A working theory of Bezae’s text

The aim of this post is to outline a working theory of Bezae’s text in the gospels and Acts, starting with five well-known features of Bezae’s text:1

  1. Isolation from the Greek mainstream
  2. Old Latin parallels
  3. Parallels with ancient writers
  4. East-West exchange of readings
  5. Major variation in Acts

Isolation from the Greek mainstream

Bezae’s Greek text is composed of three primary layers assembled at the end of the fourth century (c. 385), each layer representing a distinct source of mixture:2

  1. A Greek base, perhaps connected to the library at Caesarea
  2. A selective retroversion of Latin readings from various Old Latin source traditions
  3. An upper “scribal” layer (not actually by Bezae’s scribe)3

Bezae’s mixture of layers obscures its individual component traditions and hides its mainstream relationships: A significant side effect of layering is that Bezae’s text as a whole appears isolated from the mainstream tradition.4 But if we partition Bezae’s readings into natural layers, these individual layers can be related to the mainstream tradition.5 So when Old Latin and “singular” readings are set aside, Bezae has a very mainstream Greek text, with parallels to Family 1, Origen, and other texts associated with Caesarea.

Final layer combination dating to c. 385: Bezae’s three immediate layers were brought together at the end of the fourth century (c. 385) based on the following considerations:

  1. Bezae’s text must post-date the late, so-called “European” or “northern Italian” form of the Old Latin tradition (350–380) to which it is partially assimilated.6 Bezae’s close relationship with this late form of the Old Latin is most apparent in Mark.7
  2. Bezae’s upper layer exhibits a well-documented interest in enhancing Peter’s role in Acts.8 This same interest is found in Bezae’s Corrector G (see post), suggesting that Bezae’s upper layer was produced at the same time as the manuscript and exemplar, that is, close to its initial correction.9
  3. Documented tendencies detected in Bezae’s upper “scribal” layer, especially the augmentation of Peter, fit a late fourth-century context (see post).
    1. Pro-Petrine tendencies – driven by the promotion of Constantinople to second rank among apostolic sees at the Council of Constantinople (381) (see post).
    2. Anti-ascetic tendencies – a response to Jerome (380’s) or possibly Jovinian (390s)
    3. Anti-Judaic tendencies – motivated by church orders segregating Christians and Jews, reflecting marginalization of Jews in imperial code (see post)
    4. Other tendencies – all documented tendencies fit a context of c. 385
  4. The assumption that the Vulgate revision of the gospels provided some impetus for Bezae’s project as a means to legitimate the Old Latin version (384-385) (see post)
  5. Rome’s mediation in the Antiochian schism as a background for shared readings in the Latin and Syriac traditions as well as between Bezae and other “Western”-influenced Greek MSS (378-397).10

Bezae’s Greek base with Caesarean connections: Bezae’s Greek base had an independent history before being appropriated by Bezae’s producers in the final two decades of the fourth century.11 Several studies suggest that Bezae’s Greek base has connections with Caesarea in Palestine as a possible provenance.12

Bezae’s Latin column a composite of Old Latin texts: Bezae’s Latin column reflects a hybridization of Latin texts, “African” and “European,” according to the traditional nomenclature, which accounts for its independence within the Latin tradition.13

Old Latin parallels

Latin assimilation a major process in Bezan Greek text: Bezae’s distinctive parallels with the Old Latin version result from assimilation to the Latin version through a process of selective retroversion (on assimilation theories, see my survey of Bezan theories).14

Bezae’s Greek text a partial retroversion of various Latin versions: Bezae’s Greek column reflects a process of selective assimilation to one or more strands of the Old Latin version. The Greek column may not be derived directly from the Latin column (at least as written), though it shares the same readings.15 A proposed process is as follows:

  1. Multiple Latin versions were combined to create an archetypical Latin column
  2. Latin column archetype translated into Greek and used to correct a Greek base text
  3. Resulting Greek and Latin columns mutually corrected to agree, obscuring the dependence of the Greek column on the archetypical Latin column.

Parallels with ancient writers

Ancient parallels in Bezae are direct or indirect echoes of second- or third-century writers: Bezae’s parallels with ancient writers reached its text through one of two routes:

  1. Incorporation of parallels already in the Old Latin version indirectly through assimilation to this version (see post).
  2. Deliberate archaizing of Bezae’s text form to agree with authoritative ancient writers, such as Tertullian, Cyprian, Victorinus, and Irenaeus.16

The implication is that no second-century writer had access to a so-called “Western” or pre-“Western” text. Readings shared by Bezae with ancient writers and the Old Latin tradition were likely assimilated into Bezae’s text through the Old Latin tradition. Particularly vivid parallels with ancient writers and parallels not shared with the Old Latin tradition may have resulted from deliberate archaizing of the fourth-century text.17

East-West exchange of readings

Western mediation in the Antiochian schism (378-397) a plausible background for Bezae’s text: The exchange of readings responsible for Bezae’s Latin and Syriac parallels and possibly also its parallels with Codex Sinaiticus (א) in John 1–8 and with P127 in Acts occurred towards the end of the fourth century in the aftermath of the Arian controversy, when the East faced a struggle to rebuild its institutions (on East-West exchange, see post). The Antiochian Schism (362-397) provides a plausible backdrop for the introduction of Old Latin readings from West to East by representatives of Rome during the period of mediation (378-397), though important work may have occurred in Caesarea in Palestine.18 Basil of Caesarea (d. 379) explicitly sought Rome’s involvement in mediating the Antiochian schism between rival orthodox parties (Epistles 70, 92). In the subsequent period, numerous delegations were sent in both directions between Rome and Antioch, providing a promising setting for the exchange of texts.19

Major variation in Acts

Bezae’s Acts text establishes an apostolic precedent for recognition of Petrine primacy: Significant rewriting in canonical Acts is a side effect of this book’s importance for establishing apostolic precedent in general and especially recognition of Petrine primacy, in late fourth-century debates. The issue of Petrine primacy was particularly relevant in the case of Rome’s intervention in the politics of an ancient see, such as Antioch, and its rivalry with Constantinople after the Council of 381.

Is the “Western” text really “Eastern”?

It is no secret that the so-called “Western” text, as represented by the distinctive readings of its witnesses,* is not at all confined to the West, but rather may be found across a range of geographical and versional contexts, from the Old Syriac and Coptic versions to various Greek traditions associated with Eastern writers, for example, with Origen or Basil of Caesarea.

G. Zuntz put it well:

“The term ‘Western Text’ is a misnomer — everybody knows that. This characteristic form of the Sacred Text, first discovered in Codex Bezae and the Old Latin, is found in Eastern witnesses as well.” [1]

B. Aland and K. Aland observe with some hyperbole:

“The origin of the ‘Western’ text lies anywhere but in the direction its name would suggest.” [2]

But if the “Western” text is “Western” only in its Latin representatives, what are we to think about its origins? Must they lie in the East — whether in Greek or in Syriac?

In fact, if we follow Zuntz, the “Western” text consists of two layers, both with Eastern origins:

  1. First, there is a base layer containing “remnants of the original text,” which being a Greek tradition, presumably derives from the East.
  2. Then, there is a secondary layer representing “the product of retranslation, from the Syriac, of an ecclesiastical adaptation, namely, the lectionary-text of the oldest Church of Edessa,” which of course, being Edessene, must have arisen in the East as well. [3]

According to Zuntz then, the “Western” text form is essentially an “eastern” text form whose most distinctive, secondary elements derive from Syria.

Certainly, there are elements of the “Western” text that seem to demand an Eastern background. At the same time though, it is difficult to overlook the close association of some of the most distinctively “Western” readings with the Old Latin version. Yet since there is no secure evidence of a Latin version before the early third century, to propose a developed “Western” text in the second century all but requires us to place its origins in the East.

But if we postpone the time frame when the elements of the mature “Western” text form came together to the second half of the fourth century, we are now no longer constrained to an Eastern provenance. In fact, with its reputation for Nicene orthodoxy, the West had certain advantages as a promulgator of influence — and perhaps even of textual influence — to the East from the Synod of Alexandria (362) onward, especially from the perspective of the pro-Nicene party.

Thus we find Basil of Caesarea in Cappadocia [!] turning to Rome for assistance in resolving the Antiochian schism:

“to move some of the Italians to undertake a voyage by sea to visit us.” (Epist. 68; NPNF 2.8, 164)


“It has seemed to me to be desirable to send a letter to the bishop of Rome, begging him to examine our condition, and since there are difficulties in the way of representatives being sent from the West by a general synodical decree, to advise him to exercise his own personal authority in the matter by choosing suitable persons to sustain the labours of a journey … to correct the unruly among us here; able to speak with proper reserve and appropriateness, and thoroughly well acquainted with all that has been effected after Ariminum to undo the violent measures adopted there.” (Epist. 69.1; NPNF 2.8, 166)

But should we expect these worthy representatives from the West to have left their gospels in Italy? Would they have stopped at correcting the “unruly” to also correct unruly texts? Would they have spoken with proper “appropriateness” but not appropriately cite from their own orthodox readings?

Of course, we can only speculate as to the potential impact such contact might have had. Yet we may observe one thing with greater confidence, namely, that by the 370’s there was a flow of influence from West to East that was both initiated and received by sympathetic Easterners. In such a context, we can plausibly imagine Latin readings being imported to the region surrounding Antioch and the great library at Caesarea, where it seems witnesses such as Sinaiticus and Bezae might at one point have resided.

References and Notes

* I added the phrase “as represented by the distinctive readings of its witnesses” in response to James Snapp, Jr.’s insightful comment below.

[1] Günther Zuntz, “The Western Text of the Acts of the Apostles” in Opuscula Selecta; Classica, Hellenistica Christiana (Manchester: University Press, 1972) 189–215 at 189.

[2] K. Aland and B. Aland, The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism (2d ed.; Grand Rapids: Eerdmans, 1987) 67.

[3] Zuntz, “Western Text,” 189-190.

The puzzling case of Codex Sinaiticus in John 1-8

A study of Bezae’s text in John 4:1–42 is attractive for a number of reasons. Not only is John 4 the best-preserved chapter in Origen’s monumental Commentary on John, but it is also the first entire chapter of John in which both Codices Sinaiticus (א) and Bezae (D) preserve a so-called “Western” text.

In fact, the textual character of Sinaiticus in John 1-8 is one the fascinating puzzles of textual research in the gospels. Short of actual historical contact with a tradition very much like that of Bezae, it is difficult to explain the remarkable parallels shared by this normally-solid Alexandrian witness with perhaps the most famously divergent of gospel manuscripts.

In his seminal article on the text of Sinaiticus in John, G. Fee focused on John 4 as the most “Western” of the early chapters of John, a situation that has proved convenient for my own study of Bezae’s layering in John 4:1-42. [1] (For details on this study and its method, see this previous post.) In fact, nine of Fee’s thirteen singular agreements between א and D and four of his seven sparsely-attested agreements are included in my data set. (For Fee’s data from John 4:1-42, see this page.)

As seen in the plot below, with points (o, Δ, +, ×, ◊, ∇) representing D’s readings by layer, Fee’s observations corroborate well with Bezae’s layering in John 4:1-42, where as we might expect, readings attested by א, color-coded in violet, exhibit a significant level of correspondence with D’s two Old Latin layers in the two small ellipses at the lower left:

John 4 - 6 clusters with X01 - special circled

Agreements of א and D in John 4:1-42 (in violet) with layers identified as 1. “Transitional” Greek (o), 2. Greek Mainstream (Δ), 3. “Free” Greek (+), 4. Alexandrian (×), 5. African Old Latin (◊), 6. European Old Latin (∇). A selection of witnesses is listed for each reading.

Certainly, א’s correspondence with D’s Old Latin layers stands out against the remaining Greek witnesses, none of which has any significant representation in the Latin layers, while א is represented at half of the eighteen readings apportioned among the two Old Latin layers.

What is perhaps more striking though, is that the four sparsely-attested א-D agreements singled out by Fee (circled in yellow) do not fit cleanly inside any of the other six layers identified in the study, occupying an outlier position at the fringes of the layer boundaries (in fact, on the perimeters of the blue ellipses in layers 1 and 3). Remarkably then, even after we consider the Alexandrian, mainstream, and two Old Latin layers, א still attests an additional residual layer in common with D.

So from these remarkable parallels, can we say anything more about potential contexts for Bezae or its traditions? For a suggestion, we might turn to Bezae’s correctors:

As it turns out in his study of Bezae’s correctors, D. C. Parker notes a “striking” 62.5% (25/40) agreement between Bezae’s Corrector B and Sinaiticus together with its C-group of correctors, specifically the C[a] and C[b2] correctors, whom A. C. Myshrall places in the fifth to sixth centuries, apparently not long after Bezae’s production in ca 400. [2] Noting the possibility that Sinaiticus was in Caesarea at some point in the fifth or sixth centuries, Parker observes:

“there is evidence [from Corrector B] to suggest that Codex Bezae may also have been in Caesarea, or somewhere susceptible to Caesarean influence, at an early stage in its life, perhaps during the fifth century.” [3]

Parker then suggests that Corrector B had access to both a D*/d-like text and a א[c]-like text, whether combined or as individual manuscripts. [4] Of course,  this plausible coincidence of location is just the kind of evidence we need to begin building a historical framework capable of accommodating the multiple shared layers we find in these two unlikely codices in John 1-8.

Yet there remain many questions. For example:

  • Why were only the first eight chapters of John copied in this so-called “Western” text form?
  • Was this a feature of the exemplar?
  • Did the scribe start out correcting from a second “Western” exemplar and lose interest in chapter 8?
  • Did the scribe choose an unauthorized “Western” MS and proceed until discovered and apprehended?
  • Was the pericope adulterae in this “Western” exemplar at John 8?
  • Was a MS chosen specifically because it attested the pericope adulterae without considering its divergent character throughout?
  • Might dissension among producers of the MS over the pericope adulterae explain why the “Western” text form cuts off in chapter 8?

However we address these questions, the situation we find in א seems to reflect a real grappling with a text form much like that of D. This is surprising given what is often considered of D’s isolation in the Greek tradition. It may suggest though that whenever א was produced, whether in Caesarea or elsewhere, D’s tradition held a real attraction among at least some of the participants. We can only ask what brought this to an abrupt halt, as it were, almost in the act of copying.


[1] G. D. Fee, “Codex Sinaiticus in the Gospel of John: A Contribution to Methodology in Establishing Textual Relationships,” New Testament Studies 15 (1968) 23–44 at 32.

[2] D. C. Parker, Codex Bezae: An early Christian manuscript and its text (Cambridge: Cambridge University Press, 1992) 146; A. C. Myshrall, “Codex Sinaiticus, its Correctors, and the Caesarean Text of the Gospels” (Ph.D. Diss., The University of Birmingham, 2005) 102.

[3] Parker, Bezae, 144-148. But see also on the setting, D. Jongkind, Scribal Habits of Codex Sinaiticus (Piscataway, N.J.: Gorgias, 2007) 252-254.

[4] Parker, Bezae, 149.

Can we use papyri to date Bezae’s layers?

In my latest post, I pointed to a hypothesized second-century layer of mainstream readings in Bezae’s text of John 4:1-42. This layer consists of readings whose distinguishing characteristic is agreement with the majority text (‘BYZ’). Yet the agreement of most readings in this mainstream layer with the early papyrus P66 (ca 200) suggests that it must have been in existence by the end of the second century. [1]

Of course, there is nothing surprising about a second-century mainstream layer. We might observe what G. Zuntz referred to as “that great common reservoir” of readings from which all later streams flowed. [2] The difference though, is that one of the streams flowing out of the second century is demonstrably “mainstream” rather than “Western” as Zuntz supposed.

To get a sense of the second-century date by which this layer must have existed, we can plot Bezae’s readings by layer as described in detail in this earlier post. The example below depicts Bezae’s agreements in each of its readings in John 4:1-42, with readings attested by P66 highlighted in red:

[Update: note that the highlighted readings include both initial and probable secondary readings (for a plot of initial vs. secondary readings, see figure 3 at the bottom of this post). Of course, only the latter would date from the second century.]

John 4 - 6 clusters with P66

Figure 1: Bezae’s readings in John 4:1-42 partitioned in six layers with P66 readings in red

For reference, my interpretation of the layers from an earlier post is:

  1. Transitional Greek (8 readings, marked ‘o’)
  2. Mainstream / Byzantine (33 readings, ‘Δ’)
  3. Free Greek (9 readings, ‘+’)
  4. Alexandrian (5 readings, ‘×’)
  5. African Old Latin (8 readings, ‘◊’)
  6. European Old Latin (10 readings, ‘∇’)

The axes correspond to the directions of greatest diversity in the per-witness support for the readings. Intuitively, the left-to-right axis seems to register the closeness of a given reading to the mainstream layer, while the up-down axis seems to register along versional or geographical lines. (Keep in mind though that every reading in the plot is a Bezan reading. The graph merely shows Bezae’s agreements per reading.)

It is clear that the readings attested by P66 are clustered over the far-right ellipse corresponding to a (hypothesized) mainstream layer. If we assume a degree of coherence within the readings of this layer, we may conclude that the layer must have existed by at least the paleographically-assigned date of P66. [3]

To corroborate this result, we can compare the readings of another early witness, in this case Origen, coded in green:

John 4 - 6 clusters with Or

Figure 2: Bezae’s readings in John 4:1-42 partitioned in six layers with Origen’s readings in green

But Bezae’s agreements with Origen are also clustered with the hypothetical mainstream layer at the far right, which is consistent with the result for P66.

[Update: The plot below is color-coded to indicate Bezae’s initial and secondary readings in John 4:1-42:]

John 4 - 6 clusters by level

Figure 3: Bezae’s readings in John 4:1-42 color-coded by estimated local-genealogical level (yellow = initial, green = secondary, red = tertiary)

[Update: The complete list of readings per layer is downloadable as a PDF or listed here. Look for the “Level” column, which estimates whether the reading is initial or secondary according to the local genealogical principle.]

[1] I would like to acknowledge Tim Joseph, whose comment alerted me to the potential problem in designating this layer “BYZ”, which could imply that its readings are Byzantine.

[2] Günther Zuntz, The Text of the Epistles; A Disquisition Upon the Corpus Paulinum (London, 1953) 265. For Zuntz, the two streams are “Alexandrian” and “Western”.

[3] In a recent post on his blog, Timothy Mitchell mentions an article in JBL that reconsiders the date of P66.

[Correction: Timothy Mitchell has kindly pointed out my bibliographical error in footnote [3]. The P66 article is actually in Museum Helveticum:

Brent Nongbri, “The Limits of Palaeographic Dating of Literary Papyri: Some Observations on the Date and Provenance of P.Bodmer II (P66),” Museum Helveticum 71 (2014) 1-35.

The article in JBL will discuss P75:

Brent Nongbri, “Reconsidering the Place of Papyrus Bodmer XIV–XV (P75) in the Textual Criticism of the New Testament,” Journal of Biblical Literature 135 (2016) 405–437.

Thank you, Timothy!]

How much of Bezae’s text dates to the second century?

The distinctive elements of Bezae’s Greek text are typically dated to the second century or earlier. This dating is of course based on parallels with second-century writers and a “consensus” view of the tradition as essentially stabilized by the fourth century. But in some recent posts, I have questioned aspects of this view with respect to Codex Bezae.

For example:

While it seems possible to view Bezae’s text as a product of the fourth century, it appears unlikely that it would preserve no second-century influences. There must then be some elements of Bezae’s text assignable to the second century. The question is which elements.

To address this question, I suggest we follow M. W. Holmes’ layered view of Bezae’s text, as a layered sequence of traditions accumulated over time from the various influences to which the text was exposed. [1]

In a recent article in which I examined John 4:1-42 as a test case for Bezae’s layers, I found the following six layers for the 73 readings examined. Each layer consists of readings that Bezae shares with various groups of witnesses, which are unknown in advance:

  1. Transitional Greek (8 readings)
  2. Mainstream / Byzantine (33 readings)
  3. Free Greek (9 readings)
  4. Alexandrian (5 readings)
  5. African Old Latin (8 readings)
  6. European Old Latin (10 readings)

The readings may be plotted as points on a graph (see explanation) with ellipses indicating the group of readings per layer:

Figure 2: Holmes' categories superimposed over six clusters

Bezae’s readings in John 4:1-42 partitioned into six layers (numbering arbitrary): 1. Transitional Greek, 2. Mainstream / Byzantine, 3. Free Greek, 4. Alexandrian, 5. African Old Latin, 6. European Old Latin. Readings labeled as: x = neither Alexandrian or Byzantine, o = Alexandrian, triangle = Old Latin, diamond = Byzantine.

To sequence the readings, I identified several witnesses with identifiable dates:

  1. P66 (200)
  2. Origen (254)
  3. P75 (iii)
  4. Cyprian (258)
  5. a B א (iv)

Then each reading is dated according to the earliest attesting witness or 400, whichever is earlier:

John 4 - 6 clusters by date BW

Bezae’s readings in John 4:1-42 by earliest datable attestation.

From the sequenced layers, it can be seen that layer 2 (Bezae’s agreement with Mainstream or Byzantine witnesses) attests by far the most readings datable to 200, while the highest proportion of readings datable to 200 belongs to layer 4 (Bezae’s agreement with Alexandrian witnesses). Otherwise, layer 3 (Free Greek) and layer 1 (Transitional Greek) have just two and one reading(s) attested in 200 respectively.

The two layers in which we cannot demonstrate attestation in the second-century are the two Old Latin layers (5 and 6) where we are most conditioned to expect second-century attestation. This shows how completely theoretical our dating schema is for “Bezae” and its Old Latin relatives. Given that unambiguous parallels with early writers such as Irenaeus are so sparse and can be easily explained another way, it seems there is little that can secure the so-called “Western” tradition in the second-century.

[1] M. W. Holmes, “Codex Bezae as a Recension of the Gospels” in Codex Bezae: Studies from the Lunel Colloquium, June, 1994 (ed. D. C. Parker and C.-B. Amphoux; Leiden: Brill, 1996) 123–160 at 127.

Is Codex Bezae’s text deceptively ancient?

Dating from the mid-380’s, not long before Codex Bezae’s paleographically-assigned date of ca 400, Ambrosiaster’s thinly-veiled attack on the Vulgate in his Commentary on Romans (chapter 5, verse 14) must be considered highly-relevant to the problem of Bezae’s origins as a bilingual codex, due especially to its concern with the interaction between the Greek and Latin traditions (see my working translation). Ambrosiaster criticizes the Vulgate for its dependence on divergent Greek texts whose pedigree cannot be verified, while at the same time, arguing for the genuineness of the Old Latin version whose reliability (he claims) can be verified with reference to the citations of ancient writers, such as Tertullian, Cyprian, and Victorinus (all three of whom are named explicitly).

Of course, Ambrosiaster could not have been unaware of discrepancies between Latin texts. His concern then must be on the latter point that a text is only as reliable as the authorities that certify its authenticity. Ambrosiaster apparently believes that he can distinguish between corrupted texts and the ancient text on the basis of their respective agreement with the citations of a select pool of authoritative writers. The implication here is that corrupt texts can be corrected on the basis of the appropriate selection of citations.

Now if Ambrosiaster’s method were applied to an actual document, the result would be a text with a series of quite deliberate parallels to the citations of a selection of old writers, embedded within a text that otherwise reflects the period of its production. Taken at face value, such a text might appear very ancient indeed, which is of course the intention. It might even appear as though the ancient writers cited in the text had themselves depended on this later text form.

While we know that stabilization of the Old Latin version was at the time a desideratum, the question is whether Ambrosiaster’s approach would ever have been sufficiently representative to have been applied in practice. But if such a scenario is plausible, the implications for Codex Bezae are momentous. When Bezae parallels a citation of Irenaeus, we can no longer be sure whether it is Irenaeus who used an early text form resembling Bezae’s text or whether Bezae’s producers considered Irenaeus a suitable authority for the correction of their text.

While we must remain ever aware of the limitations of our sources, the uncanny yet somewhat selective appearance in Bezae’s text of parallels with early writers appears at least consistent with Ambrosiaster’s criterion of authenticity (as it were), a connection that seems intriguing enough to pursue for potential insights into possible contexts for Bezae’s origins.

Article Now Available on Textual Layers in Codex Bezae

I recently published an article on textual layers in Codex Bezae found here:

Analyzing Textual Stratification in the Greek Gospel Text of Codex Bezae: Comparing Three Approaches to Layer Extraction in John 4,1–42 (2015)

It is one thing of course to suggest that Codex Bezae attests a mixed text with readings from multiple sources combined into its final text. But it is quite another matter to identify and extract these sources in a systematic and repeatable way. The basic method was proposed by Michael Holmes in a 1996 essay (“Codex Bezae as a Recension of the Gospels”). Holmes then successfully demonstrated this method on the text of Matthew. The consistency of his results speaks for itself.

My goal in this paper is to identify more accurate and efficient techniques to extract Bezae’s layers based on Holmes key insights, namely

  1. That the same witnesses are often found together in support of distinct subsets of Bezae’s readings and
  2. That any group of readings supported by essentially the same selection of witnesses represents a ‘layer,’ which we can treat as a distinct element of Bezae’s tradition.

The motivation is to repeat Holmes’ proof-of-concept on the text of Matthew with other full-scale applications in other parts of Bezae. The article uses complete IGNTP transcriptions for a small part of John.

From the abstract:

It has been suggested that Codex Bezae’s Greek column (D) attests a stratified text, consisting of distinct layers of readings that reflect its historical contact with different traditions. Using John 4:1-42 as a case study, this paper compares three methods of partitioning D’s readings by layer: first, Holmes’ (1996) method based on patterns of agreement; second, a proposed method based on the levels of D’s readings in local genealogies; and, third, a proposed method based on multivariate clustering.

The result shows that Bezae’s readings tend to bifurcate cleanly between two main layers, a mainstream layer and an Old Latin layer.