Μετὰ δὲ ταῦτα ἀνέδειξεν ὁ κύριος ἑτέρους ἑβδομήκοντα °[δύο] … (Luke 10.1)
Ὑπέστρεψαν δὲ οἱ ἑβδομήκοντα °[δύο] … (Luke 10.17)
1. Introduction
I recently had the privilege of examining several folios of P45 at the Chester Beatty Library in Dublin.Footnote 1 This first-hand inspection confirmed my suspicion that most modern critical editions incorrectly cite the papyrus at Luke 10.17. In this uniquely Lukan passage manuscripts are evenly divided over whether Jesus sent out seventy or seventy-two disciples, an unresolved textual problem that occurs once in 10.1 and again 10.17. While the first instance in P45 is lost due to fragmentation, the second is visible although its unfavourable position at the deteriorated edge of the papyrus has caused some confusion. Personal examination verified that, contrary to the presentation of standard critical texts, P45 should be listed in support of ἑβδομήκοντα (70) rather than ἑβδομήκοντα δύο (72) – an error that has been observed by others but never sufficiently rectified.
As P45 is possibly the earliest extant witness to the passage, this revision of evidence invites a renewed look at the textual problem. Previous attempts to work out this ‘insolvable’ problem have relied almost exclusively on appeals to the perceived symbolic meaning(s) of the numeral(s), mostly drawn from traditions in the Jewish Bible. Unsurprisingly, this approach has only led to an impasse. Both seventy and seventy-two have numerous possible symbolic meanings that could have motivated the evangelist to use them in Luke 10, and simply choosing one over another as ‘the more Lukan’ seems hopelessly unprovable. In spite of the obvious limitations of this reliance upon numerical symbolism, little reflection has been given to other, more promising, paths of research. Specifically, no serious consideration has been given to the scribal techniques of number writing and, in particular, the transcriptional factors that led copyists to alter numerals.Footnote 2
In a brief treatment of this variation unit, Bruce M. Metzger remarked on the transcriptional probabilities involved in the problem and saw no hope of a resolution:
The factors bearing upon the evaluation of internal evidence, whether involving transcriptional or intrinsic probabilities, are singularly elusive. It is likely that in most of the early manuscripts (as in P45 and P75) the numeral was written with letters of the alphabet (either ο̅β̅ or ο̅). It was easy, therefore, for either number to be accidentally altered to the other.Footnote 3
Yet is the matter really so simple? Were copyists equally as likely to mistake ο̅ for ο̅β̅ as they were to mistake ο̅β̅ for ο̅? The purpose of the present essay is to address this particular question and explore the transcriptional dynamics in greater detail. As we will see, a survey of related numeral corruptions in New Testament manuscripts reveals a clear tendency among scribes (or possibly readers) that favours one of these options over against the other. Accordingly, in what follows we will (1) point out the correct wording of P45 at Luke 10.17, (2) review previous attempts to solve the textual problem, and then (3) draw attention to a valuable body of transcriptional evidence that points in favour of ἑβδομήκοντα δύο as the earlier wording.
2. The Witness of P45 at Luke 10.17
P45, or P.Beatty i, is an invaluable manuscript of the Gospels and Acts that is dated to the third century ce. Its precise wording at Luke 10.17, however, has been the epicentre of disagreement and substantial confusion. Two factors have undoubtedly contributed to, if not totally created, this problem: (1) the number is written as an alphabetic numeral (ο̅ or ο̅β̅) rather than a longhand word (ἑβδομήκοντα or ἑβδομήκοντα δύο),Footnote 4 and (2) it occurs at the very end of its line of text, right on the fragmentary edge of the papyrus leaf.
Frederic G. Kenyon's 1934 publication of P45 recorded the wording in question as ο̅β̅, Greek shorthand for ἑβδομήκοντα δύο (= 72),Footnote 5 and critical editions naturally followed suit.Footnote 6 In 1953, however, papyrologist Colin H. Roberts saw a different value and proposed that the actual wording was ο̅ϛ̅ (= 76), the latter character being a stigma or digamma – an obsolete letter still used in Koine Greek as the number ‘six’.Footnote 7 (I know of no critical editions that cite P45 to this effect.) Yet a third value was seen in 1959 by Bruce M. Metzger, who corrected both previous proposals in an essay on the textual problem as a whole. In his own words,
[t]he present writer has examined this passage in P45 under natural and artificial light, and has assured himself that the Greek character which follows the letter omicron (standing for ‘70’) is neither β, as Kenyon supposed, nor ϛ, as Roberts thinks, but merely the diplé, or space-filler (>), which scribes would use occasionally in order to bring an otherwise short line even with the right-hand margin of the column. In fact, by consulting Kenyon's volume of Plates of P45 anyone can see the similarity between the disputed character and the diplé which appears on the same folio near the top of the column.Footnote 8
Indeed, Kenyon's edition of plates is sufficiently clear for one to observe the similarity between the character in question (fol. 11v, l. 20) and the line-fillers that occur on the same page (10.8, fol. 11v, l. 5; see Figures 1 and 2) and the one prior (9.56 and 58, fol. 10v, ll. 19 and 23).Footnote 9 Further, the nearby word καταβιβασθηση contains two betas that can be used for comparison, and there is very little likeness between these betas and the second character (10.16; see Figure 3).Footnote 10 Roberts’ suggestion of stigma/digamma is equally untenable; this reading is otherwise unattested among other witnesses, and the character itself bears no real resemblance to a stigma/digamma (ϛ or ϲ versus >).Footnote 11 Other scholars have since affirmed Metzger's solution of ο̅ followed by a diple.Footnote 12 After personally inspecting the folio with all three options in mind, I too am persuaded that Metzger's proposal is by far the most likely. (In addition to the figures included here, digital images of the papyrus are now available online and may also be consulted.)Footnote 13
Luke 10.17 in P45 (fol. 11v, l. 20). © The Trustees of the Chester Beatty Library, Dublin. Photographed by CSNTM.
Comparative line-filler (fol. 11v, l. 5). © The Trustees of the Chester Beatty Library, Dublin. Photographed by CSNTM.
Comparative betas (fol. 11v, l. 18). © The Trustees of the Chester Beatty Library, Dublin. Photographed by CSNTM.
Metzger's proposal of ο̅/70 appeared to win the day. Beginning with the Nestle24 (1960), P45 was moved to support the shorter reading, ἑβδομήκοντα, and Metzger in fact took personal credit for this change.Footnote 14 Yet this victory was short lived. For reasons not entirely obvious, in NA26 (1979) and UBS3 (1975) P45 found its way back in support of the longer reading, ἑβδομήκοντα δύο. And so it has remained up to the present in the NA28 (2012) and UBS5 (2012), albeit with the added note of caution ‘vid’ (= ut videtur, ‘seemingly’).Footnote 15 Other modern editions such as Hodges–Farstad (1982) reflect this curious revision. It is not clear why the witness of P45 was (again) reversed, but we have reason to believe that its presence there is problematic and misleading – the ‘vid’ notwithstanding.
3. The External Evidence
Identifying the erroneous citation of one of the two earliest witnesses leads naturally to a reconsideration of the textual problem of the seventy(-two). To date, no consensus has been reached about the preferred wording, and when scholars do opt for one over the other rarely is it asserted with confidence. This is largely due to the fact that the external testimony for this variation unit is balanced. After revising the wording of P45, the Greek and versional evidence is presented in the NA28 at Luke 10.17 in the following way:
ἑβδομήκοντα δύο: P75 B D lat sys.h sa boms
ἑβδομήκοντα: P45 א A C K L N W Γ Δ Θ Ξ Ψ 0115 f 1. 13 33 565 579 700 892 1241 1424 2542 𝔐 f i q syc.p bo
Although the shorter reading is backed by several early manuscripts and the majority of witnesses, any reading supported simultaneously by P75, B and D (plus some versions) is typically given special consideration by most text critics. On the other hand, however, the support for the shorter reading is also early and widespread (e.g. א, A, C). This arrangement of witnesses led the editorial committee of the Greek New Testament to conclude that ‘the external evidence is almost evenly divided’.Footnote 16
Evidence from the Church Fathers provides valuable context for the present discussion, and, at the same time, introduces many of the difficulties specific to this variation unit.Footnote 17 One important caveat is that most of these Fathers attest to an awareness of a tradition involving ‘seventy’ or ‘seventy-two’ followers of Jesus, but few of the citations that follow are direct quotations of or comments on Luke 10.1/17 itself (which perhaps explains the relatively modest list given in NA28: Irlat, Cl, Or, Ad). Further, no patristic writers seem to acknowledge the existence of the textual problem at this point. That said, both ‘seventy’ and ‘seventy-two’ appear in patristic sources as early as the second century (the following examines Fathers cited in editions at both 10.1 and 10.17).
To begin with, both IGNTP and UBS5 list Marcion as a witness to ‘seventy’, an inference made from Tertullian's statement in Adversus Marcionem, in which he suggests the following: adlegit et alios septuaginta apostolos super duodecim. quo enim duodecim secundum totidem fontes in Elim, si non et septuaginta secundum totidem arbusta palmarum? Footnote 18 Tertullian here draws an interesting parallel between the Old and New Testaments: just as Jesus chose twelve disciples and seventy others, the nation of Israel was led out to Elim – where they found twelve fountains and seventy palm trees (Exod 15.27; Num 33.9). While this passage does make clear that Tertullian is a witness to ‘seventy’, it is nonetheless difficult to ascertain the degree to which this represents the text of Marcion, and so caution ought to be exercised here.Footnote 19
Irenaeus is also cited as a witness to ‘seventy’, since on two occasions he makes reference to ‘the seventy others’ in addition to the twelve disciples.Footnote 20 These references have, however, come down to us in Latin translation (hence Irlat). Finally, UBS5 lists the Diatessaron as a witness to ‘seventy-two’. Evidence for this comes from the commentary by Ephrem, which twice makes reference to the ‘seventy-two’ others (Comm. 5.18; 14.14), though both references appear in contexts other than the pericope in question. In his comment on Luke 10 itself, Ephrem simply states that the followers were sent ‘two-by-two’ without further specification.Footnote 21
Turning to third-century witnesses, Clement at one point quotes a passage from the Epistle of Barnabas and parenthetically refers to that author as ὃ δὲ τῶν ἑβδομήκοντα, or ‘one of the seventy’.Footnote 22 Tertullian also, as mentioned above, makes a reference to ‘the seventy’ in addition to the twelve apostles, and suggests the connection to the seventy palms at Elim. In the writings of Origen, we find evidence of both ‘seventy’ and ‘seventy-two’. In two separate passages (extant only in Latin), Origen draws the very same parallel cited by Tertullian: the seventy palms at Elim are mirrored by the seventy in Luke.Footnote 23 Elsewhere, however, Origen is a witness to ‘seventy-two’. In his commentary on Romans (also dependent on Latin translation), he proposes that Paul's description of Adronicus and Junia – nobiles in apostolic – probably meant that they were ex illis septuaginta duobus.Footnote 24
UBS5 also lists Hippolytusarab as support for ‘seventy’, though this must be qualified sufficiently. ‘Hippolytusarab’ here refers to a collection of fragments of an Arabic commentary on the Pentateuch, mediated through a Syriac translation of an original Greek, and not certainly from the hand of Hippolytus himself. One of these fragments mentions that the ‘seventy evangelists’ spent forty days preaching in the same way in which the Israelite spies sent into Canaan spent forty days scouting out the land (Num 13.25).Footnote 25
Regarding patristic sources from the fourth and fifth centuries, those witnessing to ‘seventy-two’ are Adamantius,Footnote 26 Apostolic Constitutions,Footnote 27 AmbrosiasterFootnote 28 and Augustine.Footnote 29 Somewhat more numerous are witnesses to ‘seventy’: Eusebius,Footnote 30 Basil of Caesarea,Footnote 31 Cyril of AlexandriaFootnote 32 (who makes a connection with the seventy elders chosen by Moses in Exod 24.1), Theodoret,Footnote 33 AmbroseFootnote 34 and Jerome.Footnote 35 Finally, we should note John Chrysostom, who, though listed by UBS5 as a witness to ‘seventy’, actually attests both readings.Footnote 36
In all, this survey of the patristic and manuscript evidence has shown that while both values have early and widespread support, ‘seventy’ seems to have ‘enjoyed a somewhat wider currency than seventy-two’.Footnote 37 One further observation is also important: as early as Tertullian, the number of those sent by Jesus was frequently bound together with Old Testament traditions of the number ‘seventy’. The number of Jesus’ followers was not simply random; it was prefigured in the palm trees at Elim or in the elders chosen by Moses. As we will see, the perceived symbolic import of the numeral continues to loom large in most modern discussions of the textual problem.
4. Problematic Appeals to Numerical Symbolism
In attempts to decide between the two readings with such tightly balanced external support, scholars have relied almost completely upon considerations of the symbolic meanings of the two numbers. That is, it is presumed that the evangelist intended some figurative meaning in the sending of the seventy(-two), and one simply needs to identify the value that best expresses that symbolism to find the earlier wording.Footnote 38 This has, however, proven extremely difficult. Below I list some of the major occurrences of the numbers in Jewish tradition (though this is not exhaustive), first of ‘seventy’:Footnote 39
-
• 70 nations descended from Noah (Genesis 10, Masoretic Text)
-
• 70 offspring of Jacob (Exod 1.5; Deut 10.22)
-
• 70 palms at Elim (Exod 15.27; Num 33.9)
-
• 70 elders chosen by Moses (Exod 24.1, 9–10; Num. 11.16, 24–5),
-
• 70 sons of Jerubbaal (Judg 9.2)
-
• 70 sons of King Ahab (2 Kings 10.1)
-
• 70 years of exile prophesied by Jeremiah (Jer 25.11–12; 29.10)
-
• 70 priests of Bel (Bel and the Dragon 10),
and then of ‘seventy-two’:
-
• 72 nations descended from Noah (Genesis 10, LXX)
-
• 72 princes and nations of the world in 3 Enoch (3 Enoch 17.8; 18.2–3; 30.2)
-
• 72 translators of the LXX (Letter of Aristeas 35–51)
This abundance of possibilities has led to several contrasting proposals. For example, Sidney Jellicoe argued that Luke's fondness for the Septuagint in scriptural quotation indicates that ‘seventy-two’ is the more likely original reading: ‘Just as the seventy-two emissaries of Aristeas had, by their translation [of the Greek Pentateuch], brought the knowledge of the Law to the Greek-speaking world, so the Seventy(-two) are divinely commissioned to proclaim its fulfilment in the Gospel message.’Footnote 40
On the other hand, Kurt Aland argued that, given the ‘overwhelming’ examples of ‘seventy’ in the Old Testament, ‘it is astonishing that the reading ἑβδομήκοντα δύο occurs at all’, which would mean that the latter is to be preferred because of its relative scarcity. Scribes and readers would be more likely to ‘normalise’ the comparatively unfamiliar number (72) to the one with plentiful examples in Jewish tradition (70), not the reverse.Footnote 41 Still others prefer to see ‘seventy’ as a recapitulation of Moses’ appointing of the seventy elders (Num 11.16), especially given other Mosaic themes in the narratives of Luke-Acts.Footnote 42 This view, however, would be complicated if Eldad and Medad were added to the ‘seventy’ (11.29).Footnote 43
Another common view is that Luke's aim was to anticipate the future mission to all the nations of earth (i.e. the Gentiles), suggesting that the seventy nations named in Genesis 10 (MT) form the most likely background.Footnote 44 However, that the same passage in the LXX lists not seventy but seventy-two names makes it possible that the evangelist used ‘seventy-two’ for the same reason.Footnote 45 In fact, this numerical variation in the Old Testament tradition (much like that of Num 11.16 noted above) seems to be the very reason why the same bifurcation exists in witnesses of Luke's Gospel: scribes (or readers) dependent upon the MT tradition tended towards ‘seventy’, while those dependent upon the LXX tended towards ‘seventy-two’. The coincidence is indeed striking. In any case, this would still not assist us in identifying the prior wording in Luke.
More recently, at least one scholar has appealed to Greco-Roman backgrounds in order to identify relevant numerical symbolism. Drawing primarily from the work of Livy, it has been suggested that ἑβδομήκοντα is the better reading because of its symbolic value of ‘people who have been designated or have committed themselves to a common cause and have begun acting towards it’.Footnote 46
Many of these suggestions seem plausible, but which is the most probable? As it stands now, the NA28 text simply reads οἱ ἑβδομήκοντα [δύο] – the square brackets indicating that ‘textual critics today are not completely convinced of the authenticity of the enclosed words’.Footnote 47 This uncertainty does not imply that discussions of possible symbolism are without merit, but it is clear that no single explanation has garnered widespread confidence.Footnote 48 It is in fact doubtful that scholars will be able to identify securely just one of these symbolic meanings as the ‘most Lukan’ in nature. Furthermore, the Achilles’ heel of this reliance upon numerical symbolism is that one can never be sure which party intended to exploit the symbolism in question: the evangelist, a pre-Lukan tradition, a scribe, a reader, or Jesus himself? That is, if it is decided by a modern reader that ‘seventy’ (or ‘seventy-two’) seems more consistent with Lukan themes (and is therefore earlier), one can never escape the possibility that an early copyist or reader – according to the same perception – altered the prior wording to fit that theme.Footnote 49
In short, two major text-critical criteria are at a gridlock: the external evidence is evenly split, and intrinsic probability – with respect to the author's use of numerical symbolism – is frustratingly inconclusive. Regarding transcriptional probability, one basic observation has been made: immediately after the variation unit in question (in Luke 10.1), the text reads καὶ ἀπέστειλεν αὐτοὺς ἀνὰ δύο δύο (some manuscripts omit one δύο). It has been suggested that the presence of δύο (or β̅) in such close proximity, perhaps on the following line, could have caused a scribe to add it accidentally to ἑβδομήκοντα. While this does seem possible, the opposite error also presents itself: the original δύο/β̅ could have been omitted accidentally from ‘seventy-two’ because δύο δύο (or just δύο) follows in the immediate context. Metzger himself regarded both as possibilities, but more information would be needed to choose one as the more likely option.Footnote 50
Thus, aside from this basic observation, there has been no in-depth investigation of the transcriptional nature of this textual problem. In particular, the study of scribal behaviour specifically related to the writing of numerals has not been explored. For instance, in what ways were numerals corrupted in textual transmission? Can it be demonstrated that scribes were prone to adding extraneous digits to round numbers? Were compound numbers equally subject to alteration and/or omission? Answers to such questions could reveal actual patterns of scribal tendencies that might inform the textual problem in Luke 10.1 and 17. And to this we now turn.
5. Numeral Corruptions and Transcriptional Probabilities
The textual problem in question is most naturally viewed as either (1) the omission of a digit from a two-digit compound numeral, or (2) the addition of a digit to a single-digit round number. An examination can be made, therefore, of both types of errors among Greek New Testament manuscripts. So, for instance, where single-digit, round numbers occur in New Testament witnesses, how often are they corrupted through the erroneous addition of a second digit? Alternatively, how often are two-digit numerals corrupted through the omission of one digit? The basic question we seek to answer is this: were New Testament copyists more prone to omit a digit from a compound number (72 → 70) or to add a digit to a round number (70 → 72)? Fortunately, this question can be answered with some confidence.Footnote 51
In what follows, the term ‘round number’ refers to single-word numbers composed of either tens, hundreds or thousands (e.g. εἴκοσι, τετρακόσιοι etc.), and ‘compound’ and ‘two-digit’ refer to a number composed of two numerical elements, either of ones, tens, hundreds or thousands (e.g. δέκα καὶ ὀκτώ, ἐνενήκοντα ἐννέα). Although some two-digit numbers such as δώδεκα can be single words, these are nonetheless included as compounds; this is necessary because such words still contain two basic elements (δω- + -δεκα), and because their alphabetic equivalents were likewise composed of two-digits (ι̅β̅). It is not our aim to discern precisely how these corruptions occurred; I do not intend to argue that copyists were especially prone to mistakes when numbers were written as alphabetic numerals in their exemplars.Footnote 52 We will content ourselves, therefore, with simply observing the nature and relative frequency of these two kinds of corruptions.
5.1 Omissions from Two-digit Numbers
We first examine the frequency of scribal omission from two-digit numerals. In ninety-eight instances of a compound numeral in the text of the New Testament, an examination of numerous critical apparatuses resulted in the following thirteen variation units:Footnote 53
These examples illustrate that the omission of a digit from two-digit numerals is a common type of error within New Testament manuscripts. It is worth pointing out that most of these changes are pure nonsense in context. For example, in Matt. 18.12 and Luke 15.7, the departures from ἐνενήκοντα ἐννέα (99) are clearly mindless blunders; one hundred sheep minus one sheep must equal ninety-nine, not ‘nine’ as in minuscule 565 or ‘ninety’ as in majuscule 047, etc. And similarly, the omission of οκτώ from δεκαοκτώ in 047 in Luke 13.4 is equally as senseless because the numeral was written correctly in 13.11 and 13.16. This confirms that there need not be a motivation of symbolism in the alteration of the number; as the above examples show, two-digit numerals were simply prone to accidental omission.
Especially relevant to our variation unit in Luke 10 are the omissions of ἐννέα from ἐνενήκοντα ἐννέα (Luke 15.7), πέντε from ἑβδομήκοντα πέντε (Acts 7.14) and τέσσαρες from εἴκοσι τέσσαρες (Rev 5.8). In these instances, a second digit following a round number is omitted. This exactly parallels the difference between ἑβδομήκοντα and ἑβδομήκοντα δύο in Luke 10.1 and 17, where the second digit is in question.
5.2 Additions to Single-digit Round Numbers
Now we inquire whether the opposite error – the addition of a digit to a round number – was a similarly common occurrence. A survey of 104 single-word, round numerals results in the following seven variants:
There is, therefore, a corresponding principle of adding a figure to a single-digit round number, although there are fewer examples. Unsurprisingly, many of these are also nonsense in context, clearly attributable to involuntary alteration (e.g. δέκα καὶ ὀκτώ → δωδεκα και οκτω). It is important to observe, however, that six of the seven additions listed above are the same change from δέκα (10) to δώδεκα (12), a somewhat predictable alteration given the relative predominance of the more familiar number ‘twelve’. In fact, given the overwhelming frequency of δώδεκα in the New Testament, one might prefer to see this as a case of harmonisation to familiar wording rather than addition to a numeral. Furthermore, if this were truly a case of addition, one would expect to see δέκα → δεκαδύο rather than δώδεκα.Footnote 57 Nevertheless, for the sake of completeness we cannot rightly discount this as an example of addition to a round number.
The change from of ἑξήκοντα to ἑκατόν ἑξήκοντα (Luke 24.13) is thought to be an intentional correction rather than an accidental one. The evangelist states that the distance between Jerusalem and Emmaus is σταδίους ἑξήκοντα (= 60 stadia), and it seems that early patristic writers (incorrectly) identified Emmaus with ‘Amwâs/Imwas (modern Nicopolis), roughly 176 stadia from Jerusalem – hence the addition of ἑκατόν.Footnote 58 This is also the only case of addition that occurs in majuscule manuscripts. However, while this corruption is similar to our variant ἑβδομήκοντα (+ δύο), it is not a precise parallel because the added digit precedes the correct one. Therefore, what is missing from these examples is a change parallel to that of ἑβδομήκοντα to ἑβδομήκοντα δύο; that is, the erroneous addition of a second digit immediately following a round number. One might imagine that this has occurred in the change from δέκα to δώδεκα since their alphabetic equivalents would be ι̅ → ι̅β̅. Nevertheless, this does not apply because, in all of the above cases, the longhand form δώδεκα is used by the copyists rather than the numerical shorthand; the additions occur prior to the round number (δώδεκα).
We can see, therefore, that both classes of error were common within New Testament manuscripts, but omission applies much more readily to the problem at hand. Importantly, the error of omission occurs (1) more frequently, (2) in twice as many witnesses, and (3) with a greater range of values; that is, 11, 12, 14, 18, 24, 75 and 99 are affected by omission, but only 10 and 60 are affected by addition.
Confirmation of this scribal tendency towards omission comes from texts outside the New Testament.Footnote 59 Although a wider study could be conducted, a similar investigation of Greek manuscripts of the Old Testament shows that the same inclination towards omission of digits is found. In the Greek text of Genesis, for example, a survey of forty-nine occurrences of double-digit numerals yielded thirteen examples of omission, while a survey of fifty occurrences of single-digit round numbers yielded only four examples of addition (two of which, incidentally, were the familiar change from δέκα to δώδεκα).Footnote 60 Just as in the text of the New Testament, omission prevails.
5.3 ‘The Shorter Text with a Vengeance’: Numerals in P45
That copyists more frequently omitted from numbers than added to them coheres with other recent analyses of scribal habits. Numerous studies of scribal behaviour in early Greek manuscripts have confirmed the fact that copyists in general were usually more prone to omit text than to add, an observation that has essentially qualified a hitherto fundamental axiom of textual criticism which is to prefer the shorter reading, or lectio brevior potior.Footnote 61
This raises an important issue that has not yet been mentioned. Studies of the scribal habits of P45 have shown that the copyist was not only prone to the omission of text in general but especially numerals in particular.Footnote 62 On several occasions, the scribe of P45 conspicuously omitted numerals from the text, creating through omission what have been recognised as singular readings. A singular reading is usually defined as wording that is found in no other known Greek witness, strongly suggesting that none other than the scribe in question created it. Within P45 specifically, the scribe omitted both πέντε (Mark 6.41a) and δύο (6.41b) from the narrative of the feeding of the 5,000.Footnote 63 Further, the scribe omits a four-word phrase containing two more numbers, creating yet another singular reading: κατὰ ἑκατὸν καὶ πεντήκοντα (6.40). Whether these (and other) omissions were intentional or accidental is a matter of ongoing debate.Footnote 64
Regardless, this inclination in P45 to omit numerals is relevant to the textual problem in Luke 10.1 and 17, especially as this papyrus is the earliest extant witness to the shorter reading of ἑβδομήκοντα. Might the lack of δύο in P45 simply be another instance of the scribe overlooking a numeral, just as we saw in Mark 6.40 and 6.41? By way of contrast, the same tendency of numerical omission is not evident in P75, the earliest witness to ἑβδομήκοντα δύο; the scribe consistently wrote numerals correctly. Nevertheless, Joseph Verheyden has suggested that it was the scribe of P75 who, because of ‘the ample use of abbreviations’, probably confused ἑβδομήκοντα for ἑβδομήκοντα δύο (ο̅ → ο̅β̅).Footnote 65 Yet this proposal runs directly counter to two separate lines of evidence. First, the scribe of P75 – who regularly used numerical shorthand in the body text – consistently employed them without error. In other words, their abundance in P75 simply illustrates that the copyist was indeed comfortable with them and capable of using them effectively. And second, it was the scribe of P45 who employed only a handful number-abbreviations in the text of the papyrus (usually preferring longhand forms) and committed several glaring numerical omissions.Footnote 66 On these grounds, the opposite of Verheyden's suggestion is undoubtedly more likely: of the two scribes, it is far easier to envision that of P45 as the culprit.
Given the tendency of P45 to mishandle numerals and its early date of origin, it is tempting to suggest that the initial loss of δύο occurred in this papyrus. The strength of this proposal is that it would locate the source of the shorter reading in an early papyrus already known for its faulty numerals. This solution, however, is not without its problems. Since the shorter reading is also found in א, A and C (etc.) – none of which are thought to have a direct genealogical relationship to P45 – one would have to say that the same omission occurred in at least two separate instances, which is not impossible, but does not seem entirely likely. But two further points should be noted: first, patristic testimony shows that the tradition of ‘seventy’ dates well into the second century (and P45 dates to the third); and second, it is highly unlikely that the scribe simultaneously omitted the beta (= δύο) and took the effort to add a line-filler immediately after the omicron.Footnote 67 It seems, then, that P45 is not the culprit. None of this, of course, invalidates the observation that the longer reading appears to be the prior wording, it simply means that the ἑβδομήκοντα probably predates the witness of P45.
Let us now return to Metzger's argument cited at the beginning of this study. He remarked that the transcriptional probabilities of Luke 10.1/17 were ‘singularly elusive’ given that it was easy ‘for either number to be accidentally altered to the other’.Footnote 68 After a closer look at the nature and frequency of numerical corruptions in early manuscripts, however, we are now in a position to refine this idea. New Testament copyists were far more likely to omit a digit from a compound number than they were to add a digit to a round number, and when it concerns the second of two digits in a compound number, the tendency is always towards omission rather than addition.
6. Summary and Conclusion
In summary, two conclusions are possible. First, as Metzger observed over fifty years ago, P45 should not be considered as support for ἑβδομήκοντα δύο in Luke 10.17. Maximally, it could be added in support of ἑβδομήκοντα (perhaps with ‘vid’?), but in the least it should no longer be regarded as a witness to the longer reading.
Second, the revised testimony of P45 led to an illuminating study of scribal habits with respect to numerals. While appeals to perceived numerical symbolism for the recovery of the earlier wording seem to be at a stalemate, the actual patterns of numeral corruption within New Testament manuscripts add valuable data to the equation. Generally speaking, compound numerals were frequently corrupted through the omission of digits, whereas erroneous addition occurred far less often by comparison (thirteen variation units compared to seven) and affected fewer numerical values. Finally, it is striking that there are several occurrences of a compound numeral being corrupted through the omission of the second digit, but there are no clear examples of an addition of a second digit following a round number. This means that, on transcriptional grounds, it is easier to explain the change from ἑβδομήκοντα δύο to ἑβδομήκοντα than it is the reverse.Footnote 69
Nevertheless, we must admit that transcriptional probability is not the only relevant criterion to be utilised in such textual problems, and it must be balanced by other considerations. To be sure, even if the omission of numerical digits is the more common transcriptional error, this does not require that such a principle must apply to the variation unit in Luke 10.1/17. To conclude, however, in light of the fact that both the external evidence and intrinsic probability are inconclusive, the scribal tendency towards omission in the transcription of numerals is an important observation that should be considered in future discussions of this textual problem, and it is one that might just tip the scales.
