Our goal in producing this edition was to provide a basis of comparison between the decisions of other text-critical efforts and the results of what Robinson calls “reasoned transmissionalism,”‡ which we consider to be the most rigorous Byzantine-priority approach. Our edition therefore agrees with RP in all places where its text has virtually unanimous support from Byzantine witnesses and in most places where such support is divided. While a comprehensive defense of the Byzantine text and the assumption of its historical priority is beyond the scope of this essay,§ we would like to be as transparent as possible regarding our departures from RP and to encourage readers to engage in more detailed interactions with individual variants, whatever text-critical framework they have adopted. We have produced this commentary with these goals in mind.

Where possible, we have used the apparatus of NA 28, the twenty-eighth edition of the Nestle-Aland critical text,* to present collation data pertinent to considerations of external evidence. For the readings of f 35, which has many majority readings, but departs from the RP text in a number of these variation units, we have used Pickering's collation.† Since a number of the variants covered here are not noted in the NA 28 apparatus, we have collated the NA 28 consistently-cited Greek witnesses ourselves where necessary; we present our results in the sections where we discuss external evidence. We have endeavored to incorporate existing literature into our discussions, but for many of the variants treated here, little to no published work has been done.

As a normal transmissional model would predict, most of the divided-majority variants covered here hinge on common scribal errors. In many cases, change in either direction is likely, and the text along different developmental branches could easily have switched back and forth between both readings at separate times. For this reason, divided majority variants can rarely be handled with certainty, and even when they can be, they are rarely descriptive of essential genealogical relationships. Accordingly, this commentary is not intended to serve as an authoritative guideline for weighing manuscripts or groups of manuscripts; it is simply an attempt to detail the evidence for why we departed from our base text where we did.

In our treatment of each variation unit, we will list both the RP text reading and the reading adopted in this edition (SRS), and offer our assessments of both in terms of internal and external evidence. Every SRS reading covered here coincides with the RP margin reading and therefore represents an alternate Byzantine reading.

On intrinsic grounds, ἕτερος is slightly more common in Matthew. Within the Byzantine text, it occurs seven times,† while δεύτερος only occurs three times.‡ From what we can tell, δεύτερος has a more specific function in Matthew's usage than ἕτερος does: In 22:26, it is used in a sequence seven items long, where ordinal designations serve to distinguish the items; in 22:39, it is used to emphasize a difference in priority (the commandment cited being the “second greatest”), where ἕτερος might imply equality; and in 26:42, the ordinal term is used, as it was in 22:26, to mark one event in a sequence of more than two (the first being in 26:39 and the third being in 26:44).

In the passage at hand, the man has only two sons, so there is no need to clarify that the other son is the second one. A distinction of age might demand such clarification, but the sons' ages are irrelevant to the parable. Likewise, a difference in priority does not seem to fit Matthew's intent here, because up to this point in the parable, the audience has had no reason to consider either son the better one. The best intrinsic argument for δευτέρῳ is that the sons were simply identified by the order in which the father spoke to them, but this detail is so obvious that δευτέρῳ could be attributed to a scribe just as easily as it could to Matthew. The semantic difference between both choices is nuanced enough that either is plausible, but ἑτέρῳ fits at least slightly better with Matthew's usage.

On transcriptional grounds, context is especially important for this variant. The parable of the two sons does not appear in any of the other gospels; therefore, harmonization does not lend itself immediately as an explanation for either reading. Some scribes might infer a superficial association with Lk 15:11-32, but since the parable there differentiates the sons as νεώτερος (younger) and πρεσβύτερος (older), it is also not likely to be a source of harmonization with respect to this variant.

Nevertheless, the near context does provide a clue: In Mat 21:28, one son is identified with πρώτῳ (first). While Matthew may have had his own reasons for using ἑτέρῳ two verses later, as we have established, the sequence πρώτῳ … δευτέρῳ would probably come more naturally to a scribe's mind than the sequence πρώτῳ … ἑτέρῳ. Such a change is so subtle that a scribe copying the longer phrase all at once might make it without thinking. Moreover, the similarity of the two readings would surely have made it easier for correctors to miss the change.

On the other hand, the similarity of the two words might have led a scribe to drop a couple letters from δευτέρῳ accidentally, resulting in ἑτέρῳ. The problem with this scenario is that since the omitted letters are not adjacent, two independent omissions would have to have occurred. While such a scenario is still possible, it is surely less probable than a change from ἑτέρῳ to δευτέρῳ.

Turning to external evidence, the case for ἑτέρῳ becomes somewhat stronger. The non-Byzantine support for δευτέρῳ, which includes the majuscule B, the textual family f 1, the important minuscule 33, the second correctors of majuscules א and C, two Coptic versions, and a few other witnesses, seems somewhat isolated to witnesses with Alexandrian ties. The support for ἑτέρῳ is not detailed explicitly in the negative apparatus of NA 28, but based on what is not included in the support for δευτέρῳ, it is clearly as early and considerably more diverse. For this reason, we find ἑτέρῳ preferable to δευτέρῳ on external grounds as well.

Intrinsically, the RP text has multiple compelling arguments for its priority. The Byzantine texttype is agreed on Matthew's inclusion of καὶ in verses 37 and 39, where similar phrases are employed. Matthew also follows οὕτως ἔσται with καὶ in 12:45. Matthew's style in this context, if it can be inferred from these few other cases, favors the longer reading.

Turning to transcriptional arguments for the RP text, homoeoteleuton following ἔσται could easily explain the shorter reading. Indeed, in the similar phrases in verses 37 and 39, smaller groups of MSS do omit the conjunction in one or both cases. While the witnesses that omit in both places might reflect assimilation to an inherited omission in either location, the earlier individual omissions are best explained as independent scribal slips. Omissions in general were common scribal errors,§ and the occasion for a skip from one αι to the next would make an omission all the more likely here. The small size of the conjunction and its subtlety in the sense of the passage make the case for accidental omission even stronger.

We now turn to arguments in favor of the RP marginal reading. On intrinsic grounds, we note that Matthew uses οὕτως ἔσται without an accompanying καὶ in 12:40, 13:40, 13:49, and 20:26, so an argument based on mere frequencies may be moot; however, there are other factors behind Matthew's usage that might favor the shorter reading. One possibility is that the καὶ in verses 37 and 39 modifies not ἡ παρουσία, but οὕτως ἔσται. While the former usage imparts the sense that the parousia will be another event like the ones described in the immediate context (the lightning flashing across the sky and the days of Noah), the latter usage is more cumulative in force (“in addition to being like the lightning, the coming of the Son of Man will also be like the days of Noah”). In this case, it would make sense for the first οὕτως ἔσται not to be followed by καὶ, because there are no other comparisons in the preceding context. This is not an immediately obvious way to read the conjunction, which may explain why so many scribes departed from it, but it seems to serve the deeper purpose of elevating the significance of the event already described.

Another case for the marginal reading on intrinsic grounds could be made from synoptic parallels. The Markan parallel lacks any phrase similar to the one in question, but in Lk 17:24, the first οὕτως ἔσται, which corresponds to the sign of the lightning as it does in Matthew, has the undivided support of the Byzantine texttype in its exclusion of καὶ.* In Lk 17:26, the second οὕτως ἔσται, which corresponds to “the days of Noah” as it does in Matthew, is followed by καὶ, again without any recorded variation among the extant evidence.

How we interpret this evidence depends on the textual relationship between Matthew and Luke, but most cases lead to the same conclusions. If we hold that Luke used Matthew as a source, or that Matthew and Luke drew their shared material from a common source, then it is intrinsically more likely that Luke copied the shorter reading from Matthew, or that Luke copied the shorter reading from a shared source, and Matthew had every opportunity to do the same. Alternatively, if Matthew used Luke as a source for this passage, then we would have reason to believe that Matthew copied Luke's wording here. The scenario under the assumption of Matthean posteriority is particularly plausible, since intrinsic and transcriptional probabilities in this case are easy to distinguish: The author's use of Luke would run counter to the tendencies of later scribes who more often harmonized Luke to Matthew.†

On transcriptional grounds, the longer reading could be explained as the result of assimilation to the parallel phrases in verses 37 and 39.‡ For scribes familiar with the οὕτως ἔσται formula repeated in this section, it would be easy to supply an additional word to the formula that was not originally part of it, perhaps even without thinking about it. Furthermore, the originality of the shorter reading here would better explain the omission of both conjunctions in verses 37 and 39 as an assimilation in the opposite direction. If the absence of καὶ in verse 27 was original, then an accidental omission in either of the later verses would be enough to tempt subsequent scribes to omit the remaining καὶ for consistency.

In terms of external evidence, the NA 28 apparatus is unfortunately silent on this variant. To fill this gap, we have examined most of the NA 28 consistently-cited witnesses at this unit. The resulting data, combined with Pickering's citation for f 35, appears in the following table:

We thus opt for the shorter reading here, both on the internal grounds previously established and on the external strength of the agreement between part of the Byzantine text and early Alexandrian and Western witnesses.

In terms of transcriptional probability, the apparent syntactic triviality of this type of variant and the frequency with which it occurs throughout the gospels together suggest that scribes introduced these changes largely without thinking about it. Knowledge of scribal habits will therefore be relevant to our evaluation of this variant. Based on Wilson's study of singular readings,§ omissions of the article are over 1.66 times more frequent than additions of the same, so we have some statistical justification for preferring the longer reading.

To be fair, the longer reading might be an assimilation, deliberate or subconscious, to the use of the article with Herod's name in verses 17, 18, 20, and 22. Yet with regard to the possibility of intentional assimilation, the article's manifest unimportance in this context makes it unlikely that any scribe would have bothered to insert it deliberately. Moreover, the absence of the article before Herod's name in verse 21, where most MSS appear to be in agreement, would not have been left unchanged by scribes intent on adding the article here. Subconscious assimilation, meanwhile, is problematic on the grounds that the variant in question occurs before all of the other uses of the article with Herod's name in this passage. Any scribe who had accidentally copied the article by glancing at one of its later occurrences would also have been at risk of omitting a larger phrase through homoeoteleuton, but we know of no documented instance of such an error here. The remaining possibility, that a scribe glanced up at his exemplar only to copy a one-letter article before resuming in the correct place, strains the imagination. Of course, the phrase ὁ βασιλεὺς Ἡρῴδης does occur in verse 14, but the article there is clearly applied to βασιλεὺς and not to Ἡρῴδης, so if anything, this phrase should have led absentminded scribes to omit the article here. Transcriptional probability therefore points in the direction of the longer reading.

Finally, we turn to the external evidence. The NA 28 apparatus again offers no data on this variant, so we have provided the following data based on our own examination, Pickering's collation of f 35, and Greenlee's transcription* of codex 0269:

While both readings have early support, the more diverse part of the MS evidence, both within the Byzantine texttype and outside of it, favors the longer reading. Given our assessment that the shorter reading likely arose by accident, it would be reasonable to conclude that most textual families preserved the longer reading, while minorities in several of them fell prey to the same easy mistake independently.

Intrinsically, the most important question is whether or not δὲ fits the context and Mark's intentions in this passage. On the surface, it seems inappropriate, as δὲ is most commonly adversative in its sense, and there is no obvious contrast in the preceding verses that would warrant this sense; Jesus has just praised the woman's sacrifice in anointing his body for burial, and he now adds that her act will be remembered wherever the gospel is preached. Likewise, Jesus is not making a broad change of topic when he makes this remark, so the use of δὲ as an introductory conjunction also seems like a poor fit.

On the other hand, δὲ can be used to continue or expand a previous idea, as it is in the Matthean and Lukan genealogies of Jesus and in Mat 16:18. Mark may have intended to use the conjunction in this narrower sense. Alternatively, he might have used δὲ with its common adversative sense, but with respect to a less conspicuous element in the preceding context. Specifically, Mark may have been conveying that although the woman could only do so much, what she did would nevertheless be remembered. In either case, the conjunction's nuanced usage speaks to its originality; if a later scribe were to supply a conjunction, either subconsciously or deliberately, he surely would have reached for one with a more clearly connective sense, like καὶ or γὰρ.

This point is also relevant to the discussion of transcriptional probability. While the inclusion of a conjunction does make for a smoother transition into Jesus' profound concluding remark, other conjunctions could serve this purpose more readily than δὲ does. If anything, the common adversative sense of δὲ might have presented enough difficulty to scribes that they considered the phrase to be preferable without any conjunction at all. Other scribes may have simply substituted a more fitting conjunction. We know from Pickering's collation data that a small proportion of later MSS have γὰρ in place of δὲ; this might have been motivated by the difficulty of δὲ in this context, or it might have been an assimilation to Matthew's usage, which employs γὰρ with ἀμὴν λέγω ὑμῖν on several occasions.

Furthermore, there are a number of other explanations for the shorter reading. The simplest is that δὲ was accidentally omitted; the omission of a two-letter word was easy enough, and for a scribe copying from a majuscule exemplar, a skip from ΔΕ to ΛΕ by homoeoarcton would make this even easier. More remotely, the shorter reading might have been an assimilation to nearby occurrences of the phrase without the conjunction in Mk 13:30 and 14:18, 25, 30, or a harmonization to Mat 26:13.

On external grounds, the evidence is so closely divided that the NA 28 apparatus's silence on this variant is surprising. To fill this gap, we have collated most of its consistently-cited witnesses at this unit, along with Pickering's data for f 35, in the following table:

Both Byzantine readings therefore have early attestation and support from diverse textual families, making the external evidence effectively ambiguous. We must therefore rely on the internal evidence, which, by our estimation, favors the longer reading.

For intrinsic probability, we must consider Luke's potential stylistic motivations in this passage. The RP text reading takes the more standard word order for this phrase, which would be suitable if Luke did not wish to draw extra attention to it. In contrast, the placement of the adjective outside the prepositional phrase to which it belongs would make the word μου far more noticeable to a reader. If the writer's intention was to highlight the centurion's humility, then this word order would be quite fitting. Luke clearly does draw attention to this aspect throughout the passage, so the RP margin reading seems perfectly plausible stylistically.

A further consideration is the relationship of the Lukan passage to its parallels. Mark lacks a parallel pericope, and the similar story in Jn 4:46-54 does not recount the words of the centurion, but the Matthean parallel to the phrase in question occurs in Mat 8:8. If Luke used Matthew as a source in writing his gospel, then it would not have been unusual for him to follow Matthew's word order in Mat 8:8, which is μου ὑπὸ τὴν στέγην according to both RP and NA 28. Of course, later scribes would also have been more familiar with Matthew's wording, so it is difficult to differentiate intrinsic likelihood from transcriptional likelihood under the assumption of Matthean priority. If Matthew used Luke as a source, then Matthew's reading could be taken as evidence that he copied the emphatic word order from Luke. If Matthew and Luke used a shared source for this material, then we would have reason to believe that at least Matthew's word order was derived from this source, and in this case, Luke's might also have been. The lingering problem in any case is the abundance of differences between Luke's version of the speech and Matthew's.† Whichever direction of dependence we assume, it is clear from these differences that the evangelists themselves could easily have been responsible for this variant.

Just as complicated is the question of transcriptional probability. Luke's focus on the humility of the centurion is so obvious that scribes would have no trouble noticing it. Those familiar with the passage might therefore have been tempted to transpose the unembellished RP text reading into the RP margin reading for the sake of conformity with the overall sense of the passage. Yet while this is certainly a possibility, there is an even simpler explanation in favor of the RP margin reading: Scribes copying phrase-by-phrase might have read μου ὑπὸ τὴν στέγην in their exemplar, and then misremembered the words in a more standard order as they copied out the larger phrase.

Another argument in favor of the RP text reading is that the RP margin reading is a deliberate harmonization to Mat 8:8, but this argument is lacking for two reasons. First, how likely is it that a difference in word order would merit such an emendation? And second, if harmonization was at work here, then why do we not see it at work among the same witnesses in all of the places where it would be more noticeable? In examining the NA 28 consistently-cited witnesses for this passage, we have found that several MSS do adopt one or more of these harmonizing changes at a time, but curiously, none of them adopts all changes together, and in some cases, the harmonizing change is inexact (e.g., 579 reads εἰπὲ λόγῳ μόνον in Luke, while in Matthew, it reads μόνον εἰπὲ λόγῳ).

These points suggest a simpler explanation: Scribes did harmonize to Matthew, but not intentionally. More specifically, scribes copying larger phrases from memory might have inadvertently supplied the Matthean word order because Matthew's gospel was more popular and therefore more familiar to them. This, at least, would account for the scattered and partial nature of the harmonizing variant readings in the centurion's speech. In addition, the only auxiliary assumption required for this explanation is that as early as the second century (the date of the earliest extant witness to the RP margin reading), Matthew's gospel was popular enough to have such an influence on scribes. While this explanation remains slightly more demanding than the competing argument in favor of the RP margin reading, we consider it plausible enough to leave transcriptional probability ambiguous.

We are therefore left with a somewhat balanced picture of internal evidence. Luke could easily have written ὑπὸ τὴν στέγην μου, but he also had good reason to use μου ὑπὸ τὴν στέγην. Transcriptionally speaking, ὑπὸ τὴν στέγην μου was the more common word order, and many scribes may have adopted this word order inadvertently. On the other hand, μου ὑπὸ τὴν στέγην could have been a conscious assimilation to the sense of the larger passage, or it might have been the result of subconscious harmonization to the parallel passage in Matthew.

On external grounds, even early witnesses are divided, so it is curious that the NA 28 apparatus does not account for this variant. The following table presents our own collation data, combined with Pickering's collation of f 35, Tischendorf's transcription of codex C,‡ checked against Lyon's corrections,§ and Tregelles's transcription* of codex Ξ, checked against Greenlee's corrections†:

While both readings clearly have early support, the RP margin's reading clearly has the advantage in terms of diversity. Indeed, it finds support from early and late Alexandrian witnesses, the Western texttype, and various Byzantine witnesses. Therefore, while we admit that the case is far from certain, on the basis of mildly favorable intrinsic probability and diverse external evidence, we feel that the RP margin reading is the better choice.

On intrinsic grounds, we consider whether Luke himself had any reason to draw additional attention to this phrase. The reunion of the son and his father is certainly a pivotal moment in the story, and with good reason. The son's sin, which was not only against heaven, but also in his father's sight, has put their relationship at stake. Will his father consider him worthy even to be called his son? Will he only take him back as a hired hand? These are real possibilities to the son, and their impact is magnified by the fact that formerly, he enjoyed his father's love and generosity as a son. Luke might have employed the phrase “his own father” precisely because it gets to the intimately personal root of the son's yearning and guilt.

Luke therefore had good reason to use the emphatic term, but one might make the following objection: Wouldn't he have placed ἑαυτοῦ in its usual emphatic position between τὸν and πατέρα? To answer this question, we must look at Luke's general usage. In the RP text, we see the article + reflexive pronoun + noun construction used in Lk 11:21, 13:34, 14:26, and 14:33. Meanwhile, the article + noun + reflexive pronoun construction occurs in Lk 2:39, 12:36, 12:47, 15:5, and 16:5. To our knowledge, Luke does not use either of these constructions in Acts.‡ The data is sparse, but to the extent that we can infer from it, we see that Luke had no problem placing ἑαυτοῦ after the noun it modifies.

Turning to transcriptional probability, the case is similar to those of the other divided Byzantine variants we've reviewed: Scribes with an eye for detail might consider changing an original αὐτοῦ to ἑαυτοῦ for emotional emphasis here, but this requires us to assume not only that scribes would consider such a small change worth making, but also that they would be willing to make it intentionally. A more likely explanation is that ἑαυτοῦ was original, but scribes either dropped the first letter accidentally or subconsciously changed the more emphatic phrase to the more common one while copying the line from memory.

Finally, we consider the external evidence. The NA 28 apparatus lists the MS support for both readings as follows:

To add to this, Pickering cites f 35 in favor of ἑαυτοῦ. To summarize, the RP text reading seems to have support from part of the Byzantine tradition, the lectionary text, and witnesses commonly identified as Western and Caesarean, while the RP margin reading has the support of f 35, the early Alexandrian tradition, and a few prominent Caesarean witnesses. We conclude that on external grounds, the RP margin reading is acceptable, and on internal grounds, it is preferable.

Robinson's extensive work collating MSS of the PA and studying the passage's textual history§ has undoubtedly informed his decision to favor the μ 5 profile. On his reputation as a scholar, we are confident that between the full detail of his collation and his argumentation, he will offer a thorough and compelling defense of this decision. At the time of writing, however, his research has not been published. Robinson has expressed that he finds the μ 5 profile the most compelling on both internal and external grounds, and he has summarized the transmissional considerations behind his decision to part ways with Κ x in the PA as follows:

We agree that liturgical factors played a role in the inclusion and exclusion of the PA and that certain scribes copying non-lectionary MSS were influenced by the lectionary tradition. But how extensive was this influence, both in terms of how many scribes it reached and in terms of how much change it introduced to the text? While it might be acceptable to suppose that at one time, the exemplars of most MSS lacked the PA, it still seems a stretch to say that the majority of scribes copying them turned to the lectionary text, rather than some other source, to fill the gap. Indeed, Robinson has pointed out that the most dramatic lectionary element within the PA is the truncated version that dominates those lectionaries that have it,† but this shortened form has little in common with the μ 6 text. Lectionary influence, therefore, seems to have more to do with the inclusion and the exclusion of the passage as a whole and less to do with the amount of textual variation occurring within the passage.

There may well be other factors involved that make μ 6 secondary to μ 5, but we will likely have to wait for the publication of Robinson's research before we can address this issue adequately. Given the limited amount of information we have in the meantime, and given our support for the Κ x family everywhere it is defined outside of the PA, we consider it our safest option at this time to favor the μ 6 profile in the PA. We have therefore tentatively adopted the readings of the RP lower text for this passage, except for the following locations where we have swapped the reading of the RP lower text with that of its margin:

Starting with intrinsic probability, John's stylistic preferences are either ambiguous or heavily in favor of ἴδε, depending on which works we consider Johannine. In the Byzantine text of the fourth gospel, excluding this variation unit, ἰδοὺ occurs 4 times, while ἴδε occurs 18 times. Neither term appears in the epistles attributed to John, so these texts do not change the picture. In Revelation, however, ἰδοὺ is used almost exclusively. Hence, if the same author composed the fourth gospel and Revelation, then he may have used ἰδοὺ and ἴδε interchangeably. Of course, this is already at least somewhat true in the gospel by itself, as John does use ἰδοὺ occasionally throughout it, and in particular, he uses it one verse after the variant in question. So even if the fourth gospel and Revelation share the same author, the change in usage might be accounted for as a change in writing style over the years between the writing of both works. To the extent that John's usage in the gospel reflects his style at the time he authored it, ἴδε seems the reading he was more likely to have used.

Another concern on intrinsic grounds is authorial intention. Since there is no doubt under the assumptions of Byzantine priority that the related phrase in 19:27 is ἰδοὺ ἡ μήτηρ σου, it is reasonable to suppose that John would have used the same word in both statements. Yet it should also be observed that these statements do not express two parallel truths, but rather, two complementary sides of the same truth: The beloved disciple is to care for Mary as a son cares for his mother. The use of distinct words conveying the same idea might be fitting in this usage. Additionally, while John records Jesus as addressing Mary with the vocative γύναι, he records no comparable address to the disciple in the next verse, so he was evidently not interested in presenting the statements as word-for-word parallels. We conclude that while John would have had a very easy reason to use ἰδοὺ given the context, he might have had more subtle reasons for using ἴδε.

In connection with these intrinsic factors, transcriptional probability clearly favors ἴδε. If John had originally employed ἰδοὺ in both this unit and the related unit in verse 27, then later scribes would have had little occasion, intentional or accidental, to change either occurrence. Moreover, had they intended to change ἰδοὺ to ἴδε, they surely would have done so in both cases and not just in one. In contrast, if John wrote ἴδε here and ἰδοὺ in verse 27, then scribes could easily have assimilated one phrase to the other, either intentionally (for stylistic reasons) or inadvertently (possibly supplying the expected word while copying from memory). This explanation finds further support in the fact that assimilations in both directions are attested in the MS evidence.

Finally, we turn to external evidence. Given our previous points on the intrinsic and transcriptional evidence, we are interested not only in the MS evidence for the variant at hand, but also in the evidence for the related variant in the next verse. Based on the NA 28 apparatus, Pickering's collation of f 35, and Welsby's collation of f 1 in John,‡ we have the following support for all combinations of these readings:

Of the two Byzantine readings, ἰδοὺ … ἰδοὺ certainly has earlier and more widespread support. Of course, if the supplement to majuscule D has accurately preserved the codex's original reading, then both readings can be assigned early origins. As for diversity, the agreement of f 35, part of the Κ x group, and a prominent Western witness is weaker than the agreement of part of the Κ x group and some Caesarean or late Alexandrian representatives, but we would not consider it much weaker. Moreover, the fact that f 35 preserves the non-assimilating combination ἴδε … ἰδοὺ also merits serious consideration if this textual family is supposed to be recensional and inclined towards smoother readings. These external criteria therefore favor ἰδοὺ for the variant in question, but not decisively.

As our discussion to this point has made clear, the evidence in this variant does not lend itself to a straightforward interpretation. Based on the clarity and strength of transcriptional probability, we prefer the RP margin reading ἴδε here, but we hasten to add that a fuller collation of witnesses in this passage could convince us to reverse our decision.

The primary arguments for and against the RP text reading καυθήσωμαι concern its grammatical validity in Paul's usage and in the usage of the scribes who copied his writings. The future subjunctive has been criticized as a superfluous form more fitting with the Greek of the Byzantine era than with the Koine of Paul's day.* Yet more recent linguistic analysis has suggested that the tense's apparent grammatical absurdity may be exaggerated, and besides this, this form can be traced back to the fourth century through patristic citations.† The RP text reading, therefore, might not have been too far outside of realm of Paul's usage. Indeed, a case could be made on transcriptional probability that Paul originally wrote καυθήσωμαι, but early scribes concerned with proper form changed this to the similar, but more appropriate future indicative καυθήσομαι. Alexandrian scribes, meanwhile, might have made a one-letter “correction” from this same form to καυχήσωμαι based on Paul's style.‡ In terms of external evidence, καυθήσωμαι has no support from early extant MSS, but, as noted above, it is found in a number of early patristic citations.

The RP marginal reading καυθήσομαι avoids the grammatical problems associated with the future subjunctive, but as it is indicative rather than subjunctive and directly follows ἵνα, it poses a potential grammatical problem of its own. Still, this problem is much more superficial than the problem of the future subjunctive, and there are other places in the Byzantine text where Paul has arguably employed the ἵνα + future indicative construction.§ Indeed, the presence of ἵνα provides a simple and compelling transcriptional explanation for the reading καυθήσωμαι: Early scribes less acquainted with the grammatical norms of their day* and later scribes for whom the future subjunctive was acceptable usage could easily have changed the future indicative to a subjunctive, perhaps even without thinking about it, after seeing ἵνα. In terms of external evidence, καυθήσομαι enjoys the support of the fifth-century majuscules C and D, manuscripts in the Latin tradition going back to the second century, and a wide array of patristic citations going back to the second century.† In addition, Pickering cites f 35 as supporting καυθήσομαι, and his summary statistics from the Text und Textwert collation for this variant list 50.6% of MSS as supporting καυθήσομαι and 44.7% as supporting καυθήσωμαι.

While recent study has effectively challenged the notion that Paul would never have used the future subjunctive καυθήσωμαι, we find more convincing evidence of priority in the earlier extant support for καυθήσομαι. Additionally, we have trouble accepting the transcriptional argument for the development of καυθήσομαι from καυθήσωμαι, as this would attribute a deliberate, grammatically-motivated change to over half of the manuscript tradition. What the evidence does show is that καυθήσωμαι surely arose much earlier than the time of Byzantine Greek.

We prefer καυθήσομαι on the basis of its early and diverse external support and due to the simplicity of its argument for transcriptional priority. In light of the evidence suggesting an early origin for καυθήσωμαι, it would be plausible to suppose that an original καυθήσομαι gave rise to καυθήσωμαι before the third century, likely by inadvertent means, and that not long after this reading began to propagate, it fell into the hands of an educated scribe who emended it to καυχήσωμαι for grammatical and stylistic reasons. Later scribes versed in Byzantine Greek might then have provided multiple independent avenues of change from καυθήσομαι to καυθήσωμαι, which would help explain the large presence of the latter reading in the MS tradition.

Intrinsically, the case seems straightforward: Since the writer did not use the double augment in 2 Cor 11:1, since he does not appear to be quoting the work of any other author, and since we are not aware of any contextual rule that would have led him to use the double augment here, the simplest conclusion is that he wrote ἀνείχεσθε consistently. One might argue that because ἀνέχω was rare in the NT and because its conjugation in the imperfect was even rarer, the writer might not have had an established spelling for this form and therefore switched back and forth without thinking about it, but given the proximity of the two occurrences, such a scenario seems unlikely.

The best case that can be made for the spelling with the double augment is transcriptional. If the original spelling had been ἠνείχεσθε, then the immediate context of the chapter, with its frequent use of the present tense form ἀνέχεσθε and the imperfect tense spelling ἀνείχεσθε, would have created several potential avenues for assimilation. The minority reading ἀνέχεσθε, which we regard as secondary on Byzantine-priority principles, demonstrates the plausibility of this explanation. In the opposite direction, one could ascribe the use of the double augment to inadvertent modernization of spelling on the part of one or more early, influential scribes. The biggest difficulty with this explanation is the same one described in relation to the intrinsic evidence: The two occurrences of the verb in this form are just a few verses apart, so it would be odd for the scribes responsible not to have noticed the inconsistency they had created. Transcriptional probability is therefore somewhat unclear. The easier explanation is this case is that in favor of ἠνείχεσθε, but the explanation in favor of ἀνείχεσθε coheres better with the intrinsic evidence.

The external evidence, meanwhile, points plainly in the direction of ἀνείχεσθε. While a large part of the Byzantine tradition and f 35 read ἠνείχεσθε, the NA 28 apparatus lists only two of its consistently-cited witnesses, F vid and Ψ, as supporting this reading.‡ In contrast, the RP marginal reading ἀνείχεσθε is supported by the following Greek witnesses: P 34, א, D 2, G, H, K, L, P, 0121, 0243, 81, 104, 365, 630, 1175, 1241, 1505, 1739, 1881, and 2464. Given the age and the overwhelming diversity of its support, ἀνείχεσθε is clearly superior to ἠνείχεσθε on external grounds. This, in combination with strong intrinsic evidence and acceptable transcriptional evidence, compels us to prefer the RP margin reading.

On intrinsic grounds, ἡμῶν fits well with the context, as John speaks in the first person in the preceding two verses. But the second person seems to fit in a deeper way. John's primary intent in this letter seems to be to reassure his readers of their good standing (although this intent does not preclude reminding them to stay on track, as in 2:1 and 5:16-21), so the emphasis ὑμῶν places on the readers' faith (and hence, their experience of victory) is particularly fitting as John begins to wrap up his letter. While this may seem to depart from the usage of the first person in the immediate context, it is worthwhile to remember that the verse numbers and paragraph breaks of modern editions were not a part of the original text; if an edition moved the second half of verse 4 to the beginning of the next paragraph, this passage, including its shift to address the readers, would serve as a reasonable transition from the topic of victory in faith to the topic of the object of that faith.§

On transcriptional grounds, either reading could have given rise to the other through itacism, but for scribes who copied from a lector or who committed to memory longer passages from their exemplars and then recited these passages to themselves as they copied, the preceding verses would have made ἡμῶν the more likely choice by far. Outside of this inadvertent source of error, a minority of scribes might have intentionally “corrected” an original ὑμῶν to ἡμῶν to assimilate to the preceding verses. Lectionary influence may have been an additional factor behind the adoption of ἡμῶν, as the first-person plural would have been more appropriate for public reading. Piety might even have led some scribes to adopt ἡμῶν, so as to leave no question of John's own faith. Given the ease of the transition from ὑμῶν to ἡμῶν in all cases, it is perfectly justified to expect this change to have taken place multiple times independently, which would explain the widespread MS support for ἡμῶν.

Turning finally to external evidence, we see from the NA 28 apparatus that much of the early MS evidence backs ἡμῶν.* Pickering also lists f 35 in support of this reading. Yet ὑμῶν does find early support in the fifth-century majuscule 048 and in some MSS of the Vulgate, and it has reasonably diverse support among later witnesses, as well. According to Pickering's summary statistics on the Text und Textwert collation for this variant, 56.4% of witnesses read ἡμῶν, but 43.2% read ὑμῶν.† Given the intrinsic likelihood of an original ὑμῶν, the transcriptional probability of numerous independent changes to ἡμῶν, and an otherwise normal model of transmission, this picture is not too unexpected: The secondary reading has taken over the majority, but a strong proportion of witnesses has still managed to preserve the original reading. To the contrary, if ἡμῶν had been original, then it would be quite surprising, under the same transmissional assumptions, to see as much support for the harder reading as we do.

It should go without saying that our text-critical decisions (especially at Lk 7:6, the PA, and Jn 19:26) reflect our estimation of the evidence presently available to us, and as such, they are subject to refinement as new hypotheses, arguments, and evidence become available. It is our heartfelt desire that the NT textual criticism community will find a renewed interest in the Byzantine text and the variants that divide it, and as a result, engage in fresh efforts to fill these gaps.