The Angel and the Sweat Like “Drops of Blood” (Lk 22:43–44): ∏69 and f13

Claire Clivaz University of Lausanne, Switzerland

It is impossible to write except by making a palimpsest of a rediscovered manuscript. Umberto Eco Introduction Eldon J. Epp recently asserted that “the greater the ambiguity in the variant readings in a given variation unit, the more clearly we are able to grasp the concerns of the early church.”1 I agree with this principle and submit that the famous text-critical problem regarding the authenticity of Luke 22:43–44—the agony of Jesus on the Mount of Olives, which mentions the angel and the sweat “like drops of blood”—has not yet yielded its store of information about early Christianity, particularly in Alex- andria. This article has the modest purpose of presenting two technical notes on the complex question of the external evidence (from the manuscripts) bearing on Luke 22:43–44. These fi ndings provide some of the groundworkfor a reappraisal of the Lukan account of Jesus’ prayer on the Mount of Olives.2 This article has the modest purpose of presenting two technical notes on the complex question of the external evidence in the manuscripts bearing on Luke 22:43–44. I begin by reviewing recent research on the disputed verses, and I shall argue that it is necessary to re-examine the external evidence of Luke 22:43–44. In the second section, I shall consider ∏69, a mid-third century manuscript that omits not just Luke 22:43–44, but the entire detailed content of Jesus’ prayer in Luke 22: 42–45a. I suggest that we can make sense of this larger omission by treating ∏69

1Eldon J. Epp, “Issues in : Moving from the Nineteenth Cen- tury to the Twenty-First Century,” in Rethinking New Testament Textual Criticism (ed. David Alan Black; Grand Rapids: Baker Academic, 2002) 60. 2Such a reappraisal is the task of my nearly completed Ph.D. dissertation on Luke 22:39–46 (directed by Prof. Daniel Marguerat, University of Lausanne, Switzerland; written in French).

HTR 98:4 (2005) 419–440 420 HARVARD THEOLOGICAL REVIEW as a witness to a Marcionite edition of Luke’s . In the third section, I shall examine (f13), a class of manuscripts that appear to transfer Luke 22: 43–44 to a position after Matt 26:39. I shall demonstrate that the transposition of these verses in the manuscripts of f13 cannot be used to prove that Luke 22:43–44 stems from another literary source.

Reconsidering the External Evidence for Luke 22:43–44 At the beginning of the last century, Adolf von Harnack campaigned strongly for the authenticity of Luke 22:43–44, emphasizing that there was no evidence for the omission of these verses before 300 C.E.3 The discovery of ∏75, a third-century manuscript that omits the verses, reopened the question of their authenticity.4 In the last twenty-five years, the authenticity of the verses has been postulatedby Jerome H. Neyrey and strongly defended by Raymond E. Brown.5 But the most infl uential position remains the one advanced in 1983 by Bart D. Ehrman and Mark A. Plunkett, who consider the verses as a later interpolation.6 The authorsunderlined thecom- plexity of the matter, saying, “no one argument yields a defi nitive solution. Rather, the cumulative force of a group of arguments must be assessed, and even then the critic is left with a probability-judgment.”7 This cautious position seems to have

3Adolf von Harnack, “Probleme im Texte der Leidengeschichte Jesu,” Sitzungsberichte der königlich preussischen Akademie der Wissenschaften zu Berlin 11 (1901) 253. 4Joseph A. Fitzmyer, “Papyrus Bodmer XIV: Some Features of our Oldest Text of Luke,” CBQ 24 (1962) 178. See also Bart D. Ehrman and Mark A.Plunkett, “The Angel and the Agony: The Textual Problem of Luke 22:43–44,” CBQ 45 (1983)407 n. 20. The“ ∏75-effect” reserves surprises: in The Text of the New Testament, Kurt and Barbara Aland offer two different dates for ∏75, the third century or “the early second century” (The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism [trans. Erroll F. Rhodes; 2d ed.; Grand Rapids: Eerdmans; Leiden: E. J. Brill, 1989] compare 57, 101 with 91). 5Jerome H. Neyrey, “The Absence of Jesus’ Emotions—The Lucan Redaction of Lk 22:39–46,” Bib 61 (1980) 153–71; Raymond E. Brown, The Death of the Messiah: From Gethsemane to the Grave: A Commentary on the Passion Narratives in the Four (2 vols.; ABRL; New York: Doubleday, 1994) 1:184–90. See also Jean Duplacy, “La préhistoire du texte en Luc 22:43–44,” in New Testament Textual Criticism: Essays in Honour of Bruce M. Metzgerr (ed. Eldon J. Epp and Gordon D. Fee; Oxford: Clarendon Press, 1981) 78–86; and Joel B. Green, “Jesus on the Mount of Olives,” JSNT 26 (1986) 33–47. Recent publications include Jerome Murphy-O’Connor, “What Really Happened at Gethsemane?” Bible Review 14 (1998) 28–39, 52; Michael Patella, The Death of Jesus: The Diabolic Force and the Ministering Angel (CahRB 43; Paris: Gabalda et Cie, 1999) 9–15; Rainer Riesner, “Versuchung und Verklärung (Lukas 4,1–13; 9,28–36; 10,17–20; 22,39–53 und Johannes 12,20–36),” TBei 33 (2002) 197–207; Christopher M. Tuckett, “Luke 22,43–44. The ‘Agony’ in the Garden and Luke’s Gospel,” in New Testament Textual Criticism and Exegesis (ed. Adelbert Denaux; BEThL 161; Leuven: Leuven University Press, 2002) 131–44; and François Bovon in the forthcomingfinal volume ofhis commentary on the end of Luke’s Gospel. Prof. Bovon kindly shared with me his manuscript commentary on Luke 22:39–46. He has published three volumes ofhis commentary on the Gospel of Luke (up to Luke 19:27) in French (Genève: Labor et Fides, 1991–) andin German (Zürich: Benziger Verlag; Neukirchener-Vluyn: Neukirchener Verlag, 1989–); the first volume has also appeared in English (Minneapolis: Fortress Press, 2002). 6See Ehrman and Plunkett, “The Angel and the Agony,” 401–16. 7Ibid., 416. CLAIRE CLIVAZ 421 been abandonedby Ehrman himself, who assertedin a lecture deliveredin 1997, that the chiasmus in Luke 22:40–46 is “nearly impossible to overlook.”8 Although he had admitted in 1993, “one can fi nd a structure wherever one looks.”9 Many scholars have been convinced by the literary argument made by Ehrman and Plunkett,10 which they reinforced with a theological argument.11 I suggest that the widespread acceptance of their 1983 article can be traced by examining the shift in the UBS editorial team’s assessment of the disputed verses. In 1983, UBS3 hesitantly included the verses, rating the decision with a “C” on a scale of A to D, but in 1994, UBS4 recommended omitting the verses, rating that decision with an “A.”12 Enthusiasm for Ehrman and Plunkett’s literary and theological arguments, however, seems to have overshadowed the lingering questions posedby the exter- nal evidence. The general attitude toward the textual evidence for the verses that describe the angel and the bloody sweat seems to be summed up in two statements: fi rst, Luke 22:43–44 is found, “generally speaking,”13 in Western witnesses but not in Alexandrian; second, “the arguments over external evidence are unable to decide the issue,” even if theygive to the shorter text a “slight edge.”14 This second, and prudent, observation, however, sometimes gives way to a very positive estimation

8Bart D. Ehrman, “Text and Traditions: The Role of New Testament Manuscripts in Early Chris- tian Studies. Lecture One: Text and Interpretation: The Exegetical Significance of the ‘Original’ Text,” TC: A Journal of Biblical Textual Criticism 5 (2000) §38. Online: http://rosetta.reltech.org/ TC/vol05/Ehrman2000a.html. 9Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early Christological Controversies on the Text of the New Testament (New York: Oxford University Press, 1993) 205. On the criticism of the chiasmus argument, see also Green, “Jesus,” 36; and Patella, The Death of Jesus, 11: “Here, specifically, Ehrman and Plunkett’s ‘concise chiasmus’ would still standby extending D to include the strengthening angel and sweating ofblood.” 10See, for example, John Nolland, Luke (3 vols.; WBC 35; Dallas: Word, 1993) 3:1081; Craig A. Smith, “A Comparative Study of The Prayer of Gethsemane,” Irish BiblicalStudies 22 (2000) 101; and Gregory E. Sterling, “Mors philosophi: The Death of Jesus in Luke,” HTR 94 (2001) 396. 11In their article, Ehrman and Plunkett posed as the key question in this debate, “Which read- ing is more readily explained as originating in the theological climate of the second century?” (“The Angel,” 407). Ehrman pursued this theological approach in 1997, saying that “the Jesus of Luke goes to his death calm andin control. . . . Only the longer text of 22:43–44 stands out as anomalous” (“Text and Tradition,” §42). Sterling developed this point of view in 2001 (“Mors,” esp. 397–98); see also David C. Parker, The Living Text of the Gospels (Cambridge: Cambridge University Press, 1997) 158. 12See The Greek New Testamentt (ed. Kurt Aland et al.; 3d ed.; Stuttgart: United Bible Societies, 1983) 305, and The Greek New Testamentt (4th ed.; Stuttgart: United Bible Societies, 1993) 297. Tuckett also noted the change (“Luke 22,43–44,” 131 n. 2). Also, the decision of the online The- saurus Linguae Graece not to include the verses should be mentioned: Luke 22:43–44 disappeared from the Lukan Gospel in the online edition (http://www.tlg.uci.edu). 13Ehrman and Plunkett, “The Angel,” 402; see also Brown, The Death of the Messiah,1:181; Sterling, “Mors,” 396. 14Ehrman, Orthodox, 188. See also Brown, The Death of the Messiah, 1:181; S. P. Cowe, “Christological Trends and Textual Transmission: The Pericope of Bloody Sweat (Luke 22:43–44) in the Armenian Version,”inText and Context: Studies in the Armenian New Testament, May 22–28, 1992 (ed. Shahe Ajamian and Michael E. Stone; Atlanta: Scholars Press, 1994) 36; Smith, “A Comparative,” 101; and Tuckett, “Luke 22,43–44,” 133. 422 HARVARD THEOLOGICAL REVIEW of the external evidence. Thus, in a recent article, Craig A. Smith wrote in the main text of “strong external evidence” for omitting the verses, after admitting in a footnote that the external evidence is inconclusive.15 And Kurtand BarbaraAland affi rm that “the external evidence leaves no doubt that these verses were added to the original text of Luke.” Nevertheless, they feel the need to reinforce this conclu- sion by insisting on the transfer by f13 of Luke 22:43–44 to a position after Matt 26: 39: “This kind of fluctuation in the New Testament manuscript tradition is one of the surest evidences for the secondary character of a text.”16 The Alands’ remark shows the strategic role that f13 can play in the evaluation of the external evidence, a problem that will be pursuedin the third section of this article. One item of external evidence bearing on Luke 22:43–44 has failed to receive the attention it deserves. Fragment 0171, found at Hermopolis Magna, dates to approximately 300 C.E. and contains the end of Luke 22:44.17 Recently, Michael Patella, Christopher Tuckett, and François Bovon18 have rightly insisted on the witness of this fragment. Yet the fragment has been neglected by several critics who study the disputed verses. In spite of its early dating, Brown,19 Cowe,20 and even Ehrman in his analysis of 199321 simply fail to mention it. More surprisingly, David Parker underlines the importance of 0171 by describing it as “of consider- able interest,”22 but does not discuss the fragment in relation to Luke 22:43–44.23 These facts reveal that 0171 is simply not as well-known or well-considered as it deserves to be; Joseph A. Fitzmyer even confused it with the twelfth-century minuscule 1071, which contains Luke 22:43–44 as a correction.24

15Smith, “A Comparative,” respectively 113 and 101 n. 9. Ehrman shows the same ability to shift quickly from expressing reservations about the external evidence to strongly affirming its decisive weight (Orthodox, 188, 189). 16 K. and B. Aland, The Text of the New Testament, 2d ed., 310. 17Ibid., 123; James Neville Birdsall, “A Fresh Examination of the Fragment of the Gospel of St. Luke in ms. 0171 and an Attempted Reconstruction with Special Reference to the Recto,” in Philologia Sacra. Biblische und patristische Studien für Hermann J. Frede und Walter Thiele zu ihrem siebzigsten Geburtstag (ed. Roger Gryson; Freiburg: Herder, 1993) 212. The Lukan fragment of 0171 was published in two stages: Papiri greci e latini della Società Italiana della ricerca dei papiri greci e latini in Egitto 1:2 (Florence: E. Ariani, 1912) 2–5; and ibid., 2:124 (Florence: E. Ariani, 1913) 22–25. The end of Luke 22:44 was not included in PSII 1:2, but appears at the begin- ning of PSI 2:124, col. 1 verso. 18See Patella, The Death of Jesus, 10; Tuckett, “Luke 22,43–44,” 131; and Bovon (see n. 5, above). 19Patella, The Death of the Messiah, 1:180. 20Cowe, “Christological Trends,” 37. 21In Orthodox Corruption, Ehrman summarizes the most important and earliest witnesses, but omits 0171, which allows him to conclude that Luke 22:43–44 “are omitted by the earliest Greek manuscripts” (188). Ehrman and Plunkett had of course mentioned 0171 in their detailed analysis in 1983 (“The Angel,” 402). 22Parker, The Living, 159, with reference to Luke 22:51; the author cites 0171’s reading in sup- port of omitting Luke 22:62 (ibid., 160). 23Ibid., 157–59. 24JosephA. Fitzmyer, The Gospel according to Luke: Introduction, Translation, and Notes (2 vols.; New York: Doubleday, 1985) 2:1443. C. M. Tuckett noted the error (“Luke 22,43–44,” 131). CLAIRE CLIVAZ 423

The misfortune of 0171 is probably due to its classification by Kurt and Bar- bara Aland as a “paraphrastic text, category IV.”25 In spite ofits antiquity, 0171 is relegated to category IV because it shares certain similarities with the Western tradition.26 Yet Stanley E. Porter has recently sounded a warning about the excesses of the Alands’ categorizations, which establish the “twenty-sixth edition of the Nestle-Aland [as] the ‘original text,’ ” and avoid dealing with the complexity of the papyri.27 And Kurt Aland himself recognized that “0171 moves away as much from the D-text as from the ‘Standard-Text,’”28 andhe underlined the differences between 0171 and the D-text in Luke 22:47b, 51.29 Aland compared 0171’s par- ticular way of reading to that of ∏69, saying, “like 0171, ∏69 follows its own way too.”30 Also, Aland refused to explain the presence of Luke 22:44 in both D and 0171 by postulating a common source, arguing instead that the verses are known by many church fathers.31 Consequently, when 0171 “follows its own way” in Luke 22:47b, 51, 62,32 the fragment cannot be devalued simply because it shares certain similarities with the D text.33 Thus, the question we must pose is: Could 0171 represent “another”34 Egyptian voice, as does ∏45, for example? Are there valid reasons to classify ∏45 in category I as “free text,” and 0171 in category IV as “paraphrastic”?35 J. K. Elliott correctly stressed that there are also similarities between ∏45 and the D-text,36 and Barbara Aland recently affi rmed the value of the

25K. and B. Aland, The Text of the New Testament, 2d ed., 123. 26See the statement about 0171 in the first edition of K. and B. Aland, The Text of the New Testament, 1987, 1st ed., 122: “Category IV because of affi nity to D.” The second edition (1989) doesn’t need any more to justify the “category IV,” a fact that signals the progressive devaluation of0171. 27Stanley E. Porter, “Why So Many Holes in the Papyrological Evidence for the Greek New Testament?” in The Bible as Book. The Transmission of the Greek Textt (ed. Scot McKendrick and Orlaith A. O’Sullivan; London: British Library Press; Grand Haven: Oak Knoll Press, 2003) 174. 28“0171 entfernt sich vom D-text wie vom ‘Standard-Text’ ” (Kurt Aland, “Alter und Enstehung des D-Textes im Neuen Testament. Betrachtungen zu ∏69 und 0171,” in Miscellània papirològica Ramón Roca-Puig [ed. Sebastià Janeras; Barcelona: Fundacio Salvador Vives Casajuana, 1987] 51). 29Ibid., 52. 30“Wie 0171 geht auch ∏69 ganz eigene Wege” (ibid., 59). 31Ibid., 56. 32J. N. Birdsall demonstrated the absence of Luke 22:62 in 0171 (cf. “A Fresh Examination,” 212–27). 33In my opinion, James K. Elliott is correct in saying that “the increased number of early papyri has not yet enabled us to point to any significant corpus of Greek material that seems to have been a precursor of the D-type text. Only ∏38 and 0171 deserve some recognition in this context” (“Codex Bezae and the Earliest Greek Papyri,” in Codex Bezae. Studies from the Lunel Colloquium. June 1994 [ed. David C. Parker and Christian-Bernard Amphoux; Leiden: E. J. Brill, 1996] 182). 34My use of the term “another” is determinedby the fact that the “strict” Alexandrian text serves as the classic point of reference in New Testament textual criticism, with ∏75 as a main representa- tive for the Lukan Gospel. 35K. and B. Aland, The Text, 2d ed., 1989, resp. 99 and 123. 36Elliott, “Codex Bezae” in Parker and Amphoux, eds., Codex, 169–70. 424 HARVARD THEOLOGICAL REVIEW

Chester Beatty papyri,37 although she limits their weight by noting that they refl ect “the situation of the New Testament in a particular region of Egypt in the third century.”38 I concur with this point of view, if it concerns each papyrus, even the famous ∏75, which omits Luke 22:43–44. In other words, I question the assumption that certain papyri should represent de facto the essence of a pure text, while other papyri are regarded as preserving “historically contingent” readings. We should recall that every manuscript potentially reflects or has been affected by historical contingencies.39 New Testament researchis just arriving at the point where scholars can appreciate the complexity and the diversity of Christian voices in Egypt in the first and second centuries C.E. Thus, one of the main challenges in current research on the textualhistory of Luke is to resist the prevalent fascination with ∏75. We should keep in mind the opinion expressed by Joseph A. Fitzmyer just after ∏75 was published: “It is surprising how frequently [∏75] disagrees with ∏45, the Chester Beatty Papyrus I, which contains part of Luke and which is roughly a half century older than it. . . . The fact that both ∏45 and ∏75 stem from Egypt in the period prior to the great parchment uncial mss., and yet do not agree in their peculiar readings, gives evidence of a fl uctuating state of the text in that country in the early period.”40 Unfortunately, ∏45 does not preserve Luke 22, and we can only deplore this lacuna. But if we take into account 0171, it raises the question of a diversity of opinions on Luke 22:43–44 within the Egyptian tradition, a question that is also raised by Codex Sinaiticus, which originally included the verses.41 The

37B. Aland, “The Significance of the Chester Beatty Papyri in Early Church History,” in The Earliest Gospels: The Origins and Transmission of the Earliest Christian Gospels—The Contribu- tion of the Chester Beatty Gospel Codex ∏45 (ed. Charles Horton; JSNTSupp 258; London: T&T Clark, 2004) 108. 38Ibid., 120. 39Thus, Epp campaigns for taking into account all the data offeredby the Oxyrhynchus papyri, and for considering the “New Testament manuscripts in their cultural, intellectual context” (“Issues in New Testament,” 71, 60). 40Fitzmyer adds: “Lest we be tempted to think of ∏75 as containing some sort of a ‘mixed text,’ it is well to recall here the observation of J. N. Birdsall about the inadequacy of applying to these early texts categories and classes which were set up on the basis of the fourth century mss.” (“Papyrus Bodmer XIV: Some Features of Our Oldest Text of Luke,” CBQ 24 [1962] 175). 41a* and a2 give the verses and a1 suppresses them. One wonders why Sinaiticus is often counted among the manuscripts that omit Luke 22:43–44, rather than among those that include it (cf. K. Aland, “Alter,” 59 and recently C. A. Smith, “A Comparative,” 101 and 113; Roselyne Dupont- Roc, “Le texte du Nouveau Testament et son histoire,” in Introduction au Nouveau Testament. Son histoire, son écriture, sa théologie (ed. Daniel Marguerat; 3d ed.; Genève: Labor et Fides, 2004) 495. A careful reading of Helen and Kirsopp Lake’s facsimile (Codex Sinaiticus Petropolitanus: the New Testament, the Epistle of Barnabas and the Shepherd of Hermas preserved in the Impe- rial Library of St. Petersburg [Oxford: Clarendon Press, 1911]) shows impressive evidence for the original presence of the verses in the manuscript, and gives elements for going further in the external evidence’s exploration, whichis beyond the scope of this article. It can at least be underlined that, if a* and a2 contain Luke 22:43–44, they do not contain other classically disputed passages, such as Matt 16:2–3; Mark 16:9–20; John 5:4; 7:53–8:11; Rom 16:24 (cf. William H. P. Hatch, The Principal Uncial Manuscripts of the New Testamentt [Chicago: University of Chicago Press, 1939] commentary on plates XV and XVI). CLAIRE CLIVAZ 425 following discussion of ∏69 will illustrate that a distinction between “Western wit- nesses” and “Alexandrian witnesses” does not suffi ciently account for the diversity of the external evidence bearing on Luke 22:43–44 in Egypt.

Listening to What ∏69 Has to Say ∏69: A Neglected Papyrus Scholars have not given adequate attention to ∏69, a small papyrus of the mid-third century discovered at Oxyrhynchus.42 Joseph van Haelst described it as containing an “aberrant text, very different from that of the major uncials.”43 ∏69 is not cited in the critical apparatus44 of NA27, and appears quoted with or without videtur.45 Eminent scholars, such as Bart D. Ehrman and Raymond E. Brown, have confused it46 with ∏66, as Christopher Tuckett has noted.47 Moreover, the Alands considerably altered their opinion of ∏69 between the first and the second editions of The Text of the New Testament. In the first edition (1987), ∏69 is described as a “very free text, similar but unrelated to D; category I because ofdate.”48 In the second edition (1989), however, ∏69 is described as “a very free text, characteristic of precursors of the D-text; therefore category IV.”49 Thus, the Alands’ re-evaluation of ∏69 classifi es it as an example of the D-text and therefore as a less important witness overall: it is demotedfrom category I to category IV. As I notedin the previous section, however, Kurt Aland had written in 1987 that “like 0171, ∏69 follows its own way too,” independently of the D-type.50 Scholarly estimation of the worth of both0171 and ∏69 has varied, depending on a common interpretative frame in which these texts are associated with the D-type and are not very well known. Certainly ∏69 has not received the careful treatment it deserves as an independent textual witness to the problem of Luke 22:43–44. I preface my attempt to remedy this defi cit with a few general remarks. E. G. Turner noted that on the recto—which presents Luke 22:41, 45b–48—“the inkis almost effaced and the text is readable only with diffi culty.”51 I have only

42E. G. Turner, “2383. Gospel According to St. Luke XXII,” in Oxyrhynchus Papyri XXIV (ed. E. Lobel et al; London: Egypt Exploration Fund, 1957) 1–4; K. and B. Aland, The Text of the New Testament, 2d ed., 1989, 100. 43J. van Haelst, Catalogue des papyrus littéraires juifs et chrétiens (Série “Papyrologie” 1; Paris: Publications de la Sorbonne, 1976) 154. My translation. 44 Novum Testamentum Graece (27th ed.; Stuttgart: Deutsche Bibelgesellschaft, 1993) 235. 45 For example without videtur: Parker, Living, 157; Smith, “A Comparative,” 113. 46Ehrman, Orthodox, 188; Brown, The Death of the Messiah,1:181. 47Tuckett, “Luke 22, 43–44,” 132. But C. M. Tuckett himself asserts on the same page ofhis article that ∏69 gives Luke 22:43–48 andd omits Luke 22:42–44 (cf. “Luke 22,43–44,” 132 and n. 8). 48First edition of The Text of the New Testament, 1987, 100. 49Second edition of The Text of the New Testament, 1989, 100. 50K. Aland, “Alter und Enstehung,” 59. Quoted in n. 28 of this article. 51Turner, “2383,” 1. 426 HARVARD THEOLOGICAL REVIEW been able to consult photographs52 of ∏69,but they bear out Turner’s remark and show that the new transcription of Luke 22:40–41 proposedby Comfort and Bar- rett is quite optimistic.53 In any case, the editions of Turner and of Comfort and Barrett agree that the Lukan features of the text are certain (note wJsei; livqou bolhvn in Luke 22:41 and ajpo; th`~ luvph~ in Luke 22:45). If some particulars are disputable (for example, whether Judas approaches Jesus to kiss him, or whether having approached Jesus he does in fact kiss him [Luke 22:47b]),54 both editions agree about two striking details. First, Peter—and not Jesus—is the one looking in Luke 22:61;55 and second, ∏69 omits a long passage, from levgwn Pavter (Luke 22:42) until ajpo; th`~ proseuch`~ (Luke 22:45a).56 This last feature has only been adducedin discussions about Luke 22:43–44’s external evidence by the Interna- tional Greek New Testament Project and C. M. Tuckett,57 insofar as I have been able to discover.58 The problem of the status of Luke 22:43–44 is of such important weight in the history of exegesis that scholars have approached ∏69 as a witness to that problem alone, instead of inquiring about its apparent omission of Luke 22:42–45a. Turner explained the long omission by positing that “the scribe’s eye traveledfrom proshuvceto v.41 to proseuch`~ v.45. Such an omission by homoioteleuton is plausible only if the intervaljumpedis not too great,”59 and so Turner concluded that vv. 43–44 were not in the scribe’s exemplar.60 Since Turner, only Kurt Alandhas discussed the omission,61 proposing another explanation: “The omission of 42-45a has no mechanical origins (one finds no trace of homoioteleuton, etc.), and there is here no

52I would like to warmly thank the Institut für Neutestamentliche Textforschung in Münster (Germany), where I examined the photographs, for its generous welcome andfor assisting my research; I am gratefulin particular to Luc Herren and Dr. Michael Welte. 53Philip W. Comfort and David P. Barrett, eds., The Text of the Earliest New Testament Greek Manuscripts (Wheaton. Ill: Tyndale House Publishers, 2001) 471–72. This re-edition has been received by critics with reservations; see, e.g., http://rosetta.reltech.org/TC/vol06/CB2001rev.html. 54Turner judges the latter reading “very uncertain” (“2383,” 3 n. 13). 55Even if it remains uncertain whether Peter is looking at Jesus or at the cock (cf. Aland, “Alter,” 59). 56Comfort and Barrett imprecisely speak of the omission of “Lk 22:42–44” (The Text, 471), but Luke 22:45a is missing too, until ajpo; th`~ proseuch`~, as their own edition shows (cf. The Text,472; cf. too Turner, “2383,” 2 and 3, nn. 3 and 4; Aland, “Alter,” 57; Elliott, “Codex,” 171). 57The Gospel According to St. Luke; part 3 of The New Testament in Greekk (vol. 2; Oxford: Clarendon Press, 1984) 2:189–91; Tuckett, “Luke 22,43–44,” 132 n. 8. S. P. Cowe mentions the omission of Luke 22:42–44 by ∏69 in a footnote, but not the omission of Luke 22:45a (cf. “Christological Trends,” 36 n. 5). 58Except for the edition of Comfort and Barrett (The Text, 2001), I find no other works on ∏69, as these mentioned in James K. Elliott’s bibliography (cf. A Bibliography of Greek New Testament Manuscripts [2d ed.; Cambridge: Cambridge University Press, 2000] 34). I thank Eldon Epp for his help on this point and on the bibliography for f13. 59Turner, “2383,” 3 n. 4. 60His argument has not always been correctly understood; S. P. Cowe speaks of a homoeoarchton (“Christological Trends,” 36 n. 5). 61We can only regret that Comfort and Barrett did not take into account the remarks of K. Aland, even in their second edition (cf. The Text, 471). CLAIRE CLIVAZ 427 negligence or error, but a conscious action.”62 Apparently, Aland was not convinced by Turner’s postulatedhomoioteleuton, and so he offers two decisive arguments in favor of a conscious omission.63 First, he notes that the individual and free character of ∏69 in Luke 22:61 (Peter is looking, not Jesus) makes the deliberate omission of Luke 22:42–45a plausible; second, Luke 22:45a, which ends the scene, would also be omitted consciously (∏69 recto l. 4), and a kaiv would introduce the following text, according to Turner’s reconstruction.64 Aland’s purpose here is not to discuss ∏69 as a witness bearing on Luke 22: 43–44,65 but to stress the absence of Luke 22:42–45a, and so to classify ∏69 as “paraphrastic,” like the D-text.66 But if the omission of Luke 22:42–45a in ∏69 is seen as voluntary, it offers no definitive information regarding the presence or absence of Luke 22:43–44 in the exemplar usedby ∏69; therefore, ∏69 is not to be treated as a witness to the inclusion or omission of vv. 43–44,67 but represents, in my opinion, a thirdd and alternate view of the Lukan pericope of the Mount of Olives. That is, ∏69 reflects a textual tradition that consciously omits the longer passage of Luke 22:42–45a (or Luke 22:42, 45a). In the following paragraphs, I shall develop Aland’s argument for the conscious omission of these verses in ∏69, and I shall briefl y indicate the implications of this change in viewpoint. First, the plausibility of a conscious break after ajpo; th`~ proseuch`~ (Luke 22: 45a) is reinforced by the presence of other witnesses to this break. Those witnesses include many lectionaries,68 particularly the f13 lectionary (Vat. gr. 12177)69; the mar- ginal note next to Matt 26:39 in the Ephrem Codex (f. 62 recto); the minuscule 713 (f. 99 verso);70 and the margins of two minuscules of f13—the minuscule 13 (upper

62“Die Auslassung von 42–45a hat keine mechanischen Ursachen (von Homoioteleuton usw. fi ndet sichkeine Spur), es handelt sichhier auch nicht um Flüchtigkeit oder Versehen, sondern um bewusstes Handeln” (“Alter,” 59). My translation. 63Ibid.,59. 64Luke 22:45b in ∏69, recto, l.4: kai elqwn proß touß maqºhtªaß eu (Turner, “2383,” 2). 65Kurt Aland leaves the question unanswered (ibid., 59): “Natürlich musste er dabei ausser V. 42(–44) auch V. 45a auslassen.” 66See “Alter,” 59–60. Aland postulates links between 0171 and ∏69, in order to join them in opposition to a “strict” Alexandrian tradition. At the same time, he recognizes that they cannot really be compared, as they share almost no verses in common, and even seem opposed on Luke 22:42–45a (cf. ibid., 60). One wonders why he affirms that “both don’t have Lk 22, 47b” (ibid.,60), since the manuscript of 0171 PSII I:2 verso stops before the end of Luke 22:47: the only manifest similarity between 0171 and ∏69 is the kaiv before ejlqw;n in Luke 22:45. 67∏69 is usually counted as strong support for the omission, cf. Bruce M. Metzger, A Textual Commentary on the Greek New Testament (Stuttgart: Deutsche Bibelgesellschaft, 1987) 151; Brown, The Death of the Messiah, 1:181; Smith, “A Comparative,” 113. 68See l10, l12, l32, l48, l70, l150, l184, l211, l299, l859, l890, l950, l1127, l1231, l1579, l1599, l1627, l1634, l1642, l1633, l1761 (as listed in The New Testament in Greek: The Gospel according to St. Luke, 190). This fact has already been notedby Brooke F. Westcott and Fenton J. A. Hort, The New Testament in the Original Greek: Introduction and Appendix (Cambridge/London: Macmillan and Co, 1882) 2:Appendix, 65; Brown, The Death of the Messiah, 1:181 n. 4. 69J. Geerlings, ed., The Lectionary Text of Family 13 According to Cod Vat Gr 1217 (Gregory 547) (Studies and Documents XVIII; Salt Lake City: University of Utah Press, 1959) 38. 70713 transferred Luke 22:43–45a after Matt 26:39: the inclusion of 45a pleads for a liturgical influence. 428 HARVARD THEOLOGICAL REVIEW margin off. 117 recto),71 as well as the indications of the corrector of 230 (f. 163 verso),72 who put asterisks in the margin next to Luke 22:43–44, precisely until ajpo; th`~ proseuch`~ in Luke 22:45a. At this point, I would like to draw a distinc- tion between a source-critical analysis and a literary analysis of the passage. In my opinion, these witnesses cannot be used to argue on the level of source criticism, in order to prove the existence of a distinct source for vv. 42–45a; ∏69 omits Luke 22: 42–45a, but the other manuscripts I have cited are dealing with Luke 22:43–45a. Moreover, Westcott and Hort, and Brown correctly underlined the liturgical purpose of the lectionaries;73 indeed, some other lectionaries include the full text of Luke 22:43–45, until elqw;nj pro;~ tou;~ maqhta;~,74 whichdemonstrates the flexibility with whichliturgical usage can determine the limits of a passage. Thus, the witnesses that recognize a break at Luke 22:45a are evidence for a literary pattern—that is, they testify to a common narrative pattern that was recognized by readers of different periods. Indeed Luke 22:42–45a can be read as a micro-narrative, according to the quinary scheme employed in narrative criticism.75 An initial situation is indicated in v. 42a (parevnegke tou`to to; pothvrion ajp∆ ejmou`), while v. 45a describes the resolution of the situation.76 Consequently, a break after ajpo; th`~ proseuch`~ in 45a is logicalfrom a narrative point of view; and ∏69,Cmg, 13C, 230C, and 713, along with the lectionaries mentioned in n. 68, indicate that ancient readers recognized the plausibility of this break, which should be understood as a trace of a common reading pattern, and not the existence of another source. Second, current exegesis of Luke 22:42a frequently overlooks the fact that for readers in antiquity, Jesus’ demand that the cup pass from him was the most shock- ing element in the Gethsemane story.77 Ancient readers expected that an individual

71The minuscule 13 belongs to the National Library of Paris (Paris gr. 50); I have consultedit on microfilm at the Institut für Neutestamentliche Textforschung in Münster. The upper margin is unfortunately cut, but it is possible to read the end of Luke 22:43–45a. Minuscule 13 presents w[fqh de; in the main and original text, instead of the complete Luke 22:43–44. 72The minuscule 230 belongs to the National Library of Madrid; I have consulted it on microfilm at the Münster Institute. 73Westcott and Hort, The New Testament, 2:Appendix, 65; Brown, The Death of the Messiah, 1:181 n. 4. 74The New Testament in Greek: The Gospel according to St. Luke, 190. 75For an exposition of the quinary scheme in narrative criticism, see Daniel Marguerat and Y. Bourquin, How to Read Bible Stories: An Introduction to Narrative Criticism (trans. John Bowden; London: SCM Press, 1999) 43–44. 76A complication of the initial situation can be identified in Luke 22:44; the fact that Jesus stands up in Luke 22:45a indicates equally in this case the resolution of the situation (see the reading of Cmg, 13C, 230, 713 and severallectionaries). 77I defended this opinion in a recent paper: “A Sweat Like Drops of Blood” (Luke 22:44): At the Crossing of Intertextual Reading and Textual Criticism,” SBL Seminar Papers 2004. San Antonio, http://www.sbl-site.org/PDF/Clivaz_Sweat.pdf. Celsus of course (cf. Contra Celsum 2.24; 7.53), but Prophyrius too (cf. Contra Christianos, frag. 62, 3–6), and even Origen (cf. Mart. 29.32–37) attest that the request that the cup be removed was the most shocking element in the Gethsemane story for readers in antiquity. CLAIRE CLIVAZ 429 confronting death should make a noble speech, or at least maintain a dignifi ed silence that could be interpreted as proof of that individual’s sensibility.78 It is thus highly significant that Celsus concludes his attack against the prayer at Gethse- mane by saying that “certain of the Christian believers, like persons who in a fit of drunkenness lay violent hands upon themselves, have corrupted the Gospel from its original integrity, to a threefold, and fourfold, and many-fold degree, and have remodeled it, so that they might be able to oppose negations to the objections.”79 So we have here a very early (secondhalf of the second century C.E.) attestation of different versions of Jesus’ prayer on the Mount of Olives. What Celsus could mean, when he speaks of those who “oppose negations to the objections,” is that some scribes were omitting the word on the cup, which was the most shocking element (cf. n. 77). This strategy of “negating the objections” by omission would be practical only in a type of Christianity that preserved a single gospel, as did Marcion. I want to suggest80 that ∏69 could be read as a fragment of Marcion’s redaction of the Gospel of Luke; in other words, a Marcionite background for ∏69 could supply a plausible context for its omission of Luke 22:42–45a. Indeed, such an omission is what we might expect on the basis of Tertullian’s well-known remark that Marcion did exegesis “with a knife.”81

∏69: Looking for a Marcionite Frame of Plausibility for the Omission of Luke 22:42–45a There is no extant manuscript of Marcion’s gospel. The modern reconstitution by Adolf von Harnack remains consequently a reference point in the study of Marcion. Harnack, in the supplements to the latest edition ofhis study of Marcion’s gospel, emphasized that all three verses of Lk 22:42-44 are not attestedby Marcion and concluded that “43f. (the angel and the agony at Gethsemane) is authentic in Luke, understandably deletedby Marcion.”82 But in the first edition ofhis study of Marcion (1870), Harnack had been more attentive to a possible omission of Luke 22:42–44

78See, for example, Euripides, Hippolytus 902–920 (Kovacs, LCL). Hippolytus reproaches his father Theseus for remaining silent in front of the corpse of Phaedra: “Silence is no use in mis- fortune!” (911); Theseus answers: “O foolish mankind, [you do not understand] how to teach the senseless to be sensible!” (916–20). 79Contra Celsum II, 27 (trans. Henry Chadwick; 2d ed.; Cambridge: Cambridge University Press, 1980). For Celsus, the main problem is the saying on the cup. 80I thank François Bovon for havinggiven me the initial suggestion for this hypothesis. The prudent verb “to suggest” signifies here that I am working in a frame of plausibility, which is, for me, the only way to deal with history. 81Tertullian, Praescriptio 38.7–9. 82Adolf von Harnack, Marcion. Das Evangelium von fremden Gott. Neue Studien zu Marcion [Darmstadt: Wissenschaftliche Buchgesellschaft, 1985 (1st ed.; 1924)] 234* (“43f. [Der Engel und die Agonie in Gethsemane] ist bei Lukas echt, von M. begreiflicherweise gestrichen,” 234* foot- note on 43f; my translation). In the main part of the edition of 1924, Harnack says that Luke 22: 39–40 is not attested but cannot have been entirely lacking, and that Luke 22:42–46 probably was suppressed (ibid, 57). 430 HARVARD THEOLOGICAL REVIEW by Marcion, because Hahn and Ritschl had already suggestedit.83 Coming back to the perspective of the 1870 edition, we should consider the plausibility of the omis- sion of Luke 22:42–45a (or 22:42, 45a) in connection with the Marcion data. First of all, neither Tertullian nor Epiphanius reports on Marcion’s argumentation about Luke 22:42, or gives any indication whether he retained or omitted the verse. But if Marcion did indeed omit the verse, the Fathers’ silence about such a striking omission would not be surprising, because they chose to challenge Marcion’s inter- pretation of those verses that he had retained,in order to strengthen their arguments against him.84 Second, Tertullian says nothing about the prayer in Gethsemane, but Tertullian does not speak about all the scenes that Marcion included. For example, Tertullian implies that Peter’s denial was containedin Marcion’s gospel, but he does not comment on the passage.85 Epiphanius86 fortunately mentions and comments on the presence of Luke 22:41 in Marcion’s gospel: “In kneeling down [Jesus] knelt visibly, and did so in a perceptible manner. But if he did it perceptibly, then he performed the act of kneeling in [human] fashion. Therefore the Only-begotten did not sojourn among us without fl esh.”87 Epiphanius is arguing in support of Je- sus’ humanity, and so we may conclude that if Epiphanius hadfound the passages about the cup (Luke 22:42) or the angel and the sweat like drops ofblood (Luke 22:43–44) in the copy of Marcion’s gospel that he consulted, he would have based his argument for Jesus’ humanity on both or either of those passages, instead of on the account of Jesus’ kneeling in Luke 22:41. Third, the omission of the verse about the cup—a verse that, as I have noted, was regarded by second-century readers as a sign of Jesus’ weakness—seems to be congruent with Marcionite Christology. From the data presented by Tertullian, Irenaeus, and Epiphanius, we know that Marcion had a strong concern to reject

83Adolf von Harnack, Marcion, der moderne Gläubige des 2. Jahrhunderts, der erste Reformator (Die Dorpater Preisschrift [1870]). Kritische Edition des handschriftlichen Exemplars mit einem Anhang [ed. Friedemann Steck; TU 149; Berlin: W. de Gruyter, 2003] 152, ll. 10–12. 84Tertullian, Contra Marcionem 4.6.2 (cf. too 4.1.1; 4.6.1); Epiphanius, Panarion, section 3.42.11, 1 and 15. 85See the allusion to Luke 22:33–34 in Contra Marcionem 4.41.2,but the silence about Luke 22:56–62 in Contra Marcionem 4.41.1–2. 86The Panarion of Epiphanius has not received much attention from recent scholars. There is only one complete English translation of the text, in the Nag Hammadi Studies collection (Frank Williams, trans., The Panarion of Ephiphanius of Salamis, Book I [Sects 1-46] and Books II-III [Sects 47-80, De Fide] (2 vols.; Nag Hammadi and Manichaean Studies 35–36; Leiden: Brill, 1997, 2d ed. [1st ed.1987 ]). Scholars are still influencedby Harnack’s choice to read Marcion essentially on the basis of Tertullian (cf. Marcion. Das Evangelium von fremden Gott, vi). The similarities between Epiphanius’s report and ∏69 should lead us to consider at least in part the work of Epiphanius about Marcion, cf. Panarion 3.42.11,16 (Williams, 2:291): “This collection of brief memoranda on my subject, taken from authentic copies of Marcion in the form of notes, was written by myself, word for word, as an outline.” 87Ibid., 3.42.11,17, Elenchus 65 (Williams, 2:311). CLAIRE CLIVAZ 431 any sign of assimilation between the Demiurge and the “Father.”88 The prayer of Luke 22:42, withits mention of a “Father,” would have posed a grave problem for Marcion if it were possible to assimilate this “Father” with the Demiurge. Indeed, this assimilation would have been coherent in Luke 22:42 from a Marcionite view- point, because the idea that God can tempt is linked with those who believe in the Demiurge.89 Luke 22:42 can indeed be read as an account of Jesus being tempted to renounce his death; such a reading raises the question of whether knowing the will of God presents a temptation for Jesus. Because, for Marcion, the real God cannot be the tempter, he cannot admit any tension between the real God’s will and the will of Jesus—“unless,” as Tertullian ironically points out, “Marcion presumes a Christ not submitted to the Father.”90 All of these observations allow me to conclude that Marcion could not retain Luke 22:42 in his gospel without compromising his position; the suggestion of disagreement between Jesus’ will and the Father’s will, and the issue of Jesus entering into temptation by avoiding the cup, could have prompted readers to conclude that the real God—the Father—was the same as the Demiurge. Furthermore, if Marcion omitted the ambiguity and the tension of the prayer’s content, then his omission of the resolution of this tension in Luke 22: 45a is also logical.91 Finally, a Marcionite frame for ∏69 could explain its second major peculiar- ity: Jesus is not looking at Peter in Luke 22:61, but Peter is looking (∏69 verso, l. 24), which I understand as a way to avoid here any suggestion that Jesus is paying attention to Peter;92 Marcion tried to avoid anything that could magnify Peter’s status. Tertullian indeed affirms that Peter and the disciples represent for Marcion faith in the Demiurge; for example, Peter, who does not know what he should say at the transfiguration (Luke 9:33), is influenced by his Jewish tradition and ignores the true nature ofChrist.93 For Marcion, Peter does not understand anything about the person of Christ, whichis confi rmedby the fact that, according to Epiphanius,

88See about Luke 10:21–24, Contra Marcionem 4.25.1–7 (SC 456:315 and 319–21). Tertullian and Epiphanius agree on this point, cf. Panarion 3.11.17, Elenchus 22 (Williams, 2:297); Elenchus 34 (Williams, 2:302). 89See Irenaeus, Adversus Haereses 1.27.3; Epiphanius, Panarion 3.11.17, Elenchus 60 (Wil- liams, 2:308–9). 90“Nisi Marcion Christum non subiectum patriinferret” (Contra Marcionem 4.39.7 [SC 456: 480]. A. von Harnack thought Marcion was a modalist, because of such affirmations of Tertullian, but the question is today the subject of lively dispute (cf. Tertullien. Contre Marcion. Livre IIII [ed. René Braun; SC 399; Cerf: Paris, 1994] 7 n. 2). 91See the literary argument in the preceding section. 92In the Gospel of Luke, Jesus takes care of his disciples, praying for Peter (Luke 22:32) and still thinking of them on the cross (23:49); neither passage is attestedin Marcion (cf. Harnack, Marcion, 57 and 58). For a narrative reading of Luke 22:62, see Claire Clivaz, “Quandle récit fait parler le discours (Lc 22, 28–30). Vers un redéploiement du portrait lucanien des disciples,” in La Bible en récits. L’exégèse biblique à l’heure du lecteurr (ed. Daniel Marguerat; Genève: Labor et Fides, 2003) 381–84. 93Contra Marcionem 4.22.4–5. See also the commentary of Claudio Moreschini and René Braun, eds., Tertullien. Contre Marcion, livre IV (SC 456; Paris: Cerf, 2001) 280 n. 1. 432 HARVARD THEOLOGICAL REVIEW

Marcion omitted the last attempt of Peter to defend Jesus in Luke 22:49–51.94 Andif there is a strong distance between Jesus and Peter, Jesus cannot bear to look at him in Luke 22:61, even to arouse repentance: nothing can be expected from Peter. Thus, I propose that the omission of Luke 22:42–45a (or Luke 22:42, 45a) is intentional, not accidental, and has to be taken into account in the discussion of the evidence of Luke 22:43–44, as a third way of reading the prayer on the Mount of Olives. The Marcionite background I have sketched reinforces this proposal, by offering a frame of plausibility for this specific omission. As Kurt Aland remarked, ∏69 “follows its own way” and cannot be assimilated to a specific textual tradition. Consequently, scholars cannot continue to use ∏69 as a second early witness that omits Luke 22:43–44, in the same category as ∏75. Taking into account Celsus’s remark about the many changes in the textual traditions of the prayer at Gethsemane, we must abandon as dualistic and reductive such categories as “docetic/anti-docetic” or “Western tradition/Alexandrian tradition.”95

Family 13 and the Transfer of Luke 22:43–44 after Matt 26:39: A Proof ofWhat? Our re-evaluation of the external evidence for Luke 22:43–44 has yet to consider the possibility that the two verses stem from another source, a question that is raised by the transfer of those verses to a position after Matt 26:39 in f13. Neither Kurt and Barbara Aland nor Michael Patella hesitates to cite the transfer of the verses describing the angel and the bloody sweat in f13 as proof of the existence of another source—a source later than Luke for the Alands,96 a source prior to Luke for Patella.97 These opposite viewpoints suggest that the transfer of the verses is not signifi cant in and of itself, but must be evaluated carefully in order to determine whether it indicates the existence of another source for Luke 22:43–44. After a general presentation of f13, I shall evaluate the transfer of the verses by analyzing the data in the relevant manuscripts of f13, by comparing this transfer with another one foundin f13 (the transfer of the pericope of the adulteress), andfi nally by in- vestigating what the Historia passionis Domini can tell us about the verses. The Alands98 include the following manuscripts in f13: 13, 69, 124, 174, 230, 346, 543, 788, 826, 828, 983, 1689, and 1709; but 1709 does not contain Luke.99

94Epiphanius, Panarion 3.42.11,17, Elenchus 67 (Williams, 2:311). 95S. P. Cowe, “Christological Trends,” 37: “There was great contemporary diversity of opinion on the matter [i.e., Christology] and, particularly in Alexandria, a greater openness to accommoda- tion with certain facets of Gnosticism not directly opposed to revelation.” 96Aland, The Text,2d ed.,1989, 310: “it is one of the surest evidences for the secondary character of a text” (cf. part 1 above). Metzger shares this opinion (A Textual Commentary, 151). 97Patella, The Death of Jesus, 10: “The emendation in Matthew by f13 and Lectt raises the issue of a parallel account.” 98Aland, The Text,2d ed.,1989, 129. 99Text und Textwert der griechischen Handschriften des neuen Testaments, IV.1. Das Mark- susevangelium (ed. Kurt and Barbara Aland; ANTF 26; Berlin: W. de Gruyter, 1998) 19. CLAIRE CLIVAZ 433

Frederik Wisse,100 building on von Soden, gives a slightly different list: 13, 69, 124, 174, 230, 346, 543, 788, 826, 828, 837, 983, and 1689; but 837 does not contain Luke. Minuscules 174 (1052 C.E.) and 230 (1013 C.E.) belong for Wisse to group L: Jacob Geerlings, however, indicates that 174 and 230 are “in Luke as well as in Matthew members of the Ferrar Group, although both have been extensively corrected according to the Byzantine text current in South Italy.”101 Minuscules 174 and 230 present Luke 22:43–44 only in Luke, but 230 has asterisks from Luke 22:43 until 22:45a, as well as a reference to Matthew in the margin; and 174 has a reference to Luke in the margin after Matt 26:39. Thus, both manuscripts seem to be aware of a tradition that either associates the two verses with Matt 26:39 or transfers them to that location, and therefore show solidarity with f13 on this point. In light of this observation, the list of the Alands and the fact that 174 and 230 are among the four oldest manuscripts off f13, I propose the following list of f13 manuscripts concerning Luke: 13, 69, 124, 174, 230, 346, 543, 788, 826, 828, 983, and 1689, all of which I consulted on microfilms at the Institute of Münster, except for 1689, lost “since the war of 1914–1918 and [for which] we have only von Soden’s rather incomplete statement ofits readings.”102 Wisse and Geerlings agree that 826 (twelfth century) is representative of the entire group;103 for Wisse, 543 (twelfth century) is representative too, while 13 (thirteenth century), 346 (twelfth century) and 788 (eleventh century) are “core members.”104 The critical apparatus of NA27 states that f13 omits Luke 22:43–44 from Luke and transfers the verses after Matt 26:39.105 But a look at Text und Textwert der griechischen Handschriften des Neuen Testaments106 shows a more complicated situation for the manuscripts in question: whereas five members of f13 present the verses only after Matt 26:39 (69, 124, 543, 788, 826), four other members pres- ent them both in Luke and after Matt 26:39 (346, 828, 983, 1689107); one presents

100Frederik Wisse, The Profile Method for the Classification and Evaluation of Manuscript Evidence as Applied to the Continuous Greek Text of the Gospel of Luke (Grand Rapids: W. B. Eerdmans, 1982) 106. 101Jacob Geerlings, Family 13 (The Ferrar Group): The Text According to Luke (Studies and Documents XX; Salt Lake City: University of Utah Press, 1961) vii. Wisse specifies, however, that “there are some indications to locate the archetype of Gr 13 in Southern Italy or Sicily” (The Profi le Method, 107). 102Family 13 (The Ferrar Group:. The Text according to Mark with a Collation of Codex 28 of the Gospels (ed. Kirsopp and Silva Lake; Studies and Documents XI; London: Christophers; Philadelphia: University of Pennsylvania Press, 1941) ix. 103Wisse, The Profile Method, 106; Geerlings, Family 13, 8 (Geerlings precise 826 is not an archetype). 104Wisse, The Profi le Method, 106. 105See “f13 om. hic et pon. p. Mt 26, 39,” Novum Testamentum, 235. Also K. and B. Aland, The Text, 2d ed., 1989, 310: “[Luke 22:43–44] are found after Matt. 26:39 in the minuscule family 13.” 106Text und Textwert der griechischen Handschriften des Neuen Testaments. IV/3/2 Das Lukas- evangelium (ed. Kurt Aland,Barbara Aland, and Klaus Wachtel; ANTF 31; Berlin: W. de Gruyter, 1999) 111–12. 107Aland, Text und Textwert does not mention the lost 1689. I include it in this category following the indications of Geerlings, who in turn follows von Soden (Family 13, 133). 434 HARVARD THEOLOGICAL REVIEW the verses in Luke alone, but with asterisks (230); and two present the verses in Luke alone, without annotation (13C108, 174). Text und Textwertt reports that 13* omits Luke 22:43–44, but William H. P. Hatch remarks that 13* includes w[fqh de (the first two words of Luke 22:43) in the main text.109 This mass of confl ict- ing reports led me to examine anew the microfilms of f13, and I am able to affi rm that no minuscule of Family 13 ignored that the angel and the sweat like drops of blood were in Luke. The manuscripts of f13 that do omit the verses from Luke 22 nevertheless indicate a linkbetween the verses and Luke, either in the main text after Luke 22:42 (13*, 826), or in the margin of Luke 22:42 (13C, 69, 788), and/or in the margin of Matt 26:39 (124, 543, 788, 826).110 The manuscripts of f13 that preserve the verses exclusively in Luke contain either a reference to Matthew in the margin of Luke 22:43–44 (230), or a reference to Luke in the margin of Matt 26:39 (174). In the other manuscripts, the verses are present in both gospels, with a reference to Luke in the main text of Matt 26:39 (346) or in the margin of Matt 26: 39 (828), or with a reference to Matthew in the margin of Luke 22:43–44 (983).111 The most striking feature in this set of manuscripts is the presence of w[fqh de in the main text of 826 (Wisse’s and Geerlings’s representative of f13, f. 161 recto) and of 13* (core member of f13,f. 117 recto), andin the margin of Luke 22:42 with a reference to Matthew in 788 (core member off f13, f. 144 recto): this last detail allows us to infer that w[fqh de may serve as an abbreviated reference to Luke 22: 43–44. The third core member of f13, 346 (twelfth century), reproduces Luke 22: 43–44 both after Luke 22:42 and after Matt 26:39, referring to Luke in the main textt after Matt 26:39 (f. 43 recto). These observations demonstrate that the manuscripts of f13 asaclass cannott be summarized by the expression “f13 om. hic et pon. p. Mt 26,39” as in the critical apparatus of NA27 (see n. 105), and I hope that this statement can be corrected in a future edition. Indeed, only five out of twelve manuscripts in f13 omit the verses from Luke 22, while every manuscript off f13 attests a connection between the verses and the Gospel of Luke. The presence of w[fqh de after Luke 22:42 in the main text of a representative member (826) and of a core member (13), and in the margin of another core member (788), leads us to envisage a conscious transfer of the

108It is impossible to know if 13 had Luke 22:43–44 after Matt 26:39, because there is a lacuna at this place. 109William H. P. Hatch, Facsimiles and Descriptions of Minuscule Manuscripts of the New Testament (Cambridge: Harvard University Press, 1951) 206 n. 7. 110See 13: f. 117 recto; 69: f. 68 recto; 124: 47 recto; 543: f. 37 verso; 788: f. 52 recto and f. 144 recto; 826: f. 53 recto and f. 161 recto. The only difficult case is the margin of 69, which is very hard to read, and I warmly thank Prof. John Duffy for his help. As far as it can be read, 69 presents in the margin of f. 68 recto luvpiv, with a sign that refers to the end of Luke 22:42: that could refer to the signs of Jesus’ grief (Luke 22:43–44). A modern hand added in the margin “Mat. C. 26.” In any case, Geerlings affirms that “the scribe of 69 in my judgment copied carelessly and hastily [. . . and] used a manuscript closely related to 788 as archetype” (Family 13, 23). 788 has w[fqh de in the margin of Luke 22:42 (f. 144 recto). 111Cf. 174: f. 32 verso; 230: f. 163 verso; 346: f. 43 recto; 828: f. 44 recto; 983: f. 145 verso. 1689 is lost. CLAIRE CLIVAZ 435 verses from Luke to Matthew, a conjecture that is supportedby the fact that a third core member (346) includes a reference to Luke in the main text after Matt 26: 39.112 This conjecture is further supportedby an examination of our earliest wit- ness to the transfer of Luke 22:43–44 after Matt 26:39, found in the margin of the fi fth-century Ephraim Codex (C) at f. 62 recto. Luke 22 is not preserved in C, but the indications in the margin of Matt 26:39 are most instructive. Tischendorf gave an edition of this marginal note. The beginning is unreadable, but the preserved characters (. . . ouka kf. spg: wfqi de, etc.)113 include a clear reference to Luke. The scribe even indicates the Eusebian number of the Lukan passage, and quotes from Luke 22:43 to Luke 22:45a: as we have seen, this interruption has to be con- nected with the tradition ofliturgical reading (cf. second part). Three little crosses can be seen in this marginal note that highlight the anxiety of Jesus andhis sweat “like drops of blood,” and can help us to understand the concern of the corrector. This marginal note in C at f. 62 recto114 has not been dated by NA27. Returning to Tischendorf’s remarks, we can learn that the little crosses are a reminder of the text’s use in liturgical readings, and can be traced to the second corrector, who livedin Constantinople in the ninth century (C3).115 The evidence of the ancient uncialC confirms that the transfer of Luke 22: 43–44 (or 43–45a) after Matt 26:39 was a conscious transfer from Luke, and not an insertion taken from a hypothetical source.116 The placement of the crosses117 and the decision to end the passage in Luke 22:45a instead of Luke 22:44 in C3 support the idea that the transfer from Luke to Matthew was for liturgical pur- poses. C. Renoux underlined that possibility already in the fourth century, when he reported that the ancient Jerusalemite liturgy gave priority to Matthew in the readings for Holy Tuesday, andhe noted the preponderance of Matthew in general

112543 (twelfth century) is the last f13 member underlined by Wisse as representative and core member: 543 is a very clear minuscule, and the indications of the Eusebian canons are carefully reproduced beside the text, by the same hand. Luke 22:43–44 are present only after Matt 26:39 (f. 37 verso), but clearly linked with Luke (lo in the margin before w[fqh de, same hand, and mq at the end of the transferred verses, same hand). Jesus’ prayer contains a few crosses in the main text, similar to those in the margin in C, as discussedbelow: as in Cmg, these details allow us to understand the transfer of Luke 22:43–44 as the result of piety andliturgicalinfluence. 113Constantinus Tischendorf, Codex Ephraemi Syri Rescriptus (Lipsiae: Sumptibus et typis Bernh. Tauchnitz Jun., 1843) 25, n°46 supr. 114NA27 does propose a date for another marginal note of C at John 5:3, which it attributes to C3.See Cmg on p. 77 (NA27) for the marginal note next to Matt 26:39, and C3 on p. 260 (NA27) for the marginal note next to John 5:3. 115 Cf. Tischendorf, Codex Ephraemi, 1843, 24 and 20. The second corrector is usually called “C3,” because of a preceding corrector C**, contemporaneous with C*, and possibly C* him/herself (cf. ibid., 22; cf. also Aland, The Text, 2d ed. 1989, 108). 116Apart from C3, f13, and some lectionaries, the transfer of Luke 22:43–44 (or 43–45a) after Matt 26:39 is present only in 713, which gives Luke 22:43–45a after Matt 26:39 (f. 99 recto-verso), and also shows a liturgical influence. 117Similar crosses can be seen in 174 (f. 111 verso), 124 (f. 47 recto) and 543 (f. 37 verso). 436 HARVARD THEOLOGICAL REVIEW in the liturgy.118 The desire to read Matthew as a complete gospelin a liturgical frame is consequently not surprising, given the evidence of the f13 lectionary (see n. 68, above) and the transfer of our verses—including Luke 22:45a—by C3, 13C, 230C and 713. Raymond E. Brown has written that “Luke 22:43–45a was read on Holy Thursday (between Matt 26:21–39 and Matt 26:40–27:2); when it became customary to read 22:39–23:1 as a pericope on Tuesday of the last week before Lent, 22:43–44 was omitted from it to avoid duplication.”119 The desire to avoid duplication seems to be confi rmedby the presence of the two words w[fqh de in the main text of two important manuscripts of f13, 13* and 826; after beginning to copy Luke 22:43, the scribe apparently decided not to continue. Finally, a Matthean reading of Luke in f13 can be seen in other places, as the insertion of the Matthean explanation for Judas’s kiss in Luke 22:47 attests.120 Another well-known feature of f13 is its transference of the pericope of the adulteress from the Gospel of John to a position after Luke 21:38. Comparing this transfer to the transfer of the Lukan verses, David Parker has given equal weight to the transfers of John 7:53–8:11 and of Luke 22:43–44.121 But to what extent should these two transfers be compared? I would argue that the two transfers are very different. First, I have argued that the transfer of Luke 22:43–44 has a liturgi- cal origin and that the manuscripts that make the transfer are fully conscious of the Lukan background of the verses. Second, whereas the angel and the bloody sweat have as their only other location Matt 26:39, the pericope of the adulteress has, on the contrary, been transferred to many different locations in the gospels of Luke and John: after John 7:52; John 7:36; John 21:25; Luke 21:38; and Luke 24: 53.122 Third, the Lukan style of Luke 22:43–44 indicates that the verses are closely related to the Lukan Gospel, even if they embody a secondary imitation of Lukan style, whereas the style of John 7:53–8:11 is not Johannine.123 Fourth, according to Eusebius, Papias mentioned a story that could have been that of the adulteress, and that was preservedin a noncanonical gospel, the Gospel of the Hebrews.124 But we have no such attestation of the presence of the equivalent of Luke 22: 43–44 in a noncanonical document, if we read carefully the relevant quotation from the Historia passionis Domini (f. 32r), an unedited Latin manuscript from the fourteenth century.125

118Charles Renoux, “The Reading of the Bible in the Ancient Liturgy of Jerusalem,” in The Bible in Greek Christian Antiquity (ed. Paul M. Blowers; The Bible through the Ages 1; Notre Dame: University of Notre Dame Press, 1997) 397–98. 119Brown, The Death of the Messiah, 181 n. 4. 120Geerlings, Family 13, 134, and 22 on the Matthean influence in Luke 22:47 in f13. Even the scribe of ∏75 seems to have had Matthew in mind, reading Luke and John (cf. Comfort and Barrett, The Text of the Earliest, 504). 121Parker, The Living Texts, 158. 122Ibid., 96. 123Tuckett, “Luke 22, 43-44,” 134. 124Eusebius, Eccl. Hist. III, 39. 125The location of this manuscript is unknown, and Prof. Bovon is trying to find it. I hope it will CLAIRE CLIVAZ 437

The most complete edition of this extract is foundin the Synopsis Quattuor Evangeliorum: Sequitur Luc. 22. Apparuit autem ei angelus de celo confortans eum. Quali- ter autem angelus Christum in agonia sue oracionis confortaverit dicitur in Evangelio Nazareorum. Et idem ponit Anselmus in planctu suo. Constans esto domine modo enim venit tempus quo per tuam passionem redimendum est genus humanum in Adam venditum.

According to Luke 22. So an angel appeared to him, strengthening him. And the words by which the angel strengthened Christ in his struggle in prayer, are reportedin the Gospel of the Nazarenes. And the same is also adduced by Anselm in his lamentation: Be constant, Lord, for now comes the time in which through thy passion mankind sold in Adam will be ransomed.126 Kurt Aland connects this passage to the Gospel of the Hebrews,but Wilhelm Schneemelcher classifi edit as a reference to the Gospel of the Nazarenes.127 The Latin text shows clearly that the Historia passionis Domini does not attribute Luke 22:43–44 itself to an apocryphal source, but rather as a developmentt of this Lukan passage, in which the angel’s words of encouragement are included. Indeed, Fitzmyer, Brown, and Tuckett128 wrongly report the presence of Luke 22:43–44 in the Gospel of the Nazarenes, without noting that the apocryphal source actually contains a development of Luke 22:43–44. This nuance has been notedby Jean Duplacy,129 who indicates that Theodore of Mopsuestia also mentions the words of the angel.130 The abstract of the Historia passionis Domini also mentions a text

be published one day. The abstracts of this manuscript editedby Alandhadbeen transmitted to him by Bernhard Bischoff, “brieflich mitgeteilt” (cf. Kurt Aland, Synopsis Quatuor Evangeliorum. Locis parallelis evangeliorum apocryphorum et patrum adhibitis ediditt [Stuttgart: Deutsche Bibelgesell- schaft, 19954]585). Albertus Frederik J. Klijn mentions also certain data from the Historia Passionis Domini in Jewish-Christian Gospel Tradition (Supplements to Vigiliae Christianae XVII; Leiden: E. J. Brill, 1992) 142–46. These data had been directly transmitted to him by Bischoff. 126Latin text from Aland, Synopsis, 457. My English translation completes the partial translation in New Testament Apocrypha (ed. Wilhelm Schneemelcher; 2d ed.; Louisville: Westminster John Knox Press, 2003) 1:163. 127Ibid. Without a complete edition of the Historia passionis Domini, which seems to describe five other passages from the same apocryphal gospel, it remains difficult to confirm this attribution. 128See Fitzmyer, The Gospel according to Luke, 2:1443; Brown, The Death of the Messiah, 1: 180; Tuckett, “Luke 22,43-44,” 140 n. 39. 129Duplacy, “La préhistoire du texte en Luc 22:43-44,” 84. 130Fragmenta ex libris Contra Appollinarium (PG 66:995bc). 131In this text classifi ed among Anselm’s spurious works, one can read: “Secunda, quia tanta angustia eum invasit, quod guttas sanguineas sudavit de toto corpore […]. Et quando sic oravit: Pater, si possibile est, transeat a me calix iste, verum tamen non mea voluntas, sed tua fiat, angelus Domini apparuit ei confortans eum et dicens: Constans esto, Domine, modo genus humanum re- dempturus es” (PL 159:273A). A note in PL 159 col. 271C indicates: “Collatus est cum ms Thuano 519 ms Planctus beatae Mariae virginis ad Anselmum de passione Domini,” which explains the 438 HARVARD THEOLOGICAL REVIEW of Anselm, which I have been able to identify as the Dialogus Beatae Mariae et Anselmi de Passione Domini.131 That the Gospel of the Nazarenes already witnesses a development of Luke 22: 43–44 provides further evidence for the antiquity of the verses,132 anddiminishes the possibility that they are a later interpolation in Luke, in view of the time that it would have taken for a literary topic to be received, developed, and rewritten. I have moreover found seven other texts that refer to the angel’s words: four quotations and three allusions. The four quotations present a meaning similar to the quotations of the Historia passionis Domini and of Anselm: Ephiphanius in Ancoratus 37.7; in Panarion section 5.62.7; in a passage in the Catenae Graecorum patrum133; and another in the Syriac Book of the Holy Hierotheos 2.21. The three allusions to the angel’s words present a meaning similar to the allusion of Theodore of Mopsuestia (see n. 130), and the three are found in: a scholion on LXX Ps 68:14–15;134 in De coelesti hierarchia 4.4 of Pseudo-Dionysius; and probably also in the Questions of Bartholomew 1.9. All of this evidence for later developments of the angel’s words to Jesus can only recall John 12:29: “others said, ‘An angel has spoken to him.’ ” Adolf von Harnackdesignated John 12:29 as “the Johannine transformation of Luke 22:43,44.”135 The number of apocryphaldevelopments and secondary tradi- tions that aim to supply the angel’s words must lead us to reconsider von Harnack’s proposition. Whether John is dependent on Luke or not, John and Luke may both be referring to the same tradition about the angel at Gethsemane.

Conclusion In the first part of this paper, I underlined the complexity of the external evidence bearing on Luke 22:43–44. Paying attention to 0171 leads us to conclude that Ehrman’s statement that Luke 22:43–44 “are omitted by the earliest Greek manu-

designation of “planctu suo” for this work of Anselm in the Historia passionis Domini f. 32r. Klijn, extrapolating from Wilmart’s information, had proposed as possible reference Quid dedit capite meo in Migne, PL 187, 1133 (Klijn, Jewish-Christian Gospel Tradition, 143): but the PL’s number is wrong, and a lookback to Wilmart leads to the Liber de Passione Christi et doloribus et planctibus matris ejusm of S. Bernardus in PL 182, 1133 (André Wilmart, Auteurs spirituals et texts dévots du Moyen-Age latin. Études d’histoire littéraire [Paris: Études Augustiniennes, 1971, 2d ed., 517, n.1). However, we reach here a dead end again: the Liber de Passione does not speak at all about the prayer on the Mount of Olives. 132Schneemelcher gives 180 C.E. as terminus ad quem for the gospel of the Nazareans, referring to a testimony of Hegesippus (The New Testament Apocrypha, 1:159). 133John Anthony Cramer, ed., Catenae Graecorum Patrum in Novum Testamentum. Tomus II in Evangelia S. Lucae et S. Joannis, Oxonii: E Typographeo Academico, 1844, p. 159. 134This scholion has been attributed to Athanasius of Alexandria (PG 27:309), as well as to Origen (Analecta sacra spicilegio Solesmensi parata [ed. Jean Baptiste Pitra; 8 vols.; Paris: A. Jouby et Roger, 1876–1891] 3:86). It belongs more likely to Anastasius Sinaïta (seventh century). I thank Prof. Gilles Dorival for his help in this identification. 135von Harnack, “Probleme im Texte,” 254. 136Ehrman, Orthodox Corruption, 188. CLAIRE CLIVAZ 439 scripts” is wrong.136 This point is reinforcedby the fact that a* and a2 contain the verses, by the fact that we do not have the recension of Luke 22 by the “free” ∏45, and by the fact that ∏69 cannot be counted as a witness to the omission of these two verses alone. Developing the Alands’ reading, I have demonstrated in the second part of this article that ∏69 omits Luke 22:42–22:45a (or 22:42 and 22:45a, which remains hypothetical to determine). Thus, ∏69 must be regarded as a witness to a thirdd textual tradition within the Lukan pericope of Jesus’ prayer on the Mount of Olives, whichis a tradition that fi nds a plausible context in what we can deduce about the nature of Marcion’s gospel. All of these observations lead us to reject the traditional scholarly association of the presence of the angel and the bloody sweat with the Western tradition, as well as the association of their absence with the Alexandrian witnesses. The absence of Luke 22:43–44 is first attested in the Alexandrian tradition (cf. ∏75), but the early presence of the verses is attested in both the Alexandrian and Western traditions. Moreover, we cannot forget the third tradition attestedby ∏69; Celsus’s remark137 prevents us from reducing the textual problem of Luke 22:43–44 to a dualistic perspective. Finally, in the third part of this article I showed how we can still obtain valuable information about early Christianity from the study of the external evidence for Luke 22:43–44. The transfer of the verses to a position after Matt 26: 39 by f13 cannot be used to prove that the angel and the bloody sweat stem from a non-Lukan source, but rather demonstrates the influence ofliturgical usage on the textual tradition. Indeed, no manuscript off f13 fails to acknowledge that the passage came from Luke; the marginal note next to Matt 26:39 in C, already records the debt and witnesses a liturgical transfer. Finally, our admittedly limitedknowledge of the Historia passionis Domini does not indicate a non-Lukan origin for Luke 22:43–44. Rather, this Latin manuscript, along with other witnesses, testifies to an apocryphal tradition, linked with a Jewish-Christian gospel, that developed Luke 22:43 by putting words in the angel’s mouth. Consequently, no other source is presently available to attest a non-Lukan origin for Luke 22:43–44. The collectedinformation, however, emphasizes the importance of the angelin the reception history of the text. Indeed, a knowledge of the tradition about the speaking angelis probably already refl ectedin John 12:29. While Raymond E. Brown also pointed to the angel as a major issue in the debate over Luke 22:43–44,138 Ehrman and Plunkett correctly stated that problems about the angel “leave unanswered why both verses [Luke 22:43–44] are consistently absent in witnesses to the shorter text.”139 However, I would also argue that the anti-docetic argument could explain the interpolation of Jesus’ agony and sweat in

137“Certain of the Christian believers [. . .] have corrupted the Gospelfrom its originalintegrity, to a threefold, andfourfold, and many-fold degree, andhave remodeledit, so that they might be able to oppose negations to the objections” (Contra Celsium II, 27, quoted above in the second part of this article). 138Brown, The Death of the Messiah, 1:184. 139Plunkett, “The Angel and the Agony,” 408. 440 HARVARD THEOLOGICAL REVIEW

Luke 22:44 pretty well, but it does not explain why Luke 22:43’s description of the angel should also have been added. Thus, it is necessary to fi nd an explanation for the omission or interpolation of Luke 22:43–44 that deals with all the elements of Luke 22:43–44. Moreover, in the case of an omission, it remains to be demonstrated what could have led the orthodox community of Alexandria to delete the verses. Ehrman and Plunkett have rightly objected to Duplacy’s hypothesis, asking “How is it that Docetists came to exercise such influence on the Alexandrian text [in order to suppress the verses]?”140 The next challenge in the study of these verses, then, is to consider how the historical and sociological components present in the Alexandrian context might have determined their omission or interpolation; until now, the debate over Luke 22:43–44 has clearly remained too exclusively on a theological level.141

140Ibid., 408 n. 20. 141Ibid., 407: “Which reading is more readily explained as originating in the theological climate of the second century?” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.