by Maurice A. Robinson
Principles of Internal Evidence
The basic principles of internal and external evidence utilized by Byzantine-priority advocates are quite familiar to those who practice either rigorous or reasoned eclecticism. At least one popular principle (that of favoring the shorter reading) is omitted; other principles are cautiously applied within a transmissionally-based framework in which external evidence retains significant weight. The primary principles of internal evidence include the following:
Prefer the reading that is most likely to have given rise to all others within a variant unit. This principle fits perfectly within a primarily transmissional process; it is utilized by both rigorous and reasoned eclectics, and is the guiding principle of the Nestle-Aland “local-genealogical” method. For Byzantine-priority this principle has great weight: it is extremely important to attempt to explain the rise of all readings within a variant unit. The eclectic model continually evaluates variant units in isolation, attempting to determine in each individual case that reading which seems most likely to have produced all others within that variant unit.
The Byzantine-priority principle, on the other hand, insists on not taking a variant unit in isolation from the remainder of the text, but always to ask how the reading which appears to be superior in any variant unit fits in with a full transmissional overview. Such a procedure involves the readings of all the units in near proximity: how they developed, were perpetuated, and grew into their relative proportions among the extant data. This procedure elevates the overall value of this principle and serves as a check against excess in application. The principle is not negated, but modified.
The textual researcher always must ask whether the reading that initially appears to support the rise of all others in a given variant unit is equally that which by its transmissional history remains most likely to have given rise to all other readings in the surrounding text as a whole. If one initially assumes a reading with extremely weak transmissional support to be original, a sufficient explanation must be provided as to how other competing readings could have derived from the first, and also how such readings could have ended up in transmissional relation to neighboring variant units.
When such explanations become problematic, this in itself becomes presumptive that another reading in a given unit may in fact have been the source of all competitors, and that the researcher should reexamine the case instead of accepting what at first appeared most plausible when viewed in isolation. Only thus can a final candidate be established within each variant unit–“reasoned transmissionalism” at work.
The reading which would be more difficult as a scribal creation is to be preferred. This internal canon is predicated upon the assumption that a scribe would not deliberately produce nonsense, nor make a passage more difficult to understand. If a more common word stood in an exemplar, a scribe would not normally substitute a rare word. Yet scribes do produce nonsense accidentally, and at times may even obfuscate a plain and simple reading for unknown reasons. There needs to be a transmissional corollary of qualification: difficult readings created by individual scribes do not tend to perpetuate in any significant degree within transmissional history.
This principle can be demonstrated in any relatively complete apparatus by examining the many singular or quasi-singular readings which were never or rarely perpetuated. The same can be said for readings in small groups of MSS, whether due to family or sub-texttype ties, or by coincidence. Transferring the corollary to the primary principle, the more difficult reading is to be preferred when such is found in the transmissional majority of witnesses rather than when such is limited to a single witness or an interrelated minority group.
The reasoning behind this assumption is obvious: while a minority of scribes might adopt any difficult reading for at least a time, the chances are slim that the vast majority of scribes would adopt such a reading were a simpler one originally dominant from the autograph. The researcher still must demonstrate on internal grounds that the “more difficult” reading is in fact such, as well as the transmissional likelihood of that reading having been original within that variant unit.
Readings which conform to the known style, vocabulary, and syntax of the original author are to be preferred. While this principle is valid, its application in modern eclectic praxis is fraught with difficulties. Other factors, including transmissional history, need to be considered before a final stylistic determination can be made in regard to a given passage. Merely because kai or euquj are “characteristic” in Mark or oun in John does not mean that one automatically should prefer such a reading over the alternatives. Stylistic criteria taken in isolation can easily lead to wrong decisions if the degree and quality of transmissional support are not equally considered. A basic assumption is that scribes in general would be unlikely to alter the style and vocabulary of a given author when copying that which lay before them.
Further, in any given instance, a minority of scribes might create an intentional or accidental variation which either conforms the text to a writer’s style, or which moves the text away from an author’s normal style. Transmissional criteria serve as a check and balance against mere stylistic, syntactical, content, and vocabulary considerations, allowing one to arrive at a more certain result. Attention to transmissional considerations prevents a naive acceptance of a variant solely due to stylistic conformity, especially when such is dependent upon favored MSS which fluctuate stylistically within a given book
For example, what does one do with oun in John? Certainly this word is distinctive of Johannine style, and on thoroughgoing eclectic principles perhaps should always be preferred (although structural considerations might alter such a decision). Modern reasoned eclecticism seems to prefer oun only when supported by favored MSS, even if such support is limited.
On a transmissional-historical basis, oun when found in limited perpetuation among a small minority of witnesses would be ruled out due to lack of a reasonable amount of transmissional support.
Modern eclectic methodology cannot satisfactorily distinguish a Johannine from a non-Johannine oun on the basis of either internal criteria or a small group of favored MSS. There needs to be a transmissional criterion for authenticity, since cases such as this cannot be resolved by an appeal to style, to limited external evidence, or to the reading that may have given rise to the others. Transmissional considerations offer a better solution in such cases than do eclectic methodologies. Similarly, how would one handle variation between de and oun in John? That gospel actually uses de more frequently than oun (de Byz 231x, NA27 212x; oun Byz 201x, NA27 200x), even though oun is “stylistically Johannine.” De thus cannot be ruled out when opposed by oun. The optimal (and only) solution is a reliance upon all external evidence, coupled with a solid view of historical-transmissional considerations.
Readings which clearly harmonize or assimilate the wording of one passage to another are to be rejected. That scribes engaged in some harmonization or assimilation to parallel passages or contexts can be demonstrated repeatedly within the pages of a critical apparatus. Colwell noted that harmonization to parallels in the immediate context occurs more frequently than to remote parallels.
Yet, one must carefully guard against the assumption that verbal identity where parallels exist is presumptive evidence against authenticity. Merely because harmonization or assimilation could occur at a given location, one must not assume that scribes would harmonize whenever possible. Nor is scribal harmonization when it does occur more characteristic of the Byzantine-era scribes than any other.
Once more, transmissional aspects remain the primary basis for decision. The apparatuses demonstrate that most of the numerous cases of harmonization or assimilation did not perpetuate in any great quantity. While scribes did harmonize at various places, and that frequently enough, the vast majority of scribes did not accept or perpetuate such alterations to any significant degree. Even if parallel locations were known from personal familiarity with scripture, most scribes would not adopt or add to the text that which was not in the exemplar before them. Harmonization simply did not occur on the grand scale.
It would be a transmissional absurdity to assume numerous “harmonization-prone” scribes adopting a few dozen harmonizations into their Byzantine MSS while failing to continue the process in hundreds of other places where scribes had produced more plausible and attractive harmonizations–none of which were incorporated into the main stream of transmission. The question can be framed precisely: were scribes more likely in any given instance deliberately to revise the text in the direction of harmonization, or would they generally tend simply to copy and preserve what lay before them? The answer is provided only by examining the data in the apparatuses which evidences transmissional reality. One will find that most of the time scribes would maintain and preserve the text of their exemplar. When harmonization or assimilation did occur, it was sporadic.
The MSS which systematically harmonized to parallel passages were few (the scribes of Codex Bezae and various Caesarean witnesses are more typically harmonistic than what is alleged against Byzantine scribes). While certain Byzantine readings may appear to harmonize at various points, it would be a fallacy to charge the Byzantine scribes with a harmonistic tendency for the following reasons:
(a) the Byzantine MSS fail to harmonize in most situations;
(b) the alleged harmonizations within the Byzantine Textform are relatively infrequent;
(c) alleged Byzantine harmonization often fails to conform precisely to the parallel passage; and
(d) the Byzantine scribes fail to harmonize in hundreds of places where a minority of supposedly earlier MSS had created highly persuasive and attractive harmonizations.
Readings reflecting common scribal piety or religiously-motivated expansion and alteration are secondary. From a transmissional-historical aspect, this principle is viewed somewhat differently from that which is commonly held. Pious expansions or substitutions made by a single scribe or a small number of scribes are unlikely to gain acceptance within the manuscript tradition. Were this not the case, one would see a continual expansion of divine names and titles: “Jesus” becomes “Jesus Christ,” then “the Lord Jesus Christ,” then “the Lord and Savior Jesus Christ.” “Lord” would become “Lord Jesus” or “Lord God”; “Spirit” would become “Holy Spirit,” and so forth.
While such alterations and expansions can be demonstrated to have occurred frequently within the manuscript tradition, such cases remain sporadic, localized, and shared among only a small minority of scribes. Most NT scribes did not engage in wholesale pious expansion. Conversely, when a minority of witnesses might lack one or more appellatives, this does not indicate pious expansion by all other witnesses. The shorter reading may be due to accidental omission triggered by common endings (homoioteleuta) among the various nomina sacra within a phrase. One cannot presume that the majority of scribes would adopt piously-expanded readings on a merely coincidental but not systematic basis under normal transmissional conditions.
A minority of scribes, however, might easily expand deliberately or omit unintentionally. Were pious expansion indeed typical and dominant, one would wonder why most such cases were not adopted by the transmissional majority. One cannot have it both ways–scribes either conform to certain patterns en masse, or they practice certain habits on a primarily individual and sporadic basis. Since most vagaries produced by individual scribes remained unadopted within the transmissional tradition, there should be no doubt regarding the actual situation. An example of “limited perpetuation” is provided in 1Cor 5:5 (nomina sacra in caps):
th hmera tou KU NA27 P46 B 630 1739 pc Tert Epiph
th hmera tou KU IU P61 vid Y vgst
th hmera tou KU IU XU D pc b Ambst
th hmera tou KU hmwn IU XU A F G P 33 104 365 1241s 1881 al a vgcl syp, h** cop Lcf
While modern eclectic advocates might argue that all readings beyond the shortest (that preferred by NA27) are “pious expansions,” such an approach is too simplistic and ignores the transmissional and transcriptional probabilities that point clearly to the Byzantine Textform as the reading from which all the others derived.
The MSS comprising the Byzantine Textform (basically in NA27) did not adopt the remaining “natural” expansions found in other witnesses (KU IU XU or KU hmwn IU XU). Yet, had NA27 been original, it would be peculiar if nearly all the Byzantine-era scribes were to stop at KU IU without further embellishment, especially when such was found in supposedly “earlier” MSS from the Western and Alexandrian traditions. This argues strongly that the vast majority of Byzantine-era scribes did not create or perpetuate pious expansions, but simply preserved the text which lay before them in their exemplars.
It is transcriptionally more likely that the small minority of Alexandrian and Caesarean MSS (P46 B 630 1739 pc) reflect simple homoioteleuton from the Byzantine reading, skipping from -U to -U. A minority reading created by transcriptional error is far easier to accept than to rationalize such a shorter reading as the source from which only a partial expansion was made by the Byzantine majority.
The primary evaluation of readings should be based upon transcriptional probability.
This principle goes back to Westcott and Hort, and has no inherent weaknesses. Scribes did make errors and deliberate alterations, and readings need to be categorized and assessed according to their conformity to such scribal tendencies. Other methods apply this principle inconsistently, more or less commensurate with the preferences of the critic; the application of this principle thus becomes unfairly biased. A transmissional aspect needs to be recognized: an error or deliberate alteration made in a single MS or a few MSS is unlikely to be perpetuated in quantity.
The many singular and quasi-singular readings which exist demonstrate the unlikelihood of a transcriptionally-based scribal creation extending much beyond any MS or MSS which first produced it. The chances that any sensible alteration subsequent to the autograph would extend beyond a small group of localized witnesses would be slim. Indeed, such readings as characterize minority texttype witnesses generally remain small and localized. That any deliberate alteration or transcriptional error would gain the cooperation of scribes so as to dominate the entire stream of transmission is a null proposition: scribes demonstrably did not engage in such a practice on the grand scale. Earlier exemplars would serve to nullify the growth and widespread dissemination of most scribal alterations, thus holding in check the unbridled mass of minority variants. An important corollary follows:
Transcriptional error is more likely to be the ultimate source of many sensible variants rather than deliberate alteration.
Many variant readings have their root in transcriptional causes. While this principle includes all cases which produce pure “nonsense,” it also includes many in which the end result in some way “makes sense.” Sensible readings may arise from the simple omission of a letter, syllable, or word; so too readings produced by haplography, dittography, homoioteleuton or other forms of transcriptional error.
Even an error that produced a nonsense reading may result later in other sensible variants, created in an attempt to correct the earlier error. When examining any variant unit, one first should consider whether transcriptional factors could have caused one or more of its readings. A more plausible solution will arise from this approach than from an assumption of the less frequent deliberate alteration. While many readings can only be explained as due to intentional alteration, the primary principle remains of seeking first a transcriptional cause for variant readings. Many readings could be due to either accidental transcriptional error or intentional alteration; one always must weigh the evidence before settling on one cause over another.
Neither the shorter nor longer reading is to be preferred.
The reasoned eclectic principle here omitted is the familiar lectio brevior potior, or giving preference to the shorter reading, assuming all other matters to be equal–a principle which has come under fire even from modern eclectics. Not only can its legitimacy be called into question, but its rejection as a working principle can readily be justified. The net effect of such a principle is to produce an a priori bias on insufficient internal grounds which favors the shorter Alexandrian text. The underlying premise is faulty: it assumes that scribes have a constant tendency to expand the text, whether in regard to sacred names, or by a conflationary combination of disparate narratives, lest anything original be lost. Yet scribal habits as exemplified in the extant data simply do not support such a hypothesis.
Had the later scribes done according to all that has been claimed for them, the resultant Byzantine Textform would be far longer than that currently found: divine titles would be extensively expanded, parallel passages would be in greater harmony, and a universally-conflated text would dominate. Such simply is not the case.
The problem as usual is a text-critical leap to a conclusion refuted by a careful examination of the extant data. While scribes did engage in various practices which would produce a “longer” text, such occurred only on an independent, haphazard, and sporadic basis. Such minority scribal expansions can readily be discerned in any critical apparatus (even among Byzantine-era witnesses) and rejected on the basis of their minority support. Scribes simply did not expand or harmonize the text en masse, and any principle of internal evidence which suggests and is dependent upon the contrary becomes self-refuted by transmissional evidence.
The converse principle–that the longer reading should be preferred–is equally rejected. A few may argue thus, such as A. C. Clark and C.-B. Amphoux, who favor the Western type of text, but such no more can be applied mechanically to the text than can the “shorter reading,” despite any apparent logic or plausibility which may be adduced. Such a principle simply will not work within a transmissional framework. Further, it has a similar bias favoring the Western text, just as the “shorter reading” favors the Alexandrian text. Elements which reflect “normal” transmissional considerations should not be overthrown or negated on the basis of a built-in bias within a text-critical principle.
Part Six will be published July 13th.
This excellent article is reprinted with permission of TC: A Journal of Biblical Textual Criticism .
© TC: A Journal of Biblical Textual Criticism, 2001.