The New American Standard Bible (NASB) has “Servant” in Acts 3:26 where the King James Version (KJV) has “Son.” The Greek word παῖς can mean “servant” or “son.” Most differences between versions are translational, but 5% are textual. NASB has “fruit of the Light” in Ephesians 5:9 where the New King James Version (NKJV) has “fruit of the Spirit.” This is a Greek question because the English difference results from two different Greek words. One is a corruption because only one can be original (though some textual scholars hope to change this assumption). Although manuscript differences occur in virtually every verse of the Greek New Testament (NT), the main text editors agree on 95% of the selections because most of the variants involve spelling (e.g., Δαυείδ or Δαβίδ for Δαυίδ) or abbreviations for the nomina sacra (e.g., Ις or Ιης for Ιησοῦς), and many variants are rejected as nonsense, mistakes or rare. Most of the 5% that textual scholars still disagree on is sub-translational, like word order, or relatively minor, like synonyms, a pronoun for a noun, or “Jesus” for “the Lord Jesus.” So the number of significant text disputes is only a fraction of 1%, though degree is on a scale.

Even though textual differences cause a small percentage of the differences in our English versions, they are important, for the Word of God is life to the perishing. Why do we sometimes hear about wide agreement among scholars over recovering the original NT, and sometimes that scholars have given up hope of ever recovering the original? When we emphasize agreement, we talk about the 95% or 99%. When we focus on differences, we look at the 1%. Are scholars getting closer to resolution? Some scholars believe they are getting a little closer, but others believe the decisions will remain subjective permanently. Minor skepticism coming from NT textual criticism must weaken the faith that we possess the whole exact original NT to a minor extent, but does not need to weaken faith in Christ, the Word of God, or who he is.

Part I. History, Schools and Outlines

1. Historic Shifts in Scholarly Positions

The path from the original NT to the typical English NT follows three steps. First, the autograph was copied successively into many Greek manuscripts. Some of the 5,000+ survivors (many fragments and all except the shortest with errors) are edited back into a few printed critical Greek editions, which attempt to represent the original. Then, the critical Greek editions are translated into many contemporary language versions.

The Greek NT scholars of sixteenth-century Europe were acquainted with the manuscript families available to us in the twenty-first century, though not extensively, but felt confident that the Greek church had substantially preserved the original Word of God in the Textus Receptus (TR) editions, based on a set of Byzantine manuscripts, which differed from the Vulgate in numerous places. The biggest change in the history of textual criticism so far occurred in 1881, when Westcott and Hort (WH) published a shorter critical edition based on a completely different and smaller set of older Greek manuscripts (now commonly called Alexandrian) containing over 6000 differences from the TR. The resultant Revised Version and its successors came to replace the KJV extensively. These scholars believed that they, like Hilkiah of old, had recovered the original Word of God.

In the twenty-first century, the majority of textual scholars still prefer the Alexandrian manuscripts over the Byzantine ones (even though they have a conflicted relationship with the nineteenth-century principles that support their preferences), but they no longer believe that it will be possible, barring miraculous new discoveries, to recover all the original words of the Bible. For example, Reuben Swanson, one of the most eagerly-read modern critical scholars, an advocate of the Alexandrian text over the Byzantine text, and a creator of a kind of successor to the Nestle-United Bible Societies Greek NT (NU), states: “To believe that we can reconstruct out of fragmentary and late material ‘the original pure text’ is thus a delusion.... There can, therefore, be no agreement among critics as to which reading may have been original.”[1] He calls it simply “an impossible task.”[2] According to Swanson, the preferences of the scholars, like everyone else’s, are subjective. He opts instead for a massive, uncommentated, full-disclosure manuscript-information-dump which, however, does not facilitate a traditional translation. Following the same cautious trend the International Greek New Testament Project has started to publish a critical edition, the huge New Testament in Greek,[3] in which they also decline to commit on the disputes in the NU manner. They use the 1874 TR as a collating base for the papyri readings. A modern trend is to back off from claims of having the solution.

Reformation-era Bible translations such as the Luther Bible and later the KJV were translated from the TR series of editions. Western NT scholars today frequently accept the opinion that the critical editions of Nestle Aland (NA) and United Bible Society (UBS) more accurately approximate the autograph than do the Byzantine TR or Greek Orthodox Patriarchal edition. NA and UBS now have the same Greek edition (NU), and all the popular versions except NKJV are translated from that one Greek edition. The NU critical edition is based mostly on two parchment uncial manuscripts from the fourth century, Vaticanus and Sinaiticus, supported at any given place by a small set of other Alexandrian uncials (capitals) and papyri. The Byzantine family has no popular English translations besides KJV and NKJV. The Orthodox have a modern Greek translation from the Patriarchal edition. So, from the version perspective, the current worldwide Greek edition conflict largely boils down to NU (Alexandrian) vs. TR or the Patriarchal edition (Byzantine).

2. The Three Schools of Textual Criticism

Textual criticism currently has three schools, popularly known as thoroughgoing eclecticism, reasoned eclecticism and Byzantine priority. Thoroughgoing eclecticism has published the works of the International Greek NT Project, but that is not an edition in the sense used here because it provides a multiplicity of texts rather than a final Greek selection. Swanson’s works fall in the same category. Various eclectic editors have published results (usually mostly Alexandrian) that may influence Greek scholars. However, no popular version is based on them, since there is no institutional continuity and they keep changing. The two Byzantine priority editions have had virtually no impact on the Bible market so far. The Hodges and Farstad Byzantine majority edition was used in the footnotes of the NKJV, but the TR was the basis for the NKJV. The TR has been static since 1894 and no longer has a recognized school, and there is no published school for the Patriarchal edition. Therefore, neither the Byzantine nor the thoroughgoing eclectic school will be the subject of this article.

The school of reasoned eclecticism, the school that produces the NU, is a misleading misnomer. The school will be called the Alexandrian priority. It is no more reasoned than the other schools, and it is the least eclectic. It is over 99% Alexandrian, a result that could not be produced by any other method than excluding its rival text types. The Alexandrian priority with its NU edition is currently the dominant effective scholarly enterprise, with the Institute for New Testament Textual Research headquartered in Munster, Germany, its alliance with the UBS, and close adherence to the 1881 WH edition. The Alexandrian priority is the main subject of this article. All schools involve selections by vote of the most qualified readings. The main difference between schools involves voter qualifications.

The salient characteristic of the Alexandrian Bible is shortness. The English Bible reader’s three main options are to choose an Alexandrian Bible as the original, or to choose a Byzantine TR Bible as the original, or to use either one but regard them both as reconstructions of non-original early Bibles whose original cannot be scientifically determined. The third option, although closest to the evidence in my opinion, is the least satisfactory or popular with textual critics and lay readers alike.

3. The Author and the Target Audience of This Article

Anyone who is interested in why NU editors pick one manuscript variant over another may read this article without knowing Greek. The target audience that I have in mind is thousands of people like myself who read the Greek NT in an NU edition and look down at the bottom of the page in the apparatus to see what other variants occur and what witnesses support them. When the reasons for the NU selection are unclear, the Textual Commentary on the Greek New Testament (TCGNT)[4] is consulted for an editor’s justification. Seminary professors used to teach that we have the original NT. We just do not know if it is above the line or below. The old Nestle apparatus used to have the TR variants below the line. However, as time passed and newer NU editions appeared, the majority of strictly TR or Byzantine variants disappeared from below the line also, in keeping with NU thinking of TR and the Byzantine text as corrupt and debased.[5]

This article focuses primarily on how the NU decisions were made, paying special attention to the words of Aland and Metzger, who made those decisions, rather than what hundreds of experts think about textual criticism in general, because those experts did not make the selections for the NU editions or publish NT editions. Therefore, the primary sources, besides the UBS4-NA27 itself and the TCGNT that lists its justifications, are The Text of the New Testament,[6] by Aland, the chief NU editor, and Swanson’s NT Greek Manuscripts,[7] that lists the variants more fully and conveniently.

I am a Greek reader, not a textual critic. My tools also include Greek NT databases and math and programming skills. Sometimes it is considered presumptuous for a non-expert to judge experts. I have great respect and appreciation for the labor, collection, transcription, classification, study and analysis of textual critics. However, it is apparent that textual critics have no special advantage in the area of clear thinking, once all the facts are before us and originals are to be selected from the variants.

Here is a typical example of scholars’ thinking problems. Aland,[8] speaking of P52, the oldest extant NT manuscript, from 125 AD, associates it with P75, the best Alexandrian papyrus. He states: “It preserves the text of the original exemplar in a relatively faithful form.” Alexandrian prioritists often use this fact as evidence of the continuity of the Alexandrian text back to the original. Metzger classified P52 as Alexandrian. The words οὐδένα ἵνα are easily legible.[9] It is a wonderful treasure reaching theoretically to within a few copies of the autograph.

What neither the experts nor Aland nor Metzger think to mention is that the 33 words or word fragments of P52 all happen to fall in a common section of the NT that is agreed on by the main Alexandrian and Byzantine manuscripts. The same 33 words are in the 1894 TR. Surprisingly, not a single one of the words in P52 is a type discriminator. Anyone who reads Swanson can see this right away. So, in spite of what the experts claim, P52 proves nothing about text types. The experts mislead people who accept their opinion into thinking that there must be something specially Alexandrian about P52. When experts with commitments come to conclusions, readers do well to read between the lines and read the source lines and think for themselves. The NU experts want P52 to be Alexandrian and think of it as Alexandrian and have the knowledge to make judgments, but textual critics as a whole can still make simple mistakes in objectivity that ordinary Greek readers can see through if they compare the available material.

4. Outline of the NU Variant Editing Stages to Reject Corruptions

The following outline may be consulted for orientation as the reader progresses through the article. Percentages are the portion of the edit work on variants completed by each stage.

Standard editing—solid basic historic common-ground among all schools (Part II)

Normalization (nomina sacra, spelling and lexical mistakes)—stage 1 (50%)
Elimination of singular readings (Aland Rule 7)—stage 2 (25%)

NU critical editing—speculative, specific to Alexandrian priority (elimination of corruptions)

External rules (evidence, criteria, canons, reasons: text types)—stage 3 (20%) (Part III)

Aland categories, the A-list (Aland Rule 6: “weigh not count”)

Category I. Alexandrian (uncials) [candidates] VATICANUS, Sinaiticus
Category IV. Western [voter] Bezae
Category V. Byzantine [non-voters]
Papyri categories (Aland degrees of copy fidelity)

Strict (P75) [candidate]
Normal (P72) [voters]
Free (P46, P66, P45, P47, most) [voters]
Paraphrastic [voters]

Elimination of the Byzantine, Western text types and free Papyri
The A-list vote (mechanical)

Internal or secondary rules (evidence, criteria, probabilities) override—stage 4 (5%) (Part IV)

Transmissional probabilities (circular and inconsistently applied)

Lectio brevior (shorter)
Lectio difficilior (harder)
Order of derivation, etc., etc.

Intrinsic probabilities (exegetical, open-ended, arbitrary and subjective)

Note: the outline applies to the variant-by-variant selection process, not to the prior establishment of the A-list. A distinction between “external” and “internal” is very regular in the text-critical literature (although content varies), but the terms “rules,” “criteria,” “probabilities,” “evidence,” “canons,” and “reasons” are used more or less interchangeably, suggesting their vagueness.

5. Outline of the Flow of the Argument to Discredit the NU Critical Editing Program

The propositions stated below without proof or explanation will be discussed in the article. They are outlined here first to help the reader identify interrelations in the extended argument flow.

The manuscript age and internal rules are not the primary selectors of variants.
The confusing, inconsistent critical rules are a smoke screen when applied to variant selection.
Any exegetical ideas may be called internal rules. Internal rules are subjective and fluid.
To the extent that variant selection depends on internal rules (a minority), it is subjective.
Actual main NU edit selections depend first on standard editing, a valid historical process.
The singular corruptions from standard editing are determined by an extreme form of vote.
NU edit selection depends second on mechanical voting and the A-list, not on rules.
Both Byzantine and Alexandrian programs ultimately depend mainly on a (secret?) vote.
But the Alexandrian vote ultimately depends on disqualifying most of the potential electorate.
Proof of vote dependency is that the vote better predicts NU results than do the critical rules.
Voting is more defensible than critical rules because it is objective, grounded on text types.
The actual main function of the critical rules is to determine a priori text types, not variants.
However, the internal rules are circular and inherently invalid.
Based on Aland Rule 7, corruption rates disprove the external rule: the older, the more errors.
Since the critical rules are inherently invalid, the A-list and the NU process are invalid.
The second factor that determines the A-list is text history.
In normal text history, the text that dominated the heartland early would dominate it later.
The critical history of the Byzantine text invents an imaginary interruption.
Since the text history is invalid, the A-list and the NU process are invalidated again.

Part II. Standard Editing: Common to All Schools (Stage 1 and 2)

6. Standard Editing, Critical Editing and Corruptions

Textual Criticism is divided into manuscript collection (a science) and selection (an art). Edit selection is divided into standard editing (a solid practice common to all schools) and critical editing (a speculative and subjective art specific to each school). When faced with differences in manuscripts, editors compare the different readings (variants) and attempt to identify the one original and discard the remaining corruptions (the changes that later scribes presumably introduced). Manuscript copyists and editors have been performing standard edits in a roughly similar fashion from the first- to the twenty-first century. Standard editing falls into two main parts: (1) normalizing spelling (including word fragments and garbled letter sequences), and (2) eliminating singular readings. After standard editing, critical editing involves deciding between readings that all make sense and have enough witness support, based on ideas about how scribes were supposed to alter their exemplars. In the NU world critical editing typically involves text types (like Alexandrian and Byzantine), theories of text history, and decision criteria or rules based on presumptive scribal habits.

7. Normalization: Standard Editing, Stage 1

Normalization is the simplest part of editing that governs the great majority of the decisions. The NU reader is largely unaware of it because most of the deviant spellings do not appear in the apparatus. Usually scribes abbreviated sacred names into nomina sacra (e.g. θς for θεός) with a bar on top. The abbreviations were somewhat irregular but the editors expand them all. They also standardize spelling. In John 1:39 all the papyri and older uncials have ραββεί, but NU normalizes it to ραββί. Most spelling deviations are phonetic. In John 7:7 some manuscripts have μισίν and some have μεισεῖν for μισεῖν, to hate. Vaticanus is the worst with μεισῖν, the only one with both vowels wrong. NU normalizes them. Vaticanus always writes παρισία for παρρισία (e.g., John 7:4). Such deviations do not appear in the apparatus and are not counted as corruptions. Not all text editors use the same orthographic standards. The TR editor removed most of the movable-nu’s before consonants while the NU editor retained most of them (ἔφαγε, ἔφαγεν). The manuscripts go either way—movable-nu is the most common difference between NU and TR. A number of other NU-TR differences, like contractions and second aorist endings, fall under such normalization differences. So some normalizations become edition differences, while most do not.

Lexical mistakes form a subtype of spelling deviations. Vaticanus has ἐσίν for εἰσίν (John 6:64), ἀσπασθε for ἀσπάσασθε (Rom. 16:7), συγγενην for συγγενῆ (Rom. 16:11) or the declensional mistake χειραν for χεῖρα. These are not phonetic equivalents, and are not even real words as written. Lexical mistakes are scribal mistakes, because God’s word is not nonsense. Even if Paul’s amanuensis had accidentally written a vowelless verb in the autograph, the next copyist would have corrected it. A scribe would not copy such items if he saw them in his exemplar. So these mistakes are useful to characterize habits of the particular scribe who copied this manuscript, not his predecessor.

8. Eliminating Singular Readings: Standard Editing, Stage 2

Eliminating singular readings has historically been the second major standardization technique after normalization. A singular reading is one that occurs in only one manuscript. However, the singular status may vary, depending on the manuscript population under consideration. A reading may be singular among Alexandrian manuscripts, or singular among pre-seventh century manuscripts. The singular reading is considered a corruption and is rejected by most editors. “Singular” is a relative term. If a new manuscript is found, a reading that used to be singular might become merely an infrequent reading. There are many extant manuscripts, and singularity shows that scribes probably did not like these readings enough to reproduce them. Sometimes the singular readings do not make sense. At other times they differed from the familiar majority. The idea is that one scribe is much more likely to make a singular corruption than many scribes are to make the same corruption.

Modern editors are extremely wary of selecting a singular variant, and correctly so, as were ancient copyists. NU has a rule against it and virtually never selects singular readings. Often mistakes or non-standard spellings are also singular. The Vaticanus scribe changed the official’s son, υἱόν, from “him” (αὐτόν) to “her” (αὐτήν, John 4:52). No other manuscript before the ninth century had αὐτήν there. No one copied it, because it was obviously a careless error, even though it makes sense within the immediate phrase. All the manuscripts except Vaticanus say that God “does not give the Spirit in measure” (John 3:34). Vaticanus dropped “the Spirit.” This corruption makes sense and is not an obvious corruption. Yet it is still classed as a corruption because it is singular. NU rejected it. Likewise in Acts 6:14, which has “and changes the customs that Moses handed down to you,” Vaticanus changed “customs” to “gentiles.“ No one copied this singular corruption, for obvious reasons.

NU rejected the Vaticanus’ νόμῳ τοῦ νοός (“law of the mind”) for νόμῳ τοῦ θεοῦ (“law of God,” Rom. 7:22) because it was singular, even though it makes sense, linguistically. Perhaps the Vaticanus scribe was using Scripture to communicate his ideas about God, the law and the mind, or perhaps he was just absent-minded. But standard editing classifies it as a corruption mechanically without indulging in guesswork. The Vaticanus scribe or his predecessor omitted the phrase “of my brothers” in Romans 9:3 (“for the sake of my brethren, my kinsmen according to the flesh”). NU rejected this variant, since Vaticanus was the only one to drop this phrase. We could speculate about whether the scribe was offended at Paul calling the Jews his brethren, but the bottom line is that we know from this and a number of other singular omissions that the Vaticanus scribe or the scribe of his Alexandrian exemplar had a habit of dropping words and shortening his exemplar. These decisions, based on the singular readings, are common ground between the NU and all the Byzantine textual scholars.

In Romans 12:20, we see the Vaticanus scribe changing the accusative case τήν κεφαλήν to the genitive τῆς κεφαλῆς. NU and all scholars judge that Vaticanus changed it. They reject the variant because it is a singular reading. We add inflectional corruption to the list of the Vaticanus scribe’s suspected bad habits or to those of his predecessors. In Romans 15:32 Vaticanus changed “by the will of God” to “by the will of the Lord Jesus.” Who knows why? Since he was the only one to have that reading, NU and all scholars rejected it. In John 4:16 the Vaticanus scribe engaged in transposition by substituting the singular σου τόν ἄνδρα for τόν ἄνδρα σου, telling us more about the kinds of corruptions he generated. Although there is a minute theoretical chance that Vaticanus alone was original and all the others changed it, NU and all scholars are convinced by the laws of probability that Vaticanus has the corruption. Ancient scribes also operated under this assumption. When Paul wrote “Timothy my fellow worker” in Romans 16:21, the Vaticanus scribe alone dropped the word “my.” The possessive idea is implied in Greek anyway. We learn that when he (or his predecessor) came to a word that he considered superfluous, he felt free to eliminate it. He did it hundreds of times, and the NU editor restored it each time. The other Alexandrian uncial and papyri scribes also made similar singular omissions in other places. That is probably why the modern Bibles are shorter. The Vaticanus scribe frequently shortened ἐάν to ἄν. He shortened “Golgotha” (Γολγοθᾶ) to “Golgoth” (Γολγοθ). At other times he dropped or added a whole clause (John 17:15, 18). NU editors do not keep those particular corruptions because they are singular.

Often the Vaticanus scribe’s singular corruptions give us a little window into his mind. When he saw the phrase “the day of that sabbath” (ἡ ἡμέρα ἐκείνου τοῦ σαββάτου, John 19:31), he changed the gender of the demonstrative pronoun from masculine to feminine (ἐκείνη) because he probably thought it was on the “day” side of the article rather than with “sabbath.” While either way makes sense and the Vulgate also applied it to “day,” NU rejected it as a corruption because most of the early scribes rejected it when they made their copies. When the Vaticanus scribe saw “seeketh not her own” in 1 Corinthians 13:5 (the “love chapter”), he apparently thought “seeketh not what is not her own” would improve the text, but all editors reject it because it is singular. Alexandrian proponents usually claim that the changes were mostly accidental, but they were often not random accidents because random changes produce gibberish. Who knows why Vaticanus changed “for in Christ Jesus I have begotten you through the gospel (evangel)” (1 Cor. 4:15) to “for in Christ I have begotten you through the angel”? He shortened it. In any case, both changes are singular, and everyone rejects them as a corruptions.

In general, as we go back in time from the Byzantine scribes to the Alexandrian ones and all the way back to the papyri, the singular corruption rate doubles and doubles again, even when we control it for numbers of text types. The older, the worse, generally, in terms of this objective measure. Not all scribal corruptions were lexical mistakes or singular readings. If a scribe dropped a word and two other scribes copied that omission and their manuscripts survived, then the omission is not singular. If NU adopts the omission, it is not considered a corruption. Singular corruptions, as defined here, are easy to identify objectively, but other corruptions are more difficult. In Luke 11:11 (“If a son shall ask bread of any of you that is a father, will he give him a stone? Or if he ask a fish, will he for a fish give him a serpent?”) NU leaves out the part with the fish and serpent based on Vaticanus, P45 and P75. Sinaiticus and all other manuscripts have it. If those three Alexandrian witnesses dropped the original clause, NU is mistaken. One hint that NU is mistaken is that all three of the omitting witnesses have singular corruptions in the adjacent words. If the minority witnesses cannot agree among themselves, should they be trusted?

A high singular rate for a scribe does not necessarily predict a high rate of other corruptions, but it is suggestive, especially when he shows repetitive patterns and a variety of types. Both normalization and the standard elimination of singular readings are edit methods that deal with a larger variant set than critical editing, which deals with corruptions other than the obvious. The NU reader does not appreciate the standard editing scope because the majority of these variant readings do not appear in the apparatus. However the scope becomes apparent in the Swanson format.

The three key points about standard editing are: (1) it is by far the most extensive part of editing and is settled before critical editing, (2) its methods are objective and its results are agreed on by the modern Alexandrian and Byzantine schools and by ancient scribes, and (3) it is the method for identifying objectively defined corruptions, which are a predictor of general corruptions and a measure of scribal fidelity for each manuscript.

Part III. External Editing (Age and Text types): The NU Primary Method (Stage 3)

9. Critical Editing: Applying the External and Internal Criteria

Virtually none of the first and second century manuscripts have survived. Only a small fraction of one percent of the third- and fourth-century manuscripts have survived, and even they are limited to one region of the empire, completely missing from the autograph heartland. Since the oldest manuscripts show the greatest degree of wildness, statisticians would not agree that the tiny survivor sample with high variance is large enough to identify the probable representatives of even the third-century manuscripts with a significant confidence level, let alone the autographs. Yet this small set of early witnesses contains thousands of corruptions. The text critics face an apparently impossible task of separating out corruptions and identifying the originals. After they have preserved, dated, transcribed and collated the manuscripts, how do they attempt their goal?

After the NU Committee has performed the standard editing, it votes on the variant phrases of NU, phrase by phrase, theoretically based on “twelve basic rules for textual criticism” given by Kurt and Barbara Aland on p. 280 of their book, The Text of the New Testament. (See Aland’s twelve rules.) Since Aland was the longest-standing member of NA, and Metzger, who was also a Committee member, wrote TCGNT, the commentary that provides the Committee’s reasons for many selections, their books will be cited extensively. Aland Rule 7 belongs to the initial standard editing process. It virtually forbids singular and rare readings. Another four of the 12 rules are general standards such as: the texts must come from real manuscripts (not conjectural emendations), only one can be original, and the manuscripts have precedence over the quotes of the fathers and the early versions. These five rules are a part of the solid common background of textual criticism, including standard editing, agreed on by Alexandrian and Byzantine editors alike. The other seven rules govern critical editing and are specific to the Alexandrian priority.

10. The External Criteria, Stage 3

Aland’s seven critical rules are particular to NU editing and are often highly disputed. Metzger calls them criteria or evidence. He also calls them probabilities, since the probability of their being applied in any given variant set is unpredictable except that they need to support an Alexandrian reading. They are divided into four external and three internal rules (or criteria or probabilities). It is immediately evident that the external criteria come before the internal criteria generally and involve the age of the manuscript and text types. The text types that Aland identifies are Alexandrian, Western and Byzantine. Metzger states: “In general, earlier manuscripts are more likely to be free from those errors that arise from repeated copying. Of even greater importance, however, than the age of the document itself are the date and character of the type of text that it embodies, as well as the degree of care taken by the copyist while producing the manuscript.” As we saw before and will see more so later, Metzger’s claim that the earlier texts are more error-free is not true, and is contradicted by Aland’s categories and by the NU selections themselves. Metzger himself is aware that most of the older papyri are not as good as the later Alexandrian uncials. But it is clear that in the Alexandrian priority text types, not age per se, are the determining fact of external evidence. Instead of text types, Aland usually uses the term categories, which is similar to text type but not identical.

Exactly how the text types or categories are applied to select variants, neither Aland nor Metzger reveals explicitly. The reader needs to gather this information inductively by examining the descriptions and the results. We get our first clue from Aland Rule 6: variants are to be “weighed, not counted.” A little reading and observation of the choices shows that this means that the Byzantine text, which Aland also calls the Imperial text, is eliminated from the candidate pool for selection. We get a hint of the Byzantines’ fate when Aland states: “They are all irrelevant for textual criticism, at least for establishing the original....”[10] If the thousands of later Byzantine manuscripts each counted as equal votes, they would swamp the few, older Alexandrian ones, and the present NU would be just a quaint reconstruction of a long-discontinued regional artifact. So the Byzantine elimination solves a central problem for the Alexandrian priority by axing the teeming Byzantine hoards with one stroke.

The word “weigh” also does not mean to assign a measure or weighting factor to each manuscript, as it would imply. There is no table of weights. It is not individual readings or manuscripts that are weighed. Rather it is the five NU categories [11] that are weighed. The categories span a continuum of text types from Category I (mainly Alexandrian) to Category V (mainly Byzantine) with Categories II and III in between and Category IV belonging to the “Western” codex Bezae and its allies. The Category V weight is zero. Metzger linked the external evidence to “the degree of care taken by the copyist while producing the manuscript.” Besides the text types, NU classifies the papyri by Aland fidelity categories: strict, normal, free, paraphrastic. Free means fast and loose, relatively speaking. The scribe takes liberties and makes mistakes.

11. The A-list (an Alexandrian subset) and the Mechanical Vote

TCGNT often mentions external and internal evidence and gives witness lists for selected and rejected variants, but does not mention categories or give a hint about the selection mechanics. Aland also does not disclose the selection mechanics. So the reader is left to deduce them from the results. In a particular case we picture several variants presented for election in a vote with qualifications for candidates and voters. Some variants, such as singular and Byzantine variants, are eliminated—disqualified as voters or candidates. Some variants can vote but not run. Some votes count more than others. The qualifications are based on the Aland categories. All of this is behind the scenes. The TCGNT reader, distracted by a sideshow of internal evidence, reads only vague reports of the voting process.

It becomes clear from observation that the effective, though unspoken, weighing rule of NU selection is that the text must come from the small aristocracy of Category I uncials, the A-list, which Aland calls “presumably the original text.”[12] The A-list always includes Vaticanus and Sinaiticus, and possibly one or two other Alexandrian uncials. Strict papyri theoretically also belong to the A-list, but P75 is the only strict papyrus of substantial size (parts of John and Luke). For pro-Alexandrians it is the crown jewel of papyri. It is closest to Vaticanus. The bulk of the papyri are classed as free and do not qualify as candidates, although the older ones can vote. There is usually only one papyrus for a given passage and sometimes none, although in John there are usually two. Older papyri are more like the Alexandrian than Byzantine. Category IV uncial Bezae and its papyri may vote among the candidates but may not run as candidates even if they are older. When the vote is tied, candidates are ranked with Vaticanus first and Sinaiticus second. The Category II uncials may vote on rare occasion. The other 90% of the Category V and III manuscripts and minuscules never vote or serve as candidates, though they may appear in TCGNT witness lists and in apparatuses as if they had some influence.

The vote will overwhelmingly go to Vaticanus if it finds support from some other voters, and next to Sinaiticus. Non-normalized spelling is not admitted to the edition, but does not disqualify a voter. The A-list vote usually settles the matter at the external stage and makes the internal probabilities superfluous. Although internal probabilities occasionally override the A-list vote, their frequency and application are unpredictable. One problem with the external criteria as described is that they are based on an imaginary text history, which makes them invalid even when they are applied objectively.

12. Type Vote Is Determinative, Not Internal Probabilities: Evidence

First Corinthians was chosen as a moderate-sized text to test the methodology described above to see how well it would match the NU edit selections. Vaticanus, Sinaiticus and Alexandrinus are the only three Category I uncials (candidates) for 1 Corinthians. P47 generally is the only old papyrus. So with the Western Codex Beza, there are five qualified voters for 1 Corinthians. Let us pause to examine the voter composition. Ninety-nine per-cent of the manuscripts are disqualified, including the Byzantine and minuscules. The three A-list manuscripts dominate the vote. At first glance it does not appear entirely rigged because two non A-list voters are included. In practice, however, they are only there to help choose between Vaticanus or Sinaiticus when those two differ. First, P46 and Bezae are the least aligned of the five, so they cannot gang up. Second, if one voter is singular or missing and the vote becomes two against two, the A-list wins because they outrank the others. Non A-list can only win the A-list vote if all the A-list are singular or missing.

First Corinthians contains over 6800 words. About 1100 words required normalization in at least one of the witnesses, so that was the most frequent editing. After normalization, about 1030 words (15%) had some variation in the five witnesses but NU only decided on 1009 of these (the rest are in square brackets). About 65% of these 1009 disputed words were settled by eliminating singular readings. The reading was “selected” in the sense that there was no more contest or vote needed. About 85% of these selection votes went to Vaticanus. When Vaticanus was singular, it went to Sinaiticus. This concluded standard editing.

The 353 words not settled in the singular elimination (stage 2) were put to the A-list vote (stage 3 or the external criteria). Three-hundred ten of the votes (88%) matched the NU selections. In other words, all but 43 of the 1009 disputed words were correctly predicted by singular elimination and the mechanical A-list program, without resorting to internal criteria. Apparently, the 43 words (4% of the 1009 words in question), required the NU internal probabilities to settle. Some of these were settled by common sense, not requiring any particular internal criteria like lectio brevior. Others were settled by some internal evidence. The rules for settling others were not recognizable. Even when some internal criterion was invoked to settle the case, the reason for picking the particular criterion over a variety of others was unpredictable. In general, the area of internal evidence is quite small, secondary and subjective.

The one result that was perfectly clear from the examination of the 1009 words was that the Alexandrian uncials won a landslide victory: 1006 words matched Alexandrian (99.7%) and 3 words did not; 891 words matched Vaticanus; 110 of the remaining matched Sinaiticus. Five remaining matched Alexandrinus, one remaining matched P46 and the last two were witnessed by Bezae alone. The landslide victory in favor of Alexandrian uncials could only result from the A-list, not from merely following the internal criteria. As we will see later, the internal evidence, if not overridden by the external vote, would often result in Byzantine victories. The Corinthians vote results are almost as lopsided as a Byzantine majority text victory based on a one-manuscript-one-vote rule. To summarize, we have a 96% predictability rate of the selected reading with singular elimination and a straight A-list vote. And we have a 99.7% predictability that the winner will be Alexandrian. The 99.7% Alexandrian rate is an independent fact while the 96% prediction rate for the particular manuscript depends on a hypothetical A-list method. A method that cranks out such results causes an Alexandrian NT but is not defensible.

A comparison of the TR deviations with the Vaticanus deviations may provide some perspective. Metzger spends several pages in the introduction to TCGNT (pp. 7-10) discussing the “blatant errors of the TR,” the debased and corrupt Byzantine text and the fact that Erasmus relied on “two rather inferior manuscripts.” Maurice Robinson, the developer of the Byzantine priority text, supplies some statistics comparing the Byzantine priority text with the TR and NU. First Corinthians has about 400 word-differences between NU (a kind of average Alexandrian edition) and the Byzantine priority text (a kind of average Byzantine edition). There are about 45 word differences between the TR and the Byzantine priority. There are about 133 word differences between Vaticanus and NU and 207 word differences between Sinaiticus and NU. Variances between two Byzantine editions do not directly compare to variances between Alexandrian manuscripts and their edition. However, it is unprofessional of the textual critics to mislead the public by persistently exaggerating the relatively small deviations of the TR (45) while concealing the comparatively large deviations of the main Alexandrian sources (133-207). This distortion has been going on since the days of WH and is repeated so often by the experts that people usually believe it. The Alexandrian proponents get better traction from rhetoric than from solid statistics. The Byzantine quality control was better than the Alexandrian quality control. Even a “poor” Byzantine manuscript was apt to be closer to its text center than the best Alexandrian manuscript was to its text center.

13. NU Seriously Misrepresents the Role of Age in the External Rules

Normally, when a student is asked to characterize the external rules, the response will be “older is more original,” or as Metzger stated: “earlier manuscripts are more likely to be free from errors.” Yet in 303 of the 1007 words (30%), P46, a papyrus about 100 years older than the Category I uncials, was different from the NU, and therefore rejected by the NU Committee. They rejected it consistently, even though it is older. They reject the oldest witness, P46, which is also the largest papyrus, because with only one vote, it could not outvote three Alexandrian uncials, and the internal rules were never invoked on behalf of the older P46 to override the mechanical vote. In only one case do Codex Bezae and the Category II uncials, supported by P46, trump the Category I uncials. The odds here are about 300-1 against the Committee selecting the older manuscript if it is not an A-list. The term that Aland uses to suggest that the papyri are not determinative is “free.” That means they are lower quality because their scribes do not stick to their exemplars very well. Since they have lots of corruptions, they are by definition not original and not reliable. In spite of what he says about older being better, Metzger knew that older papyri are not more original than later uncials in the majority of cases. But the NU experts are not forthright in informing the public of this fact. Instead they bury it under terminology like “free” without spelling out the consequences. TCGNT and the popular literature do not publicize the practice of eliminating the papyri due to its freedom. In fact, that is an understatement. TCGNT barrages the reader with internal reasons, as if they were determinative.

14. NU Seriously Misrepresents the Role of Text type in the External Rules

In an attempt to limit the appearance of a crude, sledgehammer application of Aland Rule 6 based on the category weights, the Aland rule contains the proviso that “There is no single manuscript or group of manuscripts that can be followed mechanically ... decisions in textual criticism must be worked out afresh, passage by passage.” It is hard to be delicate about how central this issue is or the extent to which the statement is untrue. Ninety percent of the 1 Corinthians text comes from Vaticanus, and 99.7% comes from three Alexandrian manuscripts, out of a pool of hundreds. Clearly Aland Rule 6 as practiced is not Aland Rule 6 as stated, which is honored in the breach even more than the rules of internal evidence. The fact is that NU does follow a small group of manuscripts rigidly, and primarily Vaticanus. After the A-list vote finishes its work, there is a little room for the internal rules to refine the variant selection.

The TCGNT reader who does not do statistics from Swanson will hardly have a clue about what causes results like those cited for 1 Corinthians. Every page of TCGNT gives internal evidence and lists of witnesses, as if those were the relevant factors. The reader will not know that the Byzantine or papyri variants backed by witness lists were not real candidates because they were categorically disqualified, or the vote or rank was stacked against them before the voting started. Aland hints at this by featuring an extensive discussion of categories in the early part of his book and relegating the 12 critical rules to the back, and he also posts the categories with each manuscript description. But he does not tell how the categories determine the vote mechanics.

It would be possible to state the opinion frankly that the Alexandrian text is generally better than the Byzantine text and therefore we are going to choose an Alexandrian text over the Byzantine text in virtually all variant units, regardless of the individual merits in the particular unit. This is what NU actually does, but they are afraid to come out and say it openly because the public would totally reject such an approach. It is like saying girls are better on the average at SAT English scores, so we are only going to let girls into college, except for a few brilliant boys. This type of class-based thinking is not acceptable. The fact that TCGNT regularly distracts the reader with discussions of the merits of the individual readings instead of stating frankly that the decisions are already mostly a foregone conclusion based on the text type is an indication that the NU scholars are aware of this unacceptability. And the straight misrepresentation of the actual text type basis for selection by Aland in his Rule 6, denouncing yet promoting elite favoritism, is an indictment of the whole Alexandrian enterprise. Some loyal Alexandrian fans will respond that maybe 99.7% is actually the result of a somewhat fair vote based on internal rules, which have not been discussed yet. The logical impossibility of this will be disposed of in Part IV when the internal rules are considered.

The critical rules have a long, rich history going back at least to Bengal in 1730. Textual criticism is a vast, complex and daunting labyrinth with thousands of books, scholarly studies and vocabulary. Yet to sort it out, NU believers can simplify it to a two-clause creed: Vaticanus is by far the best manuscript to recover the original, and Sinaiticus is its distant and main backup. Every other thought about the subject is supporting or secondary. You have to see the forest. This formula works for lay and expert NU believers alike.

15. The Papyri Ruined the External Claim That Older Is More Original

The discovery of the papyri after WH’s time ruined the claim that the Alexandrian text reflects the oldest witnesses. (Remember, the oldest witness, P46, was rejected 303 times, 30% of the time, in 1 Corinthians.) However, the Alexandrian ship is coasting on momentum (the memory of being supported by the oldest manuscripts), and people have not noticed that it ran out of steam (it is no longer actually supported by the oldest manuscript). Saying that the Alexandrian text is more original than the Byzantine because its manuscripts are relatively older but the papyri are not more original than the Alexandrian uncials, even though they are the oldest, falls flat somehow, even for the most gullible supporter. It is hard to finesse with nuances and appeal to authority.

The Byzantine text of 1 Corinthians 10:28 is: “But if any man say unto you, This is offered in sacrifice unto idols, eat not for his sake that shewed it, and for conscience sake: for the earth is the Lord’s, and the fulness thereof” (KJV). The NU lacks the second part of the verse, “for the earth is the Lord’s and the fulness thereof,” because the older Alexandrian uncials lack it. The even older P46 lacks it and also lacks “that shewed it, and for conscience sake.” But the NU does not follow the oldest manuscript P46 because the NU A-list limit trumps the rule in favor of the oldest manuscript: the papyrus does not qualify because it is free. External evidence has become straight text type, frequently bypassing age. The full reductionist statement of the external evidence rule is that NU follows the older manuscript if it is an Alexandrian uncial or Alexandrian P75; otherwise it follows the Alexandrian uncial. It is like saying that anyone might be president so long as he is a multi-billionaire former governor. It drastically narrows the pool. What the NU scholars are alleging is that the later Byzantine scribes added words to 1 Corinthians 10:28 and the earlier P46 scribe omitted words. This is one of several possibilities. It could well be true, but it is inconsistent, and the scholars have no way of knowing. It is a mechanical application of the rigged A-list program that most readers are not aware of and many would reject if they were. The NU scholars are telling us what they prefer, even if their secondary reasons (the internal evidence) are not consistent. However, that should not be confused with the right answer. They can guess an answer, but who can know it? Their experience is good, and their judgment has settled in grooves, but those do not compensate for lack of hard evidence. They have transferred their credibility from the collection and standard editing phase where it is deserved to the critical editing phase where it is not deserved. Note the vote rigging. The later Byzantine votes have been excluded because they would outvote the Alexandrian uncials. The papyri are let in because they are Alexandrian, or the later Alexandrian uncials outvote them. The text type qualifications, not age, determine the outcome.

16. Singular Corruption Rates Made the Critical Scholars Abandon the Age Criterion

So, the first obvious response to the papyri dilemma is, why do the NU scholars not just follow the WH lead and prefer the papyri now that they are the oldest? Well, that simple-sounding solution has a big drawback, besides the fact that the public might become agitated and suspicious at changing Bibles every time someone dug up something relevant in Egypt. The reason is that the older papyri were obviously more paraphrastic, condensed and error prone. The farther back the manuscripts go, the wilder they get. Aland indicates this fact by categorizing the majority of them as free. People wonder how a corruption can be proven rather than guessed at, given the disappearance of the autographs. Corruptions in general cannot be easily proven, but about half of them can be proven immediately once all the early variants are assembled: singular readings. Over half of the corruptions (rejected words) in 1 Corinthians were singular readings. And these were the most reliably identified corruptions, the set agreed on by both the Alexandrian and Byzantine priority schools. Singular readings fall into a category of presumed corruptions. Editors through the ages have systematically rejected unique readings as most suspicious. Since we have many manuscripts, even readings in only two or three manuscripts are usually not accepted by editors. In fact, most text critical rules have rejected them explicitly on principle. (See Aland Rule 7.) When Vaticanus has πα for πᾶσα in 1 Corinthians 15:39 or ἔρχον instead of ἔρχονται in 1 Corinthians 15:35, editors reject it without thinking, because they are nonsense fragments. When Vaticanus drops the “not” in 1 Corinthians 14:14, it spoils the meaning. NU editors unhesitatingly reject it because it is a singular reading (i.e., no scribes reproduced the mistake). NU readers do not know about these kinds of mistakes or corruptions because they often do not appear in the apparatus. The NU method of acknowledging the unacceptably high corruption rate in the papyri is to frankly categorize them as free and disqualify most of them as viable candidates. But the NU acknowledgment of corruption rates in the Alexandrian uncials is only tacit and partial.

So what is the big deal about finding more singular readings (corruptions) in the papyri than in the Alexandrian uncials? Well, here we have hit a vital artery, not just a surface wound. Basically, textual criticism is about corruptions more than anything else. Although singular readings are objective and other corruptions are decided by critical theories (usually the A-list vote for NU), singular readings are a major part of corruptions and an indication of them. Identifying corruptions is synonymous with recovering the original. They are two sides of the same coin. Why does singular rate predict corruption rate? Corruptions are mainly of the same kind as singulars on a continuum (adds, deletes, changes and transpositions)—only one vote away. If a singular variant finds one more vote, it marginally switches to a candidate which could become the edition. The bulk of the critical editing is done mechanically by text type vote, not by expertly applying internal rules. Generally among candidates, the top vote getter gets elected for the edition and the others become corruptions. For example, Vaticanus and Sinaiticus have μου for “my” in 1 Corinthians 9:2 while all of the other early manuscripts use ἐμῆς. NU chooses μου because Alexandrinus omits the phrase and the two A-type texts outrank P46 and Bezae. If Sinaiticus had happened to agree with P46, Vaticanus would become singular, and NU would have flipped against Vaticanus. Close votes generally look iffy. They look more iffy when we see that the votes were close only because the pool was artificially limited.

17. Singular Corruption Rates Should Cause Abandonment of the Alexandrian Text

The discovery of geometrically increasing corruption rates going back from Byzantine to Alexandrian uncials and then farther back to the papyri is a two-ended spear. We could use it to insist that the papyri should take precedence over the Alexandrian uncials because of their age, because in the Alexandrian/Byzantine contest the age counted more than the corruption rate. However I choose to use the corruption rate argument in the other direction. I agree completely with the NU scholars that the papyri corruption rate is so significant that it trumps the age criterion. Older and worse is not more original. The Alexandrian uncials exhibit more careful copy work than the relatively uncontrolled papyri, so the older, sloppier papyri deserve to be rejected as not original, fascinating as they are. Aland characterizes even P75, the crown jewel of the papyri, as “strict (although with certain characteristic liberties).” “Strict” is a relative term, and liberties are found in it that are totally unacceptable to modern standards, or even to ancient Old Testament copy standards. The scholarly community is in wide agreement on this point of rejecting most papyri as viable candidates. And those who agree with the NU edition tacitly agree to downgrade the significance of age.

So now the submerged implication from the papyri is visible. The Alexandrian copyists were twice as careless and paraphrastic and full of singular readings and divergent readings as the later Byzantines, just as the papyri show twice as many suspicious variants as the later Alexandrian uncials. Since the NU scholars reject the older papyri on the basis of corruptions, if they were consistent, they should reject the older Alexandrian uncials in favor of the more error-free Byzantine manuscripts for the same reason. Otherwise, it looks like a double standard: older is better if the Alexandrian uncial is older but otherwise not. In postmodern times of unaccountable choice, rational consistency is not as critical for persuasion as it once was, but it still counts for something.

18. History of Text Types and the Indispensable Recension Legend

At the surface, on the popular level, external criteria are about age. But that is a misleading picture because underneath are text types (Alexandrian, Western, Byzantine). But there is a core layer under the regular text types: the A-list, a subset of Alexandrian manuscripts including the Alexandrian Category I uncials and P75. The main function of the A-list vote is to elect Alexandrian Vaticanus and reject the far more numerous competitor types. Although the workings of the A-list are obscured, the prominence of Vaticanus is evident, and people need reasons for its prominence. Why is this little group better than all its predecessors, contemporaries and successors? The reasons given are subjective and inconclusive, like quality and age, but the indispensable reason is text history. And that is imaginary.

Textual criticism is awash with debates and assumptions about text types and their history, as if they were realities rather than convenient generalities. Objectors think text types are just a partial-fit abstraction of the fact that some manuscripts are more like others in some ways, with lots of muddy crossovers. Obviously they are not like biological species with scientific, operational distinctions. Text types are idealized forms of the NU categories. The idea that the few and widely divergent Alexandrian manuscripts represent a text type rests not on the few extant survivors themselves but on the assumption that there was probably a widespread set of similar manuscripts in the early days which stemmed from an ancestral prototype. The theory of text types is the necessary foundation for NU and any Alexandrian edition. The Byzantine majority methods do not depend on text types.

Indo-European scholars believe that Lithuanian is closer to the original Indo-European proto-language than Sanskrit is, even though the Sanskrit documents are older, because Sanskrit deviated from the proto-language earlier and faster than Lithuanian. In the same way, the age of the Alexandrian documents does not prove that they are closer to the original. Likewise, Lithuanian is closer to the original Indo-European than the more widespread English is. In the same way, the numerical majority of the Byzantine documents does not prove that they are closer to the original. However differently they may have evolved from the original, neither Lithuanian, Sanskrit nor English derived from each other. Nor are any of their sentences the same as the original. NT editions are patchworks, cleaned-up blends of subsets of early manuscripts, but probably none of them is identical to the original.

Was the Byzantine derived from the Alexandrian, as WH and the Alexandrian-priority scholars claim, or an independent strain going parallel back to the original, as the Byzantine-priority scholars claim? Why did the Byzantine text dominate in the autograph heartland? Why did the church choose to let the Alexandrian text die out? Why did the correctors of the papyri and Alexandrian uncials correct them to be more Byzantine? Without a powerful deflecting influence, the most populous manuscripts in the second century Byzantine homeland of the majority of the autographs would naturally become the most populous in the fifth century. The Alexandrian prioritists are under a heavy burden to justify their unnatural hypothesis. WH felt this pressure, as do most modern scholars. To answer this question, WH promoted a genealogical text type story called the Byzantine recension (sometimes called the Antioch recension) about the origin of the Byzantine family. According to the story, about the time that Constantine made Christianity a state religion and elevated the Byzantine bishops to power, someone edited an official version of the NT, and the bishops promoted it and it supplanted the original copies. (The name Lucian is often associated with the recension, because there were several of them and one of them is said to have edited a copy of the LXX.) But some of the precious near-originals survived, buried in remote corners in Egypt, which modern scholars finally and heroically traced and painstakingly resuscitated to restore the autographs. The story can be found in Aland’s book, the TCGNT introduction, Philip Comfort’s book, WH’s works—virtually everywhere in Alexandrian textual criticism.

The first problem with this Rousseauian story of oppression and ultimate victory of the downtrodden victims is that there is no mention of it in history. It was invented fifteen centuries after the facts. (The widespread protracted vocal resistance to Jerome’s Vulgate shows what really happens under such circumstances.) The second problem is that the various unresolved Byzantine branches demonstrate that even if centralized uniformity had been attempted, it was never successfully imposed. The third problem is that it was so technically improbable in that age to suppress all those monumentally expensive Bibles that many of the modern scholars have modified the story.

Past scholars, like Aland and Metzger, and some present ones, like Comfort, still espouse the Lucian recension, while others have toned it down to a process rather than an event. These variations are equally unproven. Pro-Alexandrians are obliged to answer the challenge but content themselves with stories about possibilities. It is a weakness for them. Many pastors and Bible readers have grown accustomed to a faith-based formula to guide them through the text-critical maze: scientific NU and pre-scientific Byzantine TR. A goal of this article is to disclose a paradigm shift of skepticism among newer scholars: manuscript collection is scientific but text selection for Greek editions is still a web of pre-scientific speculation. Without the missing critical first- and second-century manuscripts in the heartland from Greece to Antioch, the subject will remain pre-scientific permanently. Real science needs real evidence.

To summarize the external evidence, the heart of the Alexandrian method is the A-list. There are only two main external supports for the A-list: age and text history. We have seen that NU itself has abandoned the age advantage, and the text history is fictional. From the days of WH to the TCGNT, Alexandrians have been fooling the gullible with the same debating trick. They cannot reduce the Byzantine hoards to one late ancestor by actual historical evidence. So they present a story as a plausible, genealogical, conditional assertion, hoping that acceptance will be as automatic as belief in Darwinism. Here under the heading “External Evidence” is Metzger’s pseudo-algebraic version: “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.”[13] The conclusion is fictional because the condition is fictional. The “if” is not true because the “discovered” is certainly not true. Discovery is precisely the difference between science and fiction. This type of desperate statement should not be written. As we will see, the internal support for the A-list is no better than the external. The A-list has no objective justification. It is just a complex, authority-laden wrapping for the Alexandrian Vaticanus preference.

19. When and Where Did the Alexandrian Text Type Flourish?

The main period of known widespread prosperity for the Alexandrian text was the twentieth century. Before that it appeared mainly in the third and fourth century, in Egypt, judging by its three large strict manuscripts. In that time and place it competed with other text types. After the fourth century, when Constantine issued the edict of toleration, more Byzantine-type corrections began to appear in Vaticanus and Sinaiticus. The Old Italic versions came from a Western text, and the Vulgate came from a mixture of source types later in the fourth century. By the fifth century, the Alexandrian uncials like Alexandrinus and Ephemi Syri Rescriptus began to have a mixture of Byzantine and Alexandrian. Aland classifies them as part Category II and part V. Fifth century Codex Washingtonian has blocks of each type. The Alexandrian text was gradually sidelined over the next few centuries and mostly abandoned in favor of the Byzantine text, or, as Aland puts it “as the years passed even this text showed the corrosive effects of the Koine influence” (“Koine” being a synonym for “Byzantine”).[14]

The third-century P75 manuscript of Luke and John was the only large strictly Alexandrian papyrus, although other third-century papyri like P66 and P72 are predominantly Alexandrian. Before the fourth century the papyri are classified as Category I because they were early (except P48 is Category IV). But text types per se had perhaps not fully developed. Aland states: “The major text types trace their beginnings to the Diocletianic persecution and the Age of Constantine which followed.”[15] The papyri are full of singular readings and type crossovers. P45 and P46 are closer to Alexandrian than Byzantine, but not exactly either one. So the known prime Alexandrian manuscripts come from third- and fourth-century Egypt. After that the Alexandrian type starts to fade out until it was revived by scholars in the nineteenth century. The length of time and geographic extent that the “pure” Alexandrian type manuscripts were in use beyond third- and fourth-century Egypt is a matter of speculation. Coptic texts continued to be Alexandrian. Vaticanus and Sinaiticus have such similar scripts that Tischendorf thought they were by the same scribe. The scroll work on the colophons is of the same type, so they may have come from the same scriptorium. The fact that their texts still only match up 70% suggests that a long-lasting stable Alexandrian text type might not have existed before modern times.

Part IV. Internal Editing: The NU Secondary Method (Stage 4)

20. The Internal Probabilities or Criteria, Stage 4

The second and minor part of NU critical editing is internal. Students of textual criticism are generally more familiar with internal criteria than external. Metzger divides the internal criteria for selecting variants into transmissional probabilities and intrinsic probabilities.[16] Three of Aland’s 12 rules are transmissional. They involve confuted or disputed Alexandrian theories about how scribes transmitted the text and corrupted it:

Prefer the shorter reading (lectio brevior).
Prefer the more difficult reading (lectio difficilior).
Prefer the reading that explains the others in terms of scribal habits (derivation).

The intrinsic probabilities are open-ended criteria, involving what the critic thinks the author was more likely to have written. They include such things as style, context, harmony and the influence of the Christian community. Basically they are a stretchy set of exegetical considerations with no exhaustive limits. TCGNT readers encounter a large variety of them and are often amazed at the richness of ideas.

We have seen that the external rules make most of the decisions without resort to internal evidence and operate mechanically in a secret way that readers often do not know. What about the internal rules? They are much more messy and subjective. The internal rules have four problems, as applied to variant selection.

They are often circular or speculative.
They are often inherently invalid, invalidated by empirical discoveries since WH.
There are many criteria or types of internal evidence that often conflict with each other.
Sometimes TCGNT selects one criterion and sometimes another. NU has no principles to arbitrate among them or assign weights. Rule selection is arbitrary and unpredictable on the surface, but the A-list vote leads predictably to the Alexandrian selection, i.e., the internal reasons are a cover for the text type reasons.

Selecting variants phrase by phrase, as in TCGNT, is the public job of the internal criteria. Their first job, however, is to support the a priori text types that establish the voting structure that precedes the variant selection. In their first job the internal criteria encounter the same problems that they do in their public job.

21. Internal Evidence Is More Inconsistent and Circular Than External Evidence

In Greek, direct discourse is often determined by retention of person and, unlike English, may optionally be governed by the conjunction ὅτι, that. For example, the Byzantine manuscripts for John 18:6 say literally: “He said to them that I am,” though in English we say, “He said to them, ‘I am.’ ” The Byzantine manuscripts for John 18:6 have the optional ὅτι and the Alexandrian ones lack it, as in English. In John 9:11 the case is exactly reversed, though in either case NU chooses the Alexandrian. This kind of inconsistent application of the internal rules, although common, does not usually bother most pro-Alexandrians because there are always trade-offs, and the decision process is considered an art, not a science, anyway. (The science part of textual criticism is confined mainly to collection and standardization, not critical rules.) Regularly, the various internal criteria do not always favor the same source, so usually if one kind of evidence does not support the Alexandrian manuscripts, another one will. And since textual criticism is very complicated, the main thing is that the experts are satisfied with their choices. Still, the news that the NU scholars have quietly abandoned the rule of preferring the older manuscripts on a large scale may bring a twinge of disapproval or even disbelief on the part of pro-Alexandrians. The reason for this uneasiness is that the external age evidence has always been the only substantial non-circular foundation for the Alexandrian case in the lay mind. It was the non-negotiable reason for the WH victory in the nineteenth century.

Why are the internal arguments unable to sustain the Alexandrian hypothesis without the age rule? A shorter reading is not inherent evidence for originality. Lectio brevior is not a self-evident axiom, because obviously editors can cut out words as easily as add. NU scholars have not produced empirical evidence to demonstrate that Koine scribes had a habit of lengthening their exemplars. In fact, when the actual studies were done on the Alexandrian and papyri scribes, they supported the opposite conclusion.[17] The NU’s own decisions demonstrate that all the Alexandrian manuscripts engage in systematic shortening. Nevertheless, the scholars just assumed the rule: since Byzantine manuscripts were longer, they must have been lengthened. Saying that readings are older because they are shorter is like saying Frenchmen are smarter because they speak French. It is circular logic. It is a rationalization. Correlation does not demonstrate causation. The alleged reason is merely an oblique restatement of the hypothesis that requires proving. Shorter readings and speaking French are characteristics, not proofs. The “reason” is no more proven than the “conclusion.” Similar circular observations apply to the other internal criteria. But the age factor falls in a more objective class. Even though the oldest was not necessarily the closest to the original, there is an inherent probability in its favor. In WH’s day the assertion that the Alexandrian edition uses the oldest known manuscripts was the indispensable clincher. But back in their day the statement was still true, though it no longer is.

22. The Inconsistent Application of the Internal Evidence

We have seen statistically that the internal criteria such as lectio difficilior have been relegated to the background by the A-list in the NU selection of individual passages. TCGNT discloses secondary justifications for some of the NU Committee preferences. It is a classic study in inconsistency and rationalization, fairly amusing in places.

Where the Byzantine manuscripts have a word the Alexandrians lack, the Byzantine scribes are charged by the NU Committee with “inserting it” (Matt. 1:25) or “making a scribal assimilation to the LXX” (Matt. 2:18) or “softening the rigor of the precept” (Matt. 5:22) or making “an obvious expansion designed to heighten the impressiveness of the saying” (Matt. 6:4) or “supporting the perpetual virginity of Mary” (Acts 1:14) or “obviously a secondary development, probably connected with the beginning of an ecclesiastical lection” (Acts 3:11) or “deriving it from a list of vices” (1 Cor. 3:3). But conversely, when the Alexandrian manuscripts have a word that the Byzantine lack, the Byzantine scribes are charged with “homoeoteleuton” (1 John 2:23) or “deliberate editorial pruning of an awkward parenthetical clause” (1 John 2:23) or “omitting because the idea was theologically unacceptable” (1 Pet. 2:2) or “deliberate excision ... palaeographical oversight” (1 Cor. 7:34) or a “transcriptional blunder” (Luke 9:59). Mercy! Sometimes a poor Byzantine just can’t win, no matter whether he is shorter or longer. NU imagination has discovered an incredible variety of corruptions the Byzantine scribes committed, and even their motives. For example, when the Byzantine scribes have “God” instead of the Alexandrian “Lord” (Acts 15:40), they are guilty of “scribal assimilation.” A little later, when the Byzantine scribes have “Lord” instead of the Alexandrian “God” (Acts 16:32), they are guilty of “scribal refinement.” Is it the Byzantine scribes or modern experts who are guilty of refinement?

In individual passages the experts seem to have trenchant-sounding reasons. But in looking at the whole picture, a systematic subjectivity emerges. In Acts 20:32 the Byzantine includes “brethren” where the Alexandrian excludes it. In 1 Corinthians 15:31 the Alexandrian includes “brethren” where Byzantine excludes it. The internal evidence (lectio brevior) favors the Byzantine but the Committee chooses the Alexandrian because of the “strong external support for inclusion.” Why is the external evidence for the Alexandrian reading called strong when the oldest manuscript, P45, has the Byzantine reading? In Luke 10:21 NU chooses the Alexandrian “Holy Spirit” over the Byzantine “Spirit” without the word “Holy.” In Acts 8:18 NU chooses the Alexandrian “Spirit” without the adjective “Holy,” over the Byzantine “Holy Spirit,” in spite of the fact that the earlier papyri support the Byzantine. The alleged reason is that, in the Committee view, “the addition of τὸ ἅγιον was as natural for Christian scribes to make as its deletion would be inexplicable.” One time the Committee thinks the unreliable Byzantine scribes omitted it and the other time the Committee thinks it would be inexplicable for the trustworthy Alexandrian scribes to omit it. Who can argue with enthusiasm and confidence in the home team? It is an endearing quality in sports, but does it help get the Bible right?

A number of the mirror-opposite pairs have been listed in which the roles of the Alexandrian-Byzantine units are reversed but NU chooses Alexandrian either way. Mirror-opposite pairs do not just show the invalidity of some internal rule or other. To the extent that they are true mirror opposites, they show the invalidity of every conceivable internal evidence to achieve the NU result. To put it another way, no possible internal rules could ever achieve the NU edition. It is a point of logic. This observation is important, since the supply of internal evidence is open-ended. This pattern results from the eclipse of the internal criteria by the external A-list.

23. The Venerable TCGNT

The reader may feel it unfair to accuse the venerable TCGNT of rationalizations. However, the characterization is sometimes accurate. A rationalization is a reason that is given, but it is not the real reason. Consider a typical case, John 2:24, where NU chooses αὐτόν from Vaticanus and Sinaiticus over P75 omission and P66 ἐαυτόν. The Committee did not follow lectio brevior or either of the oldest manuscripts or lectio difficilior. They took a straight vote based on the A-list vote. The shortest reading, the omission in P75, was immediately disqualified as a singular reading. P66 was disqualified as a free papyrus. The remaining two candidates, Vaticanus and Sinaiticus, are in agreement. Based on the mechanical procedures, without resort to the secondary internal evidence, there was no question. Metzger’s list of witnesses for the potential Byzantine reading are just Byzantine or free. Why does he present this list as if they were real contenders?

He claims to be rejecting ἐαυτόν because these witnesses “clarify the sense.” This is supposed to follow the internal rule about preferring the reading that explains the other readings, which NU usually follows only if it favors the Alexandrian text. But that is not actually why he rejects it. When the Alexandrian text can be characterized as “clarifying the sense,” he does not reject it (consider mirror-opposite John 17:13, where NU chooses the longer Alexandrian ἐαυτοῖς over Byzantine αὐτοῖς even though αὐτοῖς is older and shorter—no word about clarification there). The real reason NU rejects ἐαυτόν is because the A-list Vaticanus and Sinaiticus manuscripts stand united against Byzantine and free manuscripts. NU normally chooses the Alexandrian A-list mechanically under those circumstances, regardless of the internal evidence. The real reason is the reason that operates consistently. The rationalization is the reason that operates when it supports the rationalizer’s bias. Bias is the a priori category that does not need individual reasoning. Rationalizations abound in the Commentary. One or more will likely be found in any random opening of the TCGNT.

In spite of the fact that the standard and external edit phases predict the NU decision most of the time, it must be observed that on occasion the NU editors promote some kind of internal evidence over the mechanical vote. In 1 Corinthians 15:49 the Committee bases its decision on “exegetical considerations” of what is “consonant with the apostle’s argument.” In Romans 5:1 the Committee completely abandoned the external evidence and the A-list and picked the Byzantine variant that most people think makes more sense. The case involves a choice between ἔχομεν and ἔχωμεν, where the question hangs on the difference between the indicative omicron and the subjunctive omega, which were pronounced alike, or enough alike, that there are at least a hundred readings where the two are confused. But in justifying their decision, Metzger displays a sense of humor. He claims that Paul’s amanuensis misunderstood his meaning and wrote omega when he should have written omicron. So the change-prone Byzantine scribes changed the incorrect autographic omega back to Paul’s true meaning, while faithful Alexandrian scribes accurately preserved the incorrect autographic omega. But, like the Byzantine scribes, the NU is correcting the autographic corruption of the amanuensis. In other words, they are making a good corruption. Amazing! Read it in TCGNT for yourself. The Committee has moved the goal post. Instead of recovering the original word, now they aim to recover the original meaning. Where does that lead, I wonder?

Abandoning the A-list in favor of a Byzantine reading is not common. But since this kind of exegetical override could apply to hundreds of variants, its use is arbitrary. For example, in Acts 9:25 TCGNT comes to a similar conclusion, that the original pronoun was corrupted from accusative to genitive before the earliest surviving manuscripts, even though the Byzantine has the accusative. But this time they decide to remain faithful to the external Alexandrian evidence. They retain the known corruption in the NU. Basically, they advise translators to ignore their own NU edition genitive and translate the correct case, which the translators do (see NIV, NASB, etc.). The lengths NU goes to avoid admitting that the Byzantine text is original are Byzantine.

24. Definitions of Omission and Why Modern Bibles are Shorter

Definitions of omission depend on the comparison base. There are three common definitions of omission. First and worst, an omission from the TR perspective is any word that occurs in the TR but not in the NU or other editions or manuscripts. This is the most extensive definition, because TR is long and NU is short. The Western text is even longer than the Byzantine, but definitions of omission are not used from that base. Alexandrian prioritists consider the TR definition self-serving, and it is not used in this article. The second definition of omission is: whatever is not in the NU. At first this sounds equally self-serving. However, in practice the definition is usually acceptable to all, because most of what is omitted in NU is also omitted in TR, since the chief feature of Alexandrian manuscripts is abbreviation. For example, Vaticanus omits the phrase “I brought him down to their sanhedrin” (Acts 23:28), but the editors of TR and NU both keep the phrase anyway, so they both call it an omission. This definition applies to manuscripts (primarily Alexandrian) rather than to editions. The reason that it frequently applies to Alexandrian manuscripts is that they are not only shorter, but wilder: different Alexandrian manuscripts shorten in different places. When this happens, it is evident that at least some of them shortened the original, because they contradict each other’s witness (cf. Mark 14:56). (Of course, omission is not the only Alexandrian method of shortening.) Therefore, an NU omission of a word from a manuscript is generally considered an omission of the word from the original autograph.

The third definition of omissions is singular omission. At first this definition might also appear arbitrary, but in practice it is even more universal than the second one. Virtually all singular omissions are rejected by both schools. So singular omissions are a subset of NU omissions. This is the most immediate and objective definition of omission. Within singular omissions, the definition may be refined to exclude manuscripts after a certain date (say, sixth century). The reason for this restriction is to control the populations so that the Byzantine texts will not have an undue matching advantage on account of the larger pool. This practice has been followed in the statistics for this article. This restriction does not greatly alter the results. Lectio brevior is always violated for singular omissions.

A fourth and impractical definition of omissions would be words that were in the autograph but that more than one Alexandrian scribe or a common proto-Alexandrian scribe omitted. This set of possible omissions, however large, remains speculative. Metzger toys with this idea in John 6:17 where NU puts “Jesus” in square brackets (undecided) and calls it a possible Alexandrian deletion because Vaticanus, Sinaiticus and P75 omit it. Since there are many Alexandrian singular omissions, the difference between this type and the singular definition may often only be due to chance alignment. For example, in Matthew 6:15 all the early manuscripts have the Byzantine reading “but if you do not forgive men their trespasses,” but Vaticanus and Bezae lack “their trespasses,” so NU rejects it as not being original. Likewise all early manuscripts have the Byzantine reading “seek ye first the kingdom of God and his righteousness” (Matt. 6:33), except Vaticanus and Sinaiticus, which omit “of God.” This time NU puts “of God” in square brackets, so many of the versions omit it. The difference involves the stylistic use of ellipsis rather than theology or grammar. But Vaticanus is corrupt here because it has a singular transposition of “kingdom” and “righteousness.” NU’s decision results from only two witnesses which do not agree, one of which is corrupt in that spot. It is therefore neither a singular omission nor an NU omission, although ordinary non-experts would tend to classify it as an Alexandrian omission anyway.

In connection with possible omissions, the main internal criteria are closely related variations of lectio brevior, which favor the shortened Alexandrian manuscripts. Lectio difficilior follows naturally from omissions, because when things drop out it often makes the reading harder to understand. The rule about which reading explains the development of the others also favors the shorter text, given the Alexandrian priority’s unproven assumption that early scribes added more than dropped. This rule is especially convincing to pro-Alexandrians when the two variants have two different words and a third variant has nothing. It seems more probable to Alexandrian critics that two scribes added two different words than that one scribe added one word and another changed the same word. Non-Alexandrian critics, observing that Alexandrian scribes were more prone to drop or change than add, see the probabilities otherwise.

The fact that Alexandrian scribes were habitual shorteners (by NU’s own definition of rejection) is not public information available in scholarly pro-Alexandrian literature. It is not even readily or consistently available in the NU apparatus. It is, however, conspicuously available to word counters who read the Swanson format. Alexandrian omissions are the most common type of Alexandrian corruption (as defined by NU), about as common as all the types of changes together.

25. Why Lectio Brevior Is Invalid but Its Influence Endures Anyway

Philip Comfort, a contemporary pro-Alexandrian critical scholar, gives the reason that the lectio brevior rule cannot always be trusted and a papyrus like P45 cannot be admitted to the A-list, even though it is older than the Alexandrian uncials:

According to a study done by Colwell, the scribe of P45 worked “without any intention of exactly reproducing his source.”[18] ... While copying phrases and clauses, he worked at reproducing what he imagined to be the thought of each phrase. Thus he transposed and omitted many words and deleted several phrases. Colwell said, “The most striking aspect of his style is its conciseness. The dispensable word is dispensed with. He omits adverbs, adjectives, nouns, participles, verbs, personal pronouns—without any compensating habit of addition.”[19] Another study of P45 done by Royse affirms Colwell’s observation about the scribe’s penchant for brevity.... “The scribe has a marked tendency to omit portions of text, often (as it seems) accidentally but perhaps also by deliberately pruning.”[20]

Comfort and Barrett describe the papyrus as “an abbreviated yet readable rendition.”[21] Obviously, if third-century scribes could have done this to an Alexandrian papyrus exemplar, the earlier Alexandrian scribes could have done it to their exemplars.

Yet the internal rules—particularly lectio brevior, even if compromised—are critical in understanding the visceral formative phase of the Alexandrian/Byzantine conflict. It operates effectively at the deep, unchangeable text type level rather than at the variable internal-evidence surface level. Back then, before the papyri were discovered, WH used the rule to help eliminate the Byzantine majority. WH were trained in classical Attic Greek, before the days of Koine studies and volumes of Koine papyri. In the classical world, the editor cuts out words. The shorter, leaner, more elliptical expression has more punch. WH’s followers identify with the legendary Alexandrian editors, the literary elite of the ancient world at the empire’s premier university, library and book-copying center. WH and their followers believed that the clumsy Byzantine scribes expanded the sparser original. But since Koine is now known to be a more padded, verbose language than classical Greek, if the Byzantine was original and if an edit-minded Alexandrian scribe wanted to digest it, the Alexandrian manuscripts would be the result. For example, the NU editors judged that the Vaticanus dropped 34 words in 1 Corinthians and Sinaiticus dropped 87 words, although not usually the same words. The Alexandrian scribes had the habit of condensing their exemplars. That’s why the modern Bibles are shorter. NU scholars admit this when comparing Alexandrian manuscripts with each other, but cannot afford to admit it when comparing them with Byzantine manuscripts because “older and shorter” is the basic identity of the Alexandrian priority. Bias is a deep category, no matter how honest.

26. When Lectio Brevior Got Longer: Western Non-Interpolation

A problem for the Alexandrian prioritists is that the NU edition is based on the chance whims of archaeological discoveries, subject to periodic shifts. Only a few years ago scientists discovered how to to apply a machine to read a good-sized Oxyrhynchus papyrus that had been around for 100 years but was illegible. Scholars enthuse over the possibility that it may tip the balance for some readings and cause a change in NU. In the 1930s, when the Chester Beatty P45 and P46 were published, hundreds of mostly minor Byzantine readings that experts had previously classified as late (with suitable critical rationalizations to explain them) were now found in papyri that were earlier than Vaticanus. It was awkward convincing Greek NT readers that these oldest readings were coincidental parallel corruptions, but the public relations job apparently succeeded. The discovery of the papyri challenged the Alexandrian priority because they could not reject the papyri in favor of the Alexandrian and yet maintain that the older papyri were more original than the uncials. However, when the Bodmer papyri were published, NU scholars encountered a problem that could not be hidden from English Bible readers.

The problem had been invented by WH, who had given it the bone-crushing name “Western non-interpolation.” It was the fruit of the lectio brevior tree. WH conceived the idea that because the Western text was so paraphrastic and padded, where it omitted words, those words must have been omitted in the original, even if they were amply attested in the Alexandrian texts. Scholars think like that. In other words, a Western omission trumped all. Of course, the same logic did not apply to the Byzantine omissions.

Whole verses were pruned from the Greek editions after 1881, especially from Luke. Dutifully, the WH and NU translations like ASV followed suit, which agitated the public. But when P75, the crown jewel of the papyri, was discovered to contain the western non-interpolations, the lectio brevior line was stretched to the snapping point, and the lost verses had to be restored in the next round of translations, including NIV, after the next NA edition. This kind of up-and-down ride on the lectio brevior roller coaster made the public suspect that the experts were engaging in guesswork based on the mere accident of discovery. Lectio brevior never was a valid general principle, just a principal characteristic of Alexandrian manuscripts. In reality, corruptions could be shorter or longer. We know from singular readings and from NU’s own results that Alexandrian scribes were inveterate shorteners. Lectio brevior is woven deep into the very fabric of the Alexandrian priority. Lectio difficilior is primarily just a corollary. Try dropping out a bunch of words from something and see if it is not more difficult. The sand is washed out of lectio brevior, and the day of reckoning is coming.

In Acts 2:17 Vaticanus shortened four words (“in the last days”) down to two words (“after these things”). NU scholars indicate their conclusion that it was a corruption by rejecting it. However, all schools knew that anyway, because the reading is singular. In Acts 2:44 the Vaticanus shortened “All the believers were in the same place and had all things in common” to “all the believers had all things in common in the same place.” Although it means about the same thing, Vaticanus is a singular reading; so NU rejected it. In Acts 11:13 the Vaticanus scribe found a way to shorten send (ἀπόστειλον) from ten letters to six (πέμψον). Acts 13:42 reads, “When they (Paul and Barnabas) came out, they (the people) begged them to speak to them on the following sabbath.” Vaticanus shortens it to “When they came out on the following sabbath, they came out to speak to them.” The meaning is different. In Acts 23:26 Vaticanus left off six words, and in the book of Acts it left out over 60 words, that NU restored. All these were singular omissions and hence corruptions, showing that shortening was a regular habit for the Alexandrian leader, and even more for the other scribes. When Vaticanus and Sinaiticus both lack a word, the NU edition frequently lacks it. Since they both derive from a common source, whatever words their common source dropped are missing from the NU edition. The Alexandrian priority principle that scribes padded the more concise expression to make it longer flies in the face of editor experience and Alexandrian deletion facts. Real editing typically shortens. Alexandrian prioritists are stuck. They have to believe in lectio brevior no matter what, because their text type is substantially shorter throughout. An attack against lectio brevior is a torpedo at the Alexandrian priority itself. The thing that has kept the wishful thinking afloat all these years is public gullibility toward expertise, and lack of information.

27. The Lure of Style

In John 10:22 the NU editors followed WH and chose τότε over δέ against the majority of the early witnesses. What internal criterion drew the Committee there? Style. TCGNT states: “After considerable debate a majority of the Committee preferred τότε as ‘too appropriate not to have been included originally.’ ” Does the Committee have special access to the appropriate original word of God? It turns out that the Committee was not familiar with Greek NT usage. The adverb τότε occurs over 150 times in the NT, always at the beginning of a clause or following a time expression or a clause conjunction. Τότε is never used otherwise as a postpositive (second position), but δέ is used over 2000 times as a postpositive. No clause starts with a finite verb followed by τότε, but a finite verb followed by δέ occurs over 450 times. Usage is an objective component of style. A drastic usage probability of 150-0 odds against a proposed usage may be overridden by clear external evidence, but not based on a small minority.

The idea might have been appropriate, but the word position shows evidence of tampering. The Committee apparently goofed. Maybe it sounded like good Greek in Egyptian or Attic usage, but John wrote simple Koine. It is generally clear that style is a shaky reason for deciding what the Bible said. De gustibus non disputandum is ancient wisdom. The Committee was pretty sure of its ability to assess appropriate style to find the Word of God, but others should be more skeptical. Basing choice on text type is not good, but it is better than style. Readers may console themselves with the fact that, although τότε was missing from the Vulgate and most Bibles for over 1500 years, its appearance in 1881 probably did not produce enough difference to be noticeable.

The question now is not why a particular passage is preferred, but why the Alexandrian manuscripts as a whole are elected to the decisive A-list. Many scholars prefer the Alexandrian style for its relative elegance, independent of the issue of originality. For example, in 1 Corinthians 10:28 (quoted above) the clause that the Byzantines have (Metzger calls it a “gloss”), and the Alexandrians do not have, may be considered redundant because it is a repetition of the same phrase in 10:26. Modern editors or early copyist-editors may have judged it to be dispensable. It is often easy to sympathize with WH’s preference for the trimmer reading. Normally the older, shorter reading takes precedence in NU. But if the Vaticanus is longer and the Committee likes its style, exceptions can be made. In Revelation 13:10 (“He that leadeth into captivity shall go into captivity”) the Committee violates the external evidence, lectio brevior and lectio difficilior principles, and even its aversion to TR, to side with a bewildering alignment of Vaticanus and TR against the papyrus P47, Sinaiticus and the Byzantine majority, because of a preference for “the epigrammatic style” of Vaticanus. This decision required the Committee to assume parallel “omissions” in earlier papyri, in contemporary Alexandrian and in later Byzantine manuscripts. To justify this alignment of coincidences, the Committee speculated that it must have resulted from “accidental oversight in transcription.”

It is almost comical to watch the Committee struggle when the Byzantines are more elegant and the Alexandrian manuscripts are more pedestrian. In Acts 23:30 the Byzantines have an intriguing articular preposition (“the to him”), but the papyri and Category I Alexandrian manuscripts lack this article. What will the Committee voters do? They permit themselves a small indulgence. They find one sixth-century Category II (half Alexandrian, half Byzantine) support for the article. They put the article in the edition, but in brackets, indicating that it is a hard call. Why is the longer reading with no early witnesses a hard call? Because they are drawn to the style—the very aspect that fired up WH for the Alexandrian text. So from the time of WH and before it was probably the style, as much as anything, that landed the Alexandrian uncials in in the winning A-list. Even though Vaticanus and Sinaiticus have a higher rate of obvious corruptions (singulars and mistakes) than the Byzantine manuscripts, which prejudiced so many TR lovers against them, Tischendorf and WH were impressed with the overall superiority of the Alexandrian style. It is, however, not evident that the NT authors shared this preference because Koine itself does not share it, and even the Alexandrian manuscripts contain many non-classic, pleonastic constructions that NU judged original.

28. The Bidirectionality of Style: A Case Study

In order to get a hands-on feel for the invalidity of lectio brevior as an independent proof of originality, consider a typical passage, which is longer in the Byzantine than the Alexandrian but means the same thing:

And all the devils besought him, saying, Send us into the swine, that we may enter into them. And forthwith Jesus gave them leave. And the unclean spirits went out, and entered into the swine: and the herd ran violently down a steep place into the sea, (they were about two thousand;) and were choked in the sea. (Mark 5:12-13, KJV)

NU, following the Alexandrian manuscripts, lacks the italicized words. (The translation replaces them with appropriate pronouns: “they” and “he.”) The longer Byzantine text is what an expansionary Byzantine editor could do to the Alexandrian text if it were original. Conversely, the shorter Alexandrian text is what a snipping Alexandrian editor could do to the Byzantine if it were original. The Committee often bypasses lectio brevior in favor of one Alexandrian uncial over another, but not in favor of a Byzantine reading. As Metzger explains, when Vaticanus lacks a word that the Committee considers original, “the Alexandrian text, with its usual tendency toward parsimoniousness, has eliminated ...” (Acts 3:22). “Parsimoniousness” is Metzger’s expression for the Alexandrian scribes’ habit of cutting nonessential words out of the original, according to NU. There is a grammatical hint in Mark 5:12 that the Alexandrian text might have been incompletely derived from the Byzantine by cutting. In the Byzantine “saying” (λέγοντες) is masculine, agreeing with the unusual masculine “demons” (δαίμονες, as in Matt. 8:31), but “going out” (ἐξελθόντα) switches to neuter, agreeing with “unclean spirits” (τὰ πνεύματα τὰ ἀκάθαρτα). If the Alexandrian editor cut out “demons,” he should have changed “saying” to neuter (λέγοντα) to agree with “unclean spirits” (see also Acts 8:7, Luke 4:41), but he left it masculine. This hypothetical derivation of one text from the other follows one of the three NU internal rules (Which reading best explains the others?), but in practice the NU criteria are not invoked to support Byzantine readings, even if cogent, because the readings are not Category I.

Since adding or subtracting words is inherently equally probable, we must study the scribal habits to find empirical probability. We know that the early scribes (Alexandrian and papyri) regularly shortened their exemplar, because they shortened it in different places. (These habits are obscure in the NU apparatus but stand out clearly in the Swanson format.) The known Byzantine scribes shortened or lengthened their exemplars far less frequently, although we do not know what the Byzantine scribes did before the extant evidence. The NT was popular literature, and Koine studies show that popular literature was fuller and more explicit than classical Greek. Therefore, it is possible, although inappropriate, to project these tendencies into a rule of preferring the lectio longior, but lectio brevior is even less appropriate. Whether for justifying an individual variant or the whole A-list, lectio brevior is merely a circular argument based on the relative brevity of the Alexandrian text or a preference for the streamlined style. One other point may have struck the reader. Since the difference between NU and TR in such cases is mainly style, either option is safe. Our editions may have gotten shorter and different in the last 150 years of scholarship but not much better or worse—so far. One impact from the edition conflict and the pursuit of modernity is more unnecessary disunity: “I am of Byzantine Paul, I am of Alexandrian Apollos.”

29. The Conflate Argument: Why Reason Does Not Win

Bias is common in postmodern times because evidence is effectively circular: it is what convinces, and many are easily convinced. People often do not have the habit of searching for good arguments against their own position. Let us look at how people think who have acquired a degree of textual expertise. Consider a typical pro-Alexandrian book reviewer on Amazon.com. He gives a polite commendation to a pro-Byzantine author, Van Bruggen, and includes the following critique:

And the problem of conflations is nearly always dealt with by a casual dismissal: “only eight” is a common mantra. But the problem is that it only takes one to prove a conflated text type. And the underlying assumption given by this attempt at refutation is the notion that Hort presented all of the conflations. As a matter of fact, I know of at least nine other conflations of the Byzantine text type—four in Matthew, four in John, and one in Acts—that demonstrate conclusively that it is a secondary text. That brings us to 17—and I wonder how that will be dealt with.[22]

What is a conflation? If a manuscript has “to him the Father” (αὐτῷ πατρί), and others have only “to him” (αὐτῷ) or only “to the father” (πατρί, John 10:38), the reading with both is called a conflation and is assumed by internal probabilities to be derived from the two elementary readings. This idea forms one of the internal criteria for selecting variants. WH used it to great effect to convince scholars in 1881 that the Byzantine text was late-conflate. It was one of their three biggest selling points for the Alexandrian version of text history. The reviewer, like many pro-Alexandrians, is convinced that the conflate evidence is such a crushing argument that it should finish the Byzantine version of text history off almost by itself, if the stubborn proponents could only grasp it. His confidence is pumped up because the literature he reads from the Alexandrian scholars is full of such confident claims. He objects that argumentative rejectors of the conflate rule repeatedly point out that WH only had eight examples, whereas the he knows of 17 that satisfy his definition, and presumably there must be others. Since he has read Van Bruggen’s book, he also knows that there are many hundreds of opportunities that Byzantine scribes had to conflate but did not do it. But he does not mention that in his review. Actually relative infrequency (“the mantra”) is a good logical argument against the claim that a principle exists.

The conflation example above illustrates another serious flaw in the alleged conflate evidence. The conflation example (“to him the father”) is from a papyrus, P66. It is earlier than any of the Alexandrian uncials that have the parts. The “conflated” manuscript came first and then the Alexandrian manuscripts with the parts were written later. Obviously, “conflate” does not automatically equate to “late,” as the pro-Alexandrians imply. And of course there are other Alexandrian conflations, even in Vaticanus and Sinaiticus. The picture is far messier and more inconclusive than the advocates claim. If one dismisses the P66 example as an early conflation, it weakens the whole conflate argument. Even if the Byzantine text was conflate, it could have existed prior to 200. And in general there is no inherent logical cogency to the conflate rule at all, since it is as easy to drop as to add, and NU admits that Alexandrian early scribes frequently did drop. The reviewer has heard all these points before (since they are in the book he read), but, like the scholars in general, he argues like a lawyer, not a scientist, marshaling his best arguments and keeping silent about his opponents’ good ones. Sometimes he thinks his goal is to find the truth, but in practice the goal he is focused on is convincing the partially informed that his answer is right.

This pattern has been going on for lifetimes in textual criticism, even at the highest levels. Even if courtesy toward opponents is generally practiced, high standards of weighing both sides fully and avoiding speculation are not. The critical phase of textual criticism is a mine field riddled with speculation masquerading as knowledge. Here is an example. In Acts 25:9 Alexandrinus places the words “I am standing” after “at the judgment seat of Caesar,” while Sinaiticus places “standing” before it. There is a simple transposition. But when the Vaticanus scribe saw these two sources, he put “standing” in both places to copy both, thus conflating them. Of course the preceding sentence is just a made-up story about how Vaticanus came to have its two “standings,” a story made up by a later “critical scholar” (me), which may or may not be true. Alexandrian scholars such as TCGNT would never make up this particular story because they want the Alexandrinus text to be later than the Vaticanus text, just as the Alexandrinus manuscript is later than the Vaticanus manuscript. But they make up such stories on a regular basis when it works against the Byzantine priority. TCGNT is full of them. The Alexandrian priority is built on a foundation of storytelling which cannot be verified. If readers stopped believing the stories, the Alexandrian priority would collapse.

30. Many Meaningless Minor Changes in the Early Manuscripts

One thing that strikes the reader of Swanson’s parallel Greek NT texts is the number of apparently insignificant, meaningless changes that the early scribes made in the Alexandrian uncials and the papyri—for example, transpositions. How do we know the scribes made these harmless corruptions? When two or three key early scribes had the same transposition, we do not know; it becomes the official NU Bible. But when they each made them in different places, one at a time, then by its rule requiring only one original, NU is usually bound to reject the oddball and select the majority. A vote lies at the heart of most editing in practice. These transpositions seldom make a translational difference and seem capricious: change for change’s sake, as though the early scribes were recounting the gist of a narrative, but putting it in their own words.

The modern NU editors sometimes demonstrate continuity with their stylistic predecessors by favoring random alterations. For example, in John 10:39 in the Vaticanus and P66 phrase “again him” (πάλιν αὐτόν), Sinaiticus, P45 and Bezae dropped the πάλιν while a few later manuscripts transposed, but most of the manuscripts, including the Byzantine majority and TR, retained the Vaticanus order. The NU editors, defying their preference for the Vaticanus and the older Alexandrian witness and bypassing the external and internal evidence, unexpectedly selected the later transposed variant for the NU, thereby demonstrating that they are not always slaves to mechanical rules or even to Vaticanus. It was a completely insignificant exercise of prerogative, possibly even an accident, not commented on by TCGNT. So now you know the story of how the NU edition came to read “him again” (αὐτόν πάλιν), instead of the old TR reading in John 10:39. One other observation: NU kept the 1881 WH order. The fact that critical experts a hundred years apart came to the same conclusion, even if it appears arbitrary and its reason is unknown, helps maintain the appearance of steadiness in the critical enterprise.

Part V. Conclusion

31. The Alexandrian-Byzantine Situation Today

The late age of the earliest Byzantine manuscripts and supporting evidence has always prevented the Byzantine texts from acquiring a scholarly reputation, so the majority of scholarly oversellers are on the Alexandrian side today. Regardless of which side the oversellers take, they should stop and see the inadequacy of their case. And those who are experiencing pressure to switch but do not want to should feel confident that the opponents do not have a compelling case either. Some might wonder if a compromise could not be reached and take the best of both worlds. The thoroughgoing eclectics advocate this. But they have not produced a popular conflation and probably will not, because it would be more artificial than the current texts. Even if they could devise a set of neutral scoring methods and apply them with careful evenhandedness, there are no special reasons to believe that their criteria would favor originality. NU and the Byzantine majority show that scorers can tweak criteria to insure whatever percentage-mixture they determine in advance. “The overwhelming majority of readings were created before the year 200,” affirms Colwell.[23] The evidence void spans the very period when most substantial corruptions developed. When the evidence trail grows cold, even the good detectives just spin theories. Then, since the remedies to circumvent this gap are all questionable, why do many persist in believing they have the answers? “Hope springs eternal” is the answer that agnostics give to this and to Christian faith itself, yet many believers understand that the Holy Spirit inspires faith in Christ but not in experts. They may find some consolation for their margin of unresolved uncertainty about God’s Word in the smallness of its scope.

The two camps do not stand on a level playing field. The Alexandrians have a modern edition based on massive collaboration. The Byzantines have primarily an edition (TR) that was produced at the dawn of text-critical history and naturally has some weaknesses. The story is told that Erasmus’ one Greek source for Revelation lacked the last few verses, forcing him to retro-translate them from the Latin Vulgate into Greek. Unknown to him, the Vulgate had the phrase “book of life” in Revelation 22:19, where virtually all Greek manuscripts had “tree of life,” so the KJV and NKJV say “book of life” to this day. It is a genetic marker. The Byzantine readers may avoid this kind of problem by using a Byzantine priority text by Robinson and Pierpont or Farstad and Hodges. The Orthodox Patriarchal edition of 1904 could conceivably be even more natural, since it is described cryptically as deriving mainly from one best manuscript, but it contains no apparatus or source identification. Maurice Robinson warns that it is mainly a lectionary edition. Mysteriously, it contains the Johannine comma.

If you believe that new versions are a good thing, the need for an edition for translation remains, and noncommittal formats like Swanson’s do not fill it. The “fix” is probably not a new WH text revolution or continual adjustments but the use of our current texts with awareness that they just represent two different ancient texts, not originals.

32. Theological Bias

A particularly volatile subject involves text tampering for theological purposes. Antenicene fathers like Irenaeus denounced it. KJV-only advocates often claim that heresy was rife in Egypt and that the Arian heretics came from Alexandria, although actually the Arians were found throughout the empire. It is a simple fact that some disputed passages of the Byzantine family support the doctrines of the Trinity and the divinity of Christ better than the Alexandrian manuscripts do. But the pro-Alexandrian scholars counter with an assertion that there is no doctrinal bias coming from their side, that any essential doctrines are supported in other passages, and that sometimes the Alexandrian manuscripts are even more orthodox than the Byzantine ones. The picture is mixed and too broad to cover in detail here, but two or three examples may be mentioned. Clearly the Alexandrians have the better textual case on the explicitly trinitarian passage of the Johannine comma, 1 John 5:7. It is missing even from the majority of the Byzantine texts and was not used in the records of the Nicene debate. In 1 Timothy 3:16 the Byzantine text in support of the deity of Christ is fairly easily interconvertible with the Alexandrian text which does not support it explicitly.

The antenicene fathers quote both the Byzantine John 1:18 (“only begotten Son”) and the Alexandrian “only begotten God.” Some scholars think the Byzantine scribes changed it to reflect Nicene orthodoxy. Others think the Alexandrian scribes changed it to promote a non-Nicene theology. For example, “only begotten God” also occurs in Ptolomy’s gnostic “Commentary on the Gospel of John Prologue” (180 AD), in the Second Arian Confession (341 AD), and in other gnostic and Arian writings. Most NU-based versions other than NASB and the Jehovah’s Witness version translate the Alexandrian expression gingerly with creative circumlocutions, or revert to the Byzantine wording. Some, such as the New American Bible, conflate both source texts (“The only Son, God”) to be more inclusive.

33. Providential Preservation and Inerrancy

Defenders of the TR often invoke the doctrine of providential preservation to claim that God preserved his word specially in the TR or the Byzantine family. Providential preservation is a doctrine of the Westminster Confession of Faith (1.8). The obvious problem with applying providential preservation word-for-word to the TR is that, while wide variances occurred in the Greek NT manuscripts in general and in the Alexandrian family in particular, the Byzantine manuscripts themselves also exhibited variants, and there are even a number of minor variations in the TR line. So one would need to answer which of the multiple TR’s between 1516 and 1894 was providentially preserved. Some KJV-only advocates believe that Scrivener’s 1894 edition, that has all the sources for the King James, is the one. This type of thinking appears outrageous to many because the word “preserved” seems to be alternatively used to mean a facsimile or a restoration. How much change falls under the heading of preservation? It is hard to know how the doctrine can be used reasonably to defend any particular line in an absolute sense. If “preserved” means that the phrases of the 1894 TR edition or of any TR all occurred in one particular manuscript, then the TR was not preserved, because no manuscript, even when normalized, matches the TR: it is eclectic. On the other hand, if “preserved” means that all of the phrases occurred disbursed in various manuscripts, then the NU was preserved too, because NU also has a rule against inventing emendations.

Since textual variation impacts the doctrines of inerrancy and infallibility, the denominations of the church that give priority to the Bible as the ultimate source of truth and revelation also address those issues. In the late nineteenth century B. B. Warfield proposed the doctrine of the inerrancy of the autograph, and this solution has been widely adopted. Inerrancy, however, did not extend to the copies. While it solves part of the technical problem, the believer who wants to have access to the whole infallible Word may be left with a certain degree of dissatisfaction that what was infallible is now unavailable and what is left is in some question. The usual advice is to get used to it. It is more important to have a humble recognition of the large common ground and the limited uncertainties of the editions than to choose the right edition.

34. Conclusion: A Long-term Truce between Byzantine and Alexandrian

In summary, although a few living scholars may still think that NU has restored the original, and many [24] believe that the Alexandrian NU is somehow more scholarly and scientific than the Byzantine texts, this level of confidence is not warranted by hard evidence. Students would do well to develop a healthy skepticism toward the methodologies, rules, and conclusions. The collection and classification of Greek manuscripts is wonderful, and so is standard editing. The critical-editing aspect of the critical enterprise is not so wonderful. Given the nearly complete destruction of the first- and second-century manuscripts, either the Alexandrian or Byzantine texts or both are viable choices for the well informed. There are two story lines and two popular edition options. Whichever one scholars or non-scholars prefer, there is no substantial scholarly case to reject the other. In the last 120 years the Alexandrian case has been oversold academically, while the Byzantine case con

Contents

Skeptical Trends in New Testament Textual Criticism: Inside the Alexandrian Priority School and Why Bible Change Is Coming

Dennis Kenaga