the continuous discovery of fresh material, which led to the recognition that manuscripts were to be classified according to the type of text which they contained. Bentley saw that they could be divided broadly into an earlier and a later class, viz., as we can now call them, pre-Byzantine and Byzantine. Bengel accepted this division, calling them African and Asiatic. But he went farther and perceived that the earlier was not homogeneous, and he divided it into two families represented by codex A and the Old Latin. He recognized also that his 'African' on the whole was of more value than his 'Asiatic'. Griesbach anticipated modern results with great acuteness by naming the latter class Constantinopolitan, and the two families of the former Alexandrian and Western. Not only so, but he was the first to perceive with any clearness that different families were, in some cases, represented in different parts of the same manuscript; and he even dimly detected the hardest part of the problem—the mixture of texts of different types in the texts of manuscripts. The Alexandrian and Western he held to have been types at least as early as the third century, and the Constantinopolitan not earlier than the fourth or fifth. From the work of Hug it became clearer that the Western text had a wide and early currency; and he thought that it was a corrupt text universally current in the second century, of which the existing manuscripts (except D) represented three revisions. Tregelles helped to substantiate scientifically the fact that Griesbach's Alexandrian and Western texts were earlier than, and superior to, the Constantinopolitan. Thus the increase of material, and a growing insight into true methods, prepared the way for the work of Westcott and Hort, which must be studied next.
Internal Evidence | Genealogical Evidence | Mixture | True or False readings | Application of Principals | Page ^
The principles on which their edition is constructed are set forth by Hort in an Introduction and Appendix, which form the second volume of their The New Testament in the Original Greek (Cambridge and London, 1882); and those principles were summarized at the end of both the larger and smaller editions of their text. A second and corrected impression of the former appeared in December 1881, and the latter (a reprint of the former) in 1887.
The principles are of permanent validity; and scholarship owes a heavy debt to the two great Cambridge men for their clear grasp and formulation of them. But since science never stands still, their application of them has undergone some modification in the last forty years by further study and the discovery of fresh material.
The first thing to be done is to recognize the different kinds of evidence available for determining a given reading.
The most rudimentary form of criticism, when variant readings present themselves, is to adopt the one that seems to be the most probable. But even this cannot be done without weighing two kinds of probability—intrinsic and transcriptional.
The reader may consider the context, the grammar, the style of the author and his manner in other passages; and may decide, with regard to a given reading, either by itself or in comparison with other readings, what he thinks that the author meant to say. The trained reader will be more likely to arrive at the truth than the untrained. But an author does not always express himself in the best way possible; and what the reader imagines he must have said may sometimes be an improvement, but textually a corruption, of his actual words. And it may all too frequently happen that a reader does not fully understand the mind or the circumstances or the purposes of the author, so that he may corrupt without improving his words.
This is a safer, because a less subjective, basis of criticism. There are certain causes of corruption, mentioned on p. 374, which seem generally to operate in manual transcription of any kind. What Hort calls 'observed proclivities of average copyists' allow of generalizations on which transcriptional probability can be based. And the greater the experience of the trained reader in these sources of corruption, the more safely will he deal with this class of probability. But proclivities are awkward things to judge. It is extremely difficult for the most highly trained reader to determine which of various impulses may have acted upon a scribe. That which actually did, may not have been the one, which the reader might think the strongest. On the other hand, the reading which appeals to him as intrinsically the best may be only that which the scribe felt to be an improvement, and was therefore, in fact, a corruption. But in practice the two kinds of probability are not often in antagonism, because an ancient scribe was seldom able to make an 'improvement' which to the literary and historical sense of the trained student in modern times appears really better than the original. Transcriptional probability is chiefly of value when the trained student can feel the superiority of a given reading, and yet has reasons for thinking that an ancient scribe would probably prefer a variant. In such cases, which are of frequent occurrence, the mutual aid of the two kinds of probability can be of the utmost help in connexion with other methods of criticism. But when they coincide, and no likely cause can be assigned for the existence of a variant, then other methods must be sought to arrive at a decision.
An important part of the weighing of evidence is to consider whether a witness is normally credible and trustworthy; and the inquiry into the character and antecedents of a document offers a safer criterion than the mere balancing of probabilities, because it deals more with objective fact, and less with personal surmise. One fact about a manuscript that can be determined with approximate accuracy is its date. Sometimes a scribe actually dated it; sometimes the date is fixed within more or less narrow limits by external facts or records; more often the century, at least, to which it belongs is learned from the palaeographical details of the manuscript itself. It is far from being a final criterion, because a late manuscript may have been copied immediately from an early one; but broadly speaking, the later a manuscript the greater the number of corruptions it is likely to have inherited. The date, however, can be only a general guide, and by itself is useless in determining which is the better of two variants.
Here begins the first serious labour of the textual critic.
He must not be content with deciding upon reading after reading, as they occur, on the lines of Probability.
He must do so for the entire document in such a way that he becomes acquainted with its character as a whole, intimately enough to be able to gauge its relative value as compared with that of other documents that he has similarly studied.
There is only a certain proportion of its variants on which he can, at first reading, decide from Internal Evidence. There will be many others that have left him in doubt; but on studying them again, he finds that his valuation of the manuscript as a whole helps to turn the scale in several places. If he has come to feel that the manuscript as a whole is good, he will be predisposed to prefer its readings in many instances where the Internal Evidence was not clear enough for a decision. That is an important factor in Westcott and Hort's system.
And yet no single document is free from errors.
The student may feel certain that Internal Evidence sometimes condemns a reading in a 'good' document; but where his two lines of evidence are in conflict, nothing but personal caprice can lead him to follow either, if he does not possess a further criterion. Documents are never good or bad absolutely, but only comparatively, each having its obvious slips of scribes or translators. A good text was sometimes very badly transcribed, and vice versa.
But there are further difficulties.
A document containing more books than one may have been copied from more exemplars than one, which may have been of various degrees of excellence. Or—the most perplexing of all—a document may contain a mixed text, i.e. its text may be the result of an irregular combination of two or more texts belonging to different lines of transmission. So that the words 'good' and 'bad' cannot be applied to the document as a whole, but only to this or that element which has come to it from entire lines of textual ancestry.
Lastly, the Internal Evidence of Documents decreases in utility when, with the increase of the number of documents, several of them appear on general grounds to be 'good', and yet are in disagreement with each other, in which case the student is again reduced to the uncertainties of personal judgement.
Hort places this last, after the next type of evidence, because it would naturally ocme last in the order of discovery. But logically it must be placed at this point in the evolving of safe critical method. If the general value of one document can be gauged in relation to others by an examination of all its readings in the light of probability, it must be possible similarly to gauge the general value of a given group of documents in relation to other groups.
This has two advantages over the internal evidence of a single document:
(1) If a document has a mixed text, i.e. contains elements derived from different ancestries, it can be ascertained which elements have descended to every member of the group that is being studied, and which have not. And thus the various mixed elements in a document can be studied separately as though a different document were being studied in each case, because every reading (accident apart) goes back to a previous document from which it is derived.
(2) A very small group can be found 'good', while a very large one may be 'bad'. The counting of documents can play no part at all in textual criticism. All documents that contain a reading have inherited that reading from a common ancestor. 'Community of reading', as Hort says, 'implies community of origin.' But it should be observed that community in a true reading may imply only the common descent from the autograph. 'The only kind of consent between documents that shows community of origin [sc. short of the autograph] is community in error' (Burkitt). And so we are led to the best and surest kind of evidence that critics have learnt to use.
To gauge rightly the value of a group in relation to other groups, it is necessary to know the genealogical relationship of all documents to each other. If of ten documents containing the same work nine coincide in a reading, while the tenth has a variant, the subjective weighing of probability between group and group is far from adequate. For the nine might all have been copies of the same document, in which case the choice would not be between nine and one, but between two single documents. Or the nine might all have been copied from the tenth, and their variations be nothing but corruptions, and as far as those ten are concerned the reading of the tenth is to be preferred. But the former case affords the better opportunity for the critic, because the two single documents point back to a common source, i.e. a point nearer to the autograph. Or, once more, the nine might be found to fall into two sets, five descended from one lost ancestor and four from another; and in that case the five and the four and the one resolve themselves into three single documents, and the process of tracing back would continue.
Let us suppose that there are three groups α, β, and γ; and that, in a large number of cases in which we are confronted by triple variants, the documents in the three groups are found normally or frequently arrayed together in support of them respectively. This teaches us that for a large proportion of their text the documents in each group have a common ancestor. Or it may happen that in a large number of other cases one of these groups is divided against itself, and each division of it must have had its own ancestor. Where there has been no mixture the ancestor from which the whole group has inherited a reading stands nearer to die autograph than the ancestor from which a division has inherited its reading. Again, let us suppose that x1 and x2 are copies of x, and that y1 and y2 are copies of y. Where there has been no mixture, the x's can side against the y's; x1 or x2 can go over to the y's; and y1 or y2 can go over to the x's; but neither x1y1 nor x2y2 can side together against the other two: these are cross-combinations due to mixture.
This greatly increases the complexity of the task. From a variety of causes readings were introduced into manuscripts not from their proper line of descent but from one or more other lines. In this case there is no homogeneous text that can be traced back to an ancestor but, as has been said, to different elements in a document, each element representing, as it were, a separate document whose ancestor must be traced. One result is that the ancestor of the larger or more complex group cannot necessarily be assumed to stand nearer to the autograph than the ancestor of the smaller, because readings, which had previously only a narrow distribution, may have been given, in comparatively late times, a wide extension by favourable circumstances. The first step, then, is obviously to recognize mixture when we meet it. This is done most easily in the case of 'conflate' readings, i.e. where two variants are combined into one whole, forming a third reading. It is far more likely that the third is a combination of the other two than that the other two are independent simplifications of the third. If, then, we note a considerable number of conflate readings, and find practically the same groups of documents supporting the two shorter readings and the connate reading respectively, we learn that the third group is certainly tainted with mixture, while the other two contain at least portions of two ancient texts which were eventually mixed together. The groups are seldom quite constant, though there is generally a nucleus of documents within them that is. But we feel certain that the documents, say in a group a which habitually supports conflates, witness to a later and less pure text than those from which they are habitually absent, say β and γ. But mixture does not always reveal itself in conflation. Two variants might frequently occur which could not possibly be combined into one. One of the two, say the reading of group β, is simply taken over into the documents that habitually favour conflates (group α). But since these are known, the real evidence for the two readings in the case of mixture remains as before. On the other hand, in so far as its readings are not due to mixture, the ancestor of group a was a manuscript in the same line of transmission as the ancestor of group /?, and becomes an additional witness for the β reading.
We now know the way in which the existing documents reveal their ancestry. But it remains to be seen how far this enables us to distinguish true from false readings. First, it is obvious that if a manuscript A is extant we can disregard all its descendants. Apart from mixture, all readings in which they differ from A are wrong, except in the rare cases where a scribe may have hit upon the true reading by pure accident when A is wrong. If a manuscript B with a different text has no descendants, and A has a dozen, a reading in the latter must not be reckoned as having a probability of 13 to 1: it is simply A against B. If A is lost, its descendants still have only the weight of one document against B; but we must use their evidence in such a way that we can detect the errors which have been introduced into them since A, before we compare them with B—that is to say, we must reconstruct A from them. In practice, however, we do not start from classified manuscripts; we have to discover which belong to the same families by an examination of their common readings, and thus reconstruct the ancestors of the various groups, which may disclose themselves. The following genealogical tree is given by Hort as an illustration (Introd., p. 54):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
O |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|||||||||||
|
|
X |
|
Y |
|
||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||||||
|
|
α |
|
β |
|
γ |
|
δ |
|
ε |
|
||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
a |
|
b |
|
c |
|
d |
|
e |
|
f |
|
g |
|
h |
|
I |
|
k |
|
l |
|
m |
|
n |
|
o |
0 is a manuscript descended from the autograph; and the use of this tree, therefore, will take us no nearer to the autograph than 0. If, from another set of manuscripts, we could reconstruct another ancestor P, then 0 and P would point back to their common ancestor Q, nearer to the autograph. But when we have worked back with all the available evidence to the earliest ancestors we can, and have not yet reached the autograph, it is clear that genealogical evidence can help us no farther. And yet the period between the autograph and the earliest traceable ancestor was that in which it was most easy for corruptions to appear, because the books had not begun to be considered sacred. That is the position in which we are left with regard to every book of the New Testament. The number and complexity of the Gospel codices take us farther back than those of any other book; but in the last resort we are obliged to be content with subjective considerations, and fall back on the internal evidence of documents and groups.
But it is important to understand why complexity can be a help and not a
hindrance. In Hort's imaginary tree, in all cases in which (say) the γ group agrees with that of the δε groups against the αβ
groups, the αβ readings can be rejected
(except in the cases of mixture or accidental coincidence), because, while a corruption can have occurred in an intermediate
ancestor of αβ, it is possible for γ and δε to coincide only by having received the reading
through Χ and Υ from Ο. And there is nothing (except mixture or accident) to make αβ agree
with the autograph against Ο. The same is true when any one of α, β, γ, δ, ε stands against the others, or when (by mutual mixture among descendants of Χ antecedent to α, β, γ) any two of
α, β, γ stand against the others. The consent of Υ with any part of the descendants of Χ leads back through Ο to the autograph. The same must be true, therefore, in the last generation, when any one of
the five lesser sets is divided against itself.
But lastly, if mixture comes in from another line of descent than that of Ο, quite
different conclusions may be reached. If, say, γ sides with δε against αβ, it may
be that γ and Υ have both
been affected by mixture, so that the reading of αβ may be that of Χ and of Ο. Or αβ may have received a mixture from a text independent of Ο, and this
rival to Ο may have preserved the true reading of the autograph. But these suppositions need be
entertained only when the reading in question is actually found in fairly numerous manuscripts, or when there is other good ground
for supposing that mixture from without exists.
Such are the principles underlying the work of Westcott and Hort.
Their Introduction elaborates them, but this may serve as a summary of their main argument. Their application of them to the then known documents must now be sketched.
(Cf. E. C. Colwell, J.B.L. liv, 1935, pp.211-21.)
An overwhelming proportion of the readings common to the great mass of Greek manuscripts—cursives and late uncials'—is identical with those of Chrysostom's quotations, of his fellow pupil Theodorus, and their teacher Diodorus. The first named spent the last ten years of his life first at Constantinople as bishop and then wandering as an exile; the second was at Mopsuestia; and the third at Tarsus; but all three belonged to, and worked at, Antioch. Thus the fundamental text of late Greek manuscripts generally is that which was dominant in the second half of the fourth century, which Hort calls 'Antiochian or Graeco-Syrian'. The varying degrees of corruption of our better manuscripts cannot be understood unless it is realized that this 'Syrian' text was either contemporary with, or earlier than, the oldest manuscripts that have come down to us, and that every one of them, with the exception of B and early papyri, has been to some extent affected by it. All the non-Syrian texts to be found in our manuscripts are older than the Syrian.
To prove this Hort examined:
(1) eight conflate readings;
(2) ante-Nicene patristic quotations;
(3) the internal evidence of Syrian readings,
(1) In each case two short readings are found in two different groups of documents, and they are conflated in a third. And in each case the ancestors from which the two former groups were descended were older than that of the third group. He names them 'Neutral and Alexandrian', 'Western', and 'Syrian' respectively.
(2) On patristic writings he sums up by saying, '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 which have preserved the other ancient forms of text'.
(3) The authors of the Syrian text selected or combined, with many alterations of their own, the readings of at least three earlier forms of text, an Alexandrian, a Western, and a third.
The net result is that a reading that is distinctively Syrian is worthless. And in the case of any other reading the ancestor of the Syrian text has the value of only one manuscript siding with the group that contains that reading.
Hort wrote before enough material was available for a just appreciation of the nature of the text to which was given this name, which came down from Griesbach. He holds it in lower esteem than that which it has gained in recent times. He admits that 'it is not uncommon to find one, two, or three of the most independent and most authentically Western documents' attesting 'a state of the Western text when some of its characteristic corruptions had not yet arisen, and others had'. But this means that in some cases they must attest the earliest known readings, and probably more often than he was willing to admit.
Of characteristic corruptions he names three:
(a) Readings due to a love of paraphrase,
(b) Non-biblical alterations and additions,
(c) Assimilations, e.g. between parallel passages of the Old Testament, of Ephesians and Colossians, of Jude and 2 Peter, and above all of the first three Gospels. So he concludes that, 'whatever be the merits of individual Western readings, the Western texts generally are due to a corruption of the apostolic texts'. He recognizes their merit, however, in a small number of passages, all (except Matt.xxvii.49) in the last three chapters of Luke, where he believes interpolation to have taken place in all non-Western texts, but not in the Western. He names these 'Western non-interpolations', though modern scholars would say that this class is not homogeneous and that a simpler term would be 'Western omissions'. The trend of modern criticism is to recognize a larger number of passages in which the Western text has escaped interpolation where all other texts have suffered and at the same time to recognize, more than Hort did, the possibility of doctrinal modifications upon the text especially by 'Western' scribes.
It is not unnatural that a purer text should have been preserved at Alexandria with its exact grammatical School. Readings which are pre-Syrian and non-Western find the great bulk of their support in writers connected with Egypt, especially Alexandria and neighbouring places in north Egypt. But not there only; early non-Western readings were preserved in various degrees of purity in regions remote from Alexandria. Hort held, therefore, that it was misleading to use the term 'Alexandrian' for all such readings. It must be applied to those that are normally supported by distinctively Alexandrian authorities. 'The more startling characteristics of Western corruption are almost wholly absent from the Alexandrian readings... 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.' Thus the Alexandrian text, from Hort's point of view, was mostly the result of Alexandrian corrections.
There are, then, readings which are pre-Syrian, but neither Western nor Alexandrian. They may be seen, for example, when documents normally Western attest non-Western readings in opposition to other Western manuscripts, in cases where mixture seems to be improbable. Such documents attest a state of the text when it had been only partially Westernized, and presuppose an earlier text which was not Western at all. And they can be seen most instructively when both the Western and Alexandrian texts err, 'especially when they severally exhibit independent modes of easing an apparent difficulty in the text antecedent to both'. No manuscript, version, or patristic writer preserves this text in its original purity; it can be arrived at only by a delicate comparison of pre-Syrian groups. But the nearest approach to it is to be found in B (except in the Pauline epistles and Apoc.), and next to it, but a long way after, comes א. Of other manuscripts, which again come a long way after א, it may be said in general that those which have most Alexandrian readings usually have also most neutral readings.
Modern criticism has only confirmed the fact that B contains a purer text than any other known manuscript, and that Westcott and Hort's edition contained a purer text than any that preceded it. But their argument with regard to a neutral text has been called in question, and it is usual now to put B and א at the head of the Alexandrian text, not apart as 'neutral'. More recent editors have, indeed, reached not dissimilar results by other methods. And these methods must now be studied.
Like all scientific results, those reached by Westcott and Hort were a stepping-stone to more. Their solid contribution was the safe foundation which they laid in the principle that manuscripts are to be judged not by their age, nor the numbers in which they support a given reading, but by the type of text which they exhibit, which enables them to be grouped genealogically. Salmon (Some Thoughts on the Textual Criticism of the N.T., 1897, p.5), though he criticized their work, said that it could be called epoch-making quite as correctly as that of Darwin. And to a large extent the grouping, which they mapped out, holds good today. But since their time several discoveries have been made, and experts have been continuously at work, with the result that their conclusions are undergoing certain modifications.
Their first chief conclusion, which may be said to be permanently established, was that the text, generally speaking, of later manuscripts, which superseded all earlier texts, was the result of a revision which was officially, and became universally, approved. No manuscript containing the actual text of the revision survives (unless it is 77, as Mrs. Lake's work suggests (Family II and the Codex Alexandrinus, 1936; cf. B. H. Streeter, J.T.S. xxxviii, 1937, pp.225-9.)), but in a slightly modified form it became authoritative throughout the Byzantine Empire from the ninth century. Hence the name 'Byzantine', akin to Griesbach's, is preferable to the name 'Syrian', which Hort uses (which might be thought to have something to do with 'Syriac'), or 'Antiochian' as it is sometimes called. But it no doubt emanated from Antioch, since it was used, as has been said, by Chrysostom, who worked there for nearly twenty years, and by no writer before his date. In 398 he became Patriarch of Constantinople, where the text was speedily adopted. The revision was possibly made about 300 by Lucian of Antioch, who was martyred there in 312. His authorship is supported by two statements of Jerome:
1. In his preface to his Vulgate Chronicles: 'Alexandria and Egypt in their Septuagint extol the authority of Hesychius; Constantinople to Antioch approves the manuscripts containing the text of Lucian the martyr; between these, the provinces of Palestine read the codices edited by Origen, which Eusebius and Pamphilus published. The whole world is thus divided between a threefold variety.'
2. In his preface to his Vulgate Gospels he speaks of the multitude of differences in Latin manuscripts, which must be corrected from the fountainhead of the Greek; but he will have nothing to do with the versions attributed to Lucian and Hesychius, of which he speaks very slightingly. But it must be noted that he may be referring only to the Greek Old and not to the New Testament at all. From the modern point of view the revision had deplorable results. Once an approved text had been issued, corrections were gradually made in earlier manuscripts to bring them into conformity with it; and of their descendants, the Greek manuscripts that we possess, not one (except probably B) escaped the infection, except the earlier papyri.
The revision was felt to be necessary because two centuries of corruption had produced a bewildering variety of texts. But when the unrevised elements in our manuscripts are examined, the texts are found to belong severally to certain areas. In the two centuries during which the books were only gradually becoming recognized as inspired, the manuscripts were subject to all the ordinary mistakes and corruptions of scribes, most of whom were not trained copyists but poor and often ignorant amateurs.
As soon as there were numerous copies of a book in circulation in the same area, one copy would constantly be corrected by another, and thus within that area a general standard of text would be preserved. But what we have to consider is that it is unlikely that the errors in the first copy of the Gospel of John, for example, which reached Rome would be the same as those in the first copy which came to Alexandria; and as each of these would become the parent of most other copies used in those respective cities, there would, from the very beginning, be some difference between the local texts of Rome and Alexandria. ... In this way local texts would inevitably develop, not only in the greater, but also in the smaller centres of Christianity. But along with a growing veneration for the text as that of inspired Scripture, there would come a tendency, whenever a new copy of the Gospels for official use in the public services was wanted, to lay more and more stress on the importance of having an accurate text. This would naturally result in the smaller churches obtaining new copies from the greater metropolitan sees, since these would be thought likely to possess a pure text. From these any copies in private hands in the smaller churches would be corrected. Thus the local texts of smaller churches would tend to become assimilated to those of the greater centres in their immediate neighbourhood [Streeter (The Four Gospels, 1926, pp.35 f.)].
When we look for local texts we find three, at any rate, represented in the Coptic, Syriac, and Latin versions, which would be needed for missionary work in the areas where those languages were respectively spoken, i.e. in Egypt, in Mesopotamia, and in Italy, Gaul, and Africa. And they would be translated, for the most part, from Greek manuscripts which had developed each its different type of text in its own area. The descendants of those manuscripts may have suffered, in different ways and degrees, mixture and assimilation to the Byzantine standard; but we may antecedently expect to find, by genealogical methods, types of text which may be called Alexandrian, Eastern, and Westerp.
The earliest Father whose writings afford any evidence of the text current in Alexandria is Clement. The text of his quotations is rather a puzzle when compared with that of Alexandrian manuscripts. P. M. Barnard shows that it is strongly coloured by 'Western' elements (Texts and Studies, v. 5, Cambridge, 1899.). But he had travelled in south Italy, as well as in Syria and Palestine, before he went to Alexandria. And Streeter (op. cit., pp. 57 f.) suggests that he took with him thither a Western text, and that becoming familiar with an Alexandrian text which he heard read in church, he used Alexandrian readings when he quoted from his memory of what he heard, but Western readings when he used his own manuscripts brought from Italy. Further, that his pupils may have noted some of his Western readings, which thus became perpetuated in the Alexandrian text. But this does not explain all the phenomena. There are also readings which he may have brought from the East, and which point to pre-Origenian types of which little is known.
Origen is the first biblical scholar who is known to have interested himself in textual matters. But while his work on the LXX is well known, there is no clear evidence that he revised the text of the New Testament. And yet he was not a man who would be likely to accept the popular text without critical caution. The oldest papyri suggest that the B type of text was known in Egypt before his day (But L. Cerfaux would challenge this. He thinks that there is no trace of the B-א type of text in Alexandria to be found in second-century papyri or Gnostic authors (Ephemerides Theolagicae Lovanienses, xv, 1938, pp.674-82)) and he cannot be considered the source of it. But if he used there (and at Caesarea) a type of text which compromised between the short, austere B-text and the more popular and 'gossipy' D type of text, his choice no doubt influenced many scribes who resorted later to the library of Pamphilus (Cf. Lagrange, Revue Biblique, iv, 1895, pp.501-24.). However much older than Origen the B-א type is, there is an increasing agreement among scholars that it cannot be placed on quite so high a pedestal as Westcott and Hort placed their 'Neutral' text, a transcendent text raised above local corruptions whether Western or Alexandrian. Salmon complained that the name is 'question-begging' (Some Thoughts on the Textual Criticism of the N.T., p.49.); founded on the quality of the text, it presupposes the final establishment of their theory. The strongest evidence for it comes, on their own showing, from Alexandria. But our conceptions of what they called Western have been altered by the discovery of the Sinaitic Syriac, and by the higher value which modern study places on the ancient versions, of which the Old Syriac and the Old Latin are older than our oldest manuscripts. Alexandrian scholars were on a higher level of education and training than any others, but might be 'for that reason the more exposed to the danger of treating the text of the Gospels by the same a priori methods as their heathen teachers and contemporaries treated the text of Homer' (C. H. Turner, Theology, ix, 1924, p.222 and J.T.S. xxviii, 1926-7, pp.145-58). When readings which can reasonably be regarded as 'Western' corruptions in all parts of the world have been discarded, there remain readings in the local texts of Europe and Africa on the one hand, and the East on the other, including those of early Egyptian papyri, which may, more frequently than Westcott and Hort imagined, be truer to the original than the Alexandrian.
Hort thought it possible that the group that he called distinctively 'Alexandrian' represented the 'Hesychian' revision. Bousset, on the other hand, (Texts u. Untersuch. xi, 1894, 4. 92.) suggested that B, the writing of which can have been only a few years after the revision, represented it. But the occasional Western, as against distinctively Alexandrian, readings in B render this doubtful. Whether it was the work of Hesychius or not, the text is on the lines ofOrigen's, but with a more ruthless pruning away of every trace of 'Western corruption'. That is to say, it is the early text of Alexandria refined in the crucible of Alexandrian scholarship, but not infected with the 'Alexandrianisms' found in the later local text.
The manuscripts of this later text suffered not only from the infiltration of 'Western', i.e. unrevised, elements, but from grammatical and stylistic 'improvements', such as would be natural in the home of classical scholarship. Even א is not quite exempt, but they are more frequent in C 33 Θ Δ (Mk.) Ψ (Mk.), and most frequently of all in L; they are found also in the quotations of Alexandrian writers, especially Cyril, and in the bohairic version. It was to a text of this type that Hort confined the name 'Alexandrian', a 'partially degenerate form of the B text', as though the degeneracy in this direction constituted the Alexandrianism. It is truer to say that the later Alexandrian text is a degenerate form of the earlier. It is throughout a local text, as Salmon long ago insisted.
Finally, while 'Lucian's' recension, at the end of the third century, adhered more closely to the Alexandrian than to the Western text, the manuscripts of the later Alexandrian text suffered, on the whole, more than the Western from 'corrections' to bring them into conformit