Source Greenlee Harold Introduction to New Testament Textual Criticism 2010

Don Wilkins

Collecting manuscript evidence is a laborious process, but it is a straightforward in comparison to the evaluation process. In the collection process, the goal is to gather as much evidence as possible concerning various readings of a specific text. In the evaluation process, the aim is to determine which reading has the best support as the original reading. The evaluation process is complicated by the fact that scholars disagree about some of the evaluation principles and their relative importance.

There are several ways we could approach the matter, including the composition of a history of criteria sets created by textual scholars. My coauthor Mr. Andrews has already provided us with a summary history of criteria in preceding chapters, and space does not permit me to expand on the history even if I desired to. Moreover, except perhaps for the Alands (sadly Kurt Aland has passed on), I doubt that any two scholars completely agree about the criteria. I have therefore chosen to use Metzger’s criteria as a foundational set, or rather the set most recently published by Metzger and Ehrman,[1] whom Metzger mentored. Metzger’s The Text of the New Testament is probably the best handbook on textual criticism that has been used in the classroom. While going through the ME (fourth) edition I will note differences with Metzger’s third edition,[2] the valuable outline he provides in his Textual Commentary,[3] and I will add observations from my own experience. I will also include the perspectives of other prominent textual scholars, most notably the Alands,[4] and most recently, Tommy Wasserman.[5] The Alands’ perspective seems nearly to be that of respondents to Metzger, though they do not refer to his work in their discussion of criteria. Wasserman acknowledges repeating Metzger’s list and offers comments on various criteria, all of which shows that we are in good company to take Metzger’s work as our foundation. After this, we will look at some additional criteria recommended by the Alands. I am going to describe Metzger’s work in the present tense as if he were still with us–if only that were so!

I believe I owe the reader more than just another regurgitation of the criteria that can all be found elsewhere, however.  What will be particularly different about this discussion of criteria is that I intend to provide judgments on the value and practicality of the criteria for those who are aiming at some proficiency in the practice of textual criticism, either because they sincerely want to do so or because the demands of their occupations or special circumstances require it. I am painfully aware that I risk revealing my own ignorance about aspects of the discipline that may have escaped my notice, and I strongly encourage the reader to verify or further examine anything I say that may seem mistaken.

I should also point out that I am aware of the duplication the reader will encounter in this chapter after reading the preceding chapters on scribal tendencies, textual scholars, and other subjects by Edward Andrews. When Edward approached me about contributing to this project, we agreed to write chapters independently. It is my hope that our different perspectives and emphases as conservatives will be helpful and compensate for the duplication, which may also serve to cover important details that either of us might have otherwise overlooked.

Metzger-Ehrman; the Autograph

To begin, then, the first thing anyone familiar with Metzger’s work would want to know about the Metzger-Ehrman fourth edition is how it differs from previous editions. Most of what Metzger taught in previous editions has simply been duplicated in the fourth edition, and I will comment on that in passing. The two fundamental principles remain the same: 1) the reading that best explains the origin of the others is to be preferred, and 2) the reconstruction of the history of a variant is prerequisite to forming a judgment about it.[6]  The Alands, who summarize their criteria as a list of twelve “basic rules,” essentially have these two principles under rules eight and nine. They maintain that a reconstruction of a stemma of readings for a variant is extremely important because the reading that most easily accounts for the derivation of the others most likely is the original. Under rule nine they state that variants must never be treated in isolation, but always considered in the context of the tradition.[7]

Let me flesh these principles out and briefly discuss where we stand in relationship to them today. We obtain additional help from three of the Alands’ twelve rules, the first of which (rule one) is that only one reading can be original.[8] This probably seems an obvious reference to the autograph in the view of conservatives, but in current textual criticism (and for some time now) the assumption of a single autograph for a book or epistle is old-school and arguably obsolete. Indeed, Today the goal or end target of textual criticism is increasingly debated, with considerable doubt expressed as to whether it is possible to reconstruct the autograph. The problem, which should be obvious to all, is that if changes were made to the author’s text before it was published as the initial text, we have no means of determining what the changes were. The hypothesis of the author’s maintaining one or more copies of his own work, and later editing and publishing them as equally authoritative revisions, poses a similar problem. When we encounter one or more witnesses that plausibly could be viewed as possible revisions from the author’s hand, we have no way to determine if the differences go back to the author, or were created by scribes.[9]

As we will refer to it later in our discussion of the CBGM, the goal usually identified in modern criticism is a reconstruction of the Ausgangstext or the “initial text.” For conservatives and other optimists, the initial text can be equated with the autograph, but those who are less conservative or otherwise dubious about the reconstruction of the autograph usually prefer the term or the concept Auggangstext because they aim at recovering the first “published” copy of a New Testament (hereafter NT) work that in turn became the origin of all subsequent copies–or all subsequent copies like it. Various theoretical options are left open, including the possibility that the author himself may have edited and published more than one version of it, as we just discussed. Critics who opine that the author’s secretary edited and revised his work before it was published, or that the initial text was a later work redacted by the Christian community, can put a considerable distance between the autograph and the text that was actually the origin of the whole textual tradition.

To illustrate how far this can be taken, I mention a book by John Walton and Brent Sandy, who use their thesis of the priority of [10]oral tradition to argue that differing variants were accepted as a legitimate aspect of ancient storytelling and reporting, even when they seemed contradictory by modern standards.[11] If we operated from their viewpoint, we would have to assume that all variants–at least early ones–may represent equally authoritative versions of an account. There would be no value in searching for a single autograph. Like different versions of the lyrics to “When the Saints Go Marching In,” each would be authentic, and none would be the official version, so to speak.

One of the fatal flaws in this freewheeling approach is the absence of credible support for a terminus at a point in history when variants can be excluded from consideration. Walton and Sandy leave it to the community of faith to decide and maintain that the acceptance of the canon was the cutoff point. They attach inerrancy and authority to the final, canonical form, not to the autograph.[12] The problem with this final product, which they describe as “a pristine canonical final form,”[13] is that it still is not in any kind of final form textually, as anyone can see in the apparatus of any Greek NT. Ironically, the authors would depend on textual critics to reproduce such a text. Moreover, if Christians like myself who do translation or textual criticism are still making decisions about readings–and of course we are–then the work of the community of faith continues toward reconstructing the final form of the NT.

As we will also see later in our discussion of the CBGM, however, since the initial text can be equated with the autograph in that system, there is no rational or practical reason compelling us to discard the concept of a true, traditional autograph. I prefer to hold not only to the concept but to the quest of reconstructing it as well. In fact, I acknowledge the autograph as the foundation of my own personal belief system and my work as a translator.

Beliefs and preferences aside, however, the possibility of more than one initial text would rule out the Alands’ “rule” that only one reading can be original, as a working hypothesis. There could still be an agreement in principle that the NT author produced a first draft by one means or another, but if the textual critic (hereafter just “critic”) assumed that more than one initial text might have resulted from it, he or she would have to process variant readings differently. That is, a substantive, deliberate change of the text (as opposed to a careless scribal error) would have to be considered part of a different textual tradition, at least potentially.

To see how this might play out–or unravel–I offer an example from the beatitudes. We have Luke’s account (6:20) that Jesus declared, “Blessed are the poor,” while Matthew (5:3) reports it as, “Blessed are the poor in spirit….” We do not have any extant variants reported in NA28 for Matthew 5:3, but for Luke 6:20 one or more seventh-century correctors added “in spirit” to 01 (Sinaiticus), and the phrase is found in a number of late Greek manuscripts and some versions, including a fourth-century Latin manuscript. I will discuss the criteria that we would use in this situation shortly, but if we ignore issues of the date and quality of the Luke manuscripts with “poor in spirit,” we might conclude that there were at least two authoritative versions of Luke, one with the additional phrase and one without it. We already know, after all, that two authentic versions of this beatitude existed.[14] The question would be whether we follow traditional TC criteria and reject “in spirit” in Luke, or we decide instead that the witnesses featuring it go back to another initial text of Luke. If I came to the latter conclusion as a translator, I would want to provide the reader with both versions at an equal level of authority, perhaps in parallel columns. For more complicated combinations of readings in other passages, three or four columns might be required!

In the CBGM, if I may again anticipate our discussion of it, all the variables of the initial text concept are recognized, but the system itself is programmed to trace all witnesses back to a single initial text. In fact, as a matter of simplification, the ECM2 editors operated with the goal of reconstructing the autograph (the “authorial texts”) for the General Epistles.[15] I must immediately add that the editors have not taken a theological position, such as I have, however. They also state that their hypothetical witness A “represents” the initial text and is a “potential ancestor of all other witnesses.”[16] This proves to be a purely statistical viewpoint, but I dare say that the practical outcome of the CBGM research will continue to be a single, albeit updated witness A.

However, there is a situation that virtually exemplifies what a stemma would look like in the event that there were two initial texts (and two textual traditions): the split line in the ECM2. An example can be found in 1 John 1:4, where the CBGM online tools supply the following stemma for the variant “our/your”:

1J1

(a)  ημων

(b)  η ημων

(c)  υμων

What the stemma is telling us is that reading b clearly derived from reading a, but the editors came to the conclusion that either reading (a) or (c) had equal claim as the initial text.[17] So by analogy, a critic who takes the position that more than one initial text of a book or letter may exist can, in turn, declare that a line of inheritance between two or more readings is void. In the case above, he could argue that the question mark might as well be the author or the originator of the work. This particular case is also interesting because the choice of reading (a) in GNT4/NA27 was rated as an “A” decision (highest confidence), and Metzger maintained that the first person possessive was entirely in character for John, though it would understandably cause difficulty for scribes. Moreover, the reading enjoys the combined support of 01 and 03, which would be decisive for critics favoring external support (see below). Yet, the rating was removed in GNT5, and the black diamond was added to reading c, indicating the split line. Clearly, the ECM2 editors made a judgment call to the contrary of Metzger’s, so any critic conceivably could evaluate the textual history of a variant from his or her own point of view.

THE TEXT OF THE NEW TESTAMENT_1410-2250The Alands emphasize the importance of reconstructing the history of a variant, which Metzger says should be done before we form a judgment about it. Their third rule is that criticism of the text always must begin with the evidence of the manuscript tradition. To further clarify that, the Alands make it their fifth rule that primary authority for a textual decision lies with the Greek manuscript tradition, as opposed to the versions or the Fathers.[18] What this all means is that when we encounter a variant in any NT text, we should begin by determining the manuscripts supporting the different readings, and try to put them in a stemma (the equivalent of a family tree) that shows how they related to each other over the course of time. We have just touched on that in the example from 1 John above. This can be very challenging, and one of the difficulties you will encounter is that you must do stemmata (the plural of stemma) variant by variant. We can only wish that the situation was more like distinguishing Ehrman’s remarks from Metzger’s in the fourth edition of TOTNT. In that case, we know that the latest edition is the genealogical descendant of the one chronologically preceding it, and we know that wherever the two editions agree, what is stated is solely from Metzger’s pen or keyboard. We could make the same comparison between earlier editions that were solely Metzger’s, to see what he himself changed from one edition to the next. Unfortunately, the genealogical relationships between ancient biblical manuscripts are so much more complicated that finding the original readings must be done on a case by case basis, without the assumption that one manuscript always has the earlier reading in comparison with another.

I feel reluctantly compelled to add here that while the student may receive a professor’s assignment to produce a stemma, and likely be provided the data necessary for it, the practicality of doing this for the variants present in any passage one studies routinely for exegesis, let alone for something as important as a sermon, for example, may be very low. Unlike sermon outlines, one seldom comes across stemmata that address the languages in detail, even in the better commentaries, Moreover, as I will note later in the chapter on the CBGM, plain lists of manuscripts or witnesses such as those found in the standard Greek NT’s provide virtually no assistance in constructing stemmata. What is really needed is a user-modifiable version of the CBGM that covers the entire NT. Full coverage may be realized in the next decade or so; I have no idea how flexible the tools in the CBGM will eventually become.

In the meantime, probably the most practical approach toward stemmata that a non-expert can take is to list all the readings for a variant and apply the internal guidelines or principles we are examining here in attempting to create stemmata, without great effort in accounting for all the manuscripts/witnesses supporting readings, except for the more important early manuscripts. In the example from 1 John, we could have redesigned the stemma with reading (a) at the top, supported by manuscripts 01 and 03, with readings b and c branching off separately from a. It would be a matter of priorities while recognizing that we have no knowledge of likely gaps in the stemma between manuscripts. In any case, I strongly recommend establishing good habits at the very beginning. One of these habits is to build a file or folder of stemmata, being careful thereby to save all the hard work that goes into creating them. The stemmata can always be modified as more information, or better techniques come into one’s possession. Another good habit is to keep a copy of Metzger’s Textual Commentary[19] handy. We could wish that he had commented on all the variants in the NT if that were humanly possible, but we can be grateful for what he was able to accomplish.

Metzger’s External Criteria

Let’s resume our review of Metzger’s textbook on TC. After the two leading criteria, he categorizes more specific criteria as either external or internal, to which I have just alluded. The external criteria are more limited than the internal, essentially only three: 1) the date and character of the  manuscripts and their texts, together with the care taken by their scribes; 2) the geographical distribution of witnesses that agree in supporting a variant; and 3) the genealogical relationship of texts and families of witnesses. Attached to criterion three is the famous caveat, “witnesses are to be weighed rather than counted.”[20] This originally was a legal concept referring to the evaluation of evidence and witnesses in court (in Latin, ponderantur testes, non numerantur). I do not know who was the first to apply it to the discipline of textual criticism; perhaps it was Metzger. We will return to this in a moment.

First, I want to emphasize a provision Metzger attached to the first external criterion, probably when he wrote his first edition (I have a copy of the second). He remarks that of even greater importance than the age of a manuscript (i.e. the actual material on which the words are written) is the date of the “type of text” it embodies.[21] He goes on to list the major text-types that were generally acknowledged at that time, and for clarification, we can also compare the discussion in his Textual Commentary, where he expands somewhat on his list of criteria. There, under the first criterion, he notes the importance of “the date and character” of the type of text embodied in the manuscript, and he repeats the list of text-types. He also includes the care taken by the copyist in the task of producing the manuscript.[22]

It helps, too, that Metzger points to minuscules 33, 81, and 1739 as providing better evidence than some “later or secondary uncials.”[23] The only way that minuscules could have such value is by effectively reproducing texts hundreds of years earlier than the material on which they are recorded. The same thing is done today, in effect, whenever anyone orders a facsimile of an ancient manuscript such as Vaticanus (03) or Sinaiticus (01).

If we focus on the witnesses rather than the traditional text-types, then we discover that Metzger’s provision has become a major part of the infrastructure of the CBGM, as we will see in the chapter devoted to it. Ironically, it may also have done severe damage to the main point of his first criterion, i.e. the date of the “witness.” To understand this, it is essential that we differentiate between “witness” and “manuscript” in contemporary TC terminology. They are not interchangeable terms (if they ever truly were): “manuscript” is the physical medium containing the words of the text, while “witness” is the intellectual content of the manuscript, not the physical medium providing it. “Witness” and “text” are interchangeable from a practical viewpoint, but “text” is a general term (like the content of any book) while “witness” is specifically the content of a biblical manuscript or fragment of the same, for our purposes at least.

The problem for the first criterion–if by “witness” one understands “manuscript”–is that the witness presented by any manuscript can potentially be older than that of another manuscript which itself is older than the manuscript being compared to it–possibly hundreds of years older. A simple example would be that of a ninth-century scribe copying the text of a second-century manuscript long after a fourth-century scribe copied the text of a third-century manuscript. Assuming in this example that the second-century exemplar was more reliable than the third-century exemplar, the dates of the two copies would be irrelevant, and misleading if we valued the fourth-century manuscript over its ninth-century counterpart. Only the dates of the witnesses would matter.

When we think about it, we soon realize that this is how things actually stand: it is the date (and quality) of the witnesses that really matters, not that of the manuscripts. The question becomes, is there any value in an early date for the latter at all? If we compare an early manuscript with a late manuscript, we can at least say this much: if they agree in content at variants (and there seems always to be significant agreement), the scribe of the late manuscript might have copied the early manuscript, but the reverse is impossible. At the same time, we must also concede that if the agreements resulted from the scribes of both copying the same ancestor that is no longer extant–or a combination of circumstances amounting to the same thing–then the early manuscript just establishes an earlier date for that ancestor than either of its descendant copies. If on the other hand, the texts are significantly different, I feel a higher level of confidence in the manuscript with the earlier date.

I use the term “feel” in the previous sentence because my confidence may be simplistic. For many years, like so many other conservatives I have described the Alexandrian manuscripts as the “earliest and best” in comparison to the much later Byzantine manuscripts. Yet there are in fact a few late manuscripts, such as the tenth-century ms. 1739 at Athos that exhibit an Alexandrian text and are of higher quality (from the traditional perspective) than some manuscripts that are much earlier. These manuscripts serve as real-world exceptions to the earliest-and-best comparison. Assuming that we value the Alexandrian text over others, if we prefer the reading of 1739 to that of D (05), for example, then we are acknowledging that the quality of a reading is more important than the date of the witness. Then again, whether we include 1739 among the witnesses for a reading or not indicates whether we view it as an independent witness, representing an exemplar six or more centuries older that has been lost; or just a copy of another extant manuscript (e.g. 03). If the latter, then for the reading in question 1739 would have no value to us. On the other hand, if the reading in 1739 were very close–but not identical–to that of an early Alexandrian manuscript, we might be forced to decide whether in our view its scribe introduced a worthless change of his own or one taken from another early, unknown manuscript.

So the value of a manuscript’s date is largely in the eye of the beholder. The witness of a late manuscript may be old; that of an old manuscript is old. Clearly, a reading cannot be late if it is found in the original hand of an early manuscript. For some, such a manuscript (i.e. a Greek manuscript) serves as a guarantee of authenticity, in the absence of which any Greek text is suspect. I confess to leaning in this direction. For a good many critics, however, the age of a manuscript is irrelevant. As I will point out later, dating plays no discernable roll in the programming of the CBGM, only genealogies based on internal criteria. One justification for this position is that the early manuscripts, on the whole, did not survive, and we have huge gaps in time in the historical sequence of those that did survive. Ironically, we have so many NT manuscripts, and yet so few!

The geographical distribution of witnesses, Metzger’s second external criterion, has common-sense appeal because the best explanation would seem to be that only a very early reading, ideally the original, could account for a wide distribution of a particular variant. This is often cited in Metzger’s Textual Commentary as support for a given reading. A later, secondary (i.e. non-original) variant would presumably pop up in a given part of the world and tend to have only limited distribution elsewhere. Sometimes this geographical reasoning still seems to carry some weight. It is probably fair to say that it depends on the variant reading. If the reading is one that would have been an obvious or natural choice for “correcting” a difficulty in the biblical text, then mere coincidence could account for a wide distribution.

The most problematic issue for geographical distribution is that it was linked to various inferred tendencies in style associated with major regions, the so-called categorization of “text-types.” For example, in my own work, I have often given considerable weight to a combination of strong Alexandrian and Western witnesses, principally B (03) and D (o5, 06). I expect them to disagree the great majority of the time, so when they agree, I have evaluated their readings as likely reproducing the autograph. Research done using the CBGM has all but eliminated the concept of these types as unhelpful, however, because it is very difficult to sort out and categorize the witnesses when all their variant readings are examined thoroughly. There is too little consistency, due to significant “contamination” of readings, i.e. readings that scribes have imported from other witnesses. Vaticanus (03) still enjoys top ranking, but only because it closely agrees with the ECM Ausgangstext. Other than that, the ECM editors have only retained the Byzantine category of text, using that label, which I find curious. More on this follows below.

Metzger’s third external criterion, the genealogical relationship of texts and families of witnesses, has in effect become the center of attention in contemporary textual criticism, but in a different way and for a different reason. Metzger again addressed text-types, and while he conceded that New Testament manuscripts are largely mixed texts, he believed nonetheless that one could find broad features common to closely related manuscripts.[24] Genealogy, for him, seems to have consisted of assigning witnesses to text-types based on their readings. As I will discuss later below, genealogy now is applied internally to individual manuscripts, primarily comparing them in pairs, and the concept of textual families or “types” that one sees so often in Metzger’s discussions has fallen into disfavor. His addition of the “weighed rather than counted” caveat undoubtedly was aimed from the start at the Byzantine family, the type later refined and defended by some scholars as the Majority Text.

I noted above that Metzger treated the “weight” caveat as a separate (but related) criterion in his Commentary.[25] He elaborated this into the guideline that those witnesses[26] found generally trustworthy in cases where satisfactory decisions can be reached, deserve greater weight where the decisions are difficult. This is a very important principle from the viewpoint of a traditional critic. Together with an early date, it becomes the basis for assuming the superiority of individual manuscripts such as B (03). Few textual critics, however, would approve of (or confess to) using the support of B as the sole reason for the choice of a reading. It must also be conceded that while the combination of 03 and 01 seems to point to the autograph, we are often perplexed when these two manuscripts split over readings. I would much rather face a 7/10 split in bowling. Such splits are best left to professionals, whether in bowling or textual criticism and even then there are no guarantees of successful resolution.

The Alands make Metzger’s “weight” caveat their sixth rule, and in their method, it becomes the justification for their local-genealogical approach. As such it can also be viewed as one of the foundational principles of the GBGM (to be discussed later). Instead of being used as a means to refute arguments for the Byzantine text based on superior numbers, in the hands of the Alands this rule becomes a guarantee that no single manuscript, let alone family of manuscripts, can be decisive in textual decisions. This strikes at the heart of the Westcott-Hort position that the autograph is found in the combination of 01 and 03. So while Metzger might say that Alexandrian manuscript X has been properly evaluated and found to be superior to a particular group of a thousand late manuscripts with roughly the same, inferior text, the most that the Alands assert is that “certain combinations of witnesses” may deserve more confidence than others.[27] Of particular importance, they insist that text-critical decisions must be worked out passage-by-passage.

If there could be any uncertainty about their guarded approach to all manuscripts, the Alexandrians included, the Alands state in their seventh rule that the principle that the original reading can be found in any single manuscript or version when it stands alone or nearly alone is only a “theoretical possibility.”[28] What are we to make of this? I have no doubt that they are targeting any choice of 03 (B) by itself, or in concert with 01. However, while NA27 and 28 reveal a significant shift away from these manuscripts in textual decisions, it is clear that 03 and 01 both still carry a great deal of weight, or at least they appear to. What has happened is that the passage-by-passage (local-genealogical) approach, which focuses on internal criteria, gives weight to these manuscripts as a result of the criteria that are followed. There is now little or no thought of accepting readings from these manuscripts simply because of their reputations for quality.

As to the Alands’ position on the main provisions in Metzger’s first two external criteria–date and geographical distribution–neither is mentioned in their rules. Yet it is clear that at least one of these criteria plays an important role because rules one through five and nine (half of the twelve!) emphasize the importance of external criteria over internal. Then, when we glean what we can from the rules and examine the examples provided, the dates of manuscripts and the historical sequences of readings appear to be very important in deciding readings.[29] This can be inferred from the discussion of the reading NHPIOI vs. HPIOI in 1 Thess. 2:7. The Alands note that the former reading is found in a third-century papyrus and in the original hands of a number of major uncials, with the exception of 02 (A, siding with the Majority Text).

Also revealing, in this case, is that the Alands allow for the objection that the correction to the other reading by later hands could have restored the original. Their reply is that it is unlikely that the same mistake if there was one leading to NHPIOI, would have occurred in five different manuscripts.[30] There happens to be a significant geographical separation between some of the manuscripts, but nothing is said by the Alands about this, as we might expect. However, if we eliminate geographical distribution as a criterion, then it is more difficult to rule out contamination as an explanation for the repetition of NHPIOI if it was originally an error.

Another discussion that is useful in understanding the Alands’ position is the reading EUDOKIAS vs. EUDOKIA in Luke 2:14. We can also compare this discussion with Metzger’s comments on the same variant. I leave it to the reader to study the overall issues with the passage and focus on the evaluations of external evidence by the Alands and Metzger. Briefly, the Alands state that internal support for EUDOKIAS is irrefutable, but the external evidence for the reading is much less extensive than that for the other (nominative-case) reading. In fact, they note that the latter reading is attested “only” by 01 (original hand), A, 03 (original), D, W, “and a few others.”[31] Yet this group is not that different from the preferred group for NHPIOI above. In contrast, Metzger–who specifically refers to the geographical distribution of the manuscripts (Alexandrian and Western families)–comes to the opposite conclusion that EUDOKIAS has better external evidence.[32] So again we see an indication that geography seems to play little to no role in the Alands’ decisions. Along with the date of a manuscript, the number of ancient manuscripts also appears to emerge as a factor.

However, my own experience has taught that no textual critic is capable of absolute consistency in making textual decisions. When the Alands turn to another variant in 1 Cor. 13:3, we find them stating that a combination of 01, 02, 03, an early papyrus, and a few late minuscules is “distinctly superior” to another, larger group of uncials and minuscules, without commenting on the reasons for this appraisal.[33] They may have based their conclusion entirely on the value of the papyrus (second-century), but if so we would expect them to comment as they did above on the third-century papyrus.[34] It is more likely that in this case, they fell back to a more traditional view of the evidence.

Probably most revealing for this criterion, the Alands dismiss the concept of a Western text, of which codex Bezae (D, 05) has traditionally been designated the chief witness–sometimes the only bona fide witness. They (and other textual scholars since) note that this text exhibits both “western” and “eastern” readings. Indeed, they go so far as to remark that hardly anyone in their day (1983) uses the term “Western text” without quotation marks, and they conclude with, “[s]o much for the phantom ‘Western text.'”[35]

As a matter of fact, one can examine the apparatus of NA anywhere 05 contributes readings, and it soon becomes apparent that 05 does show considerable variation in its affinities to other texts. Also, whatever one may think of the text exhibited by 05, known for its longer, periphrastic readings, it is at least as important for this text as 03 is for the Alexandrian text. This makes 05 vital for the theory of a geographical criterion: many locales have been hypothesized during the iterations of the theory, and most have been dismissed as unviable. Only three have remained: the Alexandrian, the Eastern (Byzantine), and the Western. Even so, one can justifiably add quotation marks to all three if desired.

So we might ask whether anything can be salvaged of the geographical criterion. Geography itself has become virtually irrelevant because of the mixture of readings, and critics are agreed that the terminology connected to it is now (and has been) a matter of convenience rather than truth. If the CBGM is providing a window into the future, the geographical designations are living on borrowed time. However, whether it is a hard habit to break, or there is still some validity to textual categories, not all critics are willing to abandon the concept as a criterion. In his chapter on “textual clusters,” Eldon J. Epp makes a rather impassioned argument for the Western text while conceding that the geographical distinction is of no use. He prefers to classify groups of manuscripts that tend to agree in their witnesses as “clusters,” and for both familiarity and neutrality he calls the text represented by 05 the “D-text cluster.”[36]

Renaming the “Western text” manuscripts in this way[37] may technically eliminate the offending elements of the traditional labels. However, I sense that there is a fatal categorical flaw to Epp’s argument. The geographical criterion had value precisely because it gave precedence to readings with a wider geographical distribution. Sometimes one sees such readings preferred in NA27/28 over those favored by 03 etc. I have no problem with the concept of “clustering” similar witnesses, but I fail to see how this is significantly different from sorting out witnesses by genealogy, Metzger’s third criterion. If a group of witnesses exhibits essentially the same text, the next step would be to determine how they are related to each other. As we shall see later, in the CBGM witnesses with similar readings are simply rated as “potential” ancestors or descendants.

Like Epp, Michael Holmes has some affinity for the Western text, and perhaps one could say that Holmes’ position is somewhat related to geography.[38] He is an advocate of Zuntz’s theories, principally that certain so-called Western readings predate Alexandrian readings and are preserved in the Western text, which developed other features of its own. These readings are ancient and important, and not really Western at all, predating Alexandrian readings as they do. So to Holmes, the existence of a Western text does not have the same function it did for Metzger, i.e. that of establishing a wide geographical distribution for a reading. Thus a combination of support such as B/03 and D/05 for a reading is not strong for that reason. On the contrary, Holmes points out that some would see this as an intrusion of an inferior western reading into an Alexandrian witness. Holmes would probably consider the combination that of two ancient witnesses, ignoring any bias against the Western text.

I have included Holmes here not just for his views, but also because he is the editor of the SBL Greek New Testament, which in turn is the text chosen for the Lexham English Bible. Both publications are available online and are obviously competitors with other Greek texts and translations. One would expect the SBLGNT to reflect Holmes’ textual views and feature more choices of the Western text. That is probably the case, because by one report the SBLGNT has 18 readings not found in any other critical edition of the GNT, with six of them having Western support.[39] Most of the six also have Alexandrian support, however, and they could be chosen following internal criteria, so I do not consider them convincing evidence of a commitment to the Western text. It may be that in Holmes’ case we have a highly informed and proficient textual critic with the best of reasons to preserve the Western text, yet one who does not make a real effort to do that for its own sake in his edited Greek text.

This brings us back to Metzger’s third external criterion: the genealogical relationship of texts and families of witnesses. Some may prefer to ignore the issue of textual mixture etc. and cling to the older categories of text-types found in Metzger. Others who are familiar with contemporary work in textual genealogy may instead prefer the methodology of the ECM and the CBGM. In any case, I find genealogy problematic as an external criterion. Let me state emphatically that I do not doubt its validity as a plausible criterion, but the problem I see is the basis for identifying relationships between witnesses. No matter whether the goal is to assign a manuscript/witness to a particular family or group (Metzger and, from a different viewpoint, Epp), or to identify its ancestors and descendants (ECM), it is the readings that are being compared and evaluated by internal criteria (to be discussed below).

One might attempt to save genealogy as an external criterion by arguing that “external” in this context only means “objective,” and that we are looking for agreements and disagreements between witnesses. I have little doubt that this is what Metzger meant, and it is what the ECM editors mean. Among the latter Klaus Wachtel wrote a chapter providing a somewhat subtle argument for treating the evaluation of readings as external evidence; the title he gave the chapter clearly identifies his goal, however: “Towards a Redefinition of External Criteria: the Role of Coherence in Assessing the Origin of Variants.”[40]  Neither dating nor geography plays a part in ECM/CBGM tasks, so Wachtel maintains that percentages of agreement and disagreement between witnesses–evaluated as “pre-genealogical coherence”–can serve as external evidence. He insists that these percentages can hardly be biased.[41] Technically they can be, since many variants can be subdivided or not as the editor or critic sees fit, resulting in different numbers of variants for the same passage.

Even if we all were to agree on the exact numbers of agreements and disagreements in variants for a witness, however, a high degree of subjectivity follows in the final stages of deciding genealogical relationships among witnesses. If one takes Metzger’s approach, reiterated in the fourth edition by Metzger and Ehrman, the determination has to be made at some point as to what kinds of readings and how many of them for a given witness are going to be sufficient to assign its text-type. Before the CBGM and other powerful tools were invented, this determination might be accepted as a judgment call, and not many would dispute the categories to which most manuscripts were assigned. Today these same categories are under intense scrutiny, and logical rationales need to be specified for distinguishing witnesses.

As for the ECM/CBGM, aside from the issue of counting variants, identifying agreements and disagreements is only the beginning. The determination of the genealogical direction (relationship) between individual witnesses–which is what the CBGM is all about–depends on determining this relationship for the readings of every variant in every witness examined. This appears to be done entirely by internal criteria and can be a highly subjective decision. If the quality of a witness such as B/03 is taken into account as a criterion (I can find no indication that it is), this too is based on the subjective evaluation of readings. Moreover, proof of the subjectivity is seen in the fact that there are two versions of the CBGM tools available online, and they do not always produce the same results for the same queries.[42] The reason is that version 2.0 operates on a database that has been revised over time, with decisions about some readings having been changed as the circumstances that led to those decisions in version 1.0 are reevaluated.

Summarizing the state and value of external criteria, the date of a manuscript is an objective datum and serves at least to establish a terminal date for the witness (text) contained in the manuscript. It does not, however, tell us how early the witness is nor the exemplar from which it was copied, and it is possible for the witness of a comparatively later manuscript to have been produced earlier than that of a comparatively earlier one. Also, it is a fact that witnesses are mixed (contaminated) texts, so we must concede that a witness is probably composed of readings gathered ultimately from other witnesses of varying dates. However, an early manuscript does at least verify an early date for its witness. We can and should be very grateful for the early manuscripts that have survived.

We no longer have any consensus of confidence in geography as a criterion for manuscripts/witnesses. This is due to the mixture of readings we find in witnesses, making it very difficult to assign witnesses as a whole to regions. Metzger, to the contrary, was comfortable in identifying manuscripts by region and gave great weight to readings that enjoyed what appeared to be a wide geographical distribution. In drawing his conclusions, he took into account not only Greek manuscripts but citations of the fathers and ancient versions as well. Other scholars such as Eldon Epp have argued that these sources must be included, but even he dismisses the geographical element that was so important to Metzger. I doubt that the situation will change as more research is done with the CBGM and other tools, but perhaps it may to some extent. As of this writing, we are still waiting for most of the New Testament to be analyzed with the CBGM tools.

As for genealogy, determining genealogical relationships is being redefined both in meaning and method, since it was largely connected with geography. Text-types are on their way out, except for the late Byzantine text, and instead of simply classifying a manuscript or its text as being part of a “family” (Byzantine, Alexandrian, Western), the relationship between one manuscript and others similar to it is being sought. The outcome is analogous to a family tree–often more complex–and though it is touted as an external criterion, it is really based on internal criteria. I note too that the distinction between “external” and “internal” in this case if it is legitimate, is one of objectivity (external) vs. subjectivity (the internal).

In view of all these factors in external criteria, I would suggest that the only truly external criterion remaining to the student or professional minister is that of the date of a manuscript, and I emphasize that I am referring in this case only to the manuscript. One can easily obtain the date and current location of a Greek manuscript from appendices in the NA or UBS Greek texts, as well as online from the Muenster Institute, which also provides supplementary information at http://ntvmr.uni-muenster.de/manuscript-workspace. What one does with the date is mainly dependent on one’s viewpoint. We would expect a late manuscript to exhibit a Byzantine/Majority Text witness, and an early one to be Alexandrian essentially, but we want to be alerted to exceptions if possible. At present, looking up a manuscript’s “Potential Ancestors” in the online CBGM tools will give us the answer for one of the General Epistles, because on the one hand it reports the percentage of agreement with the Majority Text, while on the other it shows how close the manuscript is to the “A-text” (explained later), which usually agrees with Alexandrian witnesses. However, it will be some time before the CBGM covers a significant amount of the NT. Until then, this kind of information is only obtainable sporadically in commentaries and articles.

Internal Criteria

We move on now to internal criteria, which are more numerous and at least as debatable. If not already, it will soon become clear why those familiar with textual criticism consider it largely an art. There is a great deal of data to process, but the decisions to be made often are subjective and vary from one critic to another.

We first need to address the fundamental principle briefly mentioned earlier: that the reading best explaining the origin of the others is to be preferred.[43] It is usually assumed that this principle branches out into the detailed internal criteria which in turn represent it. The “harder reading” criterion that we will discuss below is a prime example. The result is that the principle may not be recognized as having its own practical application. However, there are times when it does, i.e. in the choices of variant readings when no particular criterion is cited. Instead, we can expect to see the commentator or textual critic simply describing the steps that may have led a scribe to pen the preferred variant. I will provide an example below in the last chapter, a reading in James.[44] For now, we will discuss the individual criteria.

Metzger subdivides the internal criteria into two groups: 1) transcriptional and scribal issues, and 2) issues in determining the author’s choices. At the top of the list in the first group–and rightly so both for its importance and the difficulty in employing it–is the criterion that the harder reading should be preferred (to which the shorthand Latin term lectio difficilior is applied). This criterion is founded on the assumption that scribes were reasonable, intelligent persons who had great respect for the Scriptures and for their work. When we encounter two or more witnesses (texts) that disagree with a reading, creating thereby a variation unit, our job as textual critics are to reverse-engineer the process of reasoning that led one or more scribes to change the original.[45] Along with our assumptions about the character of the scribes, we begin with the assumption that one of the witnesses exhibits the original reading that was changed into the others. Since the scribes knew they were copying Scripture if anything they should have been most reluctant to change what they saw in their exemplars; so something about the original reading must have been unacceptable to them. If therefore, the original reading is exhibited among the witnesses, we expect one reading to be “harder” to make sense of than the other reading(s). In effect, a competent scribe would “correct” what appeared to be a mistake, as opposed to changing something that read well into something more difficult.

One of the questions associated with this criterion is what would have made a reading difficult for a scribe? Some problems are trivial, such as misspellings, examples of which have already been provided earlier by my friend Mr. Andrews (see chapter V). Others deserve special attention and will be discussed below as subcategories of this criterion. For misspellings, however, we need to determine whether the offending text makes sense without correction. If not, then we need to have a plausible explanation for how the apparent error came about, such as the doubling of a letter or the omission of one. Another source of errors in Greek, as mentioned in chapter V, is the similarity in sound of certain vowels. In practice, you will not have to be very creative because the choices made by the editors of the Greek text in comparison to the other readings should reveal what most likely happened.

The main difficulty we encounter with the lectio difficilior criterion is an additional provision that has traditionally been attached to it: a reading meeting this criterion can be rejected if it is so difficult as to be impossible. This is an attempt to keep nonsense readings from prevailing. It is also important to note that neither this criterion nor the added provision applies to a reading that is a mistake, such as a misspelling or grammatical error by a scribe. If that is what has happened, then we ought to be able to explain how the mistake occurred, as I said above.

The problem with the provision, as you might guess, is that too often the determination of unacceptable difficulty for a reading is a subjective judgment call. The reading “found” in 2 Pet. 3:10 which I will discuss later[46] is a classic example. No mistake is possible, and to the creator of the CBGM the reading is so difficult that he resorts to a reading found in the Syriac, which commends itself as a reliable version; but this reading happens not to be found elsewhere in Greek. Other solutions have been suggested that do have Greek support, and I agree with a few others in preferring the very difficult reading as not being impossible.

Moreover, as I have done textual criticism using NA, on a number of occasions I have found difficult readings with excellent manuscript support rejected in favor of easier readings. I could only conclude in some of these cases that the rejected readings with superior support were considered too difficult to be plausible. Identifying these variants should be possible even for a student of textual criticism. The features to look for are a reading that is very difficult, yet has strong manuscript support. Expect to find 03 and 01 listed, except in those passages that are missing from 03.

Second in the list of internal criteria is that the shorter of variant readings (lectio brevior) should be preferred, though Metzger allows a number of exceptions, and Ehrman does not comment on Metzger’s original discussion other than to add references to other works on the subject.[47] One very clear exception would be a case of homoeteleuton as was discussed earlier.[48] So we would exclude a reading that was shorter by virtue of an accidental omission. However, considerable doubt has been cast on the criterion by James R. Royse, who wrote a dissertation on scribal habits and later published it as a book entitled Scribal Habits in Early Greek New Testament Papyri.[49] Royse examined six papyri in great detail and came to the conclusion that early scribes tended to omit rather than add material, with some exceptions. He reaffirms his position more recently in a work I have already mentioned, citing others who have reached similar conclusions.[50]

I do not think this criterion must be jettisoned, but we need to consider the conditions under which it could be valid. It is mainly intended to eliminate readings that in some respect are the invention of a scribe. Some critics like to describe these as “expanded” readings. One immediately thinks of glosses, i.e. brief phrases explaining something in the text that are mistaken by a later scribe for a recommended reading, and used to “correct” or “complete” the text.

Conflate readings are typically cited as an example of spurious longer readings, and probably correctly so. The rationale for rejecting these readings is that a scribe would either have been instructed or just inclined, not to omit anything that could be original to the autograph. We also assume that in this case, the scribe had before him at least two exemplars with differing readings and that his task or personal goal was to somehow combine their contents into one. Acts 20:28, discussed earlier,[51] is a good example with the conflated reading, “church of the Lord and God.” If the scribe’s choice was between “Lord” and “God,” however, we can only wonder whether he realized that combining “Lord” and “God” guaranteed that the reading would be erroneous. I fear that we sometimes grant ancient scribes too little ability to think through what they were doing. How, in this case, do we give reasonable consideration to the possibility that omission was the culprit and not conflation? The best answer is that if the “conflate” reading was original, then we have a large number of manuscripts in which “Lord” was accidentally omitted, and another significant group in which “God” was omitted. Unless nearly all the members of each group were faithfully copying an ancestor (the CBGM term for this is “coherent”), we would conclude that coincidence does not adequately account for the omissions, and conflation better explains the longer reading.

The Alands caution that a shorter reading which looks like an edit in a witness that in turn varies from other witnesses in the tradition should be viewed with suspicion.[52] However, I suspect that this determination is beyond the resources of most readers. The good news is that it is not difficult to piece together what the editors of a critical Greek NT were thinking in this situation. One can see in the apparatus whether the shorter (or shortest) reading has been selected for the text; if it also has the support of leading Alexandrian manuscripts, then the brevity of the reading may have been an additional reason for its preference. The difficulty of the reading will almost always be a major factor, often overruling everything else. If one can find no other reason for the choice of a reading in the text other than its brevity, then the “shorter reading” criterion was probably decisive.

A third criterion in the list is that when parallel passages differ, scribes would most likely bring them into agreement, i.e. harmonize them; therefore the original reading probably is the one that disagrees with the accepted text. For a quotation from the Old Testament in the New Testament, readings that match the Greek Old Testament (Septuagint) are more likely secondary compared to those that do not. In the Synoptics and other parallel New Testament passages, readings that agree with their parallels should generally be suspect as secondary, “corrected” by scribes.[53] Strangely, the Alands include this criterion with the shorter reading, suggested evidently by their observation that neither criterion should be applied mechanically.[54] Other than that, the two criteria have nothing to do with each other.

It might be tempting to follow the parallel passages criterion mechanically, automatically choosing the readings that differ, because it is hard to imagine a scribe altering a passage that agrees with its parallel. There is a chance, however, that an alteration in a passage is due merely to a scribe’s faulty memory of it. Wasserman notes that Vaticanus (03) is known for accuracy in avoiding harmonization, and thus one cannot assume a scribal error in the few cases where parallel passages do agree.[55]

A fourth and final criterion offered by Metzger on the list related to scribes is that they would be more likely to attempt to improve the Greek of the passage before them, than to make it rougher than it was–common sense, really. The last three criteria can probably be viewed as a subset of the lectio difficilior criterion, but this fourth one especially can be seen as such. Metzger does not use “improve” or similar language; he instead lists various ways this might be done, and Ehrman does not comment on Metzger’s discussion. The alterations include the substitution of more familiar words, the replacement of rough grammar and diction with Atticizing improvements, and the addition of pronouns, conjunctions, etc. to smooth out the text.[56]

The Alands do not address this criterion in their own list of rules, and Wasserman challenges the validity of it, citing other scholars who have challenged it as well.[57] I would not completely rule out the substitution of more familiar words, because they could begin as a gloss in the margin, or even more so between lines. A scribe not possessing the sharpest mind might mistake the gloss for an alternate reading of the term in the text, instead of a gloss.

With due respect to Metzger, however, I have to agree with others in doubting the introduction of Attic elements–i.e. words and grammar from Classical (Attic) Greek–by scribes. I especially take issue with those scholars who profess to find Attic style and grammar inflicted upon the text by scribes who were trying to bring the New Testament up to the standards of golden-age Athens. In my experience, this is the judgment of those who usually have not read enough Greek, particularly Attic, but also Greek from the other time periods. If one desires to find ancient Greek dialects, I suggest reading Homer, and old and new Ionic, to obtain an idea of what a Greek dialect really looks like. Once Attic was developed, it basically became the Greek language. I see Attic grammar persistent in Greek from the golden age through to Byzantine Greek.[58]

So it might seem ridiculous to some that the writers of the New Testament would exhibit any familiarity with Attic Greek–and I refer to the actual authors (Peter, Paul, John, Luke, of course, et al.), not merely their amanuenses–but I think that in reality if you knew Greek at all in the first century, it was basically Attic Greek. The biggest difference would not be found in the dialect or the grammar, but in literary vs. common style. Literary could be long, ponderous sentences with one or two main verbs and numerous verbal modifiers, such as we find in Luke’s introductions. We also see long, complex constructions in Paul and Peter. We even see the optative mood used in indirect discourse and conditions, which are among the more complex grammatical constructions. For the most part, NT Greek is the common style, but we do find examples of a more literary style.

Considering these objections, I think the criterion can usually be ignored for decisions, especially when the critic using it describes the NT writer as being incapable of writing “good Greek.”[59] Sometimes a reading is rejected simply as being beyond the literary ability of the biblical author. Here again, I would question whether the critic is sufficiently acquainted with the language so as to make such a judgment, because most who do are not, in my opinion. Other arguments for or against a reading should be examined instead, or at least given substantial weight in opposition to this criterion if they point to a different reading.

Metzger’s second group of internal criteria, those concerned with the author’s choices, are listed as six and termed “intrinsic probabilities.”[60] The first, though not explicitly stated this way, is that variants consistent with the diction, style, and theology of the author overall in the book are preferable. Metzger did not originally include the author’s theology as an element of this criterion.[61] I would caution the reader that the criterion can be difficult to apply. In the first place, the more material we have in a book, the better. I am not happy with the criterion in any book much shorter than Romans or the Corinthian letters; I question whether we have enough of the author’s work to make an informed judgment of his style and lexical preferences. There is also tension between this criterion and the lectio difficilior. Scribes quite possibly would be familiar with the styles of biblical authors and might alter or prefer variants to match their conceptions of a given author’s style. In other words, to any one scribe, a particular variant could have been the easier reading because it seemed more consistent with the author’s style. Modern critics may make the same mistake. It is best if the reading that is harder for other reasons also happens to be consistent with the author’s style. As for the author’s theology, I do not dispute that Ehrman (with Metzger’s concurrence) has raised a valid point. Details in theology can be very hard to objectify, however, and I suspect that only broad cardinal doctrines will be of use. Otherwise, the critic may find him- or herself having to prove first that A or B is part of the theology of the author in question before it can be used for this criterion.

The Alands address the author’s diction and style (and the context), as well as his “theological environment,” but only negatively. That is, they caution that these factors can never be the sole basis for a decision, particularly in opposition to external evidence.[62] We are apparently left to conclude that the criterion does have limited value in conjunction with others. Wasserman makes a much clearer point, in agreement with others, that it can be difficult to determine what is genuine to the author’s style and what has been altered by a scribe to better match his perception of the author’s style.[63] I always come back to the question of whether I have enough material to summarize the author’s style. If I am doubtful about that, then the word or construction in question needs to be very unusual, not something likely to be repeated, though it must be repeated at least once to be part of a style.

The second criterion in this list is that the variant must be consistent with the immediate context. Here again, however, tension may occur with the lectio difficilior. While some readings may seem disjointed or out of place with the preceding or following context, scribes may have added conjunctions, or other modifiers that they imagined must have been omitted by error, with the result that a smooth variant is actually secondary. I suggest two things to consider: 1) a great deal of experience in Greek is needed to make a good decision in this case, so consulting one or more commentaries that focus on the problem is essential; and 2) if an alteration of the autograph has occurred, it should be possible to reverse-engineer it. That is, there should be a plausible way to explain how the autograph could have been changed by the scribe to the reading in question without doing mental gymnastics or serious surgery with quill and ink. Otherwise, if the smoothing elements appear to be authentic, they probably are.

The third criterion in Metzger’s list is an extension of the first: that a variant is preferable if it is consistent with the usage of the author in other books written by him. What I said about the first criterion applies to this one as well, but on the whole, I think we have to allow for the possibility of more flexibility in the author’s usage in other books. It may not prove to be so; i.e. we may find complete continuity between one book and the other, or limited continuity. We do not know what may have changed in the author’s frame of mind as he moved from one book to another.

The last three criteria in this group were originally limited by Metzger to the Gospels.[64] The limitation is not mentioned with Ehrman as co-author so that the criteria evidently apply to any book in which their conditions would be met.[65] I do not want to make too much of this. If the change was intentional, the purpose probably was to allow the last criterion to apply beyond the Gospels; the other two focus on the Gospels in any case.

To continue, then, the fourth criterion, not stated as explicitly as I do here, is that when reports of Jesus’ teachings have variants, a variant is preferable which is most consistent with Aramaic idiom. This assumes that Jesus most often spoke his native language, but of course, we do not know how often he may have spoken in Greek. Some might say “never,” but I am confident that Jesus would have used Greek with Romans such as Pilate and the centurion. I do think that he would have spoken to fellow Jews in Aramaic, and of course, we have important quotations of Jesus in Aramaic. In applying this criterion, we need to bear in mind the ethnic composition of Jesus’ audience. In practice, the reader may be entirely dependent on commentators familiar with Aramaic.

One comment that I strongly encourage anyone unfamiliar with Aramaic to be on guard against is, “The detail that we find here in the Greek is irrelevant because it is translating an original Aramaic word (or construction) incapable of expressing this precision,” or words to that effect. A commentator who says this may or may not be correct because a speaker in any given language can usually find a way to express a concept for which the language is not well suited. However, I see this as a subtle challenge to the accuracy implied by inspiration, and even if one excludes the implications of inspiration, I think we can assume that the report of what was said is accurate. I believe that the biblical author, aware of the importance of his work, would have taken pains to clear up any ambiguities in what the speaker said before doing his translation. The alternative is to doubt the accuracy of any and all reports of speeches and conversations.

The fifth criterion is the priority of the Gospel of Mark. It has long been acknowledged by New Testament scholars in the great majority of circles and schools that Mark’s was the first Gospel, and that Matthew and Luke used and edited Mark’s Gospel. It usually seems to be the case that when there are parallels to a passage in Mark, he tends to have the harder reading in sense. His is also the shortest gospel.

Having said all that, I personally reject Markan priority, confessing that I do so for theological reasons. I actually believe that the four gospel writers wrote independently of each other, not editing the work of their fellow evangelist and that the incredible agreement among the synoptics was due to the miraculous work of the Holy Spirit. Even if I accepted Markan priority, however, I would not consider it of much practical help as a criterion for textual criticism. True, it would be at least hypothetically useful for choosing the most factual account of an event or quotation variously reported among the three synoptics. Nevertheless, I do not see the assumption of Markan priority providing helpful clues in determining Matthew’s or Luke’s autograph. At best I only know that I can expect them to differ with Mark, probably in ways that will mitigate or smooth out what Mark has to report. That is far too general an assumption to be of much use.

The sixth and final criterion in the list is the possible influence of the ancient Christian community (notably second to fifth cent. C.E.) on the formulation and transmission of the passage in question. Now we are referring mainly to theology, and theology underscored. Metzger has stressed the need for the textual critic to be thoroughly acquainted with early Christian theology and heresies.[66] It is only to be expected that scribes would be influenced both by their own theology and by others with their own theological agendas. When they encountered variant readings that posed theological problems, they would naturally be tempted to choose a reading that was not hampered by difficult or heretical theology. A reading that does come with troublesome theological baggage is another form of lectio difficilior.

Ehrman is of course well known for expanding on this issue in his book, The Orthodox Corruption of Scripture.[67] As an example we can refer to his discussion of the Greek word EAN in 1 John 2:28, where EAN (“if”) is replaced with hOTAN (“when”) in late manuscripts. Ehrman sees in this a clear case of scribal intervention to eliminate any appearance that John might have had doubt about the return of Christ.

Theology would never be the only factor to consider in making a textual decision, however, in this case both the manuscript evidence and the harder reading (lectio difficilior) criterion favor EAN (“if”). If we check English translations, we may wonder why so many have “when” instead of “if,” seemingly exhibiting the reading that should be rejected according to the criteria. It happens that the standard Greek lexicon (BAGD) recommends “when” for this verse. We could actually stop there, with nothing more to prove. As someone privileged to have had much more experience reading Greek than most New Testament scholars, however, I can tell you that “if” is actually correct as the translation. That is why some late scribes changed the Greek found here. But John used an unusual construction, not one that questions the truth of the return of Christ, as Ehrman suggests. Many of the personal “if” constructions we see in the New Testament seem designed to draw out a confession of faith from the readers, and that may be the case here.

I think it is imperative to add one more internal criterion that in effect was addressed by Metzger in connection with the external criteria: the quality of manuscripts and their witnesses. This is a summary judgment both of the readings of a manuscript’s witness, and of the overall care taken by the manuscript’s scribe.[68] One could, therefore, subdivide this criterion into two criteria, but in practice critics routinely speak of the quality of a manuscript like 03 (B) as if the manuscript and its witness were the same–indeed, some critics insist that they are, so I think it is reasonable to treat them as such in this case. I classify the criterion as internal because decisions about readings are mostly internal, and they play a larger role than the care taken by the scribe. Also, a careless scribe will exhibit his carelessness in the errors he makes, so we can expect the quality of readings to be consistent with the scribes care.

The point of this criterion is that manuscripts like 03 and 01 lend weight to a reading simply by their support. How relevant this will be will depend on the method used in the decision process (see below). The fact that the “best” manuscripts sometimes disagree at a variation unit must be acknowledged, though. Probably no other fact so clearly establishes the truth that TC guidelines cannot be used mechanically. One may be tempted to cut the Gordian knot by ruling that when 03 and 01 (or some other reliable combination) disagree, 03 is to be preferred. In that case, however, there are variation units where other criteria will favor 01. There are also passages where only 01 provides a reading. You will discover that many decisions are easy, many others are not, and some will (or should) rob you of sleep.

So Which Method?

We have already seen that some criteria of textual criticism act in opposition to others. The same is true for textual critics. Most can be loosely categorized according to criteria that they favor.

Thoroughgoing Eclecticism (G. D. Kilpatrick, J. K. Elliott)

In this approach, internal evidence alone is used to make decisions. The term “eclecticism” basically refers to deciding readings on a case-by-case basis. Those who take this approach tend to view all external evidence as unreliable. They claim without fear of dispute that reasoned eclectics by comparison use the same internal criteria, but that the latter inconsistently abandon these criteria and resort to external criteria when the need arises to obtain the results they want. This is a minority view and is criticized for not recognizing the value of the external evidence.

In practice, thoroughgoing eclectics will agree with reasoned eclectics (described below) in most cases, because the preferred manuscripts usually support readings also favored by internal criteria.

Reasoned Eclecticism (B. M. Metzger, the Alands, Others)

In this method, internal and external evidence are given more or less equal weight. The distribution of weight varies because among many reasoned eclectics there is a tendency to emphasize internal criteria. For example, the superiority of manuscript B (03) is acknowledged, but in practice, it is not always clear whether a reasoned eclectic is taking into account the age and quality of B, or is only thinking of the readings it exhibits.

In any case, reasoned eclecticism differs from its thoroughgoing counterpart in that some readings are preferred to a greater or lesser degree for the manuscripts that contain them. This has been the method of the committees of the Nestle-Aland edition and United Bible Societies edition of the Greek New Testament.[69] For the greater history of the NA text, the editors also preferred the manuscripts of the Alexandrian family of texts as being the best and most faithful in preserving the original reading, following the appraisal of Westcott and Hort. The editors viewed the Western family of texts, while early, as paraphrastic, adding and removing words, clauses, and whole sentences. The Byzantine family is much later than the Alexandrian and Western families and is known for its smoothing out rough readings, the combining of two or more readings, and the harmonization of parallel passages. It continues, for the most part, to be viewed as such by the editors.

Beginning with NA26, there was a concerted effort under the leadership of Kurt Aland to depart from the conclusions of Westcott and Hort and evaluate readings on a case-by-case basis both externally and internally. The results were usually the same as previous decisions, but neither 03 nor 01 nor their combination was treated as a decisive factor. Aland insisted that this was not eclecticism and grudgingly conceded that it could be called the “local-genealogical method.”[70] Today it is common to use this term synonymously with “reasoned eclecticism,” whether entirely accurate or not. The editors of NA28 have redefined reasoned eclecticism in regard to external evidence, as we saw above,[71] and in the process have in effect replaced the local-genealogical method with it.

Majority/Byzantine Text Priority (M. Robinson, Z. Hodges, A. Farstad)

Under this method, the Byzantine text family is given priority over all others and is considered the best and most faithful in preserving the original reading. Some confusion probably is inevitable between this and the Majority Text, a term possibly coined by Kurt Aland and further advanced by Zane Hodges. The vast majority of extant manuscripts exhibits this form of text which has variously been called “Byzantine,” “Koine” (pronounced coin-áy), and “Syrian.”[72] I omit here the traditional defense of the text (often associated with the Textus Receptus) based on divine preservation, which can be found elsewhere and no longer is used by scholars advocating the text. Hodges, in his paper “A Defense of the Majority Text”[73] argued that statistically, superiority in numbers of extant manuscripts was evidence of historical priority.

More recently, Maurice Robinson has taken up what, to me, seems a variation of the same argument by requiring a strong “transmissional history” of a text family as proof of historical priority.[74] In essence, whatever else may be said in favor or against a reading, it needs above all to appear in a substantial number of manuscripts to be viable. Along with this Robinson also protests that any viable Greek text should exist somewhere in a sequential form (i.e. not fragmented) for at least short sections, e.g. a few verses.[75] In making this point, he brings to light what can theoretically be viewed as a weakness of any eclectic text (one in which readings are decided case-by-case): it is virtually impossible that even a short sample of the text would ever be found. The mind may then be tempted to jump to the conclusion that no such autograph could ever have existed!

I have to agree with the majority of critics, however, that these arguments based on numbers are fatally flawed. In the first place, going back to one of Metzger’s original criticisms, mere numbers do not prove superiority:

For example, if in a given sentence reading x is supported by twenty manuscripts and reading y by only one manuscript, the relative numerical support favoring x counts for nothing if all twenty manuscripts should be discovered to be copies made from a single manuscript, no longer extant, whose scribe first introduced that particular variant reading.[76]

 In the case of the Byzantine text, we have better reason than usual to expect that scribes would have reproduced just what they found in their exemplars. For all that, even Robinson concedes that no two Byzantine manuscripts will agree exactly.[77] In the second place, the Byzantine text consistently–though not without exception–presents the easier readings in variation units. Critics are agreed that scribes will tend to change harder readings to easier alternatives, not the reverse. As a result, we would only expect to see more continuity in the “transmissional history” of the Byzantine text than in other witnesses with harder readings.

Robinson gives the reader the impression that he and his colleagues evaluate individual readings, and that may be true, but for the most part I think it is fair to say that those who favor the Byzantine text provide for themselves the relatively easy task of identifying those readings in the apparatus and simply selecting them as the autograph.[78] They can be found usually by the symbol  for “Majority Text” in NA (except the General Epistles of NA28) or by “Byz” (GNT, ECM2). Technically there is a difference between the Majority Text and the Byzantine text, which is now being accentuated by the editorial committee for NA28. I think it is fair to say that most people, the uninitiated and scholars alike, think of the Majority Text and the Alexandrian manuscripts as opposites. The committee defines Majority Text (or MT) readings as those supported by a numerical majority of manuscripts, which can include Alexandrian readings. However, they classify Byzantine readings as those supported by the majority, excluding their own hypothetical A-text, which usually agrees with Alexandrian witnesses. I will have more on this later.

Documentary Approach (F. J. A Hort, E. C. Colwell, P. Comfort)

In this approach, preference is given to the traditional external evidence, particularly the date and quality of manuscripts, both individually and in groups. It is often assumed that the combination of 03 (B) and 01 presents the autograph, even if they stand alone against all other manuscripts. Manuscript 03 is accorded the greatest weight, and if other criteria do not pose major problems, the support of 03 alone may tip the scales in favor of a reading.

The early papyri are also highly valued. However, their fragmentary nature limits their usefulness mainly to verification of the reliability of the leading early manuscripts with full texts (more or less). P75, for example, is famous for establishing an early date for the witness (text) of 03.

This approach bears considerable resemblance to reasoned eclecticism and even to the Aland’s local-genealogical method in the results obtained. It is interesting that the Alands stress the importance of external criteria over internal in half of their “rules,”[79] while they also make it their seventh rule that no single manuscript or small group of manuscripts (no doubt having 03 and 01 in mind) when standing alone has a real chance of providing the original reading.[80] Only infrequently do 03 and 01 stand alone, however, so the great majority of the time documentarians who also give serious (but not primary) consideration to internal criteria can get along rather well with reasoned eclectics who give priority to external criteria. The only problem with this picture is that the latter group may be dwindling as “external evidence” comes under criticism or is redefined.

On the other hand, most documentarians, among whom I could include myself (more or less), my coauthor as well, assign a great deal of weight to the “harder reading” (lectio difficilior) criterion and are always uncomfortable with a suspiciously easy reading regardless of its external support. If or whenever they reject a reading with strong Alexandrian support because it clearly does not commend itself as the source for the other readings, they cross the line and become reasoned eclectics for the moment or even thoroughgoing eclectics. The difference in outcomes between documentarians and the others lies, however, in the fact that a documentarian will find it very difficult at best to reject any reading that has the combined support of 03 and 01. Like those who prefer the MT, documentarians, for the most part, have a relatively easy job of choosing readings with the support of those two manuscripts.

Difficulty arises for the documentarian whenever 03 and 01 disagree, or “split” over readings, or when the lectio difficilior clearly is a competing reading. Thankfully this does not happen often, but there often is no satisfying solution when it does. The best approach, in my opinion, is humility: make a choice with the knowledge that one is on shaky ground, and remain open to reviewing the choice. If either 03 or 01 has the harder reading, choose it. Otherwise, as a true documentarian, consider the possibility that the harder reading is an error and try to determine whether 03 or 01 is the more likely source of the other readings.

If one wishes to remain true to the traditional documentary approach, I advise him or her to consider obtaining a copy of NA25, which will more conveniently display choices based on the approach. The documentarian may even want to consider the text “frozen” at this stage. Future editions of NA and UBSGNT will undoubtedly present readings more consistent with internal criteria. I realize that NA25 may be unobtainable; if so, obtain a copy of NA26 or 27. You can probably find one at a library if nothing else. In the appendices, you will find the invaluable “Editionum Differentiae.” This is a list of all the places where these editions differ from others, including NA25. Make a convenient copy of the list, even if you can purchase the text. It will provide you the means of easily determining the text of NA25, and unfortunately, the list is not included in the current NA.

If the reader understands and highly values the harder-reading criterion and other good internal criteria, however–the same that led to high valuations of the Alexandrian manuscripts–I doubt that the future NA editions and the CBGM upon which they are based will pose too much of a problem. Moreover, the apparatus should still provide the information necessary to make informed decisions from the documentary perspective.

Examples of New Testament Textual Criticism

The following examples will illustrate how textual criticism can be done from the different perspectives. None illustrates the use of the CBGM, which is discussed later and for the present is only available for the General Epistles. I am taking a pragmatic approach here, trying to minimize the amount of work necessary for the reader to make an informed decision. It has been my experience that pastors especially, and even people studying the Bible for their own enrichment, more often than not lack the time to do a thorough job of textual criticism on a passage. As a result, they easily succumb to the temptation to neglect it altogether.

As a minimum of preparation, I recommend obtaining a copy of Metzger’s Textual Commentary that I have cited before. His eyewitness accounts of committee decisions are very helpful, even though many decisions are left out. A Greek text with apparatus showing the different readings is absolutely necessary, of course, and you can choose between NA and UBSGNT. I use NA because its apparatus includes most all of the variation units, and I need to know where and what they all are. UBSGNT has a much easier apparatus to read, but it does this by avoiding abbreviations and symbols that can be difficult to follow.[81] I also recommend some kind of workspace, anything from pencil and paper to a computer spreadsheet, as the reader prefers. You may like something elaborate, or not. I would start with a simple layout of rows to list different readings and two columns, “pro” and “con” or something like that. My good friend Mr. Andrews prefers to list the criteria point by point under each reading. You may find that better, especially until you are familiar with the criteria. Let’s begin with a verse in Matthew.[82]

MATTHEW 6:33

Is it “seek first the kingdom of God and his righteousness” or “seek first his kingdom and his righteousness”?

Matthew 6:33 New King James Version (NKJV)

33 But seek first the kingdom of God and his righteousness, and all these things shall be added to you.

Matthew 6:33 New American Standard Bible (NASB)

33 But seek first His kingdom and His righteousness, and all these things will be added to you.

There are other variants besides those shown above, but they do not have comparable support. The reading followed by the NKJV and, somewhat surprisingly, the ESV is that of the MT/Byz text. For those who prefer that text, the decision is clear. The reading chosen for the NASB basically is found in 03 and 01. The words “kingdom” and “righteous” are switched in 03 relative to other manuscripts, but like 01 it lacks “of God.” So the decision is also relatively easy for documentarians.

This case is one of undoubtedly many that illustrate the different approaches of documentarians and reasoned eclectics. The latter will give greater weight to internal criteria, often ignoring the quality of individual manuscripts or even groups of them. Metzger discusses this variant unit in his Textual Commentary and reveals that the committee was essentially divided between two internal criteria: the shorter reading and the author’s style. The minority of the committee was attracted to the shorter reading and felt that “of God” would have seemed a natural addition to a scribe. The majority of the committee, however, was attracted to the longer reading because Matthew seldom employs βασιλεία (“kingdom”) without a modifier (the instances in 8:12; 13:38; 24:7, 14 were regarded as special exceptions), and attributed the absence of a modifier in several witnesses to accidental scribal omission. In view of these conflicting interpretations, Metzger concluded, it was thought best to include the words in the text but to enclose them within square brackets.[83]

I always have reservations about the author’s style as a criterion, but the Gospel of Matthew is a sufficiently long work (as NT documents go) to consider evaluating style, provided that there are multiple occurrences of the word or phrase in question. The argument of the NA committee here can easily be checked if the reader has access to any adequate computer application to manipulate the Greek text. In fact, an English search tool should be sufficient in this case, and I leave to the reader to make his or her own decision about the accuracy of the committee’s conclusion.

As an aside, I note that dissenting opinions are sometimes signed in Metzger’s Commentary, and it would have been helpful to know where Kurt Aland stood on the issue, given that the shorter reading was also, the earlier reading. This would seem to favor it in the Alands’ local-genealogical method.[84]

We have already seen that a number of critics take the position that scribes more often omit than add, so the shorter reading criterion carries little weight in this decision, and consequently it is even more difficult that it might appear from Metzger’s discussion. In seeing how this issue has filtered down to Bible translations, we can conclude that the ESV translators settled on a reasoned eclectic choice in this variation unit, resulting in their agreement with the NKJV.

MARK 1:37

Did Mark use good Greek?

Mark 1:37 New American Standard Bible (NASB)

they found Him, and *said to Him, “Everyone is looking for You.”

The problem here is not at all evident because it is completely hidden behind English that appears normal, and it is not covered in Metzger’s Commentary. Those familiar with NASB style know that the asterisk (*) indicates a peculiarity in Greek, a present tense used for the past, but this is not the issue. If we ignore it, the original Greek essentially has, “and they found Him and said to him.” I mention this variation unit because it illustrates the issue of Attic Greek that we discussed earlier. In “classical,” Attic Greek (I would say conventional ancient Greek), subordination was used to connect verbs, a grammatical style called hypotaxis. English more often uses parataxis, a structure connecting verbs together with conjunctions, and ancient Hebrew/Aramaic used parataxis heavily. The Greeks preferred to make one verb the “main” verb grammatically and attach others to it as participles or other verbal forms, a practice that was foreign to Hebrew/Aramaic style.

In the reading of 03 and 01 for 1:37, Mark has the two verbs cited above connected by the conjunction “and,” just as the translation indicates. In the MT and other manuscripts, however, “found” has the form of a participle, as one would expect of conventional Greek. Though 03 and 01 are nearly alone in their support of the paratactic reading,[85] a documentarian will somewhat reluctantly choose their reading lacking reasons to do otherwise. This kind of external support means nothing to a thoroughgoing eclectic or a reasoned eclectic taking the local-genealogical approach (recall the Alands’ position on a small group of manuscripts). However, internal criteria would favor the paratactic reading as well. Mark’s Gospel is a relatively substantial work, and he seldom uses hypotaxis instead of parataxis in these narratives.[86] We can assume that the MT reading was the work of scribes who perceived Mark’s parataxis as non-standard Greek and were attempting to improve it. So it comes as no surprise that the reading of 03/01 appeals to documentarians and reasoned eclectics alike. Those favoring the MT will most likely choose its reading.[87]

There are a few instances in Mark’s Gospel where he uses conventional hypotaxis, and I think they are sufficient to indicate that he could write in that style when he desired. I see his parataxis and other unconventional features of his Greek as the result of translating eyewitness accounts in the forms that he received them. The same stylistic variations can be seen in Luke, only to a much greater extent, and we know that Luke was capable of writing literary Greek.

Romans 14:19

Did Paul write “let us pursue” or “we pursue”?

Rom 14:19 English Standard Version (ESV)

So then let us pursue what makes for peace and for mutual upbuilding.

Rom 14:19 New American Standard Bible (NASB)

So then we pursue the things which make for peace and the building up of one another.

This is one of the more difficult variation units that one may encounter, depending on one’s preferred method or approach. It should be an easy decision for MT advocates because the MT has “let us pursue” (cf. the NKJV), and that makes good sense in the context. But when it seems that the moment is right for exhortation, 03, 01, and a number of other witnesses read “we pursue” instead, and Paul does not deliver another command until verse 20 (“Do not tear down…”).

This is a convenient opportunity to call attention to the GNT rating system. The choice of “let us pursue” has the lowest level of confidence, a ‘D’ rating. Many have criticized the rating system as being too subjective or simiply inadequate, but I think it is useful in revealing what the editors thought, whether one agrees with them or not. In this case, Metzger also makes the extraordinary statement in his Commentary that the reading “we pursue” has “slightly superior uncial support.”[88] One wonders if he had his tongue in his cheek when he wrote that, considering that the reading is supported by 01, 02, 03, and others. It might be a little stronger if it also had the support of 06 (D), which agrees with the MT, but the decision is an easy one for documentarians if only external evidence is considered.

Metzger notes in the same place that Paul’s style fits the simple “we pursue” here, but there are two factors we must consider that complicate the issue. First, the two forms of the verb in question might have sounded nearly alike to a scribe. This fact would tend to make the manuscript evidence of far less significance since an error–if there was one–could have crept in very early after the autograph. The second factor is that of the harder reading (lectio difficilior). Clearly “we pursue” is the harder reading because we expect an exhortation instead. That, taken together with its superior external support, seems to make it the better choice. But Metzger reports that the committee felt that the context required “let us pursue.” In essence, they rejected the lectio difficilior in this case as too difficult.

The case for this reading being too difficult can legitimately be made because it is easy to explain it as a misspelling of the verb based on the similarity of sound that I mentioned above. Even deliberating as a documentarian, I would have to take this into serious consideration. To base the decision entirely on the manuscript support might be irresponsible in this case. In the end, I am still inclined toward the “we pursue” reading for two reasons: 1) it is difficult, but it does not seem impossible in the context; 2) as Metzger pointed out, Paul’s style better fits it.[89] With the latter, I also take into account that fact that Romans is a book of relatively substantial length, long enough to provide evidence of the author’s style.

1 Corinthians 13:3

Did Paul refer to martyrdom by fire?

1 Corinthians 13:3 New American Standard Bible (NASB)

…and if I surrender my body to be burned

1 Corinthians 13:3 New Revised Standard Version (NRSV)

…and if I hand over my body so that I may boast

I close this chapter with one more example from Paul, another challenging one, but also one that is especially interesting. It is an easier decision for advocates of the MT than for documentarians, though the MT is problematic. We will probably agree that Paul is referring to martyrdom, and granting that, the phrase “to be burned” clearly makes more sense than the phrase about boasting. The two Greek verbs in question have almost identical spelling, differing only at one letter. However, the letter in the one word would not normally be confused with the corresponding letter in the other, so we can start with the assumption that the scribe or scribes who first made the change inferred that the one verb was an error for the other. If so, it follows that the verb for boasting was changed to that for burning rather than the reverse.

The MT has a future subjunctive verb, which may mean nothing to you, but as a matter of fact, no such form exists in Greek! It turns out that a good number of other manuscripts have almost the same verb, but in the future indicative–which does exist–and the difference between the two is a single letter: a short ‘o’ (omicron) vs. a long ‘o’ (omega). Not to bore you with grammar, I hope, I need to add that the subjunctive is needed in conventional Greek for the construction Paul used here. Nevertheless, the future indicative is sometimes used in the same construction elsewhere. Knowing that we can reverse-engineer the thinking that led to the reading in the MT: a scribe found the future indicative, considered it a mistake for the subjunctive, and changed part of the spelling but forgot to delete the rest that stood for the future tense.

So I think those who favor the MT can concede that a simple mistake was made in that text, but that the intent of the scribe was the word for “burn,” which is also the meaning that seems to make much better sense. As we have already observed, easier readings are typical of the MT.

Documentarians are faced with very convincing manuscript support for the verb that means “boast.” The same uncial manuscripts that Metzger called “slightly superior” above for the reading in Rom. 14:19 are in this case called “early and weighty” by him, as they should be.[90] There is no doubt about this being the harder reading; it is inconceivable that if burning were the reading of the original, it would be changed to “that I may boast.” The only question is whether it is too hard to be viable.

Some might protest that Paul could not boast postmortem in any case, but I think he could have referred to last words prior to execution, if not sooner. So the question mainly becomes whether the pejorative connotation associated with boasting can be mitigated here. It is possible that Paul is referring to a justified boast, as he seems to in 2 Cor. 8:24, Phil. 2:16, 1 Thess. 2:19, and 2 Thess. 1:4. I suspect it is also possible that he did not mean to refer to such boasting. Perhaps he intended to portray martyrdom as it was accepted by some with wrong attitudes. As we think these possibilities through, we are working with internal criteria: the author’s usage elsewhere and the context.

As a translator for the NASB, I especially like this one because it gives me the opportunity to have a slice of humble pie. I am always very reluctant to put a lectio difficilior reading with the best external support in the category of being too difficult to be viable unless there is a good explanation for its creation as an error. In this case, I think “boast” is viable and should, therefore, be the preferred reading, but the opinion of the other translators prevailed. Oh well, no one is perfect (myself especially). I must at least concede that the great similarity between the two words in spelling and thereby in sound as well could have provided an opportunity for the creation of “boast” as an error.

I hope these examples have revealed how textual criticism can be done practically, focusing on the criteria that are not only relevant for each variant unit but also for the particular approach used. I confess to having said relatively little in reference to thoroughgoing eclectics. At the risk of oversimplification, I remind the reader that in this approach the manuscript evidence does not matter, and only internal criteria count. In any case, I believe future work in textual criticism will necessarily focus on the CBGM, which we will discuss later.

          [1] Bruce M. Metzger and Bart D. Ehrman, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 4th ed. (New York: Oxford University Press, 2005).

          [2] Bruce M. Metzger, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 3rd enl. ed. (New York: Oxford University Press, 1992). Also identified earlier as TTNT-M.

          [3] Bruce M. Metzger, A Textual Commentary on the Greek New Testament; a Companion Volume to the United Bible Societies’ Greek New Testament (4th Ed.), 2nd ed. (London: New York, United Bible Societies, 1994).

          [4] Kurt Aland and Barbara Aland, The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism, Paperback ed., trans. Erroll F. Rhodes (Grand Rapids: W.B. Eerdmans, 1995). Second edition identified earlier as TTNT-A.

          [5] Bart D. Ehrman and Michael W. Holmes, eds., “Criteria for Evaluating Readings in New Testament Textual Criticism; by Tommy Wasserman,” in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd ed., Studies and Documents, vol. 46 (Leiden; Boston: Brill, 2013), 579–612. Identified earler as TNTCR.

          [6] Metzger and Ehrman, Text 4th Ed., 300–301.

          [7] Aland and Aland, Text, 281.

[8] Ibid., 280.

[9] I use “witnesses” here to refer properly to the texts, not the manuscripts. Age is not a decisive factor because one can find an early text in a late manuscript.

[10] Cf. chap. III above. It is certainly true that oral tradition thrived in biblical times, but Walton and Sandy take extreme positions in support of their thesis, using arguments that do not bear up under scrutiny.

[11] John H. Walton and Brent Sandy, The Lost World of Scripture: Ancient Literary Culture and Biblical Authority (Downers Grove, Illinois: IVP Academic, an imprint of InterVarsity Press, 2013).

[12] Ibid., 63, 67, 299.

[13] Ibid., 67.

[14] I do not take the position that one writer corrected the other; I believe that Christ taught either version on separate occasions.

          [15] Institut für Neutestamentliche Textforschung, Novum Testamentum Graecum: Editio Critica Maior; Part 1: Text, 2nd ed. (Stuttgart: Deutsche Bibelgesellschaft, 1997–2013), 30*.

          [16] Ibid., 32*

[17] Of even greater significance, perhaps, is the fact that all such “split lines” are regarded by the editors technically as lacunae (ECM2, Part 1, 34*). By comparison, the Alands would call such cases “insoluble ties,” and would only resort to that conclusion with great reluctance (Aland and Aland, Text, 280).

[18] Aland and Aland, Text, 280.

          [19] Bruce M. Metzger, A Textual Commentary on the Greek New Testament; a Companion Volume to the United Bible Societies’ Greek New Testament (4th Ed.), 2nd ed. (London: New York, United Bible Societies, 1994).

[20] Metzger and Ehrman, Text 4th Ed., 302. This is treated as a separate criterion in Metzger’s Textual Commentary.

          [21] Metzger, Text 2nd Ed., 209.

          [22] Metzger, Textual Commentary, 11*-16*.

            [23] Metzger and Ehrman, Text 4th Ed., 302.

[24] Metzger and Ehrman, Text 4th Ed., 212.

[25] See Metzger, Textual Commentary, 12*-13*.

[26] Ibid., 12*.

[27] Aland and Aland, Text, 281.

[28] Ibid., 281.

[29] See ibid., 280 ff.

[30] See ibid., 284.

[31] Ibid., 289.

[32] Metzger and Ehrman, Text 4th Ed., 327-28.

[33] Aland and Aland, Text, 289.

[34] See ibid., 231.

[35] Ibid., 55.

          [36] Bart D. Ehrman and Michael W. Holmes, eds., “Textual Clusters: Their Past and Future in New Testament Textual Criticism; by Eldon Jay Epp,” in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd ed., Studies and Documents, vol. 46 (Leiden; Boston: Brill, 2013), 519–77 passim.

          [37] Epp likewise calls the Byzantine family the “A-text cluster,” the Alexandrian the “B-text cluster,” and the questionable midway group the “C-text cluster.” See ibid., 556.

          [38] See “Reasoned Eclecticism in New Testament Textual Criticism; by Michael W. Holmes,” in ibid., 771–802.

          [39] See  “Critical Editions and Apparatuses of the New Testament; by Juan Hernández Jr,” in ibid., 696.

[40] D. C. Parker and H. A. G. Houghton, eds., Textual Variation: Theological and Social Tendencies?: Papers from the Fifth Birmingham Colloquium on the Textual Criticism of the New Testament, 1st Gorgias Press ed, Texts and Studies, 3rd ser., v. 6 (Piscataway, NJ: Gorgias Press, 2008), 109-27.

[41] Ibid., 126.

[42] The tools are found at http://intf.uni-muenster.de/cbgm/index_en.html. Once there, the user can choose between the versions. I highly commend the ECM editors for their transparency in making both versions publicly available.

[43] See above, p. 222.

[44] See below, p. 375 f.

[45] See the Glossary for this term. I have no reluctance to equate “original” with “autograph” here, but I acknowledge that some may insist on “original” as the text from which all other known copies ultimately came. I am also making the general, faith-based assumption that we have the autograph contained in a combination of extant early manuscripts.

[46] See below, 341-58.

[47] Metzger and Ehrman, Text 4th Ed., 302-303.

[48] See above, 141.

          [49] James R. Royse, Scribal Habits in Early Greek New Testament Papyri, New Testament Tools, Studies, and Documents, vol. 36 (Leiden ; Boston: Brill, 2008).

          [50] See Bart D. Ehrman and Michael W. Holmes, eds., “Scribal Tendencies in the Transmission of the Text of the New Testament; by James R. Royse,” in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd ed., Studies and Documents, vol. 46 (Leiden; Boston: Brill, 2013), 461–78.

[51] See above, 150 f.

[52] Aland and Aland, Text, 281.

[53] Metzger and Ehrman, Text 4th Ed., 303.

[54] Aland and Aland, Tex, 281.

            [55] Ehrman and Holmes, “Criteria; by Wasserman,” 589.

[56] Metzger and Ehrman, Text 4th Ed., 303.

[57] Ehrman and Holmes, “Criteria; by Wasserman,” 589-92. Actually Wasserman treats the substitution of unfamiliar words with more familiar as a separate criterion, which is reasonable.

[58] This will not win me any friends, but I dare say that I have found classical grammars (particularly Smyth) more helpful in addressing questions about NT Greek than NT grammars. The relatively few adjustments that must be made to accommodate NT Greek style are a small price to pay for better explanations and categories that one finds in classical grammars.

[59] I will provide an example (Mark 1:37) below, however, where external evidence indicates that the stylistic reading was a secondary attempt at improvement.

[60] Metzger and Ehrman, Text 4th Ed., 303-304.

[61] Metzger, Text 2nd Ed., 210; duplicated in the third edition.

[62] Aland and Aland, Text, 280.

[63] Ehrman and Holmes, “Criteria; by Wasserman,” 592.

[64] Metzger, Text 2nd Ed., 210.

[65] Metzger and Ehrman, Text 4th Ed., 303. The phrase “in the Gospels” found in the third edition is conspicuously missing.

            [66] Metzger, Text 3rd Ed., 217.

[67] Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early Christological Controversies on the Text of the New Testament, Updated and with a new afterword (Oxford: Oxford University Press, 2011), 233.

[68] See above, 228, 231.

[69] With NA28 and GNT5 the committees are shifting toward a de facto form of thoroughgoing eclecticism, in that traditional external criteria play little or no role in decisions. The Alexandrian manuscripts (chiefly 03 and 01) still are highly rated, however.

[70] Eberhard Nestle et al., eds., Novum testamentum Graece, 26.,  Aufl., 7. Dr., Großdruckausg (Stuttgart: Dt. Bibelges, 1979), 43*.

[71] See above, 236 f.

[72] As Constantinople [or Byzantium] became the center of the Greek-speaking church, the local text there was to become the dominant text for the whole of the empire.

[73] The paper, undated, was available at Dallas Seminary and had the subtitle, “A Revised Edition of a Paper Originally Called ‘Introduction to the Textus Receptus’.”

          [74] See David Alan Black, ed., “The Case for Byzantine Priority; by Maurice Robinson,” in Rethinking New Testament Textual Criticism (Grand Rapids, Mich.: Baker Academic, 2002), 125–39.

            [75] Ibid., 126 f.

            [76] Metzger, Textual Commentary, 12*.

            [77] Black,  “Byzantine Priority; by Robinson,” 127.

            [78] I of course exclude variant units where the Byzantine witnesses are divided.

[79] See above, p. 232.

[80] Aland and Aland, Text, 281.

[81] The strategy behind the UBS apparatus was that the text was designed for translators, and it was thought that only variation units that affect translation need be covered. I find it helpful to have copies of both, using NA but occasionally comparing GNT with it for clarification. In the example of Matt. 6:33, one has to work through the NA apparatus to piece together what is going on, while that in the GNT immediately makes sense.

[82] I am indebted to my co-author Mr. Andrews for suggesting the first examples. Some of the observations and comments included therein originated with him.

          [83] Metzger, Textual Commentary, 16.

[84] See above, 232-33.

[85] They appear to be joined only by two late Greek manuscripts and a fifth-century Latin manuscript.

[86] I confess to making a comment here about style and encourage the reader to verify it. A light browsing of Mark should be sufficient, or perhaps a search for καί followed by an indicative verb.

[87] Here, the NKJV has “When they found Him, they said…,” which agrees with the hypotactic reading of the MT (however it also agrees with manuscript D).

[88] Metzger, Textual Commentary, 469.

[89] See ibid. Specifically, Metzger observes that the initial Greek phrase found here is always followed by the indicative in Romans (seven or eight times).

[90] Ibid., 497.