The purpose of New Testament Textual Criticism is to recover the original New Testament text. This, obviously, requires the textual critic to resolve variants. This entry gives an assortment of variants, plus descriptions of how they have been resolved by various scholars .
Entries in the document fall into two parts: Those where most if not all modern scholars agree, and "challenge readings" -- places where different scholars assess the readings differently. The first section can therefore be used to see the agreed-upon methods of interpretation; the second allows you to examine methods used only be certain scholars.
Each entry begins with a presentation of the evidence, in the fullest possible manner. The variant portion of the reading is shown in bold. All major variants are presented (with the variant preferred by the UBS editors listed first), with support listed in the usual order (papyri, uncials, minuscules, versions, fathers). The printed texts that support the reading will also be listed. This is followed by the various scholars' interpretations.
The examples in the section which follows are accepted by all, or nearly all, modern scholars. (The major exception, in most cases, is the scholars who believe in Byzantine priority.) They thus serve as good examples of the ways in which scholars work, and demonstrate the methods used.
This reading (except for the question of including or excluding EN, which is relatively trivial) can be resolved based on either internal or external evidence. The external evidence overwhelmingly favours the reading "Isaiah the Prophet;" it is supported by the Alexandrian ( B L D 33 892 1241 2427 sa bo), "Western" (D it vg), and "Cæsarean" (Q f1 565 700 arm geo) texts. In favour of "in the prophets" we have only the Byzantine text.
Internal evidence is equally decisive -- because the quotation is not from Isaiah alone, but from Malachi and Isaiah. The attribution to Isaiah is an error, and scribes would obviously have been tempted to correct it. (Neither of the parallels mentions Isaiah.) Thus it becomes certain that the original reading was "In Isaiah the prophet."
At first glance it may seem that the evidence for the longer reading is overwhelming in its magnitude. Careful consideration shows this not to be the case. The shorter reading is clearly that of the earliest Alexandrian texts (P75 B), and it is also the apparent "Cæsarean" reading (1+1582 22 700). It also has the support of the original vulgate. Thus its external support is at least as strong as, if not stronger than, that for the longer reading.
But it is the internal evidence that is absolutely decisive. The longer reading is, of course, that found in Matthew 6:9, and in Matthew there is no variation. Equally important, every one of these copyists must have known his paternoster, and they would all know it in Matthew's form (since it is at once fuller and earlier in the canon). If they found a short form in Luke, they would inevitably have been tempted to flesh it out. And under no circumstances would they ever have removed the longer words. Thus it is morally certain that the short form is original (here and in the several other expansions found in the Lukan version of the Lord's Prayer).
There are two questions about this reading: Is it part of the Gospel of John, and if not, where and how should it be printed? The fact that most of the editions include the passage in the text in some form does not address whether they regard it as original.
The external evidence against the reading is almost overwhelming; it is omitted by all significant Alexandrian witnesses except except 579, 892, and some Bohairic manuscripts (all of which are secondary texts) and the "Cæsarean" witnesses omit it or move it elsewhere. It is found in some "Western" texts, but others (including the very important a) omit, and even the earliest Byzantine texts, such as A N, lack the reading. The external evidence alone is sufficient to prove that this is no part of the Gospel of John.
Some scholars have tried to rescue the passage on internal grounds, arguing that scribes would omit it because they disapproved of mercy to an adulteress. But while this might explain its omission from a few texts, it cannot possibly explain its absence from so many -- nor why it appears so often as a correction.
It should also be noted that the passage has a style very unlike the rest of John, and uses a great many words not found elsewhere in that gospel.
This is not a statement about the truth or falsity of the story. But there can be little doubt that the story of the adulteress is no part of the original gospel of John.
This reading is interesting because it has been omitted from every critically prepared edition ever published, including even the Majority Text editions. But it is found in the Textus Receptus and the King James version.
The evidence for verse 37 is usually stated to be weak. It isn't, really; the verse has the support of the "Western" text (D is defective here, but we find it in E and the Old Latins), as well as Family 1739 (323 630 945 1339 1891). Still, it is missing from the Alexandrian text, and probably also from Family 2138. So the external evidence is slightly against the verse.
Internal evidence also argues against the verse. Its style is regarded as un-Lukan, and there is no reason for it to have been omitted had it originally been present. The best explanation for its appearance seems to be that scribes felt that the eunuch needed to make some sort of confession of faith before baptism, and so added one. Thus it seems best to omit the verse.
This reading can be approached based on either internal or external evidence. The internal evidence says that longer readings are often suspect -- at least when they are more liturgical or Christological. Thus the reading with "our Lord" is highly questionable. It has been suggested that the words are derived from verse 23 -- though there is no real need for such an explanation, as there is absolutely no reason why the words might be omitted had they originally been present.
The external evidence points the same way. Although the longer reading has the support of most parts of the Alexandrian text ( C 81 1506 family 2127 bo), the words "our Lord" are omitted by P46-B-sa, by the "Western" text (D F G 629 Old Latin and all the best Vulgate witnesses), and by Family 1739 (1739* 630 2200). Thus the plurality of text-types also stand against the reading. We can be confident that the words "our Lord" are spurious.
The external evidence here is rather split; a large part of the Alexandrian text, including A 33 81 436 bo, read "dead"; they are supported by the entirety of Family 2138. "Unproductive," however, also has good Alexandrian support (B 1175 sa), as well as many of the better Family 1739 manuscripts (322 323 945 1739). (The reading "empty" of P74 may have been suggested by KENE in the preceding clause.)
If the external evidence is divided, the internal evidence is clear. In verses 17 and 26, we read that faith without works is dead. And there is no variation in either of those verses. Since assimilation to local parallels is an extremely common sort of corruption, we may feel confident that the reading "dead" is a corruption, and "unproductive" original.
This reading illustrates well the danger of applying rules over-critically. The canon "prefer the shorter reading," if applied without discretion, might lead us to prefer reading #2. This is simply a mistake. The shorter reading obviously arose due to homoeoteleuton (the preceding clause also ends with TON PATERA ECEI). When one observes that the longer reading is also supported by the best representatives of all the text-types (Alexandrian: A B 33 and the Coptic versions; Family 2138: 614 630 1505 1611 1799 2138 2412 2495 and the Harklean Syriac; Family 1739: C 323 1739 Origen; also the vulgate), it becomes clear that the longer reading is original.
The readings in this section are not universally accepted by critical editors. However, there seems to be no reason in these instances to depart from the accepted readings of the UBS/GNT editions (which usually, but not always, follow the readings of Westcott and Hort). They are thus offered for further guidance, with the note than some editors will produce different results by different methods.
The evidence of text-types here is clear: The "Caesarean" text reads Jesus Barabbas; all other texts omit Jesus. On this basis we are inclined to omit Jesus, but we must look at internal evidence to determine the history of the passage. And it is clear that the reading Jesus Barabbas can explain the reading Barabbas, but not vice versa. Origen himself shows this; although most of the manuscripts he knew read Jesus Barabbas, he preferred Barabbas. Many other scribes must have felt this way, meaning that the reading Jesus Barabbas is almost certainly original.
The readings in this section illustrate points where critical editions are very divided. They are presented to illustrate the difficulty of resolving certain readings.
These readings, like many others in the Synoptic Gospels, can only be considered together. The setting is the naming of the Twelve, and the evidence for each reading is set out in this table:
|Reading||Matt. 10:3||Mark 3:18|
|B 69 788 826 892 983 185 2211 aur c ff1 l vg sa meg bo Jerome Augustine [UBS WH Merk Bover Vogels Souter]||A B C E F G H (K Daddaion) L D(* Taddaion) Q P S 0134 f1 f13 28 33 157 565 579 700 892 1010 1071 1079 1241 1243 1342 1424 1505 1546 2427 Byz aur c f l vg sin pesh hark sa bo arm geo goth eth slav Origen [all editions]|
|D d (k) m Origenlat [Tischendorf NEB]||D a b d ff2 1 q r1|
|Qaddaios o epiklhqeis Lebbaios/Thaddeus called Lebbaeus||13 346 543 828 547|
|Lebbaios o epiklhqeis Qaddaios/Lebbaeus called Thaddeus||C(*) E F G K L N W X D Q P S f1 28 33 157 565 579 700 1010 1071 1079 1243 1342 1424 1505 1546 Byz f pesh hark palmss (arm) geo (eth) slav [Soden Hodges-Farstad TR]|
|Judas Zelotes||a b g1 h q (palms)|
|Judas of James (and transpose)||sin|
Despite the confusion of readings here, it is obvious that, in both Matthew and Mark, the original reading must be either Thaddeus or Lebbaeus. The conflate readings in Matthew are obviously an attempt to combine the two names.
But which is original?
In this case, the easiest place to start is Mark. Although internal evidence doesn't really apply here (neither name has any particular significance, since this particular disciple doesn't ever do anything), the external evidence clearly favours "Thaddeus." This reading has the support of every Alexandrian, "Cæsarean" witness, and Byzantine witness; the supporters of "Lebbaeus" are all "Western." While we cannot be certain in such a case, the reading "Thaddeus" seems much the stronger of the two.
So what does this say about Matthew? Here the matter is much less clear, since only the Alexandrian text unequivocally supports "Thaddeus." Ordinarily we might suspect that this variation arose because Matthew and Mark had different readings. This is, in fact, why the NEB chose the reading it did.
But look at the situation again. In both gospels, we find "Thaddeus" supported by the Alexandrian witnesses (with some supporting evidence), while we find "Lebbaeus" exclusively in "Western" witnesses. In other words, each of the two main text-types had its own reading, which it used consistently. There is no confusion in the witnesses, merely disagreement.
This argues that only one reading is original; one or the other text-type (for some unknown reason) altered both lists. And if this is the case, it is almost certain that it is the "Western" text which did the adapting. We therefore, and with much hesitation, adopt the reading "Thaddeus" in both passages.
The readings in this section were selected by Robert Waltz to conform with my views on textual criticism. Note that most of these examples will be rejected by the majority of scholars.
I am what is called a "historical-documentary" scholar -- that is, I start by examining the manuscripts and searching for early text-types. Only after I have determined the text-types do I turn to variants. If all the text-types agree, well and good. If not, I try to construct a local genealogy to explain the variants.
It should be obvious that, in order for this method to work, the history of the text must be known in the greatest possible detail. In Paul, for example, I find four basic non-Byzantine text-types: P46/B (P46 B sa), "Alexandrian" ( A C I 33 bo; also 81 1175 etc.), "Western" (D F G Old Latin; also 629), and family 1739 (1739 0243; also 0121 1881 6 424** 630 2200 etc.). In the Catholics there are three: "Alexandrian" (p72+B, , A+33+81+436+bo), family 1739 (C 1241 1739; also 323 945 1881 2298 etc.), family 2138 (614 630 1505 1611 1799 2138 2412 2495 Harklean etc.). In the gospels my results are incomplete, and in Acts and the Apocalypse they are barely begun; therefore I concede that my results there are tentative.
It's rare to see the evidence so nicely divided as this. The Alexandrian text clearly supports OUTWS, the "Caesarean" KRAXAS, and the Western (with some minor variations) OUTWS KRAXAS. Critical editors have hastened to adopt the Alexandrian reading, perhaps explaining the presence of KRAXAS as coming from Matthew 27:50. But this verse isn't really parallel; if it had been harmonized, why was KRAXAS the only word to show up? Given that the three early text-types differ, we must ask ourselves which reading best explains the others. Is the Western/Byzantine reading conflate? Possibly -- but if so, it is a remarkably early conflation. It also produces a difficult construction. It is easier to believe that the longer reading is original, and that the Alexandrian and "Caesarean" copyists separately shortened it.
This complex reading requires careful analysis. In looking at text-types, it is clear that the "Western" text included the longer reading. The "Caesarean" manuscripts are divided, but even so, it is clear that the type omits (since the reading is missing from family 1, family 13 (part) 700* arm geo). The evidence of P45 for a reading such as this is little help; this is just the sort of phrase it likes to omit. This leaves the Alexandrian text. Which is distinctly divided; B C 33 579 892 bo include the reading while p75 L 070 1241 sa omit. If we consider the "phases" of the Alexandrian text, however, we find that the earlier (P75 sa, though not B) and the latest (L 070 1241) omit; only the middle phase ( C 33 579 892 bo) includes the words. Thus the evidence of text-types stands slightly against the reading.
The internal evidence is also slightly mixed, since this passage has no exact parallels. However, the partial parallels in Matt. 5:15 and Mark 4:21 are probably enough to account for the addition here. It is hard to see how the phrase could have been lost; perhaps it was haplography, or the loss of a line from a manuscript with about sixteen letters per line, but both explanations are far-fetched. Thus both the evidence of text-types and internal evidence are against the reading; it is better to omit the phrase.
To my mind, this reading shows clearly the danger of assessing readings starting from the internal evidence. It gives the critic too much chance to be imaginitive.
This reading is settled instantly on the evidence of text-types. Clearly the "Western" text omitted the reading (so * -- here "Western" -- D it). So too, clearly, did the "Caesarean" text (family 1 22 565 arm geo). But so too, evidently, the earliest phase of the Alexandrian text, since the words are missing from p75 sa. There really isn't any reason to look at internal evidence (though it's worth noting that it is indecisive); the words should be omitted.
At first it might seem that the evidence of text-types would favour POLLOI/many. This is true in part; clearly this is the reading of the Alexandrian text and of family 1739. But the "Western" text favours LOIPOI/[the] rest, and p46 and B are split. (The Byzantine text is also split, but this has little effect on out deliberation except to explain why 6 and 630 defect from family 1739.) Although the external evidence favours many, the margin is very slight; we must look at internal evidence. And this clearly favours [the] rest. Either word could easily have been confused for the other, but which is more likely to survive? Obviously many. Scribes would not approve of Paul lumping all other preachers -- including themselves! -- as God-peddlers. The fact that the reading [the] others survived at all is a strong testimony for its originality. And Paul was certainly willing to use such extreme language (note his condemnation of everyone except Timothy in Phil. 2:21). While the matter cannot be certain in the face of the external evidence, LOIPOI is clearly the better reading.
As always, I start by looking at text-types. But text-types aren't much help here. It is evident that the Alexandrian text read KAI PROSKOLLHQHSETAI THN GUNAIKA AUTOU, the "Western" text read KAI KOLLHQHSETAI THN GUNAIKA AUTOU, and family 1739 omitted. The P46/B text is divided. Thus no reading commands the support of the majority of text-types. Indeed, none of the readings can even be said to have "strong" support (though the support for KAI PROSKOLLHQHSETAI THN GUNAIKA AUTOU is strongest). So we turn to internal evidence.
In assessing this, we note that readings 1, 2, and 4 are all harmonizations, and 3 is singular and probably an error for 2. Is it possible that one of these three could have given rise to the others? Of course. But it is by no means obvious which reading of the three is most original.
On the other hand, if we assume that reading five, which omits the phrase, is original, then all becomes clear. Scribes, confronted with this quotation, would observe that the middle phrase had been left out. They would instinctively conform it to the version most familiar to them. And once the phrase was in place, there would be few further alterations.
It has been proposed that the omission in family 1739 was caused by homoioarcton. This is possible, leaps from KAI to KAI were common enough. But this would be an awfully suspicious location for it to happen... why at this place where so many other readings exist? It is also possible that the omission from 1739 came because scribes marked in some sort of correction which was interpreted as a deletion. But, again, it is such a convenient error.
Back in the nineteenth century Hort said of this variant, "A singular reading, which would not be improbable if its attestation were not exclusively patristic; the words might well be inserted from Gen ii 24." We now know that the reading is not exclusively patristic. Its support is diverse, and on internal grounds it is well-founded. Although we cannot be sure in this case, this seems to me to be clearly the best reading.
Until the discovery of P72, no one paid much attention to this variant. The fact that scribes were more likely to add than subtract TWN AIWNWN was largely ignored.
It should not have been so. Even if we ignore 69 as prone to such errors, the words are missing from family 1739 (945 1739) and from family 2138 (206 614 630 1505 1611 2138 2495 hark). This leaves, apart from the Byzantine text, only the Alexandrian text-type to support the longer reading. When we note that the earliest witnesses of this type (P72 and many Coptic manuscripts) omit,and that they are joined by the best of the Latins, the short reading becomes distinctly preferable.
Most editors have preferred the reading APATAIS, regarding AGAPAIS as an assimilation to Jude 12. If there were only two readings here, this might be logical. But there are three. We must examine the reading more fully.
As far as the evidence of text-types goes, APATAIS appears to be Alexandrian, but arguably the later form of the Alexandrian text. AGAPAIS has less Alexandrian support, but what it has is generally early (A** B sa). It also appear to be the reading of family 2138 (although the majority of that family supports APATAIS, this appears likely to be a Byzantine correction; the earliest reading is probably AGAPAIS, as in 1611 and the Harklean margin). Finally, AGNOIAIS is read by family 1739.
It is obvious that we cannot make a decision based on text-types, But we must observe that all three readings are attested in early text-types. This means that the middle reading is most likely to be original. And the middle reading is obviously AGAPAIS. It's easy to see how it could have turned into APATAIS -- and also how it could have become AGNOIAIS. Whereas it is almost impossible to see how APATAIScould have become AGNOIAIS or vice versa.
The argument that AGAPAIS is an assimilation to Jude 12 is also weakened when we recall that Jude is after 2 Peter in canonical order, that it was accepted into the canon very late, and is generally a weak epistle. Also, there is variation in Jude 12 (where A Cvid 1243 al read APATAIS and 6 424** read EUWCIAIS). Colwell has shown that assimilation of distant parallels is less common than previously assumed. So it should not be assumed here. Eberhard Nestle offered cogent internal reasons why AGAPAIS should be regarded as original in 2 Peter. Surely these offset the internal evidence of assimilation. The reading AGAPAIS belongs in the text.
Of all the New Testament books, Jude is probably the most afflicted by textual variation, and it is often difficult to decide where one variant ends and the next begins. I think, though, that this variant (add/omit EQNESIN). can be treated in isolation.
Most scholars look at this reading and say, "EQNESIN? Found only in minuscules. Forget it." The evidence of text-types says otherwise. It's true that the Alxandrian text omits the word, and obviously the Byzantine text does also. But the word is found in both family 1739 (6 322 323 424** 945 1241 1243 1739 1881) and family 2138 (614 1505 1611 2138 2412 2495 hark) -- two early and unrelated text-types. In other words, on the basis of text-types, it has as strong a claim to originality as the text without it.
Internal evidence, if anything, favours the reading. There is no text anywhere in scripture which is even vaguely parallel; the reading is unexpected and strange. Frankly, it's easier to see scribes omitting EQNESIN than adding it. It might even have been an haplography induced by the following EN QEW. I agree that it's hard to adopt a reading which completely lacks uncial support. I'm far from certain this is correct. But I think EQNESIN belongs in the text.
No doubt my advocacy of a reading ignored by most other scholars will seem surprising. Strong internal grounds have been adduced for EXETE, and it also has strong manuscript support.
However, the evidence of text-types does not favour it. A C have other readings (admittedly different readings), and Andreas also defects. Under the circumstances it can be said that all of the first three readings are old -- old enough to possibly be original. In which case the reading most likely to be original is the middle reading, ECETE. From here to the other two involves a change of only a single letter.
I admit that these are awfully thin grounds. But the evidence for the other readings is not overwhelming. When in doubt, one should follow the rules.