Direct Copying in a Group of Gospel Manuscripts with Catenae

Abstract Four of the witnesses selected for the Editio Critica Maior of Mark are witnesses to a unique combination of catena commentaries on the Gospels not found in any other manuscripts. An analysis of their text in the Gospel of Mark, using the tools of the Coherence-Based Genealogical Method, shows that they also feature an almost identical form of the biblical text that frequently diverges from both the Majority Text and all other Greek New Testament manuscripts. These four manuscripts, GA 238, GA 377, GA 807 and GA 1160, therefore, constitute a group within the textual tradition of the Gospels. This article provides the evidence that GA 377 is a direct copy of GA 807. No other instances of direct copying can be proven within the group, but their agreement raises the possibility that they are siblings. The format of the catenae may explain the high degree of homogeneity in this group of Gospels manuscripts.


Introduction
The Novum Testamentum Graecum Editio Critica Maior (ECM) of the Greek New Testament differs from earlier editions in methodology and documentary foundation.In methodology, editors employ the Coherence-Based Genealogical Method (CBGM) to evaluate readings and relate witnesses to the text of the New Testament. 1 As for documentary foundation, the text of ECM editions is based on full collations of more manuscripts than any other edition.Among the witnesses selected for the ECM Mark are four catena manuscripts with a nearly identical set of commentaries on the Gospels only found together in these four codices. 2 Closer examination of the biblical text in these manuscripts using the CBGM tools, produced by the Institut für neutestamentliche Textforschung (INTF), has led to the identification of these manuscripts as not only a distinct group based on their commentaries but also as a group within the textual tradition of the Gospels. 3Moreover, the newest of these codices proves to be a direct copy of another member of the group.The three other codices cannot be proven to be direct copies of one another, but they may have been copied from the same exemplar.The format of the catena may have facilitated the high degree of homogeneity in these codices.

The Manuscript Group
The catena manuscripts GA 238 (Moscow, SHM, Syn.gr.47; Diktyon 43672 [Mt-Mk] + Moscow, RSA, F. 1607, № 3; Diktyon 44406 [Lk-Jn]), 4 GA 377 (Vatican City, BAV, Vat.gr.1618; Diktyon 68249), GA 807 (Athens, Library of the Greek Parliament, 1; Diktyon 1097) and GA 1160 (Patmos, Monastery of St John the Theologian, 58; Diktyon 54302) have a nearly identical set of catenae on the Gospels and were included in the ECM for Mark as representatives of the Codices Byzantini because of their close affiliation to the Majority Text (MT). 5They present the full text of each Gospel with a catena in an alternating catena format. 6Georgi Parpulov classified these catena manuscripts as group e.12.ii based on their shared contents. 7The group contains a nearly identical combination of catenae on the Gospels shown in Table 1, according to their classification in the Clavis Patrum Graecorum. 8he catenae in this group of manuscripts are not only identical, except for one type in one Gospel (the Theophylact commentary on John in GA 1160), but this combination of catenae on the four Gospels is also not found in any other catena manuscripts.The catena they share in Matthew (C112.4) is an expansion of the commentary attributed to Peter of Laodicea (C111).9 It is only attested by five tetraevangelia and five other codices with one or two Gospels.10 These manuscripts, therefore, comprise nearly all the complete Gospels manuscripts of that type.The pseudonymous catenae attributed to Victor of Antioch in Mark (C125.1) and to Titus of Bostra in Luke (C130) appear in dozens of codices but rarely together.11 There are only eight witnesses combining these two works in one document, including the four codices listed above.The catena in John (C147.6) is unattributed and falls within the codices singuli category as it was initially only known through GA 377. 12 It is only attested in GA 238, GA 377, and GA 807.These manuscripts stand out among the hundreds of Greek New Testament catena manuscripts on the Gospels for their unique combination of catenae.
The form of the biblical text in these manuscripts also demonstrates that they form a distinct group.Hermann Freiherr von Soden first wrote about the remarkable similarity of these manuscripts, dubbing them 'die Dublettencodices' because of their shared commentaries and the form of their biblical text.He further elaborated that 'they are duplicates among themselves'. 13The CBGM tools developed by the INTF facilitate a comprehensive analysis of all four manuscripts in the Gospel of Mark.The CBGM's General Textual Flow diagram maps the witnesses based on their relationship to their nearest potential relative.The textual flow diagram for Mark (Figure 1) links these manuscripts in direct sequence with no other stems, showing a tight connection between their texts that is not shared with other witnesses. 14When the MT is inserted into the textual flow, their stem descends directly from the MT node, reflecting the high degree of similarity between these codices and the MT and their number of similar non-majority readings.
The CBGM's Compare Witnesses tool shows a complete collation of any two manuscripts at every variation unit and allows one to search for a manuscript's nearest relatives.The data accessed in the Compare Witnesses tool indicates exceptionally high degrees of agreement between these four manuscripts, and the codices are the most highly ranked relatives with the MT following as the fourth-ranked not-too-distant relative.The total agreement of these manuscripts is shown in Table 2, with the MT added as a point of reference for the amount of difference they have to their nearest relative outside the group.This data shows remarkable coherence between these manuscripts and one another that is even closer than their affiliation with the MT.While the agreement between the four manuscripts suggests that they form a unique group within the textual tradition of the Greek New Testament, examining their differences from the MT and one another confirms their group identity.There are 437 variation units in which at least one of the four manuscripts disagrees with one of the others or the MT. 15 In 312 of these variation units (71.4%), the four group members agree with one another against the MT.Two of these can be omitted from the following discussion because the MT reading is split, leaving 310 variation units.Table 3 shows the number of witnesses sharing this group's agreements against the MT in the ECM apparatus.The number of witnesses includes the four group members.Therefore, a reading found in only four witnesses is exclusively attested by this group, whereas a reading found in ten manuscripts is attested by these four codices and six other manuscripts.
The data presented in this table shows a clear group identity that makes these four codices distinct from other Greek New Testament manuscripts.There are fifty places where the group reads together against the entire tradition of Greek New Testament manuscripts.In nearly 25% of the group disagreements with the MT, which, again, is their next closest witness, no more than two other manuscripts share their reading.And in 40.32% of the variation units where the group disagrees with the MT, their reading is only found in fewer than ten manuscripts.It is possible to distinguish a coherent group identity in these places where the group reads against the MT, with only a small number of codices and especially in the places where they read alone.
Their group identity also can be seen in the pattern of reading in places where the group has a split reading.There are twenty-five variation units in which the group splits, with at least one of the readings exclusively found in the group witnesses or fewer than ten total manuscripts.These variation units are listed in Table 4.If one reading is also attested in the Majority Text, the siglum MT is used instead of listing individual manuscripts.If a reading is found in more than ten witnesses but not the MT, then only a plus sign (+) is used instead of listing individual witnesses.The six split dual readings are listed first and followed by the nineteen split triple readings.
The readings shown in this table fall into discernible groups.Five of the split group dual readings (Mark 2.18/18; 3.9/16-18; 7.1/22; 8.28/20-22; 10.4/8-10) entail two of the group manuscripts diverging from the rest which have a reading from the MT or shared with many other manuscripts.The textual flow of these readings indicates that the readings only found in the pair of manuscripts are genealogically later than the earlier reading attested in more manuscripts.In these cases, the readings only attested by group members, found in the Reading Two column, likely pertain to changes from their archetype.The other split group double reading is Mark 11.23/54-62.In this case, the reading b1 found in GA 377 and GA 807 is shared with GA 1689.However, even though the text of their reading is the same, the ECM labels the reading found in GA 1689 as b2 because, in the CBGM's General Flow Diagram, GA 1689 is not connected to GA 377 and GA 807 by any ancestor within a range of five witnesses. 16This indicates that even though GA 1689 shares the reading with the two group witnesses, it emerged independently.The other group members, GA 238 and GA 1160, are the only witnesses to their reading.This variation unit is a probable group reading against the MT that was changed in some members of the group.It is noteworthy that five of these six split double readings involve GA 377 and GA 807 splitting from GA 238 and GA 1160.This pattern is relevant to later in the paper which will demonstrate that GA 377 was copied from GA 807.GA 377 thus inherited these differences from GA 807.In the one case where GA 377 disagrees with GA 807 (Mark 2.18/18), it corrected an unusual variant toward the MT.
The split group triple readings comprise many probable group readings against the MT.In four cases, the lone manuscript reading against the group agrees with the MT: Mark 2.9/26, 5.32/6, 9.21/32-34, 14.70/43.While the textual flow according to the CBGM indicates that Reading One, found in the lone group witness, is antecedent to the reading in the other three witnesses, except in the case of Mark 2.9/26, it is possible that the lone group witness conformed its text to the MT instead of the other three changing.Moreover, in three of these variation units, GA 238 reads against the rest of the group with the MT, a tendency that puts it at the top of the group's stem in the textual flow diagram.In the fourth case, GA 377 reads alone with the MT.As will be seen, this manuscript was directly copied from GA 807, clarifying that this was a change in conformity with the MT from a reading exclusively attested by the group.This suggests that though the group reading in these examples usually is considered a later development, in most cases, the lone manuscript has deviated from a group reading.At Mark 3.7/30-44, 5.9/10-12 and 8.3/14, the lone group manuscript reads against the others in a reading attested in other codices, and in one case, Mark 5.9/10-12, the other three group members also read with other witnesses.In the first two variation units, the lone manuscript reading is genealogically earlier according to the CBGM.Yet, in these cases, the lone group witness has likely diverged from a reading characteristic of the group.At Mark 3.7/30-44, GA 377 reads with three other manuscripts which have ἠκολούθησεν instead of ἠκολούθησαν.As a direct descendent of GA 807, the manuscript's scribe made a phonetic change from the form of the text only found in the other three group members.This is a clear group reading against the MT from which one codex split.GA 238 splits from the group at Mark 5.9/10-12 with only GA L2211.The two codices have no connection to one another, and the CBGM splits the readings into groups c1 (GA L2211) and c2 (GA 238).Compared to the other group members, GA 238 omits τὸ, bringing it closer to the MT and the Ausgangstext.The omission of the article from a rarely attested reading shared by its nearest relatives does not preclude the possibility that its reading was an incidental deviation from the group reading.The MT here reads σοὶ ὀνομά, so there are three differences between the reading found in three group members: the transposition of the personal pronoun, a change of case in the personal pronoun and the addition of the article.The reading shared by GA 238 and GA L2211 is identical in form to the group members except for the presence of the article, and no other reading includes the case change in the personal pronoun from dative to genitive.Thus, in a codex that leans toward the MT more often than the others, the omission of the article from the group reading seems more likely to have arisen from the longer τὸ ὀνομά σου than for the other three group members all having expanded the shorter reading.Thus, Mark 5.9/10-12 was initially a group reading against the MT with one witness deviating from their text.Then in Mark 8.3/14, GA 377 reads τοὺς οἴκους, a reading shared with ten other manuscripts in the d1 group.The other three group witnesses exclusively read οἴκους.The CBGM separates GA 377 from the rest of the codices which added the article by making it the only witness in the d2 group.Their classification highlights that though they share the same text, the reading in GA 377 arose independently from the other codices.Therefore, Mark 8.3/14 can be understood as an exclusive group reading against the MT, even though one of the codices has a different text.
Finally, in four cases, Mark 4.35/2-18, 5.38/20-30, 6.17/18-26, and 11.2/42-46, one of the group members has a singular reading against the rest of the group.In each of these cases, the singular reading probably diverged from a reading shared by the group's archetype.In Mark 4.35/2-18, the four manuscripts are the only ones which changed καὶ λέγει αὐτοῖς to ἃ ἔλεγεν αὐτοῖς.The imperfect form of λέγω is found five other times in Mark 4, perhaps explaining the variant found in this group.GA 238 singularly hearkens back to the standard reading by adding λέγει αὐτοῖς after ἐν ἐκείνη τῇ ἡμέρᾳ ὀψίας δὲ γενομένης.While the addition of λέγει αὐτοῖς does bring the reading closer to the MT, it is still a nonsense reading because 'he said to them' is repeated twice.The structure of the catena, however, likely explains the unusual addition found in GA 238.In all four of these manuscripts, ἃ ἔλεγεν αὐτοῖς ἐν ἐκείνη τῇ ἡμέρᾳ comprises the last seven words of a biblical lemma spanning Mk 4.30-5.After these words comes a block of commentary, which occupies eleven lines of text in GA 238.Then the next biblical lemma starts ὀψίας δὲ γενομένης.From here, GA 377, GA 807, and GA 1160 all proceed with what Jesus said: διέλθωμεν εἰς τὸ πέραν.GA 238, on the other hand, inserts λέγει αὐτοῖς before the contents.As these are the only four codices to have ἃ ἔλεγεν αὐτοῖς, this is an exclusive group reading perhaps drawn from their archetype.The gap between writing 'Jesus said' and the contents of what he said due to the scholium breaking the verse, the scribe of GA 238 inserted λέγει αὐτοῖς again-influenced more by the structure of the alternating catena manuscript than conformity to the MT. 17 The other three singular readings are more typical scribal changes: a sound change (Mark 5.38/20-30), the addition of an article (Mark 6.17/18-26) and the transposition of two words (Mark 11.2/42-6).In these three readings, at least one other manuscript shares their reading, so they cannot be called exclusive group readings.The deviation from the group still explains the singular reading, indicating a place where there would be uniform group agreement against the MT apart from the habits of a particular scribe.The examination of places where GA 238, GA 377, GA 807 and GA 1160 agree together against the MT and in twenty-five of the places where they disagree, demonstrates that these codices form a distinct group within the textual tradition of the Greek New Testament.
Even though Mark is the only Gospel for which an ECM edition has been produced and, therefore, the only one for which the CBGM tools can be used, other resources at least establish a common profile for the other Gospels. 18As noted above, von Soden found that the biblical text in the four group manuscripts frequently differed from the other Koine type codices, and he produced short lists of some of their readings. 19The INTF's Manuscript Clusters tool, which aggregates data from the Text und Textwert (TuT) volumes, lists these codices as their nearest relatives in selected test passages. 20In Matthew, the four witnesses diverge in only one variation unit where all the codices are extant in the Teststellen. 21The Luke-John portion of GA 238 has never been microfilmed, so it was not considered in TuT for either Gospel.GA 1160 contains a different type of commentary in John, and its form of John does not mirror GA 377 and GA 807.In Luke, GA 377, GA 807 and GA 1160 only differ from one another in one Teststelle each. 22The profile in John for GA 377 and GA 807 is intriguing.They agree in 146/151 Teststellen but only agree with the MT at 82.1% and 84.3%, respectively.Significantly, these two codices agree with one another and against the MT in 23/25 test passages. 23While these two codices appear to have a strikingly non-majority form of the biblical text in John, according to the Teststellen, two other studies indicate the remarkable consanguinity between GA 377 and GA 807 in John but with closer agreement to the MT. Bruce Morrill conducted a complete collation of all available continuous-text Greek New Testament manuscripts in John 18.In John 18, GA 377 agrees with the MT in 93.9% variation units, and GA 807 agrees at the slightly higher rate of 95.2%.These figures correspond to their overall agreement with the MT in Mark.The two codices align with one another in 97% of variation units and, significantly, share fifteen of sixteen non-majority readings. 24Drawing on Morrill's data, Andrew Edmondson compared the CBGM with phylogenetic techniques for relating manuscripts, using the text of more than 600 manuscripts in all of John 18.The results in John 18 mirrored those found in TuT for John 1-10. 25While complete data on the texts of these manuscripts in the other Gospels is not available, the results of the current studies indicate that the copying relationships found in Mark are mirrored in the other three Gospels.
This examination has shown that the four Gospel-catena manuscripts GA 238, GA 377, GA 807 and GA 1160 form a distinct group of Greek New Testament manuscripts.They have a unique combination of catenae not found in any other manuscript, containing all four Gospels and a biblical text that frequently diverges from the MT as a group against all other manuscripts.Parallels for this group exist in other catena manuscripts.Family 0150 contains a highly similar text of the Pauline epistles with the commentary of John of Damascus in alternating format and potential instances of sibling manuscripts or direct copying. 26The catena group within Family 1 have the same types of catenae on the Gospels in alternating format and a number of distinct readings which differ from other Family 1 manuscripts; two direct copies have been identified in this group. 27And the manuscripts with Nicetas of Heraclea's catena on John, all copied in alternating format, also form distinct groups based on their textual affinity. 28Since alternating catenae were almost exclusively copied from other alternating format manuscripts, these likely descend from another alternating catena codex.The process of copying the biblical and catena texts in discrete sections may have led to the preservation of a distinctive form of the biblical text because this was not in a format easily compared with continuous-text manuscripts. 29That these four manuscripts agree with one another, in nearly every instance, across the test passages in Matthew, Luke, and John and in nearly 5500 variation units in Mark presents the likelihood of not only one event of direct copying but potentially multiple incidences of this rare circumstance.

GA 377 a Direct Copy of GA 807
As the youngest manuscript, being copied in the sixteenth century, GA 377 is the most likely candidate to be a direct copy of one of the twelfth-century codices.Alan Taylor Farnes developed a methodology for identifying Abschriften which evaluates potential parent and child manuscripts from multiple textual and codicological perspectives, including their overall agreement, dual agreements, history, palaeographical irregularities, corrections, and codicological correspondence. 30GA 377 proves to be a direct copy of GA 807 in each of these criteria.
Remarkably, GA 377's overall agreement with all three group members in Mark is higher than some other manuscripts shown to be Abschriften. 31Despite their textual affinity, GA 1160 can be eliminated as the exemplar for GA 377 because it has a different type of catena and form of the Gospel text in John.Between the remaining two group members, each of the sources used earlier when establishing their group identity indicates that the closest potential ancestor for GA 377 is GA 807. 32Again, these two manuscripts agree in 99.22% of places in Mark-disagreeing in only forty-three variation units. 33Such a high rate of agreement suggests the possibility that one is copied from the other.
GA 377 and GA 807 share dual agreements both within their manuscript group and the larger corpus of Greek New Testament manuscripts.The two codices have the most dual readings of any pair in their group at eighteen times.Conversely, GA 807 has only three dual readings each with GA 238 and GA 1160, the least frequent pairings.GA 807 also has the fewest independent readings against the other three group witnesses-reading on its own only three times when every other manuscript has more than ten occurrences, and GA 238 reads alone against the group fifty times.GA 377 being a direct copy of GA 807 explains these patterns.It adopted many of the readings which its Vorlage would have read against the other two group witnesses or with only one of them.On the other hand, GA 377 rarely reads with any combination of group witnesses against GA 807 while having the second most readings against the rest of the group.Thus, GA 807 rarely reads on its own or without GA 377, and GA 377 either reads with GA 807 or on its own, showing a clear line of change from its exemplar.
A closer look at GA 377 and GA 807's dual agreements demonstrates their copying relationship.The two codices have nineteen dual readings against the other members of their group in Mark.In six of these instances, their reading is attested in five or fewer witnesses.Four are exclusively found in these two manuscripts.These readings and their witnesses are listed below.Mark 3. Thus, even in these codices which have a text that is exceptionally close to the MT in Mark and to their nearest relatives, there are still unique dual readings that cannot be found in other manuscripts.
Examining the disagreements between GA 377 and GA 807 substantiates that GA 377 is a direct copy of GA 807.Farnes notes that places that are difficult to read in the parent manuscript which correspond to unusual readings or differences in the child manuscript potentially indicate a copying relationship. 34At Mark 4.31/4, GA 377 splits from GA 807 and the other group members reading b κόκκῳ instead of a κόκκον.At this word, the -ον ending in GA 807 is obscured by bleed-through so that the open space at the top of the nu appears to be connected, loosely resembling an omega.The letter form does not match other omegas in GA 807, but the bleed-through gives a plausible explanation for this difference between the two manuscripts.
The most common types of disagreement are those in which GA 377 has additions, duplications or omissions which can be explained by the process of copying from GA 807.GA 377 attests a singular reading at Mark 4.1/29 with the addition of καί after ὥστε.GA 807 happens to have the word καί one line below immediately after the final epsilon in ὥστε.The scribe of GA 377 likely erroneously added the word καί due to its placement in GA 807.
A duplication occurs at Mark 2.16/47 when GA 377 repeats the phrase ἔλεγον τοῖς μαθηταῖς αὐτοῦ τί ὅτι μετὰ τῶν τελωνῶν καὶ ἁμαρτωλῶν against the three group members and the MT.This minority reading is only attested in three manuscripts.The duplication likely occurred due to eyeskip.The phrase τῶν τελωνῶν καὶ ἁμαρτωλῶν immediately precedes the sentence beginning ἔλεγον τοῖς and is also the final wording of that textual unit.In GA 807, the second καὶ ἁμαρτωλῶν occurs exactly one line below the first.Thus, the scribe of GA 377 inadvertently returned to the καὶ ἁμαρτωλῶν before ἔλεγον instead of the one before ἐσθίει and duplicated the clause.The proximity of the two words in GA 807 explains the duplication in GA 377.
Another duplication occurs at Mark 9.12/2-20 when the scribe of GA 377 duplicated the phrase ὁ δε ̀ἀ ποκριθει ς εἰ ͂ πεν αὐ τοῖ ς Ἠλι άς με ν ἐ λθω ̀ν πρω ͂ τον.In this case, they recognised the error and struck the mistake.Again, the word πρῶτον immediately precedes the duplicated phrase and is its final word.In GA 807, both instances of πρῶτον are stacked immediately above and below one another on separate lines.The scribe of GA 377 likely skipped to πρῶτον from the previous verse and began copying from there before realising their mistake.The proximity of the words in GA 807 explains the mistake made by the scribe of GA 377.
A series of disagreements between the two manuscripts occurs due to omission in GA 377 at Mark 9.45-6.This is part of a complex passage in Mark with many similar words and two verses, Mark 9.44 and 9.46, which are identical and omitted from most modern Bibles. 35GA 377 reads against GA 807 in the omission of Mark 9.45-6. 36GA 377 leaps from σβέννυται at the end of Mark 9.44 and begins verse 9.47 with καὶ ἐὰν ὁ ὀwθαλμός.As in 34 Farnes, Simply Come Copying, 45. 35 The NA28 omits the verses entirely and they are placed in double brackets in the ECM. 36Multiple errors were made in the ECM apparatus for GA 377 in these verses and are listed in the Appendix.The situation could be reversed to claim that GA 377 omits Mark 9.44-5 and skipped forward to verse 46 because 9.44 and 9.46 are identical.the cases of duplication, the scribe of GA 377 omitted these verses due to eyeskip based on the proximity of similar words in GA 807, shown in Figure 2.
GA 807 attests verses 9.44-6 in the form in which they are found in the MT.If the scribe of GA 377 omitted verses 45 and 46, they skipped from σβέννυται, which occurs in the final third of line seven in Figure 2, to the same word in verse 46 four lines below.If verses 44 and 45 were omitted, the scribe skipped from τὸ ἄσβεστον in verse 43 (line six in Figure 2) to the same words at the end of verse 45 four lines below.In both instances in GA 807, the words τὸ ἄσβεστον and σβέννυται are stacked directly on top of one another with a medial dot punctuation mark.Moreover, the phrase immediately after σβέννυται begins καὶ ἐὰν ὁ (Mark 9.45, 47), which may also have contributed to this incidence of eyeskip.Even though it requires the scribe of GA 377 to have made a longer skip, the location of these words in GA 807 explains the omission.
The adoption of the corrections of the parent manuscript in its child is one of the most telling signs of a direct copying relationship.37At Mark 8.3/14, also discussed above, GA 807 reads οἴκους, exclusively attested by the other three group witnesses.GA 377 diverges from the group by inserting the article, τοὺς οἴκους.This variant only occurs in eleven manuscripts.The reason GA 377 differs from GA 807 and the rest of the group is that a later user added the word τούς to GA 807 and, as a copy of that manuscript, the scribe of GA 377 copied the amended text of its exemplar.Another correction in GA 807 occurs at Mark 15.42/20.The ECM records GA 807 as a witness to the reading πρὸς σάββατον while GA 377 reads προσάββατον found in the Ausgangstext and the MT.GA 807 was corrected to προσάββατον offering a direct explanation for the reading found in GA 377.
Further evidence that GA 377 is a direct copy of GA 807 are places where the sixteenthcentury scribe wrote in the exemplar.The same hand that wrote GA 377 numbered the pages of GA 807 in Matthew in the upper right corner in red ink and added folio numbers in black ink next to the chapter list for Mark (fol.79).There also are two instances where the scribe of GA 377 may have corrected the text in GA 807.The two disagree at Mark 14.36/18.In this variation unit, GA 807 was corrected by a later hand to read παρένεγκαι which is also found in GA 377.The correction ligature and handwriting are comparable to those in GA 377.Likewise in Mark 16.18/22-8, GA 807 has been corrected from the variant reading οὐ μὴ αὐτοὺς βλάψει to the text found in GA 377: οὐ μὴ αὐτοὺς βλάψη.The capital letter eta written over the epsilon-iota ligature has the appearance of the handwriting in GA 377.These, therefore, could be examples of the scribe of GA 377 amending their exemplar.Even if the scribe of GA 377 did not make these corrections in GA 807, the presence of the corrections explains the readings found in GA 377.The direct evidence of GA 377's scribe using GA 807 demonstrably confirms that the latter is a direct copy.At each point of comparison, the evidence suggests that GA 377 is a direct copy of GA 807.The ECM of Mark should therefore be updated to remove GA 377 as a witness, and it should not be considered in future editions on the other Gospels.

Other Direct Copies
The close agreement between the twelfth-century manuscripts, GA 238, GA 807 and GA 1160, raises the possibility that further direct copies might be identified even though they are palaeographically dated to the same century. 38In fact, GA 807 has even fewer disagreements in Mark with GA 1160 than the manuscript copied from it.GA 238 disagrees with GA 807 seventy-seven times, GA 238 disagrees with GA 1160 seventy-one times, and GA 807 disagrees with GA 1160 thirty-eight times.This level of total agreement would support the possibility of identifying another pair of Abschriften.
Based on their agreements, the most likely pair is GA 807 and GA 1160.However, GA 1160 is the only codex in this group to have a different commentary in the Gospel of John, making it the least likely to have been the parent of another group member.An examination of their thirty-eight disagreements renders an unclear verdict about their relationship.In sixteen of the variation units, one of the manuscripts attests a reading found in fewer than ten manuscripts, including seven singular readings (if GA 377 is omitted).These could be instances of one of the copyists altering a reading from the other or simply differences from their direct copies.The location of the disagreements in either manuscript does not explain the change in either direction, many of the variants involve common sound changes or iotacism, and they do not correspond to corrections or irregularities in either codex.These factors indicate that GA 807 and GA 1160 are most likely not Abschriften.
GA 238 has a similar number of disagreements with GA 807 and GA 1160.Acknowledging that the low-quality microfilm images do not allow as thorough an investigation of that manuscript compared to the colour images of GA 807 and GA 1160, the same criteria mitigate the likelihood of it being in a direct copying relationship with either manuscript.GA 238 and GA 1160 have eighteen split-group double agreements in which they read against the text found in GA 377 and GA 807.This is one fewer than GA 377 and GA 807, and the instances of dual agreements were an important point for validating that GA 377 is a direct copy.While in the earlier example, many of these readings were only found in those manuscripts or were rarely attested variants, in seventeen of these eighteen readings GA 238 and GA 1160 read with the MT against the rest of the group.There are no true dual readings in Mark where GA 238 and GA 1160 have a reading not found in any other manuscripts.GA 238 cannot be proven to be the parent or a copy of another group member.
It cannot be proven that there is any direct copying relationship between GA 238, GA 807 or GA 1160.Detecting Abschriften is not an easy task, and Farnes remarked that 'because establishing an Abschrift is difficult on textual grounds alone, it is difficult to determine an Abschrift of a very high-quality manuscript'. 39These three codices may have a remarkably consanguineous biblical text in Mark and many similarities in their composition, but the fact they are near-exact points toward each being a well-executed copy of another codex or a well-masked child of the other.

Conclusion
Four codices, GA 238, GA 377, GA 807 and GA 1160, constitute a textual group within the corpus of Greek New Testament manuscripts.They contain a unique combination of catenae on the Gospels in alternating format, and they have a distinctive form of the Gospel of Mark.Their many similarities raise the possibility of multiple direct copying relationships.GA 377, the youngest member of the group, has here been shown to be a copy of GA 807.It is only the twenty-second direct copy of a surviving Greek New Testament manuscript to be identified, and the ninth which also contains a catena. 40he other three codices cannot be proven to be copies of one another on textual or codicological grounds.As an Abschrift, GA 377 should be removed from the ECM apparatus for Mark and not be included in other editions. 41However, their frequent group readings against the MT and the rest of the Greek manuscript tradition warrant the others' continued inclusion in future ECM editions.In each case, the alternating catena format may have facilitated the careful copying of the lemmata from their exemplar, leading to a high degree of agreement and their familial distinctiveness.As the texts of the scholia in these types have not been analysed, one cannot say whether the content of the catena led to any unique group readings.An intriguing possibility is that the three twelfthcentury codices were directly copied from the same manuscript, which would explain how they came to have such similar Gospel and catena texts, the commentary on John in GA 1160 notwithstanding.This would also explain common codicological features which were not explored in this paper. 42Identifying this group of manuscripts and a new Abschrift provides opportunities for further research on the textual history of the Gospels, catena manuscripts and the provenance of Byzantine manuscripts.The tools of the CBGM and complete collation of more manuscripts produced for ECM editions not only advance research on the text and history of the Greek New Testament, but provide the resources to identify direct copies and new groups of manuscripts. 39Farnes, Simply Come Copying, 46. 40Farnes, Simply Come Copying, 24-8, 209-10; Houghton and Myshrall, 'Three Direct Copies', 397; Parker, 'Family 1 in the Gospel of John', 328.Houghton and Myshrall note that although Farnes suggests twenty-three Abschriften, only eighteen have been proven.Among the unproven direct copies are three manuscripts with commentaries: GA 0142, GA 0151 and GA 2110, lowering the number of catena manuscripts which are direct copies according to Farnes.The figures given here account for the three Abschriften identified by Houghton and Myshrall, which include two manuscripts with Theophylact commentaries, and one identified by Parker which is also a catena manuscript.Among the twenty-three Abschriften proposed by Farnes are eight which he indicates are only 'argued for' or 'suggested' and one pair (GA 205/GA 2886) which he considers more likely to be siblings (Simply Come Copying, 162-5).The unconfirmed codices warrant further evaluation.

Figure 2 .
Figure 2. The inclusion of Mark 9.44-46 in GA 807 (fol.110v) 41 It has been selected for the ECM of John: D.C. Parker, K. Wachtel, B. Morrill, and U. Schmid, 'The Selection of Greek Manuscripts to be Included in the International Greek New Testament Project's Edition of John in the Editio Critica Maior', Studies on the Text of the New Testament and Early Christianity: Essays in Honor of Michael W. Holmes On the Occasion of His 65th Birthday (ed.D.M. Gurtner, J. Hernández Jr., and P. Foster; NTTSD 50; Leiden: Brill, 2015) 287-328. 42On the use of decorative and codicological features of manuscripts to locate where they were copied and their correspondence to particular forms of the biblical text, see K. Maxwell, 'The Textual Affiliation of Deluxe Byzantine Gospel Books', The New Testament in Byzantium (ed.M.J. Kruger and R.S. Nelson; Dumbarton Oaks Byzantine Symposia and Colloquia; Washington: Dumbarton Oaks, 2016) 33-85; N. Kavrus-Hoffman, 'Producing New Testament Manuscripts in Byzantium: Scribes, Scriptoria, and Patrons', The New Testament in Byzantium, 117-45.

Table 1 .
Catena Types in the Group Witnesses Figure 1.Detail of CBGM General Textual Flow Diagram

Table 2 .
Percentages of Agreement between Group Witnesses and the Majority Text

Table 3 .
Attestation of Group Agreements Against the Majority Text