Codex Augiensis is a Copy of the Greek Text of Codex Boernerianus

Abstract

Scholars have long been aware of the close relationship between two ninth-century Greek-Latin bilingual manuscripts, Codex Boernerianus (GA 012, VL 77) and Codex Augiensis (GA 010, VL 78). However, assessments of the nature of this relationship differ. The present article seeks to resolve this question by comparing full electronic transcriptions of the Greek texts of these manuscripts in Romans, Galatians, Ephesians, and First Timothy. An examination of the points of divergence, including unique readings, word division, corrections and lacunae confirm that their Greek text was either copied from the same exemplar, or that one served as exemplar for the other. Close analysis of the types of errors and the way in which corrections in Codex Boernerianus are handled in Codex Augiensis proves that the latter was copied from the former. These findings indicate that, as a copy of an existing manuscript, Codex Augiensis should no longer be cited in the apparatus of the Greek New Testament.

1 Introduction

Codices Augiensis and Boernerianus are ninth-century bilingual manuscripts of the Pauline Epistles. While Augiensis is arranged with parallel Greek and Latin columns on each folio, Boernerianus is written in a single Greek column with an interlinear Latin text. Though they differ vastly in appearance, their similarities are striking, and their relationship has been widely debated.

In 1791, Christian Friedrich von Matthaei published an edition of Boernerianus and asserted that Augiensis was a copy of that manuscript.¹ In his 1859 edition of Augiensis, Frederick Scrivener agreed that these manuscripts were closely related but rejected the possibility of one being a direct copy of the other based on their word divisions and the clear independence of their Latin texts.² When Friedrich Zimmer contended in 1887 that Augiensis was a copy of Boernerianus due to the shared unique Greek forms, Peter Corssen, later in the same year, affirmed Scrivener's conclusion that both derived from a hypothetical common ancestor.³ He suggested that this was related to Codex Claromontanus, a fifth-century bilingual manuscript of the Pauline Epistles whose Greek text is related to that of Augiensis and Boernerianus. In his review of Corssen's work, Zimmer responded that Boernerianus displayed all the characteristics attributed to this common ancestor.⁴ Having examined the examples provided by Zimmer, William Smith concluded that Boernerianus and Augiensis were more likely to be cousins, separated by a generation from a shared ancestor, while some years later Corssen himself proposed more examples in support of his own position.⁵

Since then, there has been little progress towards a conclusion. William Hatch in 1951 reaffirmed Smith's position that Boernerianus and Augiensis were cousins, supplying little in the way of fresh data.⁶ Hermann Joseph Frede's 1964 survey of all the published positions concluded that the exemplar of Boernerianus was later used for Augiensis.⁷ In 2007, Walter Berschin brought together material on the history of Codex Augiensis, treating Scrivener's work as a high point: while acknowledging Hatch's claim that the manuscripts were cousins, he followed Frede's conclusion that the two manuscripts likely shared an exemplar.⁸ Most recently, David Parker observed that the relationship between these codices has still not been dealt with in a satisfactory manner.⁹

In this article, I draw on fresh transcriptions based on new digital images to demonstrate that Codex Augiensis is a copy of Codex Boernerianus. In the first section, I set out some of the evidence for their unusually close relationship, showing that there is no need to hypothesise that they are a generation removed from the same exemplar. I then consider the evidence that the Greek variant readings, peculiar to these manuscripts, derive from a bilingual exemplar in interlinear format, matching that of Boernerianus. Third, I examine the correspondence between features distinctive of Boernerianus itself, such as letter forms and corrections, and the same passages in Augiensis. These lead to the conclusion that Boernerianus itself served as the exemplar for the Greek text of Augiensis. Finally, I illustrate how Corssen's objections can be explained in the light of this conclusion and contemporary understanding of scribal practice.

The writings considered in this study are Romans, Galatians, Ephesians and First Timothy, providing material from the beginning, middle and end of the Pauline corpus in these manuscripts. The Greek transcriptions were produced by the International Greek New Testament Project (IGNTP) and the Greek Paul Project, while the Latin transcriptions of Romans and Galatians were created by the COMPAUL Project: all are published online.¹⁰ For Ephesians and First Timothy, the Latin texts were compared directly with high-quality digital images released by the holding institutions.¹¹ These images were also used to verify details not recorded in the transcriptions. As Codex Boernerianus was damaged during World War II, it was sometimes also necessary to consult Alexander Reichardt's 1909 facsimile edition.¹²

The published transcriptions were used to generate a series of collations, using a tool developed at the Institute for Textual Scholarship and Electronic Editing at the University of Birmingham.¹³ The key outputs were 1) a collation of the Greek text of Boernerianus against the Greek text of Augiensis and 2) the Greek text of Boernerianus against the 2005 Robinson-Pierpoint Majority Text (RP2005), which were exported into Excel spreadsheets. In these, I categorised the differences between the manuscripts as 1) phonetic or visual, 2) consisting of vowels, consonants or both, 3) involving abbreviations, 4) additions or omissions of words, 5) word division and order, 6) involving corrections. Each place of variation was also compared to the RP2005. The collation of Boernerianus and RP2005 was compared to the Greek transcriptions of Codex Claromontanus, Papyrus 46 (P46) and the Latin text of Boernerianus in a separate spreadsheet, where it was categorised according to part of speech and comparison between the Greek and Latin texts of Boernerianus. The organisation of the spreadsheet data by type of variation provided the material for the present paper. Though the focus of this study is on Romans, Galatians, Ephesians and First Timothy, there are also some examples from other epistles, including the list of shared lacunae.

2 The Close Relationship between Boernerianus and Augiensis

2.1 Shared Lacunae

The Greek lacunae shared by Augiensis and Boernerianus are consistent with a direct relationship between the two manuscripts. The correspondence between the Latin and Greek texts in Augiensis is also consistent with this relationship. The Greek text of Augiensis is always written in the interior column with the Latin text in the exterior. The word placement of the Latin text is often congruent with the Greek text, and marks are sometimes placed above words in the Greek column to show coordination when necessary.¹⁴ Berschin noticed simultaneous changes in the Latin and Greek styles indicating that Augiensis had multiple scribes and that the same scribe often wrote both columns.¹⁵ Berschin also observed that the Greek text was usually written to the end of the line implying that it was written first, and the Latin text was subsequently written to correspond to the Greek.¹⁶ However, Berschin's second observation does not always appear to be true as there are instances in which the Latin column seems to have influenced the layout of the Greek column. In 2 Thess 3.5 (folio 114v), while the Latin text, di et patientia xpi, stops short at two-thirds of a line, the Greek text, του θυ και ειc την υπομονην, lacks sufficient space on the corresponding line, and του χρυ is written on the next line across from an empty line in the Latin column. Rather than having xpi on the blank line across from του χρυ, it seems that the Latin text was written first, and the corresponding Greek text then went beyond what had been designated. Additionally, where the Greek text of Augiensis is lacunose, the Latin text is present. Lines in the Greek column are left empty for text to be added later. In Boernerianus, these same Greek lacunae are accompanied by lacunae in the Latin text, confirming that the Latin texts of Augiensis and Boernerianus derive from different sources. There are a few exceptions in both Augiensis and Boernerianus which will be discussed below. Next to the space left for 1 Cor 3.8–16, a lacuna shared by Augiensis and Boernerianus, Boernerianus (folio 32v) includes the marginal notation, deest in graeco, which seems to indicate that Boernerianus is the source of the shared lacunae.

Though Scrivener and Hatch only identified several major Greek lacunae shared by Boernerianus and Augiensis: 1 Cor 3.8–16, 1 Cor 6.7–14, Col 2.1–8, 2 Tim 2.12–13 and Phm 21–25, there are also other instances.¹⁷ In Rom 8.1, both manuscripts lack the RP2005 reading μὴ κατὰ σάρκα πɛριπατοῦσιν ἀλλὰ κατὰ πνɛῦμα. While Boernerianus (folio 10v) leaves just over two lines empty, Augiensis (folio 15r) leaves over four lines empty in its Greek column. Here, the Latin text of Augiensis is spread out with noticeably exaggerated spaces to fill the Latin lines. The word carnem itself inhabits a full line and is divided as car nem with a very large space between the two syllables.

While dissimilar lacunae also show the close relationship between the manuscripts, they also demonstrate that the Latin text of Augiensis guides the layout of its Greek text. Most noticeable is the doxology in Romans. Boernerianus (folio 18r) includes a space after Rom 14.23 large enough to fit seven to nine lines, which is not present in Augiensis (folio 28r). Augiensis (folio 31v) includes an eleven-line space after Rom 16.24, which is not included in Boernerianus (folio 21r). The scribe for that section of Augiensis did not have any Latin text to include after Rom 14.23, so there is no space left for the Greek or Latin texts. Unlike Boernerianus, Augiensis has Latin text for Rom 16.25–6 where the Greek text is missing. The layout of the text in Augiensis is determined by the Latin text.

There are other lacunae unique to the Greek text of Augiensis, which do not appear in Boernerianus. In Rom 8.17, in Boernerianus (folio 11r), και κληρονομοι δε χρυ and et c(o)haeredes autem xpi appear seamlessly. While the Greek text of Augiensis (folio 16r) is the same as Boernerianus, it also includes a three-word space to match its longer Latin text, et heredes heredes quidem di coheredes autem xpi. The Latin exemplar of Augiensis included this text, but the Greek exemplar did not. Similarly, in Gal 5.6, Boernerianus (folio 58v) has εν γαρ χρω ιυ ουτε ακροβυcτια αλλα and omits the phrase οὔτɛ περιτομή τι ἰσχύɛι, a result of the scribe's eye skipping from ουτε to ουτε. The corresponding Latin text neque praeputiam sed written above the Greek text also omits the phrase. Augiensis (folio 74r) has the same Greek text as Boernerianus but leaves a line and a half empty between ακροβυcτια and αλλα, anticipating supplemental Greek text to match its full Latin text, nam in xpo ihu neque circumcisio aliquid ualet neque prepucium sed. These dissimilar lacunae further demonstrate how the Latin text affects the layout of the Greek text in Augiensis, and even when the lacunae are dissimilar, Boernerianus and Augiensis still have the same Greek text.

As noted above, the Latin text of Boernerianus is rarely present when its Greek text is lacunose. In 1 Cor 13.3, Boernerianus (folio 35r) has υπαρχοντα μου at the very beginning of the line and leaves the remainder of the line empty. The scribe wrote substantias meas above the Greek text and continued the Latin text over the empty space with the Vulgate reading i(n) cibos pauperum. The same space is left in the Greek column of Augiensis (folio 50r). In 1 Tim 1.5, Boernerianus (folio 86r) omits the Greek word ἀγαθῆς but leaves a space for it, above which the Latin word bona is written. Augiensis also leaves a blank space in its Greek text large enough for the word but also includes bona in its Latin text.

Also noted above, in Augiensis, rarely are Greek lacunae preserved when there is also no Latin text given. In Eph 2.4, there is a three-word space in the Greek text of Boernerianus (folio 62v) after ην ηγαπηcεν ημαc, though nothing appears to have been omitted. Above this open space, misertus (est) nostri is written, possibly by a later hand. Vulgate manuscripts L and U read qua diligit nos misertus est nostri.¹⁸ So, it is plausible that the Latin exemplar for Boernerianus also had this longer reading, a doublet relating back to the text of P46 (folio 76r), ηλεηcεν ημαc, hence the leaving of this gap. Augiensis (folio 78r) includes a similar space in its Greek text while also leaving a space in its Latin text. Again, just after 2 Tim 2.4, Boernerianus leaves a partial line space at the end of folio 92v and a larger space at the top of folio 93r. Augiensis leaves three full lines empty in both the Greek and Latin columns (folio 124r).

In one instance, the Latin text of Augiensis partially reflects the lacuna in the Greek text. Beginning in 2 Tim 2.12, after ει υπομενομεν and si sustinemus, and ending in 2 Tim 2.13, before εκεινοc and ille, Boernerianus (folio 93r) presents a gap taking up almost two full lines. Augiensis (folios 124r–124v) includes the same empty space in the Greek text leaving almost four full lines in the Greek column. However, the Latin text of Augiensis (folio 124r) includes the first two words which would correspond to the missing Greek text, et c(on)regnauimus, but leaves the rest of verse blank. The Latin text then picks up again at the top of folio 124v at the beginning of verse 13, reading si non credimus across from an empty line in the Greek column.¹⁹ This difference in practice might be attributed to different scribal habits among the different scribes copying Augiensis, or a scribe might have written the Latin text first in both cases before noticing that there was no corresponding Greek text.

Ultimately, the Greek text of Augiensis has two limiting factors. It is limited – with very few exceptions – by the organising layout of the Latin text. It is also limited as it follows the Greek text of its exemplar. Contrary to assertions that Augiensis was copied from a single bilingual manuscript written similarly to Claromontanus, a staple in Hatch's argument, the lacunae suggest that each scribe of Augiensis took the Greek and Latin texts from two different exemplars.²⁰ While the Latin texts of Boernerianus and Augiensis differ from each other significantly, their Greek texts share a close relationship.

2.2 Latin Glosses

This close relationship between the manuscripts is also demonstrated in the Latin glosses sporadically written above the Greek text of Augiensis. For example, in Rom 8.38, Boernerianus (folio 12r) has (con)fido (ue)l cert(us) su(m) enim. Augiensis (folio 17v) has certus sum enim in the Latin column but also (con)fido above πεπιcμαι in the Greek column. It is not always clear which glosses were written by one of the scribes of Augiensis and which were written by a later hand.²¹ Scrivener counted 106 glosses, eighty-six of which match the interlinear Latin text of Boernerianus.²² Of the twenty glosses incongruent with Boernerianus, ten appear to have been directly influenced by the Latin and Greek texts of Augiensis or retain some similarity to Boernerianus. The ten remaining glosses appear to have been derived from another source altogether, five of which were written by the original hand and also tend to represent the Greek text more plainly. In four of these five instances, the main Latin text of Augiensis and Boernerianus attest the same reading, while the fifth gloss partially reflects the Latin of Boernerianus. Of the eighty-six glosses congruent with Boernerianus, those by the original hand support its presence at the copying of Augiensis. The others demonstrate its use to make corrections at another time. Those differing from Boernerianus should be expected as it has already been established that the Latin text of Augiensis came from a different source.

3. An Interlinear Exemplar

3.1.1 Word Order

The Greek text of Augiensis often reflects the strange word order of Boernerianus, which in turn is consistent with its interlinear Latin text. For example, in 2 Cor 11.21, rather than the RP2005 reading, τις τολμᾷ, ἐν ἀφροσύνῃ λɛ́γω, τολμῶ κἀγώ, Boernerianus (folio 51r) relocates the parenthetical statement to the end of the clause, τιc τολμα τολμω καγω εν αφροcυνη λεγω, like its Latin text, quis audit audio et ego in insipientia dico. Though the Latin text of Augiensis (folio 63v) follows the Vulgate word order, quis audet in insipientia dico audeo (et) ego, which also reflects the RP2005 word order, its Greek text has the same word order as Boernerianus which is unique to these two manuscripts. Differences between them in word order will be discussed below.

3.1.2 Shared Unique Readings

Boernerianus and Augiensis share many unique readings, which can also be explained by an interlinear format in the exemplar for Augiensis. For example, in 1 Cor 7.16, only in Augiensis (folio 40r) and Boernerianus (folio 27v) are the two vocatives γύναι and ἄνɛρ rendered as nominatives, γυνη and ανηρ. Jeffrey Kloha argues that this variation must be attributed to Latinisation of a shared ancestor because the vocative forms of mulier and vir match their nominative forms.²³ However, Hugh Houghton has also observed this same Latinisation of Greek words in the closely related gospel manuscript, Codex Sangallensis, which has the same interlinear format as Boernerianus.²⁴

Latinised Greek readings in Augiensis and Boernerianus often do not make sense in their respective clauses. In Rom 14.20, Boernerianus (folio 18r) originally read καταλυε. After destruere was written above it, the scribe added ιν in smaller letters to read καταλυειν, changing the Greek imperative to an infinitive. This matches the Latin usage in which the infinitive is used with noli to form a negative imperative. This Latinised Greek reading is also in Augiensis (folio 27v). In Gal 4.24, rather than αὗται … διαθῆκαι, Boernerianus and Augiensis alone share (pseudo-) singular forms, αυτα … διαθηκη (Boernerianus folio 58r, Augiensis folio 73r) with haec (ue)l ea … testamenta written above the Greek text in Boernerianus. This can be explained as Latinization of the Greek text due to grammatical confusion: haec and ea, which could be either feminine singular or neuter plural, are written above the deficient form αυτα, and the interpretation of this as feminine singular appears to have led to the false ‘correction’ of διαθηκη. In 1 Tim 1.5, rather than ἀγάπη, Boernerianus (folio 86r) and Augiensis (folio 116r) have αγαπηc. In Boernerianus, the Latin word caritas is written directly above it. The genitive case here is nonsensical but could be explained through attraction to the final c of its Latin counterpart above it. In 2 Cor 12.12, rather than τὰ μὲν, Boernerianus (folio 52r) has a peculiar double form αλλε (ue)l ταμεν aligned with sed and t(amen) (ue)l quide(m) respectively, and ταμεν is presented as a single word clearly corresponding to the two options in the Latin text. Augiensis (folio 65) has αλλε·τα·μεν, with the Greek letters written very closely together and the medial dots written in a darker ink and likely later.

The interlinear Latin text also appears to have influenced the misspelling of Greek words. In Gal 1.6, Zimmer argued that Boernerianus (folio 53v) had μαζω, because the m in miror caught the scribe's eye while writing the Greek word, a form also in Augiensis (folio 67v).²⁵ According to Smith, this only demonstrated that the scribes of Boernerianus and Augiensis were ignorant of Greek and copied letter by letter.²⁶ However, in Rom 15.9, the scribe of Boernerianus (folio 19r) initially wrote τον θμ, mirroring the ending of the corresponding Latin text above it, dm, before correcting it to τον θν. This demonstrates that the scribe had enough knowledge of Greek and sometimes enough awareness while copying to correct such a Latinisation in the Greek text.²⁷ Augiensis (folio 25r) agrees with the corrected reading in Boernerianus, τον θν, which also demonstrates that the reading τον θμ is original to Boernerianus and not a feature of a shared ancestor.

There are other similar examples: in Rom 13.14, rather than ɛἰς, Boernerianus (folio 17r) and Augiensis (folio 26r) have εν, influenced by the corresponding in; in Rom 16.15, rather than Νηρɛ́α, Boernerianus (folio 20v) and Augiensis (folio 31r) have νηρεαν, matching neream above; in Gal 3.16, rather than ὅς, Boernerianus (folio 56v) and Augiensis (folio 71r) have ου, matching quo; in Gal 3.28, rather than ἄρσɛν, Boernerianus (folio 57r) and Augiensis (folio 71v) have αρcηc influenced by the corresponding masculus; in Rom 16.15, rather than Ὀλυμπᾶν, Boernerianus (folio 20v) has ολυμπειδα and Augiensis (folio 31r) has ολιμπειδα, a transliteration of the corresponding olympiadem. In Gal 3.16, rather than ɛ̓παγγɛλίαι, Boernerianus (folio 56v) and Augiensis (folio 71r) have επαγγελειccαι, influenced by promissiones above it; in Phil 4.3, rather than γνήσιε, Boernerianus (folio 72v) and Augiensis (folio 90v) have γερμανε, a transliteration of the corresponding germane. This kind of direct influence is best explained as having originated from the proximity of the Greek and Latin texts in an interlinear format, as seen in Boernerianus, where one language being copied after the other led to instances of confusion between them by a copyist who was more proficient in Latin than Greek.

3.1.3 Word Divisions

There are many distinctive word divisions in the Greek text of Boernerianus and Augiensis, which are meant, according to Frede, to aid language study by readers with only a basic understanding of Greek.²⁸ Corssen and Scrivener concluded that the differences in word division demonstrated that the manuscripts were not derived one from the other, but from the same codex which had a continuous text.²⁹ Many of the examples given by Corssen simply show that Augiensis breaks up larger words and mis-associates letters and syllables with the incorrect words, as the scribes of Augiensis were often concerned with copying syllable by syllable rather than word by word – often erroneously.³⁰ This is observable throughout the manuscript. Corssen gives an example from in Rom 4.19, in which Boernerianus (folio 6v) has ουκατενοησεν,το and Augiensis (folio 9r) has ουκ·ατεν·ησεν·το. However, in Boernerianus, non considerauit is written above the Greek word, and con begins over ατε, which would explain the strange division in Augiensis. Also, the dot in the middle may simply be to indicate the place of the added o. Finally, the separation of the final το seems to have originated in Boernerianus.

More important than the differences are those strange word divisions shared by the manuscripts.³¹ The examples below show that the scribes of Boernerianus and Augiensis did not come to these word divisions independently, but they appear to have been influenced by an interlinear Latin text different from the Latin text in Augiensis.³² In Rom 6.9, rather than οὐκέτι … οὐκέτι, Boernerianus (folio 8v) introduces two word divisions and an itacism, ουκ ετι … ουκ ετει, with the corresponding Latin reading, non iam … non ultra, written above each Greek word respectively. While the Latin text of Augiensis (folios 11v–12r) is iam non … non, the Greek text is the same as Boernerianus. In Rom 10.12, rather than Ἰουδαίου τɛ, Boernerianus (folio 13v) has ιουδαι · ουτε aligned with the word division of the Latin text, iudaei neq(ue). Here, ιουδαι mirrors the appearance of iudaei but drops a syllable, and ουτε is a better companion to neque than τε. Augiensis (folio 20v) has the same Greek word division but adds ι, ιουδιαι ουτε, replacing the syllable present in the Latin text as if the word break had already been established. In Rom 11.19, rather than ἐξɛκλάσθησαν, Boernerianus (folio 15r) and Augiensis (folio 22v) have ει · κλαcθηcαν, omitting the prefix, introducing an itacism, and inserting a word division. In Boernerianus, the Latin word si is placed above ει, a sensible counterpart, and fracti sunt, a perfect passive verb, is placed above κλαcθηcαν, an unaugmented aorist passive. In Rom 15.6, rather than ὁμοθυμαδὸν ἐν ἑνὶ στόματι, Boernerianus (folio 18v) has ομοθυμαδονεν εν ειcτοματι. The first Greek word ομοθυμαδονεν – a conflation of ὁμοθυμαδὸν and ἐν – is aligned with the Latin word unanimes. After this, ἑνὶ has been italicised to read ενει and spaced as εν ει with the latter half connected, without space, to the following word, cτοματι. Whereas in is written over εν and ore is written – on the following line – over ματι, uno is written directly over ει or ειc, as it is likely that the scribe would equate uno with ɛἷς. Augiensis (folio 28r) has ομοθυμαδονεν · εν·ει·c τοματι with the letters εν·ει·c written very closely together with medial dots added either by the original hand or an ancient corrector. The large space between c and τοματι has also been subsequently bridged by an underline. In 1 Tim 5.11, rather than καταστρηνιάσωσιν, Boernerianus (folio 89v) has κατα cτρηνειαc ουc ειν with clear word divisions governed by both vowel changes and the Latin text. At the very end of the line, κατα cτρηνειαc is aligned with luxoriatę. On the following line, ουc ειν is aligned with fuerint in. Augiensis (folio 120r) has κατα cτρηνειαc · ουc ειν, the same spelling and word divisions as Boernerianus, and it also includes a medial dot where Boernerianus comes to the end of the line. In 1 Tim 6.14, rather than τηρῆσαί σɛ, Boernerianus (folio 91r) has τη ρηcεcαι. The differences in Boernerianus in both word division and orthography give the appearance that there is a definite article and no pronoun. While this creates a nonsense reading, the Latin phrase, ut c(on)serues, which does not include a pronoun, is written above ρηcεcαι. Augiensis has the same vowel changes and word division as Boernerianus but emphasises the latter with a medial dot, reading τη · ρηcεcαι. These Greek word divisions in both manuscripts have clearly been influenced by a misinterpretation of the Greek text and the word division in the Latin text consistent with an interlinear exemplar.

3.1.4 Corrections

The corrections in Augiensis support its being copied from an interlinear exemplar. In 1 Tim 4.2, in Boernerianus (folio 88r), a Greek word clearly begins with a Latin letter, spelled h υποκριcι.³³ The inclusion of this Latin letter in the Greek text is best explained by influence from a nearby Latin text, such as the Latin word written above it in Boernerianus itself, hypocrisi. Houghton also observes the confusion of the alphabets when written by the same scribe in Codex Sangallensis. Augiensis (folio 118v) originally read υποκριcι, but a Latin h was added to read h υποκριcι, like Boernerianus. Unlike Boernerianus, there is no clear cause within Augiensis itself for this Latin h to have been added to its Greek text, and a Latin h does not appear elsewhere in the Greek text of either manuscript. The best explanation, which was also offered by Scrivener, is that Boernerianus was the source.³⁴ The use of Boernerianus to make corrections in Augiensis is consistent with Boernerianus as the exemplar because the exemplar would have been the likely source for scriptorium corrections.

4 Features Distinctive of Boernerianus

4.1 Misinterpreted Corrections

Though it is not always clear when the corrections in Augiensis were made, there are distinctive features from Boernerianus which appear in the corrections of Augiensis.³⁵ Of the 304 corrections to the Greek of Augiensis in Romans, Galatians, Ephesians and First Timothy, 298 are toward the text of Boernerianus while only six are corrected away from the text of Boernerianus. This also strongly supports the use of Boernerianus for corrections in Augiensis. Further, the same symbols appear in corrections in the same places in both manuscripts. In Rom 8.35, the corrector of Augiensis (folio 17r) uses the abbreviation for uel to denote a correction from cτενχαωρια to cτενcχωρια. The same symbol is used to show a correction here in Boernerianus (folio 11v). Additionally, like Augiensis, the corrected text in Boernerianus is cτενcχωρια rather than στɛνοχωρία.

Whereas corrections in Augiensis show its close relationship with Boernerianus and even support an interlinear exemplar, errors made by the scribes of Augiensis in interpreting corrections in Boernerianus confirm that the latter manuscript itself served as the exemplar. In Rom 7.6, rather than παλαιότητι, Boernerianus (folio 9v) initially had παλλιοτητι before it was corrected to παλαιοτητι. Augiensis (folio 13v) has, very clearly written, the initial reading found in Boernerianus, παλλιοτητι. Normally, when the scribe of Boernerianus wrote α in the middle or end of a Greek word it was written as a Latin a, which is not easily confused with λ. However, because the scribe of Boernerianus initially wrote λ here, the corrected α takes a form which Boernerianus reserved for capital letters and is also very easily mistaken for λ. This abnormal use of the capital form of α in Boernerianus, written right next to λ, would have been very easy for a copyist to mistake for λ. This is the best explanation for the reading in Augiensis, and, given the peculiarity of this form of the letter in Boernerianus, is an indication that it was the exemplar for Augiensis. Confusion between these letters will be further discussed below.

In Rom 8.34, rather than ὑπέρ, Boernerianus (folio 11v) initially had περει, omitting υ and adding ει. With the word εντυγχανει immediately preceding it, it is possible that the scribe's eye skipped, and so the letters ει were written again. In the space between εντυγχανει and περει, υ was clearly added by the scribe afterward, but the additional ει was never deleted. The resulting reading is υπερει. This nonsense reading is written very clearly in Augiensis (folio 17r). It is possible that it appeared in a shared exemplar. However, it is far more likely that the scribe of Boernerianus made a mistake and corrected most of it but forgot to delete ει, which was then copied into Augiensis directly from Boernerianus.

In Rom 8.35, which is discussed above, the scribe of Boernerianus (folio 11v) wrote cτενcαωρια rather than στɛνοχωρία, confusing the ο for c (these are lunate sigmas as used in both manuscripts) and χ for α. The scribe then made a correction by writing (ue)l χ above cα apparently correcting the α to χ. The scribe of Augiensis (folio 17r) misinterpreted this as a correction from c to χ and wrote cτενχαωρια. As noted above, the corrector of Augiensis, using the same abbreviation for uel as in Boernerianus, then corrected the text to read cτενcχωρια, matching the correction in Boernerianus but also repeating the mistake of rendering ο as c. The misinterpretation of the correcting notation in Boernerianus, the resulting nonsense readings in both manuscripts, and the use of the uel abbreviation all indicate that Boernerianus was the exemplar for Augiensis.

In 1 Cor 14.19, rather than ἢ μυρίους, the scribe of Boernerianus (folio 36r) wrote η · υ μυριουc before deleting the υ with an underdot, which shows that the scribe was aware of the mistake. Based on the scribal habits in Boernerianus, it is unlikely that the scribe would have corrected the spelling if this additional υ had appeared in its exemplar. So, the best explanation is that the error was introduced by the scribe of Boernerianus, who then corrected it. The error is easily explained by the presence of the υ in the following word μυριουc. Augiensis (folio 51v) also includes υ but never deletes it, reading ·η·υμυριουc· like the original reading in Boernerianus. While the medial dot between η and υ was added by a later corrector, the rest of the text was written very clearly without any word division or other notations. Because this error 1) is best explained as having been made and corrected by the scribe of Boernerianus and 2) has been copied directly into Augiensis from its own exemplar, this supports Boernerianus as the exemplar for Augiensis. The correction was likely missed during the copying of Augiensis, because the underdot in Boernerianus is slightly elongated and might have been confused for a notation in the interlinear Latin text below.

In 2 Cor 3.12, the scribe of Boernerianus (folio 43v) wrote του αυτην before correcting it to τοι αυτην by writing ι down the middle of υ. As with the previous example, the scribe of Boernerianus was clearly aware of the mistaken letter and then corrected it, and so the error and subsequent correction in Boernerianus are best explained as being original to its own scribe and caused by the υ in the following word αυτην. Augiensis (folio 108r) has το · αυτην, omitting the letter altogether. If the letter had been omitted in a shared exemplar, then it would not have appeared in Boernerianus or been corrected by the scribe. Rather, the omission in Augiensis is best explained as a scribe of Augiensis misinterpreting the correction in Boernerianus as a deletion.

In 1 Tim 1.9, Boernerianus (folio 86r) initially included αλλ in the Greek text with the standard Latin reading, sed, written above it. A correction was then made by lightly striking through the lower part of the Greek letters and putting five dots underneath the Greek word. This addition of αλλ appears to have been influenced by sed, which corresponds to the postpositive δὲ in the standard Greek text. However, as this was written as τε in Boernerianus, there seemed to be nothing to correspond to the Latin sed. The hand which deleted αλλ appears to have added aute(m) over the final τε, after realising the correct correspondence of the two languages. Further, aute(m) appears to be slightly darker and more compressed, suggesting that it was added at a later point. More importantly, αλλ has a clear Latin counterpart, so there would have been no reason for the scribe of Boernerianus to have deleted it if it had been in its exemplar. Augiensis (folio 116r) has αλλ with no deletion, which means that it was present in its own exemplar, eliminating the possibility that Augiensis and Boernerianus had the same exemplar. The best explanation is that Augiensis copied αλλ from Boernerianus before Boernerianus was corrected.³⁶

In 1 Tim 2.2, rather than ἤρεμον, Boernerianus (folio 87r) initially had ηρειον, which includes an itacism and omits μ. The omission of μ is easily explained as the result of the scribe's eye skipping to the next adjective, which was also written with an itacism, spelled ηcυχειον rather than ἡσύχιον. Before completing the corresponding Latin text above ηρειον, the scribe corrected this mistake by writing the letter μ above and slightly to the left of ι, which was not otherwise marked for deletion. It appears that μ was either meant to replace ι or was simply meant as an addition while maintaining the itacism, reading ηρειμον. Augiensis (folio 117r) has ηρεμιον. Though it is possible that the exemplar of Augiensis had ηρεμιον, which would be difficult to explain, a much simpler explanation is that Augiensis misinterpreted the correction in Boernerianus and, based on its position, placed μ before ι rather than after it, thereby dividing the itacism of Boernerianus and creating a new form altogether.

Some instances are more complicated than others, such as Rom 15.22–4 (Boernerianus folio 19v, Augiensis folios 29r–29v). The Greek text from both manuscripts is transcribed in Table 1 according to the line breaks of Augiensis.

Table 1. Rom 15.22–4 in Augiensis and Boernerianus

Before discussing the noticeable differences, it is important to discuss the important similarities. In Rom 15.23, on line 3 in the table, rather than νυνὶ δέ both Boernerianus and Augiensis read νυν ειδε. In Boernerianus, this reading is aligned with nunc uero. Whereas Boernerianus usually aligns δε with autem, the Latin word uero here likely influenced the Greek word break as the scribe judged ιδε to be a better match than δε. Also, on line 3, rather than μηκέτι τόπον, Boernerianus reads μη κετει τοπον, and Augiensis reads μη · κετειτοπον. In Boernerianus, μη κετει is initially written as a single word, μηκετει, but is then divided in two by a semicolon. The same word division is made in Augiensis with a medial dot. In Rom 15.23, on line 4 in the table, rather than reading τοῖς κλίμασιν, Boernerianus and Augiensis read τοιc κλημαcιν. Likewise, in Rom 15.23, on line 5 in the table, rather than reading ἐπιποθίαν, Boernerianus reads επειποθειαν, and Augiensis reads επειπο θειαν. On the same line, rather than reading ἔχων, Boernerianus and Augiensis read εχω. This reading, which agrees with the Latin text, is also attested by Claromontanus.

In Rom 15.22 and 15.23, the scribe of Boernerianus has transposed the Greek text, απο πολλων αιτων ωcανουν πορευομαι, from Rom 15.23–4 into Rom 15.22. It appears as if the scribe's eye has jumped from προc υμαc to προc υμαc, which precedes it both times. The first section of this Greek text, απο πολλων, has been underlined for deletion, but above it is written the Latin text ex multis ia(m). The remaining erroneously added text, αιτων ωcανουν πορευομαι, appears on the following line with six small dashes in place of the Latin text to denote deletion. This misplaced text also appears in Rom 15.23 and 15.24 after the phrase του ελθειν προc υμαc.

Augiensis does not include απο πολλων αιτων ωcανουν πορευομαι in Rom 15.23 and 15.24, but only πορευομαι in Rom 15.24. In Rom 15.22, the rest of the text – apart from ελθειν υμαc – is rendered almost exactly as it is in Boernerianus, απο πολλων · αιτων·ωcανυν. The simplest explanation is that the scribe of Augiensis has misunderstood the deletion in Boernerianus but has also only copied the Greek text which matches its Latin text in the parallel column, which is consistent to what has been observed with the lacunae. The Latin text ex multis iam pręcedentibus annis is aligned with the Greek text απο πολλων · αιτων·ωcανυν.

4.2 Divergences

Divergences between the two manuscripts will be further described below, but one kind of divergence is pertinent here. While many of the divergences between the two manuscripts are caused by confusion between letters which appear or sound like one another, the most asymmetrical example of this confusion occurs between α, δ, and λ, as noted in the example from Rom 7.6. This is significant because there are two different forms of α in Boernerianus: 1) the more common form is identical to the Latin a, 2) the less common form is the capital which looks like δ and λ. Therefore, it would be much more likely for a scribe copying from Boernerianus to go from δ or λ to α than from α to δ or λ. Augiensis changes δ to α in seven different instances, but does not change α to δ. Twice λ has been changed to α. Twice α has been changed to λ. The latter change occurs in 1 Tim 2.12, where a capital α appears in Boernerianus (folio 87v), and in Gal 3.17, where a stroke is written very lightly in the non-capital form so that α in Boernerianus (folio 56v) strongly resembles λ. Additionally, Boernerianus and Augiensis otherwise agree where α has been confused with δ or λ. For example, in Rom 12.20, rather than the RP2005 reading, πɛινᾷ ὁ, Boernerianus (folio 16v) and Augiensis (folio 25r) each have πεινλο written as a single word. The simplest explanation is that Boernerianus used an uncial Greek manuscript as an exemplar, and then Augiensis was copied from Boernerianus, compounding some of these uncial errors and making additional ones based on the script of Boernerianus.

5 Divergences between Boernerianus and Augiensis

While it has been otherwise established that Boernerianus was the exemplar of Augiensis, Scrivener identified 1984 places of divergence between Boernerianus and Augiensis but counted 200 as ‘real various readings’.³⁷ Hatch argued that a few divergences, attested by other textual authorities, supported the presence of multiple textual traditions affecting the different ancestors of Augiensis and Boernerianus subsequent to their hypothetical shared ancestor. ³⁸ However, Kloha observed that the shared lacunae show that there was no such interaction with other Greek manuscripts, otherwise the lacunae would have been filled in.³⁹ Here, I will show that the divergences between the manuscripts can be explained as a combination of visual and phonetic errors within the scribal practice of Augiensis.

5.1 Internal Acoustic Rendering

Whereas Scrivener and Smith only allowed for visual copying errors,⁴⁰ Alphonse Dain acknowledged a more complicated copying process recognising four stages: 1) the copyist reading the text, 2) retaining the text, 3) dictating the text to oneself, and 4) the action of the hand.⁴¹ Dirk Jongkind elaborates on how this process might give rise to complex variation between a manuscript and its exemplar as each stage of the process ‘carries its own dangers and risks’.⁴² Jongkind explains that in the various stages the scribe might misread a text, misdivide a word, misremember word order, omit or transpose syllables and words, or transcribe things suggested by the context but are not actually there. The scribe's own grammatical preferences, internal dictionary, train of thought or various other distractions might also influence what is written.⁴³ Jongkind also writes, ‘Junack draws particular attention to the effect which a text written in scriptio continua has on the process of reading. In order to establish the sense of such a text, one has to create an (internal) acoustic rendering of the syllables, as one does not have the word-image as the basic interpretative unit.’⁴⁴ Though Boernerianus is not written in scriptio continua, this might also be applied to a bilingual text written by non-native speakers. The process of internal acoustic rendering seems to be the best explanation for having both phonetic and visual differences, which are often inconsistent and sometimes on multiple occasions within the same syllable, word or phrase. It also accounts for the misdivision of words and syllables as discussed above. Phonetic differences, visual differences, differences caused by the organisation of the Latin column of Augiensis, and other causes are enumerated in Table 2 and will be discussed below.

Table 2. The number of differences between Boernerianus and Augiensis in Romans, Galatians, Ephesians, and 1 Timothy

5.2 Phonetic and Visual Differences

In my test corpus of Romans, Galatians, Ephesians and First Timothy combined, there are 821 differences between the manuscripts, including 638 which might be considered phonetic differences. Most of these have to do with individual letters and syllables. Augiensis interchanges ο and ω 128 times and ε and η eighty-six times.⁴⁵ The clear inclination toward shortening the vowel might be explained by the scribe rendering the vowels in a Latin manner, in which ο and ε are the only options. It is also an indication of the textual flow between the two manuscripts which would be very consistent with Boernerianus as the exemplar of Augiensis. Whereas ο is lengthened to ω thirty-two times, ω is shortened to ο ninety-six times. Similarly, ε is lengthened to η nineteen times, and η is shortened to ε sixty-seven times. Seven times ι is changed to ει, fifteen times ει is changed to ι. Twelve times ι is changed to η, and twenty-three times η is changed to ι. Additionally, ι is changed to υ sixteen times, and υ is changed to ι forty-seven times. Augiensis also has an inclination toward replacing θ with non-aspirated dentals. While Augiensis changes δ to θ two times and τ to θ one time, θ is changed to δ eight times, and θ is changed to τ nine times. Rather than doubling γ or κ, the initial letter is often replaced with ν to better denote the way the sound is vocalised by a non-native speaker.

In Romans, Galatians, Ephesians and First Timothy, in addition to the 638 phonetic differences, there are also 177 visual differences between the manuscripts. The majority of these are also individual letters and syllables. Augiensis drops a vowel from between two consonants twenty times. This does not occur with η or ω. There are ten instances in which Augiensis adds a syllable, and fourteen instances in which a syllable is omitted. There is repeated uncial error in Augiensis seen in the rendering of the following sets of letters: α, δ, and λ; η, ν, and π; c and ε (due to the lunate sigmas). Of these visual differences, Augiensis has π for ν seven times but ν for π only once. The scribes change οι to ιο four times and ιο to οι once. More of these phonetic and visual errors will be discussed below, along with other differences which can be attributed to other scribal habits.

5.3 Word Order

In Romans, Galatians, Ephesians and First Timothy combined, there are only three instances in which Augiensis has a different word order than Boernerianus. In Rom 15.5, Boernerianus (folio 18v) has χν ιν, the RP2005 reading, while Augiensis (folio 28r) has ιν χν, the same word order as its Latin text, ihm xpm. As the word order in Augiensis follows the standard word order in later usage, it is an obvious change. Augiensis keeps the same word order in its Greek word order in 1 Tim 5.21. While Boernerianus (folio 90r) has χρυ ιυ, Augiensis (folio 120v) has the RP2005 word order, ιυ χρυ, even though its Latin text follows the Vulgate order, which is the same word order as Boernerianus. This difference might be attributed to scribal habit. In Rom 15.9, the RP2005 has τῷ ὀνόματί σου ψαλῶ, and the Stuttgart Vulgate – agreeing in word order – has nomini tuo cantabo. The same sequence is found in the Greek and Latin texts of Augiensis (folio 28v). Claromontanus (folio 80v) follows this word order in its Latin text, but the Greek text moves ψαλω to the beginning of the phrase, ψαλω τω ονοματι cου. Both the Greek and Latin texts of Boernerianus (folio 19r) follow the same word order as the Greek text of Claromontanus, ψαλω τω ονοματι cου and psalla(m) (ue)l cantabo nomini tuo respectively. Based on what has been discussed above, it is not surprising to find the Greek word order of Augiensis adjusted to the Latin text here. In Augiensis, there is also a very clear line written by the original hand directly above the final letter of ψαλω, which appears to be an acknowledgement of the change in word order rather than an abbreviation.

5.4 Word Omissions and Additions

In Romans, Galatians, Ephesians and First Timothy combined, Augiensis has only four Greek words not found in Boernerianus but lacks thirty-one – also an indication of the textual flow between the two manuscripts. The thirty-one words omitted are spread across twenty-five places. There are seventeen individual, monosyllabic words which can be explained by the scribe's own internal acoustic rendering as either phonetic or visual mistakes. There are also three individual, multisyllabic words and four phrases, which can all be explained by visual copying error.

The additions are also easily explained. While there is one addition in Galatians, three of these additions are in Romans. At the end of Gal 6.18, Augiensis (folio 76r) attests αμην along with all other Greek manuscripts chosen for the Editio Critica Maior of Galatians apart from Boernerianus (folio 60v). However, the previously discussed influence of the Latin text of Augiensis and the presence of amen in its Latin text together sufficiently explain the addition in the Greek text. In Rom 3.21, Boernerianus (folio 5r) attests the RP2005 reading θυ. Augiensis (folio 7r) has του θυ, an addition which can be attributed to the habit of writing the definite article before θεοc.

While one of the additions is consistent with Boernerianus as the exemplar, the other shows that Boernerianus is the exemplar. In Rom 9.31, while Boernerianus (folio 13r) has νομον δικαιοcυνηc ειc νομον, Augiensis (folio 19v) has νομον δικαιοcυνηc ειc νομον ÷ δικαιοcυνηc. Though many manuscripts, like Boernerianus, lack the second δικαιοcυνηc, it is present in the RP2005 and the Vulgate, reading iustitiae, which is also attested in the Latin text of Augiensis as iusticiae. However, this second δικαιοcυνηc in Augiensis is very clearly marked off by the obelus symbol ÷, which denotes that this reading was not present in the exemplar. The scribe deliberately made an addition to the Greek text to conform to the Latin text. In Rom 7.19, rather than ἀλλ’ ὃ οὐ θɛ́λω κακὸν τοῦτο, Boernerianus (folio 10r) lacks οὐ θέλω, reading αλλ ο κακον τουτο. Augiensis (folio 14v) also lacks οὐ θέλω but adds μειcω in its place, reading αλλ ο μειcω κακον τουτο, which is aligned with sed quod odio malum illud in the Latin column.⁴⁶ This text in Augiensis has been affected by the text of Rom 7.15 in Boernerianus. Six lines directly above Rom 7.19 in Boernerianus (folio 10r), Rom 7.15 reads αλλ ο μειcω τουτο with sed quod odio hoc (ue)l illud written over it, providing a source for the orthography of μειcω and odio. While copying Rom 7.19 in Augiensis, the scribe's eye skipped to Rom 7.15 above, which looks very similar, transposing the form μειcω into Rom 7.19.

5.5 Lexical Differences

In Romans, Galatians, Ephesians and First Timothy combined, Augiensis has six Greek lexical differences from Boernerianus, which can all be explained as visual or phonetic error. Additionally, two appear to have been corrected by the original hand. In Gal 4.18, whereas Boernerianus (folio 58r) has το, Augiensis (folio 73r) appears to have had a different reading but then was corrected by the original hand to το. Again, in Rom 4.11, whereas Boernerianus (folio 6r) has δια τηc δικαιοcυνηc, Augiensis (folio 8r) initially had και τηc δικαιοcυνηc, but the original hand corrected και to δια. The initial mistake is easily explained by the scribe's eye skipping ahead to δικαιοcυνηc. However, because και is written at the end of its line in Beornerianus, it could be better explained by the scribe's eye skipping directly above to the end of the previous line which also ends with και. Hugh Houghton and Amy Myshrall have identified this kind of eye skip between lines as an indication of direct copying.⁴⁷ The scribe then noticed the mistake and corrected it. In” Rom 5.18, Boernerianus (folio 8r) has το δικαιωμα while Augiensis (folio 11r) has και δικαιωμα. This difference is very easily explained by reading ahead to δικαιωμα. In Gal 3.28, while Boernerianus (folio 57r) has ειc την, Augiensis (folio 71v) has εcτιν. The phonetic error likely prompted the mistake in word division. In Gal 6.14 and 6.18, whereas Boernerianus (folio 60v) has κυ, Augiensis (folios 75v and 76r) has χυ. Though κυριου and χριcτου look and sound different when written out in full, the nomina sacra are easily confused if the copyist has visually mistaken κ for χ, a common copying error in New Testament textual tradition. Another explanation for the instance in Gal 6:18 is like the instance in Rom 4.11 given above. In Gal 6.18, κυ begins its line in Boernerianus, and χρυ is written directly above it at the beginning of the previous line. The scribe's eye might have skipped to the previous line so that χυ was written rather than κυ. These readings in Augiensis in Gal 6.14 and Gal 6.18 both give the same nonsense reading χυ ημων ιυ χυ.

5.6 Compounded Errors

Of the 821 divergences, 108 are instances in which Augiensis has compounded an aberration introduced by Boernerianus. These are changes in individual letters and syllables included in the visual and phonetic differences discussed above. These 108 instances reveal a two-step process. First, the scribe of Boernerianus introduced an aberration while copying its own exemplar. Secondly, one of the scribes of Augiensis reproduced the same aberration while copying from Boernerianus and then introduced an additional error, moving it even further away from the reading found in the RP2005.

For example, in Rom 13.10, rather than τῷ πλησίον, Boernerianus (folio 17r) has τω πληcειω. The scribe made a phonetic mistake in rendering ι as ει. Additionally, writing ον as ω was either a visual mistake or the scribe's harmonising the noun to the preceding pronoun. The scribe of Augiensis (folio 26r) recorded the same two changes but then changed η to ει. While the RP2005 has τῷ πλησίον, and Boernerianus has τω πληcειω, the reading in Augiensis (folio 26r) τω πλειcειω, in which η was written rather than ει, appears to have been developed from the reading in Boernerianus. This word was eventually corrected in Augiensis but only in so far as to agree with Boernerianus, which also shows that Boernerianus was used to make corrections.

In these 108 instances, it is conceivable but unlikely that Boernerianus copied perfectly from a shared exemplar from which Augiensis diverged. It is much less likely and hardly conceivable that Boernerianus copied from Augiensis and made partial corrections toward the Majority reading. The best explanation is that the scribes of Augiensis copied from Boernerianus and introduced additional mistakes.

6 Conclusion

After enumerating the differences between Boernerianus and Augiensis, Scrivener wrote, ‘Elsewhere the Greek texts of these manuscripts are identical, coinciding in the minutest points.’⁴⁸ While the different stages of the copying process can account for differences between the manuscripts, certain phenomena found in Augiensis originated in Boernerianus. This is demonstrated in unique readings – including nonsense readings, word divisions, and compounded errors. Ultimately, the way in which Augiensis has misinterpreted the corrections within Boernerianus itself confirms that the simplest explanation of their relationship is that Boernerianus is the exemplar. To hypothesise a shared exemplar [Scrivener and Corssen], or a further intermediate stage [Smith and Hatch], which contained all of the peculiarities connected to the interlinear layout of Boernerianus and its peculiar errors of spelling and word division is unnecessary [Zimmer], given that this article has demonstrated how the very few examples of differences in the Greek text can all be explained as deviations in the process of copying Boernerianus. Once the different origin of the Latin texts in these manuscripts has been recognised and removed from the consideration of their relationship, the evidence that the Greek text of Augiensis is a copy of Boernerianus is conclusive.