How “Byzantine” is the ECM?

How much has the Editio Critica Maior moved toward the Byzantine text? The question received fresh attention at the 2026 CSNTM Text & Manuscript Conference, where Nelson Hsieh explored how often the ECM and the Tyndale House Greek New Testament move from the text of NA27 toward Byzantine readings. He framed the question memorably as one of “Two Ships Sailing Toward Byzantium,” asking how far, and in what sense, these two editions have “sailed” toward Byzantine readings.¹

This topic is especially timely in 2026, as the United Bible Societies’ Greek New Testament, 6th edition, incorporates the textual changes of the published ECM volumes for Acts, Mark, and Revelation, and its text is identical with the forthcoming Nestle-Aland 29th edition.

The question is not simply how many Byzantine readings appear in the ECM, but what those agreements mean for the broader reevaluation of the Byzantine tradition. The ECM editors themselves place this development against the background of the longstanding anti-Byzantine bias in modern textual criticism. As Klaus Wachtel explains, “Since the Textus Receptus was overcome by the scholarly textual criticism of the 19th century, there [has been a] tenacious negative bias against the Byzantine majority text.”² If the ECM represents a reevaluation of that tradition, then asking how far it has moved toward the Byzantine text requires more than tallying readings: we first have to determine what counts as agreement between the Byzantine text and the initial text.

The increased significance accorded to the Byzantine tradition in the ECM is something I have discussed on several occasions.³ Most recently, in an article with Annette Hüffmeier on the history of modern critical editions, we described how the “newfound appreciation for Byzantine readings” is among five aspects that make the ECM “revolutionary in nature” and sets it apart from twentieth-century scholarship.⁴ We characterized the recovery of potentially early readings in the Byzantine tradition as one of the hallmarks of the edition and trace the increasingly important role of Byzantine readings in the ECM.

Against that background, the movement of the ECM toward greater acceptance of Byzantine readings is not what I intend to establish here. Rather, I want to revisit how agreement between the initial text and the Byzantine text is measured in the ECM. Using the data in the apparatus, I will consider several different ways of counting that agreement. Ultimately, I hope to show that the ECM may be even closer to the Byzantine tradition than previous figures have suggested. For this endeavor, I will focus on ECM Acts, though the same methods could be applied to other ECM volumes.

1. How the Byzantine Text is Represented in ECM Acts

Before a count of agreements between the Byzantine text and the initial text can be made, first the “Byzantine text” has to be defined: what exactly does “Byz” represent in ECM Acts? There exist printed editions of the Byzantine text such as Robinson-Pierpont or Hodges-Farstad, but the ECM does not simply adopt an existing Byzantine edition for comparison; each volume of the ECM constructs its own representation of the Byzantine text.⁵

ECM Acts includes 183 Greek manuscripts (including supplements that are counted separately, such as 1739S and 323S).⁶ Of these, seven serve as representatives of the Byzantine text: 1, 18, 35, 330, 398, 424, and 1241. The edition describes them as “seven nearly pure Byzantine manuscripts.”⁷ These representatives were selected for ECM Acts on the basis of Text und Textwert, where they showed the highest levels of agreement with the majority readings.

When these seven agree, the siglum Byz is used in the apparatus as a group symbol, representing the unity of the Byzantine textual tradition. In addition, 72 other witnesses are subsumed under Byz unless they differ from the group reading.⁸ This allows the edition to concentrate on manuscripts that differ most from the majority tradition.

But what happens when one or more of the Byzantine representatives departs from the others? The introduction to ECM Acts explains:

If one or two of these manuscripts read apart from the remaining ones, these latter represent the Byzantine reading. If two of the listed manuscripts differ from the rest in two different variants, the text of the rest is regarded as the Byzantine reading. If, however, two manuscripts from the list above differ from each other and from the rest, it is likely that these two represent a separate strand of the Byzantine tradition. In this case and also, if three or four differ from the remaining ones, the Byzantine witness is divided and the symbol Byz is relinquished.⁹

When the seven Byzantine representatives disagree with each other in this way, the siglum “Byz” will not be cited in the apparatus. Instead, each of the seven representatives will be cited individually at their respective variants.

2. Comparison of the Byzantine Text and Initial Text

There are 7,445 variant passages in ECM Acts, but at only 622 of these does the Byzantine text differ from the initial text—roughly 8%.¹⁰ Put differently, the initial text of ECM Acts agrees with the Byzantine text at more than 90% of the variant passages. This is why the edition itself can say that, apart from split-guiding lines, “the Byzantine is generally identical with the primary line text.”¹¹

2.1. New Initial Readings

This close relationship between these two is also evident in the new initial readings adopted in ECM Acts. Of the 52 new initial readings compared to the text of the NA28, 36 are attested by Byz (e.g., 1:10/34-36).¹² By contrast, there are only 4 places where a reading that is Byzantine in NA28 is changed away from the Byzantine text in ECM Acts (e.g., Acts 10:9/10). In other words, when ECM Acts departs from the NA28, the movement is overwhelmingly toward, rather than away from, Byzantine readings.

2.2. Split-Guiding Lines: What Counts as Agreement?

Determining precisely how closely the ECM aligns with the Byzantine text becomes more complicated when split-guiding lines are taken into account. ECM Acts has 155 split-guiding lines: places where the editors do not choose a single initial reading but offer two or more readings as equally possible.¹³

 When the text of the NA28 is one of the options at a split-guiding line, the editors designate it variant a for practical purposes, while the competing reading is assigned another letter—often b, though not always.

Here is the apparatus for Acts 3:26/6-10, where the text of the NA28 is variant a, and the alternative reading, marked with a diamond, is variant b. Byz attests variant b.

In theory, it does not matter which reading is designated variant a at a split-guiding line, since the alternatives are given equal priority. In this situation, where Byz attests variant b, can we say the initial text and the Byzantine text agree?

In the following I will give four possible approaches to handling this question to highlight the complexities of counting Byzantine agreement in the ECM.

2.2.1. Counting Only Variant a as Agreement

The first approach is the one the ECM follows. If a split-guiding line is comprised of two variants, variant a and variant b, and Byz attests variant b, does this count as the initial text agreeing with Byz?

In the ECM, the calculation works as follows: when Byz attests variant a (e.g., Acts 20:22/28), it is counted as agreeing with the initial text; when Byz attests another variant, even one included in the split-guiding line (e.g., Acts 20:14/6), it is counted as disagreeing with the initial text.¹⁴

Of the 155 split-guiding lines in ECM Acts, Byz attests variant a 20 times (e.g., 7:19/18) and is counted as an agreement between Byz and the initial text. Conversely, at the remaining 135 split-guiding lines, Byz does not attest variant a (e.g., 1:6/10; 2:7/5), and these are counted as disagreements between Byz and the initial text. Of these 135 cases, Byz attests the alternative reading included in the split-guiding line 106 times, attests another variant not included in the split-guiding line 4 times, and is itself split and therefore not cited 26 times.

In this scenario, the ECM counts 622 disagreements between Byz and the initial text of ECM Acts, of which the 106 times Byz attests the alternative reading in a split-guiding line are included. (See ECM Acts, 2:9–15 for a list of the 622 passages, where each of the 106 split-guiding lines are marked with an “S”.)

2.2.2. Counting Any Reading in the Split-Guiding Line as Agreement with Initial Text

A second approach to determining agreement between the Byzantine text and the initial text would be to count the alternative reading in a split-guiding line as the same as the initial text. This approach is valid because ECM Acts itself describes the readings offered in a split-guiding line as “equal to a” and refers to the non-a reading as an “alternative” to a.¹⁵ If these readings are truly regarded as equally possible reconstructions of the initial text, then Byz should not be counted as disagreeing with the initial text when it supports the alternative rather than variant a.

At 106 split-guiding lines, Byz attests such an alternative reading. If these are counted as agreements rather than disagreements, the number of differences between the Byzantine text and the initial text falls from 622 to 516. By this calculation, there are only 516 differences between the Byzantine text and the initial text.

2.2.3. Excluding Split-Guiding Lines from the Comparison Altogether

A third approach would be to simply exclude split-guiding lines from the comparison. This shifts the question away from how a Byzantine reading should be counted within a split-guiding line and limits the comparison to passages where the ECM presents a single initial reading.

As discussed above, of the 155 split-guiding lines in Acts, 106 are where Byz attests the alternative to variant a. There are an additional 4 passages (listed below) where Byz does not attest any of the readings in a split-guiding line. These 110 (106+4) are among the 622 differences between Byz and the initial text listed in the ECM. Therefore, removing these split-guiding lines from consideration leaves 512 differences (622-110) between Byz and the initial text among the 7,290 passages of Acts (7,445-155) with a single initial reading.¹⁶

The underlying issue is therefore not primarily the status of the Byzantine text, but the status of a split-guiding line itself: should a passage in which the editors present two or more readings as equally possible be treated as having an “initial text” for the purposes of calculating agreement? Can a split-guiding line be considered “initial text” in principle, since it is not a singular “text” but instead “texts”? If a split-guiding line cannot be regarded as the initial text, then a calculation of agreement between Byz and the initial text should arguably exclude such passages.

2.2.4. Looking Behind the Byz Siglum

A fourth approach changes not how the ECM presents the Byzantine witnesses, but how their agreement with the initial text is counted. Under the ECM’s current practice, when the “seven nearly pure Byzantine” representatives disagree among themselves, the group siglum Byz is not used; instead, the seven witnesses are cited individually in the apparatus. For the purpose of calculating agreement, however, these divided cases need not simply drop out of consideration. Rather, one can ask whether at least one of the seven Byzantine representatives supports the initial reading—or, in the case of a split-guiding line, one of the readings regarded as equally viable.

Acts 7:31/8–14 provides a good example of how this approach would work. Here the seven Byzantine representatives divide among several readings, so Byz cannot be cited as a group. I have circled the seven representatives in red:

Three of the seven representatives support variant a (1, 398, 1241), one supports b (424), and three support e (18, 35, 330). Because the seven do not agree, Byz cannot be cited as a group. Nevertheless, three of the Byzantine representatives attest the initial reading. Under this fourth approach, the passage could be counted as an agreement between (part of) the Byzantine tradition and the initial text, even though the group siglum Byz does not appear in the apparatus.

Applying this fourth approach more broadly produces a striking result. In most passages in Acts where the seven representatives divide, at least one of them still supports the initial reading—or, in the case of a split-guiding line, one of the readings regarded as equally possible. In fact, there are only 26 passages in all of Acts where none of the seven Byzantine representatives supports the initial reading or any of the possible initial readings in a split-guiding line.¹⁷

1:13/30-38
2:38/2-12
3:22/4-6
4:5/22-34
5:34/30-40
5:41/22-28
7:21/2-10
8:7/2-22
9:12/6-10
9:28/28-30
10:5/20-26
10:48/8-18
11:9/6-16
13:23/18-26
15:7/36-44
16:12/12-16
17:5/2-22
19:2/24-26
20:15/28-30
21:5/56-6/6
21:14/14-22
22:24/4-10
23:8/14-26
23:35/22-34
25:18/16
28:16/12-16

To this list can be added the 4 split-guiding lines mentioned above where Byz is cited in the apparatus, but not at one of the readings in a split-guiding line. This would bring the total to 30 passages:

2:36/18-30
10:24/2-8
24:24/22-24
26:17/26-28

Put differently, at 7,415 of the 7,445 variant passages in Acts, Byz—or at least one of its seven representatives—supports the initial reading or one of the possible initial readings in a split-guiding line. In this approach, the Byzantine text differs from the initial text at only 30 passages.

3. Summary: Four Ways of Measuring Byzantine Agreement with the Initial Text

The preceding discussion suggests four different ways of measuring agreement between the Byzantine text and the ECM initial text. The resulting figures differ because each method answers a slightly different question about what should count as agreement.

(1) The ECM’s conventional count is the strictest: at a split-guiding line, Byz is counted as agreeing with the initial text only when it attests variant a.

(2) Alternatively, if Byz supports one of the readings offered in a split-guiding line, that can be counted as agreement, since the readings are regarded as equally possible.

 (3) A third approach excludes split-guiding lines altogether and compares Byz only at passages where the ECM presents a single initial reading.

(4) Finally, when Byz cannot be cited in the apparatus because the seven Byzantine representatives disagree, if any of the seven representatives attest a reading offered as the initial text, or in a split-guiding line, this can be counted as an agreement between the Byzantine text and the initial text. This latter option is the most liberal in terms of what constitutes a Byzantine reading.

These four approaches produce markedly different numbers of disagreements:

4. Implications for Readers of the UBS Greek New Testament

The different approaches outlined above are not merely a matter of how the data is tabulated. They also affect how readers encounter the text of the ECM in editions derived from it. This is especially important for users of the United Bible Societies’ Greek New Testament, where some of the distinctions that are visible in the ECM are necessarily less apparent.

The new edition of the UBS GNT incorporates the ECM text of Acts, Mark, and Revelation, but as a compact edition it cannot reproduce all of the information available in the ECM. Of particular relevance here, the UBS apparatus does not contain an entry at every passage where the ECM has a split-guiding line. When a split-guiding line coincides with an apparatus entry in the UBS, a diamond signals the division in the ECM; elsewhere, however, the reader may have no indication that another reading was regarded as equally possible.

For example, in ECM Acts 3:25, there is a split-guiding line for ὑμῶν and ἡμῶν, and a split-guiding line at 3:26 for ἀναστήσας ὁ θεός and ὁ θεὸς ἀναστήσας. In the UBS, pictured below, there is an apparatus unit at 3:25 (with corresponding diamonds) but no apparatus at 3:26.

At Acts 3:26, the reader would not know that there is a split-guiding line and that one of the textual alternatives is attested by the Byzantine text.

5. Conclusion

So how Byzantine is ECM Acts? By the ECM’s conventional count, Byz differs from the initial text in 622 passages. I have argued that this number depends on how agreement is defined. At split-guiding lines, Byz may support a reading that the editors regard as equally possible and still be counted as a disagreement simply because it does not attest variant a, as happens at 106 split-guiding line passages in Acts. And when the seven Byzantine representatives divide, the siglum Byz disappears even though one or more of those witnesses may still support the initial text.

Once those conventions are taken into account, the relationship between the Byzantine text and the initial text appears even closer. Most strikingly, there are only 30 passages in Acts where none of the seven Byzantine representatives support the initial reading or, in the case of a split-guiding line, one of the readings offered as equally possible. I am not suggesting that 30 should replace 622. The two figures answer different questions. The comparison shows that ECM Acts can be considered even closer to the Byzantine tradition than the usual figure of 622 differences might suggest.

Two conclusions emerge from this investigation. First, determining just how far the ECM has “sailed toward Byzantium” is far from straightforward. I have identified four distinct ways of counting Byzantine agreement in the ECM. Each captures something useful, but each also leaves other evidence out of view. The result depends in part on which aspect of the data one chooses to emphasize: the group siglum Byz, the readings offered at split-guiding lines, or the individual Byzantine representatives when the group itself is divided. Looking at the evidence from these different angles shows why no single figure can fully capture the complex relationship between the Byzantine tradition and the ECM initial text.

Second, the calculations presented here suggest that the degree of Byzantine agreement in ECM Acts is much greater than its conventional numbers indicate. Because the reevaluation of the Byzantine tradition is one of the defining features of the ECM, this finding merits further study, both in Acts and across the other ECM volumes. A more comprehensive analysis of this question could refine our understanding of the place of the Byzantine text within the broader New Testament textual tradition and, consequently, clarify the role that Byzantine readings may play in reconstructing the earliest attainable text.

1. Nelson S. Hsieh, “Two Ships Sailing Toward Byzantium: How Much do the ECM and THGNT Sail Towards the Byzantine Text?” paper presented at the CSNTM Text & Manuscript Conference, May 29, 2026. ↩︎
2. (ECM Acts) Novum Testamentum Graecum: Editio Critica Maior, vol. III, The Acts of the Apostles, ed. Holger Strutwolf, Georg Gäbel, Annette Hüffmeier, Gerd Mink, and Klaus Wachtel (Stuttgart: Deutsche Bibelgesellschaft, 2017), 1:30*. See also Klaus Wachtel, “On the Relationship of the ‘Western Text’ and the Byzantine Tradition of Acts—A Plea Against the Text-Type Concept,” in ECM Acts, 3:137–48. ↩︎
3. See, for example, Holger Strutwolf, Gregory S. Paulson, and Klaus Wachtel, “Text-Critical Commentary,” in Novum Testamentum Graecum: Editio Critica Maior, vol. I, The Synoptic Gospels, book 2, The Gospel of Mark, ed. Holger Strutwolf, Georg Gäbel, Annette Hüffmeier, Marie-Luise Lakmann, Gregory S. Paulson, and Klaus Wachtel (Stuttgart: Deutsche Bibelgesellschaft, 2021), 3:8–34; Gregory S. Paulson, “Byzantine Lectionary Manuscripts and Their Significance for Biblical Textual Criticism,” in The Oxford Handbook of the Bible in Orthodox Christianity, ed. Eugen J. Pentiuc (Oxford: Oxford University Press, 2022), 112–32; Gregory S. Paulson, “An Investigation of the Byzantine Text of the Johannine Epistles,” Review and Expositor 114.4 (2017): 580–9. See also Gregory S. Paulson, “ECM Mark has Arrived,” INTF Blog, August 13, 2021. ↩︎
4. Annette Hüffmeier and Gregory S. Paulson, “Critical Editions of the Greek New Testament from the Twentieth to Twenty-First Century,” pages 657–87 in The Oxford Handbook of Textual Criticism of the Bible, ed. Sidnie White Crawford and Tommy Wasserman (Oxford: Oxford University Press, 2026), 672 and 669. ↩︎
5. For an overview of various reconstructions of the Byzantine text, see Gregory S. Paulson, “An Investigation of the Byzantine Text of the Johannine Epistles,” Review and Expositor 114.4 (2017): 580–9. ↩︎
6. ECM Acts, 2:5–6, 7. ↩︎
7. ECM Acts, 2:8. ↩︎
8. Cf. ECM Acts, 2:8. ↩︎
9. ECM Acts, 2:8. ↩︎
10. For a list of all 622 passages see ECM Acts, 2: 9–15. The number 7,445 is based on the CBGM phase 5 for Acts: https://ntg.uni-muenster.de/acts/ph5/. I have put numbers in bold that are important for the calculations below. ↩︎
11. ECM Acts, 2:8. ↩︎
12. For a list of new initial readings, see ECM Acts 1:34*–35*. ↩︎
13. For a list of split-guiding lines see ECM Acts, 1:35*–37*. ↩︎
14. This practice can be seen in the list of 622 Byzantine variants where split-guiding lines are not included when Byz attests variant a. Cf. ECM Acts, 2:9–15. ↩︎
15. ECM Acts, 1: 18*, 31*. ↩︎
16. The remaining split-guiding lines not accounted for in the number 110 consist of 25 where Byz cannot be cited because the seven representatives disagree, and 20 where Byz attests variant a. ↩︎
17. Excluding the inscriptio, subscriptio, and the orthographic readings at 5:1/14; 7:23/10; 7:42/50–62; 9:35/20; 16:26/24–28; 16:27/26–34. ↩︎