On Byzantine-Priority Theory
Stephen L. Brown
Introduction
This essay presents a brief introduction to and defense of Byzantine-priority theory, which posits that the original text of the NT is to be found in the consensus of the entire MS tradition. This essay will begin with two observations that point in this direction, briefly explicate Byzantine-priority theory and praxis, respond to certain objections to the theory, and finally point out some apparent deficiencies in the standard model of reasoned eclecticism. Note that a basic familiarity with the field is assumed in the following pages, though an attentive reader should be able to glean much from the discussion.
Observation 1: Scribal Inertia
One of the most important factors in the transmission of the NT text is perhaps the least interesting: Scribal inertia. There are many hundreds of places in the NT where scribes could have been expected to harmonize, clarify, amplify, ensure the orthodox interpretation of, or otherwise emend the text, but apparently none did so, even as great theological controversy beset the early church. Indeed, the theological battleground of Col 1:15-20 has astonishingly survived virtually unscathed at the hands of copyists. Similarly, most scribes seem to have borrowed readings from MSS other than their exemplar only rarely. Then too, excepting perhaps very small fragments, any two MSS, even if they originated in widely divergent regions and eras, will agree on the vast majority of their readings. Furthermore, based on full collations in John 18 and Jude, we can see that most minority readings are present in a small handful of MSS (see Tables 1 and 2). These data suggest that when errors came into existence, scribes did not generally circulate them widely. All these observations dovetail with the difficult conditions under which NT scribes worked and other practical factors. In short, scribes normally did little beyond copying the text that lay in front of them.
So far from recognizing this principle, NT textual criticism as normally practiced in recent history has stood athwart it. Bruce M. Metzger's textual commentary on the NT has tended to be the first and last resource for inquiry into text-critical matters for both students and scholars, and this commentary brims with claims or assumptions that scribes deliberately altered the text. These claims and assumptions often involve a strained sort of psychoanalysis.
Such thinking has begun to lead to the circulation of NT editions and translations characterized by peculiar and improbable readings.
However intriguing it is to imagine centuries of inventive scribes and a NT text at which these men took umbrage, such a scenario has little connection to reality.
Certainly, scribes did occasionally overthink, and the original text of the NT does possess some linguistic, literary, and exegetical puzzles. But to the extent that we can correctly characterize the scribal tradition as slow to emend, individual emendations can be assumed largely to have remained in the direct descendants of the MS affected.
Observation 2: Multiple Lines of Transmission
This brings us to another vital matter: Scribal inertia is particularly important to NT textual criticism because the NT text exists in a plurality of largely independent lines of transmission. A reading which sees widespread distribution in the MS tradition has in all likelihood attained this status because of its originality, while a reading with a limited presence is, by the same token, probably only a scribal misstep. We have already seen that borrowing happened only on a limited scale, but there are a number of other related points to consider.
For one thing, most or all books of the NT propagated their texts through multiple lines of transmission virtually from the moment of their release to the public. The book of 1 Peter provides one of the clearest examples of this phenomenon. According to 1:1, the letter was sent to a large number of churches (Peter addresses himself to several regions of Asia Minor, each of which would have contained a plurality of churches by that time). As Silvanus (5:12) made the rounds, he either brought with him an appropriate number of copies of the autograph, or he lent the autograph to a local copyist in each case. Presumably, a similar practice was used for 2 Peter, James, Revelation, and quite possibly for 1, 2, and 3 John and Jude. Matthew, Mark, and John likely saw their distribution commence much in the manner of classical literature generally: The author announced the completion of his work, possibly read it aloud (in this context, the reading would surely have taken place in a church setting), and allowed various learned friends to borrow, copy, and promote it. Luke and Acts were probably open letters, seeing a similar method of initial circulation with the added benefit of a wealthy and influential patron (Theophilus). Admittedly, the case is less certain with respect to those books which had their genesis in private correspondence (most or all of the Pauline epistles). Many or all of these books, however, seem with little appreciable interval to have been recognized as canonical and normative for the Christian faith.
For another thing, numerous transmissional lines can be identified from surviving MSS. Such is evident upon a perusal of the MS tradition, and recent computer-assisted analyses of collation data have allowed for identification of textual clusters with a degree of precision not hitherto achievable. While the traditional generalization that NT MSS fall into one of three or four categories (Alexandrian, Western, Byzantine, and possibly Caesarean) is not without value, the analyses just mentioned have suggested that one ought to take into account a significantly larger number of MS families. The Byzantine texttype in particular can be broken up into a number of groups, showing it to be anything but monolithic. Indeed, some of the clusters, having been isolated solely on the basis of readings, contain MSS of such geographic diversity that they themselves might be comprised of a number of independent lines of textual tradition. The vast temporal and spatial separation that exists among Byzantine MSS suggests that the Byzantine texttype does not speak with one voice, but rather represents the essential agreement of numerous sources whose shared ancestor goes back to antiquity. Readings on which the Byzantine MSS substantially agree cannot quickly be set aside.
Byzantine-priority Theory and Praxis
In light of factors like the foregoing, one ought not to overlook the fact that there is for the NT text a main stream of transmission (generally called “the Byzantine texttype” or “the Byzantine textform”) that can be readily seen to reflect the autographic text where its members are in principal agreement. It is worth quoting Maurice A. Robinson at length here:
The Byzantine Textform preserves with a general consistency the type of New Testament text that dominated the Greek-speaking world. This dominance existed from at least the fourth century until the invention of printing in the sixteenth century. Under the present theory, this text is also presumed in centuries prior to the fourth to have dominated the primary Greek-speaking region of the Roman Empire (southern Italy, Greece, and Asia Minor)—a large and diverse region within which manuscript, versional, and patristic evidence is lacking during the pre-fourth century era, yet the primary region of Byzantine Textform dominance in subsequent centuries.
From a transmissional standpoint, a single Textform would be expected to predominate among the vast majority of manuscripts in the absence of radical and well-documented upheavals in the manuscript tradition. This “normal” state of transmission presumes that the aggregate consentient testimony of the extant manuscript base is more likely to reflect its archetypal source (in this case the canonical autographs) than any single manuscript, small group of manuscripts, or isolated versional or patristic readings that failed to achieve widespread diversity or transmissional continuity. In support of this presumption is the fact that a consensus text—even when established from manuscripts representing non-dominant transmissional lines—tends to move toward rather than away from the dominant tradition.
Byzantine-priority theory posits that the Byzantine texttype is to be equated with the autographic text, and that it necessarily claims priority over all rival forms of the text. Rival forms of the text can plausibly be explained as localized phenomena characterized by common scribal failings at various points mixed with occasional text-critical emendation, interpolation, paraphrase, attempts to copy something sensible from a poorly-written or damaged exemplar, and sundry other alterations, generally limited to a small minority of witnesses.
Byzantine-priority praxis can be summarized thus:
Whenever a reading can claim support from an overwhelming proportion of known transmissional lines, no other reading can plausibly be accepted.
When the previous situation does not obtain, it is necessary to make recourse to a careful study of readings that compete for dominance, taking into account such matters as geographic diversity, authorial style, and scribal habits.
Defending the Byzantine Texttype
The theory just outlined involves the rejection of many readings long preferred by critics and the acceptance of many readings long condemned as secondary. This observation leads us to offer a reappraisal of the basic arguments used against the Byzantine texttype. These arguments are the following: (1) The Byzantine MSS stem from a secondary source or process; (2) they simply were the only MSS available to scribes when Christianity was legalized; (3) their late dates negate their value; and (4) internal evidence favors competing readings. We will address these issues in turn.
The Byzantine Texttype as a Secondary Source or Process
There have been two basic explanations advanced in an attempt to reduce the extremely large class of Byzantine witnesses to one or a few witnesses (of limited value). In what follows, we will briefly respond to these arguments.
Explanation 1: Syrian Recension
Westcott and Hort supposed the Byzantine MSS to be the descendants of a mid-fourth century recension done in Syria. Their only historical argument was a guarded suggestion that Lucian was responsible, on the basis that Jerome had charged Lucian with producing a recension of the LXX. But Jerome never said that Lucian produced a recension of the NT, and Lucianic MSS of the LXX have been found at Qumran, casting much doubt on Jerome's accusations. On a similar note, only one major effort in late antiquity by Greek-speaking authorities to promulgate a particular form of the NT text has been documented, and this effort, led by Marcion, failed to perpetuate itself. The historical case for a recensional origin of the Byzantine texttype is weak.
Just as weak is the internal case for a recensional origin. Arguments that an ancient recension of the NT would have included harmonizations, conflations, and smooth readings, as the Byzantine texttype is said to, amount only to speculation. What is known about ancient recensional activity points rather to the Alexandrian, and perhaps Western, texttypes, as bearing traces of secondary readings; the Byzantine texttype looks decidedly unlike any known form of ancient Greek recension. There are, moreover, readings in the Byzantine texttype that hardly reflect a bent towards an easy or uniform text even if such a bent could be said to have existed among Christians sufficiently learned, wealthy, and influential enough to promote a recension so successfully.
Explanation 2: Byzantine Evolution
Recent scholarship has favored the idea that the Byzantine texttype is the end result of an anonymous process, a product of textual evolution. But the principle of scribal inertia that we noticed above speaks against this theory: No theological, stylistic, or other trend seems to have led scribes to depart from their exemplars apart from scattered lapses of attention. Put simply, large numbers of scribes generally do not share erroneous readings. Moreover, in light of the existence of a plurality of transmissional lines, evolutionary developments would mainly afflict only a portion of the MS transmission, not the great majority.
Assuredly, to suppose that some exemplars were copied more often than others is hardly to make an improbable claim. But for all the importance of Constantinople, neither this city nor any other seems to have been a holy site to which individuals would travel to procure a copy of a single, venerated exemplar. A survey of monastic and other copying sites will show that NT copying was a decentralized phenomenon. There is no reason, then, to dismiss the mass of more or less Byzantine MSS as simply many copies of a faulty source, though of course the independence of testimony offered in the consensus of a textual cluster is more important than the number of MSS contained within it.
The Byzantine Texttype as the Last Surviving Text
One theory is that the Byzantine texttype gained dominance in the fourth century because it was the “last man standing”: The persecutions that raged against Christians up to the time of Constantine destroyed most Greek-language MSS, and when Christianity was legalized, the texttype that happened to have survived and to have subsequently perpetuated itself through copying was a secondary (and presumably localized) one, the Byzantine.
However, it does not appear that the persecutions were so severe as to leave only an infinitesimal remnant. So widespread a destruction would have been very difficult, given the geographical lengths to which the original form of the text (or something very close to it) would have spread by the early fourth century—locales as diverse as, say, Rome, Corinth, Antioch, Ephesus, Caesarea in Palestine, and Caesarea in Cappadocia. Note the significance of this last place in particular: The ancient underground cities of Cappadocia became places of refuge for many Christians. It appears that the Christians who fled there were well-supplied, so among their possessions must have been copies of the NT, copies which left these cities when the danger passed. One of the more heated early Christian controversies was the Donatist, and this controversy presupposes that only so many Christians cooperated with the authorities in the destruction of Christian texts.
If more than tiny fragments of the MS tradition survived, then one is faced with the improbable theory that persecutors largely destroyed MSS bearing a certain form of the text. The imperial authorities and local opponents of Christianity surely knew little or cared little about units of variation in the NT. In short, the theory of early persecution is an inadequate account of the dominant position of the Byzantine varieties of the text within the MS tradition as a whole.
The large degree to which readings of the Byzantine texttype appear in non-Byzantine witnesses also speaks against the texttype merely being a broad development from what was originally a purely local text. There are few Byzantine texttype readings that are especially difficult or peculiar in comparison to distinctively Alexandrian and Western readings; a text derived from a single exemplar other than the autograph would naturally contain various oddities.
Finally, we shall see that the Byzantine empire faced multiple periods of upheaval in which many MSS and MS repositories were lost. Yet minuscules that date to relatively late phases of Byzantine history essentially echo the earliest sorts of text found within the Byzantine realm (Basil of Caesarea, Chrysostom of Antioch, and the Gothic version ). If the Byzantine texttype of the fourth and later centuries survived into the later phases of the Byzantine empire, then it seems reasonable to suppose that the original text of the NT survived the first few centuries of its existence.
Manuscript Age
All other things being equal, an early MS has a greater claim on accuracy than a late one. But qualifications are in order: A late MS might stem from an early, well-done exemplar, while an early MS might come from a long chain of copies or from a MS marred by recensional activity, sloppy or inexact copying, or an attempt to wrench sensibility from a heavily damaged or poorly written exemplar. Then too, it seems normal for many textual variants to appear early in the life of a well-received document, so that as long as various lines of transmission come to be perpetuated, the age of a reading by itself proves little.
Thus, MS age cannot have the final say.
In the case of the NT, we have a number of relatively early MSS and a much larger number of MSS from the ninth century and onward. Those of the early MSS that are Alexandrian are subject to two criticisms: They seem to reflect recensional activity, and they seem largely to reflect regional phenomena. Those of the early MSS that are not Alexandrian attest to an early diversity, though not to chaos, in the early MS tradition. The later, largely Byzantine MSS are not subject to either objection.
Why is it that the early MSS largely come from a distinct textual group not well represented in later MSS? Climate is a major factor: Almost no MS of any document, secular or biblical, survives to the present, except in Egypt (apart from clay and other especially durable writing surfaces, not generally utilized as a channel of transmission for the NT text). There the dry climate vastly extends the life of written material. The difference of climate between Egypt and the larger Mediterranean world, of course, reflects geography; the geographic isolation of Egypt would naturally lead to a recognizably local form of the text developing over time, especially as the breakdown of the Roman Empire reduced the feasibility of long-distance travel and trade.
Another factor is historical. After the Islamic conquest, Egypt became for centuries a rather stagnant, uneventful place, standing in marked contrast to the Levant, Asia Minor, Greece, the Balklands, and Italy. In those lands centuries of persecution, plague, warfare, and civil unrest hindered those interested in propagating the NT text (in any case, an arduous business prior to the introduction of printing), so much so that for a period of time there was “a near cessation of copying manuscripts.” (Note, however, that not all lost MSS were burned; a common fate for old MSS in late antiquity was re-use, especially in receding economies. ) The sack of Constantinople in 1204 was particularly devastating to the world of literature, and while not all MSS from the first few centuries were lost in this event, no doubt many copies of important exemplars were. The relative uniqueness of the early MSS stems in part from the loss of most early representatives of the texttype reflected in the majority of MSS.
But do the late Byzantine MSS constitute the end of a long line of copying events, each of which introduced new errors? No scribe is perfect; a judicious comparison of a range of MSS is necessary in any case to eliminate errors. The main point to notice in this connection is that fewer transmissional steps than might be supposed seem to lie between many an unexceptional Byzantine MS and the autograph. Byzantine societal interest in the written page waxed and waned throughout the centuries. At times, pressing circumstances put most scribal work in an area or across most of the empire on hold: The horrific plagues from 542–544 and 558 surely killed many scribes, and the chaos of the Byzantine dark age curtailed literary activity for several centuries, meaning in this case that many ninth-century minuscules, copied as the Greek-speaking world emerged from that crisis, are likely copies of MSS from several centuries earlier. Byzantine MS copying was, in other words, spasmodic. New MSS were not continually produced from the latest generation of exemplars, but rather spurts of activity would motivate scribes to dust off and work with the old volumes of yesteryear. Furthermore, many Byzantine MSS were likely to have been made from exemplars significantly earlier than themselves when the format of these exemplars had become obsolete.
What is truly vital is not so much how old a MS happens to be, but where the text of that MS fits within a larger history of transmission. All readings that are not clearly polygenetic ought to be accounted for on stemmatic grounds. Even a glance at the diversity behind a few of the oldest Byzantine witnesses (codices A and W, the Gothic version, Chrysostom, and Basil of Caesarea) begs the conclusion that this form of the text predates the fourth century. Furthermore, the existence and nature of the MSS of the ninth, tenth, and eleventh centuries cannot convincingly be accounted for apart from the early existence of the Byzantine texttype.
Internal Evidence
Internal evidence principally concerns the identification of readings more likely to be original on the grounds of (1) intrinsic probability, which considers what the author is most likely to have written, given his style, the context of the variant, and so on; and (2) transcriptional probability, which considers how scribes were most likely to change the text they were copying. Both lines of evidence have been thought to support a basically Alexandrian text. In fact, neither does.
Intrinsic Probability
Take the first kind of internal evidence mentioned, intrinsic evidence. A few examples will show that the Byzantine texttype is not to be summarily dismissed on intrinsic grounds:
In Mat 8:28, Mk 5:1, and Lk 8:26 the sequence of readings found in many editions is Γαδαρηνῶν, Γερασηνῶν, and Γερασηνῶν, respectively, while the Byzantine reading in these three places is Γεργεσηνῶν, Γαδαρηνῶν, and Γαδαρηνῶν. Of these Γεργεσηνῶν is probably the only plausible term to use if a very specific locale is intended, and it seems reasonable to suppose that the author of Matthew, writing more than any other gospel writer with a Jewish audience in mind, would be the one most likely to employ great specificity on geography in the Levant. Γερασηνῶν implies an improbable location, while Γαδαρηνῶν refers to an important city in its area and so would provide a convenient point of reference for readers less familiar with the area, as Mark's and Luke's intended audiences likely were.
In Lk 2:14, the Byzantine reading is εὐδοκία, while the minority reading found in most critical texts is εὐδοκίας. The Hebrew-language phrases discovered at Qumran sometimes adduced in favor of the minority reading do not support it strongly, for they all include a third-person singular possessive suffix, and such would surely have been reflected in translation. James R. Edwards notes that “the word for 'favor' (Gr. eudokia, eudokein) means God's saving pleasure rather than humanity's good will whenever used in Luke (2:14; 3:22; 10:21; 12:32).” Edwards takes this usage to support the minority reading, viewing the Byzantine as ascribing the goodwill to certain men. Yet the opposite seems to be true: For the genitive εὐδοκίας to be understood as referring to God's election and not to the moral character of the men in question would seem to require some mark of specificity (τῆς εὐδοκίας, or better τῆς εὐδοκίας αὐτοῦ); εὐδοκία, however, is naturally parallel to εἰρήνη, which item (1) is itself parallel to δόξα (θεῷ), something man's efforts cannot achieve, and (2) comes from God's intervention, not man's, as the whole tenor of Luke's infancy narrative indicates (see especially 1:46-55, 66-79). Intrinsic evidence, then, favors the Byzantine reading.
Concerning the Byzantine inclusion of Acts 28:29, observe that every time Luke records a discourse of Paul, he also notices the reaction of the auditors (Acts 13:42, 48, 14:18-19, 17:32, 20:37-38, 22:22, 25:12, 26:24-32) with the possible exception of the discourse of 24:10-21 (but even there note verse 25). Thus from a stylistic perspective it seems more likely that Luke would record a reaction here than that he would not. The exact wording of the present verse is not lifted from any known source, and its very uniqueness is of a piece with the other reactions, none of which is verbally identical to another or to this verse. Furthermore, the verse looks Lukan: Forms of συζητέω and συζήτησις are rare in the NT as a whole, but occur several times in Luke-Acts.
In 1 Cor 14:21, the majority reading (P46, Ds, F, G, K, L, P, 365, 630, 1175, 1505, 1881, Byz, lat, (sy p), co, Mcion[E]) is ἑτέροις, while the minority reading (א, A, B, Ψ, 0201, 0243, 6, 33, 81, 104, 326, 1241, 1739, 2464) is ἑτέρων. Paul is here quoting from Is 28:11, and a glance at the Hebrew there will verify the majority reading.
A few MSS of Eph 1:1 (P46, א*, B*, 6, 1739, (Mcion[T, E]) omit ἐν Ἐφέσῳ. Because of the importance modern textual criticism has traditionally attached to some of these MSS, many editors and commentators have dismissed ἐν Ἐφέσῳ as inauthentic (or at least open to suspicion). Yet Black, in defense of the inclusion, notes that “2 Corinthians, Galatians, and 1 and 2 Thessalonians all lack personal greetings, yet all were written to congregations founded by Paul, as was the church at Ephesus. On the other hand, the Epistle to the Romans has more greetings than any other epistle of Paul, yet this church was not founded by the apostle ... In other words, it seems that the better Paul knew a church to which he was writing the fewer personal greetings he included.” Eadie's commentary agrees, stating that “Paul's long years of labour at Ephesus must have made him acquainted with so many Christian people there, that their very number may have prevented him from sending any salutation.”
Of course, one must take careful note of the limitations of intrinsic evidence. First, there is often a lack of scholarly agreement as to which reading intrinsic evidence supports; all must agree that intrinsic evidence is not always easy to assess. Some elements of the Greek language, in fact, might be too subtle for internal evidence in and of itself to highlight the right reading. Second, authorial style is variable. Third, ancient scribes could (like modern critics) become overly sensitive to perceived stylistic markers. Fourth, those interested in deliberate interpolation would no doubt tend to mimic an author's style. Still, intrinsic evidence often supports the Byzantine reading.
Transcriptional Probability
Turning to the second major type of internal evidence, transcriptional probability, we observe that NT scholars have long made certain assumptions about the ways scribes would tend to depart from the text of their exemplars, supposing addition, harmonization, conflations, and easier readings generally to characterize scribal error. But these assumptions amount to little more than guesswork.
A better approach lies in the use of singular and sub-singular readings. These readings, most scholars agree, are much more likely to be errors than to be original. Several NT textual critics have studied several thousand singular readings between them, with similar results: Scribes omit somewhat more often than they add (transcriptional error only accounting for around a third of these omissions), disharmonize more than twice as often as they harmonize, and make the style and / or sense harder vastly more often than they make it easier. This last statistic especially stands to reason, inasmuch as the principle of scribal inertia suggests that most variants arose inadvertently, and accidents are more likely to obfuscate than clarify the sense of a written passage. Furthermore, the intent of the various writers of Scripture was generally to be understood. The principle of preferring the harder of two or more readings is not to be abandoned altogether: It can be expected that a few perplexing passages in the original text prompted attempts at emendation.
In these passages, however, it seems more reasonable to suppose that a few related witnesses inherited the same error than to say that many scribes independently converged on one and the same alteration.
Even when one makes adjustments to the traditional canons, this aspect of internal evidence has its limitations. For example, while wrongful omissions certainly occur frequently, wrongful additions also afflict the MS tradition, and it seems unlikely that textual critics will always be able to predict with justifiable confidence under which circumstances the less common scribal blunder took place. Yet all in all, considerations of transcriptional probability provide no barrier to a favorable view of the readings of the Byzantine texttype, and tend rather to promote these readings.
Unpopular Byzantine Readings
While a full-scale textual commentary on even a small portion of the NT text is beyond the scope of the present essay, it would be worthwhile briefly to defend four readings long used to condemn the Byzantine texttype wholesale.
Mk 1:2
| τοῖς προφήταις | A, K, P, W, Γ, f 13, 28, 579, 1424, 2542, Byz, vgms, syh, (bomss), Irenlat |
| τῷ Ἠσαΐᾳ τῷ προφήτῃ | א, B, L, Δ, 33, 565, 892, 1241, syp, syh mg, co, Origpt |
| Ἠσαΐᾳ τῷ προφήτῃ | D, Θ, f 1, 700, l844, l2211, Iren, Origpt, Epiph |
Maurice A. Robinson offers a number of points in favor of the Byzantine reading, which we will summarize here. (1) Evidence for the common text is at least as old as evidence for the minority reading. (2) The Isaiah quotation nearly matches its source, while the Malachi quotation is more of an allusion; this, combined with the noticeable disparity in prominence between Isaiah and Malachi in the NT generally, could have led some scribes to conclude that Mark was only quoting one prophet. (3) All other quotations of Is 40:3 in the NT and in the Eusebian canons are specifically attributed to Isaiah, a circumstance that could have tempted many scribes, Alexandrian and non-Alexandrian alike, to assimilate here. (4) In all other places where Mark quotes Isaiah in his gospel, he does not cite his source, except in one location (7:6) where the quotation comes from Jesus himself. (5) Attempts to emend a seeming error of citation in Mat 27:9 made little impact on the MS tradition, so the situation should hardly be any different in a gospel less popular than Matthew's. (6) In Mat 13:35 only certain witnesses known or conjectured by Jerome (none of them extant) specify Asaph as the author of the psalm in question in accordance with the OT, but some of the same witnesses that have the “Isaiah” reading here seemingly erroneously supply “Isaiah” there.
Lk 24:53
| αἰνοῦντες καὶ εὐλογοῦντες | A, C2, K, W, Γ, Δ, Θ, Ψ, f 1, f 13, 33, 565, 579, 700, 892, 1241, 1424, l2211, Byz, lat, syp, syh |
| αἰνοῦντες | D, it |
| εὐλογοῦντες | P75, א, B, C*, L, sys |
The variation unit is notable not so much for its semantic content but for the question of whether the Byzantine reading is, as Westcott and Hort famously claimed, a conflation, and thus evidence of the secondary nature of the Byzantine texttype. James A. Borland makes a case for the long reading. Some of his main points appear in summary form here. (1) “If Luke penned but one of these words, praising or blessing, there is no feasible reason to explain the rise of the other lone word through the copying process.” (2) The minority readings are, by contrast, easy to explain as due to visual confusion over the two -ντες forms. (3) The minority readings are each attested by only one texttype, while the long reading is attested by a huge array of majuscules, important minuscules, and other witnesses. (4) “The Greek words for praising ... and blessing ... are characteristic of Luke's usage. In fact, [the former] is found just nine times in the NT, and seven of those usages are in Luke ... Luke uses the [latter] word ... 14 times in the [third] Gospel; twice as many as the other gospels combined.” (5) Similar constructions occur in the LXX in connection with the Temple, an important theme for Luke: Cf. Ps 135 (LXX 134); 2 Chr 5:13, 8:14, 31:2; and Neh 12:24 (LXX 24:12). External, transcriptional, and intrinsic factors all point to the long reading as the original.
Jn 7:53-8:11
The story known to scholars as the Pericope Adulterae (PA) is one of the most controverted—and lengthy—units of variation in the NT. (1) It is worth noting that many believers who conclude that the PA is not original to this gospel nevertheless find the content more or less unobjectionable and historically plausible. For example, Tregelles says of the PA, “Though I am fully satisfied that this narration is not a genuine part of St. John's Gospel ... I see no reason for doubting that it contains a true narration. There is nothing unworthy of the acting of the Lord Jesus in this history.” This general assessment stands in contrast to much apocryphal material, which includes, among other highly peculiar accounts, a Jesus that is hardly recognizable to Bible-loving Christians. (2) The PA has the right statistical profile for Johannine usage according to multiple vocabulary tests, a feat that we would expect to be difficult for a scribe without computational tools. For example, consider Jn 6:3-14 and 7:53-8:11. Both narratives are similar in length, contain about the same number of exclusively Johannine words, and contain about the same number of hapax legomena (words found only once in the NT). (3) Maurice A. Robinson lists dozens of verbal links between the PA and the Tabernacles discourse (Jn 7:1-10:21), and between the PA and the remainder of John. These verbal links strongly suggest not only a Johannine style for the PA, but a uniquely fourth gospel style; the PA seems designed to sit within this particular book. One set of verbal links is particularly interesting:
It is more than coincidental that the instances of the historical present in the PA (8:3 αγουσιν; 8:4, λεγουσιν) not only reflect Johannine stylistic practice, but more importantly, these identical historical presents occur in the same order in the following narrative of the man born blind (9:13 αγουσιν; 9:17 λεγουσιν), and in both instances (8:3; 9:13, 15, 16, 40) with Pharasaic participation.
(4) Without the PA, the surrounding passages become noticeably disconnected, in that Jesus in 8:12 seems suddenly and improbably to be in his opponents' midst. (5) The originality of the PA best explains the passage's transposition to various locations in different MSS. Robinson offers the following details:
[T]he various relocations of the PA all come from the tenth century and later and were clearly lection-related, intended to maintain the Pentecost lection (7:37-52; 8:12) entire. Even the omission of the PA in more than 250 minuscules tends to reflect the same lectionary-based concern, particularly since the PA was not read during the normal liturgical cycle (the Synaxarion) ... even the MSS that include the PA with obeli or other markings generally do so to indicate lectionary practice involving the Pentecost lesson, and indicating that the PA should be skipped (υπερβαλε) and resumed (αρχου) as not pertaining to that lection.
Indeed, while 85% of NT MSS contain the PA, the majority of lectionary MSS omit it.
1 Tim 3:16
| θεὸς | א3, Ac, C2, D2, K, L, Ψ, 81, 104, 630, 1241, 1505, 1739, 1881, Byz, vgms |
| ὃς | א*, A*, C*, F, G, 33, 365, 1175, Did, Epiph |
| ὃ | D, lat |
Wallace describes the Byzantine reading as “poorly” attested, basing this assessment on (1) the supposition that the Western reading derives from the Alexandrian, and (2) the fact that “[n]ot one firsthand of any Greek witness prior to the 8th century reads θεὸς.” But these arguments are not wholly convincing. As to Wallace's first point, the Western reading here might well be derived from the Alexandrian, but such would not prove that the latter is original. The Alexandrian reading might turn out to be a secondary development in the history of the text, in which case Western dependence on it is hardly a point in its favor. Similarly, the Western witnesses might have borrowed at this point from an Alexandrian MS. The change can be simply explained: If θεὸς (abbreviated as a nomen sacrum was misread on one occasion as ὃς, it might have been misread on another occasion as ὃ (due to a damaged or poorly written exemplar or to the hasty work of a copyist), the term μυστήριον bringing to a Western scribe's mind the neuter relative pronoun. It is also possible that the Alexandrian reading is a development of the Western one, a neuter relative pronoun that clearly referred to a person being emended to a masculine relative pronoun. The Byzantine reading cannot be dismissed on these grounds.
As to Wallace's second point, note the oddity of the emphasis on the eighth century, when Wallace goes on to claim that the Byzantine reading might have come into existence “in about the fourth century.” Besides this, one of the correctors in support of the common reading is believed to lie on the early side of Wallace's threshold; the NA28 introduction assigns C2 to the sixth-century, which is within a century of the date for the first hand. If a reading can be much older than the witnesses that bear it, then reliance on MS age as a major argument is tenuous. The Byzantine reading here cannot be dismissed merely by an appeal to this external criterion.
In favor of the Byzantine reading, recall that the pastoral epistles contain an unusual concentration of passages which state or strongly imply the deity of Christ. If Paul were to use the word θεὸς of Jesus Christ in any book of the NT, this would be the one. On intrinsic grounds, therefore, the theological tenor of the book and corpus in which this verse appears both favor the Byzantine reading.
There is also the literary form of what follows the variation unit in question. Wallace identifies it as poetry by placing this passage in a section of his grammar that argues that relative pronouns often commence poetic material. Yet of the NT passages cited there, not one is indisputably classified as poetry. In the case of 1 Tim 3:16, the portion of the verse starting at ὃς / θεὸς cannot be scanned as Greek poetry, regardless of whether the variation unit at hand is included. It is also far from standard, if it is to be described as a Hebrew poem (written in or translated into Greek): Hebrew poetry rarely sticks with one tense-form or orders terms so consistently, and it is not clear that the sort of parallelism that seems to align this material into three groups of two lines is especially Hebraic. It does not seem to be poetry, then, but it also is clearly not Paul's ordinary discourse, being a series of relatively balanced lines employing asyndeton. What seems most likely is that Paul is quoting or formulating a creed. He might be supplying an excerpt, in which case, if the Nicene and / or Apostles' Creeds echo earlier models, the first word of the creed would probably be Πιστεύομεν, making it virtually impossible for the relative pronoun to be nominative. More likely, Paul is supplying the entire creed, in which case θεὸς, and not a relative pronoun, would make good sense. Genre considerations, while complex, favor the Byzantine reading.
Finally, there is the question of why Paul includes 1 Tim 3:16 at all instead of either moving straight to 4:1 or mentioning some other grounds for careful conduct (compare 3:15). A good explanation for the content of verse 16 is the Jewish technique of word-linkage: Paul has just used the terms οἴκῳ θεοῦ and ἐκκλησία θεοῦ ζῶντος, and this usage leads him to employ a creed in which the word θεὸς appears prominently. Thus, the Byzantine reading helps to explain why this passage progresses as it does. When these considerations are placed alongside the diverse and ample external support for the Byzantine reading, the case for θεὸς here becomes strong; at the very least, it cannot be said that internal evidence condemns it unequivocally.
To summarize, Byzantine readings stand up well to careful scrutiny, though it should be repeated that the problem-solving potential of internal evidence is not unlimited. It would seem an improper use of an otherwise helpful tool to employ internal evidence without also considering external evidence carefully.
Objections to Modern Eclectic Editions
Not only are there strong arguments in favor of Byzantine-priority theory and reasonable answers to standard objections, but there are also a number of problems for standard eclectic theory and practice. Two of these problems are detailed below.
Problem One: Ancient Recensional Activity
Hellenistic scholars based in the library of Alexandria devised a loose but definable praxis for producing critical editions of important texts, a praxis widely adopted in the ancient world for both Greek- and Latin-language texts. This approach emphasized the elimination of perceived accretions, attention to authorial style, and piety. Part of their motivation for this recensional activity might have been the observation that a strain of MSS containing interpolations and other corruptions was in circulation, not an inaccurate observation (note the early existence of the Western texttype ).
The parallels with the Alexandrian texttype are clear: The Alexandrian texttype seems closely connected with Egypt (and plausibly with Alexandria specifically), it is characterized by the elimination of seemingly superfluous material, and the aberrations of the Western texttype might have spurred on its activity. The following are just a few examples of Alexandrian readings likely to reflect Hellenistic text-critical theory rather than a simple transmission of the original text:
In Mat 1:25, υἱόν is read by a few (mostly Alexandrian) witnesses for τὸν υἱὸν αὐτῆς τὸν πρωτότοκον, the longer expression being taken as an interpolation from Lk 2:7.
Mat 18:11 is similarly dropped in some witnesses, perhaps because it was considered an interpolation from Lk 19:10.
The entirety of Mat 23:13 (23:14 in NA28) is omitted in some witnesses as being impossible for the apostle Matthew to have written, since this passage turns a seven-fold woe into a less biblical eight-fold woe.
In Mk 9:44 and 46, ὅπου ὁ σκώληξ αὐτῶν οὐ τελευτᾷ καὶ τὸ πῦρ οὐ σβέννυται is omitted as being an interpolation from verse 48.
In 1 Pet 3:13, ζηλωταὶ is exchanged for μιμηταὶ on the basis of style, μιμηταὶ being considered a Pauline rather than a Petrine expression.
In Rom 8:1, the absence of μὴ κατὰ σάρκα περιπατοῦσιν ἀλλὰ κατὰ πνεῦμα most likely represents Hellenistic text-critical activity, early scholars suspecting the words to be an interpolation from verse 4. It is not clear why many scribes would add these words here.
In Rom 8:28, piety might explain the insertion by some witnesses of ὁ θεὸς, the realm of nature (πάντα) thereby being denied divine influence over the course of events.
Not every reading distinctive of the Alexandrian texttype bears the marks of ancient recensional activity, but many do. Indeed, while any reading that one could interpret as the result of recensional activity might be explained on other grounds (including, in theory, originality), the large number of probable alterations of the kind here described points to a recensional origin for the Alexandrian texttype. The Byzantine texttype, meanwhile, displays no pattern of probable recensional readings.
A weakness in this proposal must be pointed out: The general, or at least standard, practice among scholars at Alexandria working on non-biblical texts was not to omit altogether but to obelize (mark) passages deemed spurious. It is, however, a short step from the latter to the former, at least in the case of a carefully crafted critical text. Such a text, by virtue of the time and expense poured into it, would likely have been produced with the intent of that MS being a standard exemplar. If so, short readings of this nature can be expected to be well attested by the members of a textual group that can be said on the basis of other readings to be a tightly knitted group. Christian zeal to rid the NT text of impurities serves as a plausible, if admittedly theoretical, motive for taking the step from obelization to omission. There is, then, enough circumstantial evidence to suggest that passages might have been omitted on ancient text-critical principles, and the consistency with which such evident excisions appear in the Alexandrian MSS undermines the modern editions which rely so heavily on those MSS.
Problem Two: Neighboring Variation Units
Another problem afflicting most editions not done on Byzantine-priority principles is the presence of improbable combinations of readings liable to arise within a short passage: Eclectic editions all too often adopt in the space of a verse or two a sequence of readings for which the MS support is mutually exclusive, or nearly so. Take Lk 13:35. NA28 reads ἥξει ὅτε on the authority of D, but in the previous variation unit it sides against D on the issue of whether the word order should be ἴδητέ με or με ἴδητε. Thus, the NA28 text ἴδητέ με ... ἥξει ὅτε (comprised of two variation units separated by the word ἕως, for which no variation is recorded) is not found in any known MS. The critic who argues that (1) a MS with the text is in question once existed and that (2) this form of the text is that from which the entire MS tradition derived is on shaky ground.
This principle should not be pressed too far. As noted above, all MSS and distinctive clusters of MSS are likely to contain at least some errors; one probably should not expect the text of an edition purporting to display the autographic text to have substantial, if any, MS support over more than a few consecutive verses. But it is also clear that the NT text is extant in more than just one or two lines of transmission. If the number of transmissional lines were quite small, then the adoption of zero-support verses would be more legitimate. As it stands, however, cases like the one mentioned in the previous paragraph are highly improbable. This situation undermines the credibility of many editions.
In contrast to the various eclectic editions, a NT edition based upon the consensus of Byzantine MSS will rarely, and probably never, be subject to such criticism. Note that the Byzantine cohesion implied in this claim is not to be taken for granted: If scribes had frequently been whimsical, then the large number of MSS extant would likely leave little opportunity for a text that consistently had the support of a major proportion of the MS tradition. The fact that it is possible to avoid mutually-exclusive sets of witnesses within close-knit blocs of text suggests that unless most of the tradition descended from a single, recensional exemplar, there was an overall fidelity to scribal practice, a factor left out of sight not only in the adoption by modern eclectic editions of low- to zero-support verses and pairs of verses, but also in the adoption of readings supported by tiny and varying minorities.
Conclusions
The Byzantine-priority theorist will readily admit that much research remains to be done and that not all variation units can be resolved with equal confidence. But Byzantine-priority theory does accord well with the most salient factors relevant to NT textual criticism. One can approach the text of the NT as established on Byzantine-priority grounds with a justifiable confidence. Scribal imperfections have not left the truth of God unrecognizable in any extant NT MS. Yet even these generally minor scribal failings and idiosyncrasies can fade from view as one uncovers the common elements of the MS tradition as a whole and thereby gains the opportunity to read and reflect on the text which stands at the head of the last two millennia of MSS, printed editions, and electronic editions.
| Witnesses | Readings | Witnesses | Readings |
| 1 | 545 | 527 | 8 |
| 2 | 152 | 528 | 13 |
| 3 | 64 | 529 | 20 |
| 4 | 42 | 530 | 27 |
| 5 | 20 | 531 | 25 |
| 6 | 20 | 532 | 28 |
| 7 | 20 | 533 | 21 |
| 8 | 12 | 534 | 17 |
| 9 | 14 | 535 | 22 |
| 10 | 9 | 536 | 6 |
| 11 | 5 | 537 | 9 |
| 12 | 4 | 538 | 2 |
| 13 | 7 | 539 | 2 |
| 14 | 3 | 540 | 6 |
| 15 | 5 | 541 | 4 |
table 1: Frequency table of readings grouped by number of supporting witnesses in Jude, based on Wasserman's collation. Singular readings have a “Witnesses” value of 1, readings supported by two witnesses, a value of 2, and so on. The two peaks of the distribution are detailed here; all other numbers of supporting witnesses correspond to fewer than 10 readings.
| Witnesses | Readings | Witnesses | Readings |
| 1 | 560 | 1578 | 13 |
| 2 | 164 | 1579 | 6 |
| 3 | 87 | 1580 | 7 |
| 4 | 64 | 1581 | 9 |
| 5 | 45 | 1582 | 3 |
| 6 | 32 | 1583 | 6 |
| 7 | 30 | 1584 | 4 |
| 8 | 23 | 1585 | 7 |
| 9 | 19 | 1586 | 9 |
| 10 | 11 | 1587 | 12 |
| 11 | 8 | 1588 | 9 |
| 12 | 9 | 1589 | 10 |
| 13 | 7 | 1590 | 5 |
| 14 | 12 | 1591 | 10 |
| 15 | 7 | 1592 | 13 |
| 16 | 6 | 1593 | 12 |
| 17 | 8 | 1594 | 11 |
| 18 | 5 | 1595 | 14 |
| 19 | 4 | 1596 | 11 |
| 20 | 6 | 1597 | 10 |
| 21 | 6 | 1598 | 9 |
| 22 | 4 | 1599 | 7 |
| 23 | 4 | 1600 | 13 |
| 24 | 5 | 1601 | 9 |
| 25 | 6 | 1602 | 10 |
table 2: Frequency table of readings grouped by number of supporting witnesses in Jn 18, based on Morrill's collation. The two peaks of the distribution are detailed here; all other numbers of supporting witnesses correspond to fewer than 10 readings.