An Inductive Method for Reconstructing the Biblical Text : Illustrated by an Analysis of 1 Samuel 3

AN INDUCTIVE METHOD FOR RECONSTRUCTING THE BIBLICAL TEXT Illustrated by an Analysis of 1 Samuel 3

JAMES R. ADAIR, JR.

To my Parents, Rob and Vela Adair, Y mis Suegros, Al y Cecilia Sierra

TABLE OF CONTENTS PREFACE

. . . . . . . . . . . . . . . . . . . . . . . . . .

i

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . .

1

Research Problem . . . . . . . . . . . . . . . . . . . . Definition of Terms . . . . . . . . . . . . . . . . . . . . Outline of This Study . . . . . . . . . . . . . . . . . . .

1 6 7

1. DEVELOPMENT OF A MODEL FOR ANALYSIS OF THE WITNESSES . .

9

Chapter

Literary Analysis before Textual Analysis Textual Analysis before Literary Analysis Genealogical Method . . . . . . . Determining the Translation Technique . The Script of the Vorlagen . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

9 13 16 21 27

2. RECONSTRUCTING ARCHETYPICAL SECONDARY WITNESSES . . . 30 Method for Reconstruction . . . . . . Location of Textual Data . . . . . . . Presentation of Reconstructed Witnesses . . Septuagint . . . . . . . . Peshitta . . . . . . . . . Targum . . . . . . . . . Vulgate . . . . . . . . . Aquila . . . . . . . . . . Symmachus . . . . . . . . Theodotion . . . . . . . . Other Readings Attributed to ´ The Lucianic Recension . . . . The Hexaplaric Recension . . . Other Possible Hebrew Readings .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

30 36 37 38 53 56 60 65 65 66 66 67 69 71

3. GRAMMATICAL/STYLISTIC CHARACTERISTICS OF THE SECONDARY WITNESSES . . . . . . . . . . . . . . 72 Septuagint . . . . . . . . . . . . . . . . Limitations of Greek for Rendering Hebrew . Partial Translation Technique . . . . . . Elimination of Variants . . . . . . . . Peshitta . . . . . . . . . . . . . . . . . Limitations of Syriac for Rendering Hebrew . Partial Translation Technique . . . . . . Elimination of Variants . . . . . . . . Targum . . . . . . . . . . . . . . . . . Limitations of Aramaic for Rendering Hebrew Partial Translation Technique . . . . . . Elimination of Variants . . . . . . . . Vulgate . . . . . . . . . . . . . . . . . Limitations of Latin for Rendering Hebrew . Partial Translation Technique . . . . . . Elimination of Variants . . . . . . . . Evaluation of Partial Secondary Witnesses . . . . . Aquila . . . . . . . . . . . . . . iv

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. 73 . 73 . 77 . 85 .105 .105 .108 .113 .128 .128 .129 .132 .143 .143 .145 .150 .161 .161

Symmachus . . . . . . . . Theodotion . . . . . . . . Other Readings Attributed to ´ The Lucianic Recension . . . . The Hexaplaric Recension . . . Other Possible Hebrew Readings .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

.162 .162 .162 .162 .163 .163

4. LITERARY/THEOLOGICAL CHARACTERISTICS OF THE SECONDARY WITNESSES . . . . . . . . . . . . . .164 Septuagint . . . . . . . . . . . . . . . . . . Relevant Literary, Historical, and Theological Data Present in the Translation . . . . . . . Final Translation Technique . . . . . . . . Elimination of Variants . . . . . . . . . . Peshitta . . . . . . . . . . . . . . . . . . . Relevant Literary, Historical, and Theological Data Present in the Translation . . . . . . . Final Translation Technique . . . . . . . . Elimination of Variants . . . . . . . . . . Targum . . . . . . . . . . . . . . . . . . . Relevant Literary, Historical, and Theological Data Present in the Translation . . . . . . . Final Translation Technique . . . . . . . . Elimination of Variants . . . . . . . . . . Vulgate . . . . . . . . . . . . . . . . . . . Relevant Literary, Historical, and Theological Data Present in the Translation . . . . . . . Final Translation Technique . . . . . . . . Elimination of Variants . . . . . . . . . . Evaluation of Partial Secondary Witnesses . . . . . . . The Lucianic Recension . . . . . . . . . . The Hexaplaric Recension . . . . . . . . . Other Possible Hebrew Readings . . . . . . . Conclusion . . . . . . . . . . . . . . . . . .

. . . . .170 . . . .

. . . .

. . . .

. . . .

.170 .172 .173 .177

. . . .

. . . .

. . . .

. . . .

.177 .178 .179 .180

. . . .

. . . .

. . . .

. . . .

.180 .181 .181 .182

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

.182 .184 .184 .187 .187 .188 .188 .188

5. CONSIDERATION OF VARIANTS FROM PRIMARY WITNESSES . . .189 Variants in Masoretic Manuscripts . . . . . . . . . . . . Important Manuscripts as Identified by Goshen-Gottstein Kethib-Qere . . . . . . . . . . . . . . . . Tiqqune Sopherim . . . . . . . . . . . . . . Variants in 4QSama . . . . . . . . . . . . . . . . . Conclusions: Significant Variants in 1 Samuel 3 . . . . . . . Septuagint . . . . . . . . . . . . . . . . Peshitta . . . . . . . . . . . . . . . . . Targum . . . . . . . . . . . . . . . . . Vulgate . . . . . . . . . . . . . . . . . The Lucianic Recension . . . . . . . . . . . . The Hexaplaric Recension . . . . . . . . . . . Other Possible Hebrew Readings . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

.190 .190 .192 .192 .192 .194 .194 .196 .197 .197 .197 .197 .198

6. RETROVERSIONS OF THE SECONDARY WITNESSES . . . . . . .201 Developing a Methodology for Retroverting Translations . . . Proposed Methodologies for Retroverting Translations Orthography . . . . . . . . . . . . . . . Conclusions concerning Methodology . . . . . . v

. . . .

. . . .

. . . .

.201 .201 .212 .216

Retroversions of the Secondary and Partial Secondary Witnesses Septuagint . . . . . . . . . . . . . . . Peshitta . . . . . . . . . . . . . . . . Targum . . . . . . . . . . . . . . . . Vulgate . . . . . . . . . . . . . . . . The Lucianic Recension . . . . . . . . . . . The Hexaplaric Recension . . . . . . . . . . Other Possible Hebrew Readings . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

.217 .217 .225 .228 .229 .229 .231 .233 .234

7. EVALUATION OF THE SIGNIFICANT VARIANTS . . . . . . . . .235 The Problem of Multiple Editions . . . . . . . . . . . Conjectural Emendations . . . . . . . . . . . . . . Guidelines for Evaluating Variants . . . . . . . . . . . Evaluations of the Variants . . . . . . . . . . . . . Variants That Pertain to the Question of Multiple Literary Editions 8. A CRITICAL EDITION OF 1 SAMUEL 3

. . . . .

. . . . .

. . . . .

.237 .240 .241 .243 .268

. . . . . . . . . . . .274

CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . .281 Appendix 1. RECONSTRUCTED ARCHETYPICAL TEXTS OF SECONDARY AND PARTIAL SECONDARY WITNESSES . . . . . . . . . . . .283 Septuagint . . . . . Targum . . . . . . The Lucianic Recension The Hexaplaric Recension

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

.283 .284 .284 .286

2. ADD-OMS AND PRELIMINARY TRANSLATION TECHNIQUE TABLES .288 Septuagint . . . . . . . . . Add-oms . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Peshitta . . . . . . . . . . Add-oms . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Targum . . . . . . . . . . Add-oms . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Vulgate . . . . . . . . . . Add-oms . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation

vi

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

.288 .288 .289 .295 .296 .296 .297 .297 .299 .304 .304 .305 .307 .307 .307 .313 .313 .313 .314 .314 .316 .322 .322 .322

3. FINAL TRANSLATION TECHNIQUE TABLES Septuagint . . . . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Peshitta . . . . . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Targum . . . . . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation Vulgate . . . . . . . . . . Consistency . . . . . Segmentation . . . . Word Order . . . . . Quantitative Representation

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .325 . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

.325 .325 .331 .332 .332 .332 .332 .338 .338 .338 .340 .340 .346 .346 .346 .347 .347 .353 .354 .354

4. ALGORITHM FOR CALCULATING THE DEVIATION FACTOR . . . .356 SELECTED BIBLIOGRAPHY

. . . . . . . . . . . . . . . . . . .359

vii

PREFACE The recovery of the original text of the Old Testament (OT) is an impossible task. All the original manuscripts (mss) have long been destroyed, as have the first, second, third, and many subsequent generations of copies. The oldest more or less complete witnesses to the original text are translations into other languages, and the oldest Hebrew mss of any significant size date from more than three centuries after the authors put down their pens. Of the witnesses that survive, none present satisfactory readings in every location, and there are some crucial places throughout the witnesses where none have convincing readings. There is even good reason to question the existence of an “original text.” If, as many scholars believe, the text of most OT books grew over time through the work of redactors and scribes (not to mention earlier written sources used by the primary authors), more than one “original text” may exist for many books. In spite of the problems, however, the task of textual criticism is of continuing importance. The greatest advances in the field of textual criticism of biblical books have come in the New Testament (NT) field, beginning with the late nineteenth century work of people like Tischendorf, Westcott, and Hort.1 Several factors have led to more textual work being done in the NT than in the OT: the availability of multiple early mss in the original language; the chronological proximity of many of these mss to the autographs; the diversity of readings among these mss; the existence of several versions and patristic witnesses to the text within five hundred years of the autographs; the relative simplicity of composition of many NT books in comparison with OT books (the gospels excepted); ms discoveries and acquisitions in the late nineteenth century, including Codices Sinaiticus, Vaticanus, and Bezae, Egyptian papyri, and the Gnostic mss from Nag Hammadi; and theologically motivated decisions by Christian scholars to do NT work. The difficulties of doing textual criticism on the OT, however, do not suggest the impossibility of progress, 1 Pier Giorgio Borbone, “La critica del testo e l’Antico Testamento ebraico,” review of Critique

textuelle de l'Ancien Testament, vol. 1, Josué, Juges, Ruth, Samuel, Rois, Chroniques, Esdras, Néhémie, Esther, by Dominique Barthélemy, in Rivista di storia e letteratura religiosa 20 (1984): 251, decries the retarded development of textual criticism of the Old Testament as compared with that of the New Testament, especially the lack of a critical Hebrew text (italics original): Il Nuovo Testamento è studiato e letto in edizioni critiche, condotte con metodo filologico; anzi, si può dire che proprio l’approfondimento dei problemi relativi all’edizione dei testi neotestamentari ha promosso in maniera determinante lo sviluppo del metodo della filologia classica. Diversa è invece la situazione dell’AT: mentre esistono edizioni critiche delle tradizioni greca e latina (i LXX e la Vulgata), ed il “Peshitta Institute” di Leida sta lavorando ad un’edizione della tradizione siriaca che, se proprio non si può dir ‘critica’ in senso tecnico, è un importante passo avanti rispetto alla situazione precedente, per il testo ebraico ci si accontenta di stampare un textus receptus.

ii

and renewed interest in the text of the OT has surfaced in the last forty years or so, especially since the discovery of the mss in the Judean desert. Impetus for further study has also come from the discoveries of texts from Ugarit and Ebla and from methodological advances proposed in the past twenty or thirty years, particularly by members of groups like the International Organization for Septuagint and Cognate Studies and the International Organization for the Study of the Old Testament. It is in the light of this newfound interest in the text of the OT and in the belief that methodologies exist or can be found to illuminate more fully previous states of the text that this study is undertaken. The present work is a combination and slight revision of my Ph.D. dissertation, undertaken at Southwestern Baptist Theological Seminary (“A Methodology for Determining the Textual Variants Which Are Relevant for Reconstructing the Original Text of the Old Testament: A Case Study of 1 Samuel 3”), and my M.A. thesis from the University of Stellenbosch (“Reconstructing 1 Samuel Chapter 3”). This work was largely finished in 1992, and few revisions to the bibliography have been made, other than to note my own studies which grew out of this one but were published later. The process of combining and revising my dissertation and thesis has been an onagain, off-again affair, and the calls of job, family life, and other academic interests have turned what I once thought would be a quick process into a five-year odyssey. I have been encouraged along the way by many people, without whom this book might not have seen the light of day. Special thanks go to Johann Cook of the University of Stellenbosch, who encouraged both my work on this book and my dream of creating an electronic journal, TC: A Journal of Biblical Textual Criticism, on whose editorial board he has served since the journal’s inception in 1996. My wife Rosa has been very supportive of my efforts to produce this book, and my daughters Danielle and Elise are a constant joy, tangible reminders of the goodness of God. Finally, my education, my interest in learning, and even my very survival I owe to my parents, Rob and Vela Adair. Though they don’t read Hebrew or Greek, they keep a copy of my dissertation on a coffee table in their living room to show to friends who visit. When I see it there, I’m reminded of how much of who I am is due to them. It is to them that this book is dedicated.

iii

INTRODUCTION Research Problem The text-tradition of the OT that has played the most important role in modern exegetical, historical, literary, and theological studies, as well as that which forms the basis for most translations, is the Masoretic Text (MT). MT owes its distinction to a variety of factors: (1) it is part of the text-tradition used almost exclusively by the Jews since about 100 C. E.; (2) it is the only complete witness in the original languages to the entirety of the Jewish Scriptures (the Protestant OT) that is extant; (3) it was the only Hebrew witness to the text known to scholars before the rediscovery of the Samaritan Pentateuch in 1616 and the ms discoveries of the past century (e.g., the mss of the Judean desert and the Cairo Geniza);1 (4) it was the text Luther and other reformers turned to after their rejection of the Latin Vulgate (V).2 MT, then, has occupied and continues to occupy a privileged place among witnesses to the text of the OT. In the books of Samuel and Kings, however, scholars have long accused MT of poor quality in many places, and they have turned to the Septuagint (LXX) and other witnesses as valuable tools for understanding the text. Several recent studies have analyzed MT (or some other witness) with little or no regard for other text-traditions. James A. Sanders makes the valuable observation that each text-tradition was viewed by its tradents as the text of Scripture, and thus it was understood without reference to other texts. Thus, modern interpreters can benefit from an understanding of the meaning(s) present in a single witness.3 Nevertheless, the goal of textual criticism is “to produce a text as close as possible to the original,”4 and this process 1 Cf. Interpreter’s Dictionary of the Bible [hereafter IDB], s.v. “Samaritan Pentateuch,” by Bleddyn J. Roberts; Shemaryahu Talmon, “The Old Testament Text,” in The Cambridge History of the Bible, ed. P. R. Ackroyd and C. F. Evans, (Cambridge: Cambridge University Press, 1970) 1: 159-99; Paul E. Kahle, The Cairo Geniza, 2d ed. (Oxford: Basil Blackwell, 1959). 2 For a synopsis of the debate over the value of MT from the Middle Ages to the early twentieth

century, see Dominique Barthélemy, Critique textuelle de l’Ancien Testament, vol. 1, Josué, Juges, Ruth, Samuel, Rois, Chroniques, Esdras, Néhémie, Esther, Orbis Biblicus et Orientalis, no. 50/1 (Fribourg: Éditions Universitaires, 1982; Göttingen: Vandenhoeck & Ruprecht, 1982), *1-*40. 3 “You must learn to respect each of the witnesses before you doubt them”; lecture delivered at the University of Stellenbosch, South Africa, 17 August 1989. Cf. also Sanders, From Sacred Story to Sacred Text (Philadelphia: Fortress Press, 1987), 127-51, esp. p. 134, “All versions are to some extent relevant to the communities for which translated”; Hans Joachim Stoebe, Das erste Buch Samuelis, Kommentar zum Alten Testament, vol. 8, pt. 1 (Gütersloh: Gütersloher Verlagshaus Gerd Mohn, 1973), 31: “[Der Exeget] muß sogar jede Rezension zunächst für sich genommen und so weit wie möglich aus sich verstanden werden.” 4 Paul Maas, Textual Criticism, trans. Barbara Flower (Oxford: Clarendon Press, 1958), 1.

1

2

necessarily involves the reconstruction of a presumed “original text.” The fact that the reconstructed text is not exactly the same as the autograph should not dissuade the textual critic, nor should the assertion that the new “eclectic” text does not represent the text used in any one community deter him or her.5 Much text-critical work has focused on a specific verse or word without regard for the larger context. For example, many scholars have proposed emending the pointing of in Isa. 40:6 MT to , partially on the basis of the LXX reading .6 However, the LXX reading here does not clearly indicate that the translator had a different understanding of the Vorlage (nor a different Vorlage), for a priest, rather than a prophet, is speaking in verses 1 and 2, and the translator may have placed the words of the second half of the verse in the mouth of the priest rather than the prophet.7 Furthermore, the translation technique of LXX Isaiah has often been described as free or even paraphrastic.8 The textual critic must be especially cautious about assuming that an apparent variant in a free translation indicates a difference in the Vorlage. 9 5 Of course, the textual critic may choose to reconstruct the text used by a particular community at

a particular time rather than to reconstruct the original archetype. But the charge that a “new” text has been created which has no historical worth is simply invalid. The textual critic’s goal, by definition, is to attempt to recreate earlier, non-extant forms of the text, not to reproduce a form of the text that was actually used and interpreted by a specific community (though the latter goal is legitimate both in itself and as an intermediate step toward the recovery of a more ancient text). Since the transmission history of the OT is so complex, many have sought to reproduce a form of the text other than the original. For example, the Committee for the Textual Analysis of the Hebrew Old Testament, working under the auspices of the United Bible Societies, has identified four possible texts toward which their committee could work: (1) the original oral or written form; (2) the most primitive form attested by extant witnesses; (3) the normative Jewish text after 70 C. E.; (4) MT as determined in the ninth or tenth century. With the exception of Proverbs, Jeremiah, and Ezekiel, they have sought to reconstruct the text as described in the second stage above; see Barthélemy, Critique textuelle, 1:*69. The OT textual critic must also be aware of the overlap between textual criticism and literary criticism when dealing with the early stages of the transmission of the text, since some traditions continued to grow after their form at one stage was recorded. Cf. Shemaryahu Talmon, “The Textual Study of the Bible—A New Outlook,” in Qumran and the History of the Biblical Text, ed. Frank Moore Cross and Shemaryahu Talmon (Cambridge: Harvard University Press, 1975), 33236; Emanuel Tov, “Some Sequence Differences between the MT and LXX and Their Ramifications for the Literary Criticism of the Bible,” Journal of Northwest Semitic Languages 13 (1987): 151-60. 6 For example, D. Winton Thomas in K. Elliger and W. Rudolph, eds. Biblica Hebraica Stuttgartensia, 2d ed. (Stuttgart: Deutsche Bibelgesellschaft, 1983) [hereafter BHS] and Rudolf Kittel in Rudolf Kittel, ed. Biblia Hebraica, 3d ed. (Stuttgart: Privegierte Württembergische Bibelanstalt, 1937) [hereafter BHK]. 7 Sanders, Lecture. 8 Emanuel Tov, The Text-Critical Use of the Septuagint in Biblical Research, Jerusalem Biblical Studies, no. 3 (Jerusalem: Simor, 1981), 63; chapters 36-39, however, are relatively literal; ibid., 51. See also Isac Leo Seeligmann, The Septuagint Version of Isaiah: A Discussion of Its Problems (Leiden: E. J. Brill, 1948), passim. 9 In a discussion of the history of the indiscriminate use of textual variants, M. H. Goshen-

Gottstein remarks sardonically, “Textual critics were at liberty to plough through the collations [of Hebrew mss] and to dig up what they thought useful for their purpose. They have done so ever since”; M. H. Goshen-Gottstein, “Hebrew Biblical Manuscripts: Their History and Their Place in the HUBP Edition,” Biblica 48 (1967): 253.

3

Some scholars, while accepting the validity of textual criticism, believe that it should be done only after each of the distinct text-traditions has been examined from a literary standpoint for consistency and “fit” in the context.10 However, this approach seems to confuse the roles of textual critic, literary critic, and interpreter. Though he or she should consider the literary qualities (or lack thereof) and internal consistency of a textual witness, the textual critic is not primarily interested in an aesthetic evaluation of the various witnesses. The fact that a witness “makes sense” as it stands does not imply its originality, much less its independence from the stream of tradition.11 Moreover, the meaning of a particular reading as enshrined in a “real” text-tradition is not necessarily that of the archetype of the tradition. Many factors, both intentional and unintentional, can produce variant readings, each of which makes perfectly good sense in the context. Other scholars contend that one must begin with analysis of the text before literary analysis is of any value. For example, Emanuel Tov says, “It is absolutely mandatory to start the analysis with the textual evidence.”12 He points particularly to the necessity of determining a witness’s translation technique before evaluating its readings, and he sets forth five criteria for determining the relative literalness of the translation.13 Once the translation technique of each witness is determined, the probability of accurate retroversion into Hebrew (or other source language) can be determined to a large degree, and an approximate retroversion can be produced. Tov also discusses the problems of retroverting translated material.14 A final approach to textual criticism is the genealogical approach, which attempts to reconstruct a tree that depicts the origin of the different extant states of the text as found in the various witnesses. Vinton A. Dearing, an English professor who has edited the text of 10 Cf., e.g., David W. Gooding, “An Approach to the Literary and Textual Problems in the DavidGoliath Story,” in The Story of David and Goliath, by Dominique Barthélemy et al., Orbis Biblicus et Orientalis, no. 73 (Fribourg: Éditions Universitaires, 1986; Göttingen: Vandenhoeck & Ruprecht, 1986), 59-82; cf. p. 83, “The final question is one of textual criticism” (italics mine). See also Stanley D. Walters, “Hannah and Anna: The Greek and Hebrew Texts of 1 Samuel 1,” Journal of Biblical Literature 107 (1988): 385-412. 11 Contra S. D. Walters, “Hannah and Anna,” 410: “Internal coherence and consistency must have

at least equal weight with genetic consideration.” It is not a question of the value of this evidence, but of its relative importance in the text-critical task. Does LXX’s version of Samuel really represent a tradition entirely separate from that of MT, and was there no single original text behind the two traditions? (cf. pp. 410-11). 12 “The Story of David and Goliath in the MT and LXX,” in David and Goliath, by Barthélemy et al., 130. He details his arguments on pp. 130-34. 13 Tov, Text-Critical Use, 54-60. See below, chap. 1, for an elaboration. 14 Ibid., 97-158. For an analysis of some of the problems involved in retroverting one version of a biblical book back into the language of its Vorlage, see John Russiano Miles, Retroversion and Text Criticism: The Predictability of Syntax in an Ancient Translation from Greek to Ethiopic, Septuagint and Cognate Studies, no. 17 (Chico, CA: Scholars Press, 1985).

4

various English authors’ works, provides the most detailed explanation of this method.15 He also propounds an algorithm for determining the tree, and he discusses the use of computers in analyzing the data. The genealogical approach is extremely important in determining the origin and relationship of the various extant witnesses to classical texts and other works which exist in relatively few distinct states,16 and it is this approach which is advocated in books dealing with the textual criticism of classical works.17 The construction of stemmata (genealogical trees) for works which have hundreds or thousands of extant witnesses (such as NT mss), however, is highly problematical, and the issue of how to deal with translations (a vital matter when dealing with the witnesses to the OT) is not directly addressed by this model. In view of the diversity of approaches, it is clear that no unanimity exists concerning the proper methodology for the study of the text of the OT. Therefore, some model needs to be designed to evaluate accurately the various text-traditions. Each of these competing models will be evaluated further in the first chapter of the present study, and a working model will be developed which will then be applied to the data from 1 Samuel 3. No study of the text of Samuel would be possible without the benefit of critical texts of the various witnesses. In the seventeenth century Louis Cappel contributed the earliest study of the OT text in modern times, producing a list of variant readings and suggested emendations.18 In the following century, a fresh collation of all known Hebrew mss was produced by B. Kennicott and supplemented by G. B. de Rossi. In 1875, Frederick Field published the fragments of Origen’s Hexapla preserved in various mss, versions, and church fathers. 19 The latter half of the nineteenth century and first decades of the twentieth century also saw the maturation of Septuagintal textual studies in the work of Paul de Lagarde and Alfred Rahlfs. The two great modern editions of LXX are the Cambridge Septuagint (1906 onward, presented as a diplomatic text, usually based on Codex Vaticanus [B] or Codex Alexandrinus [A], with variants indicated in the lower margin) and the Göttingen Septuagint (1931 onward, presented as a critical, eclectic text, 15 Vinton Adams Dearing, Manual of Textual Analysis (Berkeley: University of California Press, 1959). His method is updated and revised extensively in his later book, Principles and Practice of Textual Analysis (Berkeley: University of California Press, 1974). 16 A “state” is defined as a distinct form of a text which differs from all other forms in some way.

Thus, in the modern era, an entire printing of a work, since all copies are identical, constitutes only one state. 17 For example, Maas, Textual Criticism. 18 Louis Cappel, Commentarii et notae criticae in Vetus Testamentum, Amsterdam: Jacobus

Cappellus, 1689. Cappel’s study deals with the text of the entire OT. 19 Benjamin Kennicott, ed., Vetus Testamentum Hebraicum cum variis lectionibus, 2 vols. (Oxford: Clarendon Press, 1776-80); Giovani Bernardo de Rossi, Variae lectiones Veteris Testamenti, 4 vols. (n.p.: Parma, 1784-88); Frederick Field, ed., Origenis Hexaplorum, 2 vols. (Oxford: Oxford University Press, 1875; repr., Hildesheim: Georg Olms, 1964).

5

attempting to reconstruct the earliest possible form of LXX). Neither is yet complete, but, between the two, most of the books of the OT are represented (Samuel appears in the Cambridge Septuagint).20 Rahlfs produced a complete manual edition of LXX, but it is not limited, like the Cambridge Septuagint, to the presentation of one ms, but rather presents an eclectic text based primarily on the three important uncials B, S (Codex Sinaiticus), and A. It is thus a kind of intermediary between the Cambridge and Göttingen philosophies.21 The Targum Jonathan (T) includes the targum to Samuel, and a critical edition has been produced by Alexander Sperber.22 The Peshit@ta Institute in Leiden is currently preparing a critical edition of the Syriac Peshitta (P), presented as a diplomatic text, based in Samuel on ms 7a1 (Codex Ambrosianus).23 The Benedictine Order is in the process of producing a critical edition of V, on which the manual edition of the Würtemberg Bible Society is based.24 The final primary or secondary witness (see below, p. 7) to the text of 1 Samuel is 4QSama, a ms from the Judean desert, published in preliminary form by Frank Moore Cross.25 Several studies dealing with the text of 1 Samuel in particular have appeared in the last hundred years or so. The commentaries by Otto Thenius, Julius Wellhausen, and Henry Preserved Smith in the nineteenth century, and by S. R. Driver, P. Kyle McCarter, 20 Alan England Brooke, Norman McLean, and Henry St. John Thackeray, eds., The Old Testament in Greek (London: Cambridge University Press, 1906-); J. Ziegler et al., eds., Septuaginta: Vetus Testamentum Graecum auctoritate Academiae Scientarum Gottingensis editum (Göttingen: Vandenhoeck & Ruprecht, 1931-). 21 Alfred Rahlfs, ed., Septuaginta: id est Vetus Testamentum graece iuxta LXX interpretes, Editio

Minor (Stuttgart: Deutsche Bibelgesellschaft, 1979), lv. Cf. Peter Walters, The Text of the Septuagint: Its Corruptions and Their Emendation, ed. David W. Gooding (Cambridge: Cambridge University Press, 1973), 11. It must be pointed out, however, that Rahlfs was one of the architects of the Göttingen edition and that his manual edition was intended as a preliminary edition for use by clergy and students; Rahlfs, Septuaginta, lv. 22 Alexander Sperber, ed., The Bible in Aramaic, 4 vols. in 5 parts (Leiden: E. J. Brill, 19591973). The abbreviation “T” will be used of the targums in general and of Targum Jonathan in particular when it is clear from the context to which it is a reference. In ambiguous contexts, the abbreviation will be avoided. 23 The Old Testament in Syriac According to the Peshi a Version, edited on behalf of the

International Organization for the Study of the Old Testament by the Peshit@t a Institute, Leiden (Leiden: E. J. Brill, 1972-). 24 The Benedictine edition is Biblia Sacra iuxta Latina Vulgatum versionem ad codicem fidem iussu Pii XII, ed. Cura et studio monachorum abbatiae pontificiae Sancti Hieronymi in urbe ordinis Sancti Benedicti (Rome: Typis Poliglottis Vaticanus, 1926-); it is referred to in the text as VR and in the footnotes as Biblia Sacra Romana. The manual edition is Bonifatius Fischer et al., eds., Biblia Sacra iuxta vulgatam versionem, 2d ed., 2 vols. (Stuttgart: Württembergische Bibelanstalt, 1975); it is referred to in the text as VS and in the footnotes as Biblia Sacra Stuttgartensia. The latter edition is more than a mere abbreviation of the larger work. The editors have examined the text of previous editions and their apparatuses, altering the text found in the larger work when deemed necessary. 25 Frank Moore Cross, “A New Qumran Biblical Fragment Related to the Original Hebrew

Underlying the Septuagint,” Bulletin of the American Schools of Oriental Research 132 (1953): 15-26.

6

and Ralph Klein in the twentieth century, deal extensively with the text.26 Eugene Ulrich’s study of the relationships among 4QSama, MT, LXX, and Josephus supports Cross’s theory of local texts as it applies to Samuel, detailing the general convergence of the texts of 4QSama, LXX, and Josephus as opposed to MT. Stephen Pisano’s 1982 doctoral dissertation deals with the additions and omissions in the Samuel texts of MT, LXX, and Qumran. Finally, the work of the Committee for the Textual Analysis of the Hebrew Old Testament in analyzing many significant variants in Samuel provides important data for the study of the text.27 In spite of the great amount of work done in the area of the text of 1 Samuel, however, it remains necessary to determine those textual variants that are truly significant for the recovery of the oldest possible Hebrew text. Developing a methodology by which to determine which variants are significant (Chapter 1), applying this methodology to 1 Samuel 3 (Chapters 2-5), and using the information generated to reconstruct the earliest possible text of 1 Samuel 3 (Chapters 6-8) will be the goals of this study. Definition of Terms For the purposes of this study, the term “significant variant” will refer to those variants which have some probability of representing a Hebrew Vorlage different from the base text, MT, where the variation is not purely orthographic.28 The choice of MT as the base text does not imply that it is superior to other witnesses; it is used as a base because it is the only complete witness in Hebrew. The term “(textual) witness” refers to an extant textual tradition, either complete or incomplete, whether Hebrew or some other language. It may refer to an individual ms, or it may be used to represent the presumed textual tradition lying behind a group of related mss (e.g., LXX or Lucianic mss [LXXL]). 26 Otto Thenius, Die Bücher Samuels, 2d ed., Kurzgefasstes Exegetisches Handbuch zum Alten

Testament, no. 4 (Leipzig: S. Hirzel, 1864); Julius Wellhausen, Der Text der Bücher Samuelis (Göttingen: Vandenhoeck & Ruprecht, 1871); Henry Preserved Smith, Critical and Exegetical Commentary on the Books of Samuel, International Critical Commentary (Edinburgh: T. & T. Clark, 1899); S. R. Driver, Notes on the Hebrew Text and the Topography of the Books of Samuel, 2d ed. (Oxford: Clarendon Press, 1913; repr., Oxford: Oxford University Press, 1960); P. Kyle McCarter, Jr., I Samuel, The Anchor Bible, ed. William F. Albright and David Noel Freedman, vol. 8 (Garden City, NY: Doubleday & Company, 1980); Ralph W. Klein, 1 Samuel, Word Biblical Commentary, vol. 10, gen. eds. David A. Hubbard and Glenn W. Barker, Old Testament editor John D. W. Watts (Waco: Word Books, 1983). 27 Eugene Charles Ulrich, Jr., The Qumran Text of Samuel and Josephus, Harvard Semitic Monographs, no. 19 (Missoula, MT: Scholars Press, 1978); Stephen Pisano, Additions or Omissions in the Books of Samuel: The Significant Pluses and Minuses in the Massoretic, LXX and Qumran Texts, Orbis Biblicus et Orientalis, no. 57 (Fribourg: Univeritätsverlag, 1984; Göttingen: Vandenhoeck & Ruprecht, 1984); Barthélemy, Critique textuelle, vol. 1. See also Frank Moore Cross, “The Evolution of a Theory of Local Texts,” in Qumran and the History of the Biblical Text, ed. Frank Moore Cross and Shemaryahu Talmon, pp. 306-20 (Cambridge: Harvard University Press, 1975). 28 Variants which are not significant will be termed “nonsignificant” rather than “insignificant,”

because of the broader connotations of the latter term. Some of the nonsignificant variants may not be insignificant in terms of understanding the meaning of the text or the historical development of the text.

7

In the latter case, it is equivalent to the term “text-tradition.” A “(textual) variant” within a tradition is a reading that deviates from that of another witness within the same tradition. Thus, the readings of the base text and of a divergent witness are both considered variants. More generally, where a unit of variation exists, one may speak of a variant as either the reading of MT or that which represents a Hebrew Vorlage different from MT. Variants that entail differences in length between various witnesses, ranging from one lexeme to many lexemes, are often called additions/omissions or plusses/minuses. However, this terminology is not satisfactory, because it implicitly gives preference to the base text, normally MT. In this study, all such variants will be termed “add-oms.” 29 The terms addition/omission and plus/minus, when used, will refer to deviations from the presumed archetype, not the base text. Those variants that do not differ significantly in length from the readings of other witnesses may be divided into substitutions and transpositions. All three of these types of variants are termed simple variants; complex variants are composed of any combination of simple variants. The witnesses to the text of Samuel may be divided into three groups.30 The “primary text-traditions” are those written in the original language, Hebrew (i.e., MT and Qumran mss). The “secondary text-traditions” are those translated directly from the Hebrew (i.e., LXX, P, T, V). “Tertiary text-traditions” are translations of secondary text-traditions (e.g., Old Latin [it], Coptic [cop], Armenian [arm], Ethiopic [eth], Syrohexapla [syh]). An attempt will be made to isolate all independent witnesses to a Hebrew text. For example, Lucianic witnesses, even when clearly secondary for determining the text of the Old Greek LXX, may be witnesses to an independent Hebrew textual tradition in places. Thus, all text-traditions which have been corrected at some point by a Hebrew text may also be considered secondary traditions at those points. Such texttraditions will be called “partial secondary witnesses.” Outline of This Study The first chapter of this study will examine in more detail the various models suggested above for doing textual criticism as they apply to the text of 1 Samuel. A model for textual analysis will be developed which will then be used to analyze 1 Samuel 3 in Hebrew, Greek, Syriac, Aramaic, and Latin. Chapter 2 will discuss an abbreviated method of reconstructing archetypical secondary witnesses for use in the following chapters. Chapters 3 and 4 will analyze the various secondary witnesses from 29 Adapted from Dearing, Principles and Practice, 25, where “add-omission” is used. 30 For diagrams detailing the relationships among the various witnesses, see F. E. Deist, Towards

the Text of the Old Testament, trans. W. K. Winckler (Pretoria: N.G. Kerkboekhandel Transvaal, 1978), 236; Ralph W. Klein, Textual Criticism of the Old Testament: The Septuagint after Qumran, Guides to Biblical Scholarship, Old Testament Series (Philadelphia: Fortress Press, 1974), 71; and Talmon, “Old Testament Text,” 195.

8

lexical/grammatical/stylistic and literary/theological points of view, respectively. Chapter 5 will briefly list other extant variants, gleaned from primary witnesses (variants in the Masoretic mss and in 4QSama), summarize the search for significant variants, and list all those that have been uncovered. The sixth chapter will discuss the methodology of retroversion from one language into another and will then attempt to retrovert, in whole or in part, the secondary and partial secondary witnesses into Hebrew. Chapter 7 will take all the Hebrew variants (both originally Hebrew and retroverted readings) that have been determined to be significant, evaluate them, and attempt to arrive at a conclusion concerning the earliest reading. Finally, Chapter 8 will present the results of the study in the form of a critical Hebrew text of 1 Samuel 3, including critical apparatuses.

CHAPTER 1 DEVELOPMENT OF A MODEL FOR ANALYSIS OF THE WITNESSES If one is to avoid an arbitrary or haphazard approach to textual criticism, it is necessary to have some sort of a model to follow. The problem with many attempts at textual analysis is that they are done without paying sufficient attention to matters such as translation technique, larger context, and literary characteristics of the witness. Such shortfalls are always a danger with “textual notes,” whether based on philological or textual data.1 To say this is not to deny the helpfulness of such observations in many cases but rather to remind scholars that brief, random observations must be tested with methodological soundness. The three approaches delineated in the Introduction will be discussed and evaluated in greater detail, and the insights gained from them will be used to prepare a method for evaluating the text of Samuel. Literary Analysis before Textual Analysis Those who believe that literary analysis2 should be done before textual analysis fall into two broad groups: those who do so for theological reasons and those who do so for purely literary reasons. A growing number of scholars is becoming interested in examining MT and other traditions independently of the others, viewing them as traditional texts that 1 Cf. James Barr, Comparative Philology and the Text of the Old Testament (Oxford: Oxford University Press, 1968; repr., with additions and corrections, Winona Lake, IN: Eisenbrauns, 1987), 304: “The basic assumption, that study of the relations between the Semitic languages may further the understanding of the Hebrew Bible, is incontrovertible. The trouble has not lain in comparative scholarship, but in poor judgement in its application, and in failure to see and follow out some of the general linguistic questions which are already implied in the primary use of comparative method.” Although referring only to the misuse of linguistic data, a similar statement could also be made of the misuse of textual data. Cf. also P. A. H. de Boer’s summary of H. S. Nyberg’s approach to the text: “The recension should first be determined, then the interpretation taken in hand and the amendment postponed as long as possible…”; P. A. H. de Boer, Research into the Text of I Samuel i-xvi: A Contribution to the Study of the Books of Samuel (Amsterdam: H. J. Paris, 1938), 9. 2 The term literary analysis as used in this context is usually not the older source criticism (though

the arguments are similar for this type of literary criticism) but rather an attempt to analyze the existing form of the text and discern its meaning(s). Scholars who employ such forms of literary criticism do so in a variety of ways, including reading the Bible as pure literature, structuralism, rhetorical criticism, close reading, and deconstruction. The issue here is not whether any or all of these methods are legitimate but rather whether the search for an earlier form of the text is best served by using literary analysis from the beginning. See, for example, David Robertson, The Old Testament and the Literary Critic, Guides to Biblical Scholarship, Old Testament Series (Philadelphia: Fortress Press, 1977), passim; Robert C. Culley, “Exploring New Directions,” in The Hebrew Bible and Its Modern Interpreters, ed. Douglas A. Knight and Gene M. Tucker (Philadelphia: Fortress Press, 1985; Decatur, GA: Scholars Press, 1985), 167-200.

9

10

were actually used by a worshiping community.3 An example of this approach may be seen in several recent translations of the Bible that, in the apocryphal or deuterocanonical section, render the entire book of Esther from the Greek rather than presenting only the pluses to the Hebrew in a confused and disjointed manner, as many earlier translations did. Implicit in this editorial decision is the recognition that the Greek form of Esther in its totality has served various communities, both past (Greek speaking Jews and Christians) and present (Roman Catholics and Eastern Orthodox), as sacred text.4 As mentioned above, the use of textual criticism to determine a text other than the oldest (i.e., one actually used by a specific community) is a legitimate modification of the more traditional goal of searching for the earliest text. Such a choice, though, is a theological, rather than a textcritical, choice.5 For this reason, the textual critic has no basis for contesting the prior choice of a particular literary tradition before textual criticism is done. However, some scholars, for nontheological reasons, believe that literary analysis should always precede textual analysis. In the process of examining the divergent traditions of the story of David and Goliath found in MT and LXX, D. W. Gooding demonstrates his preference for doing literary analysis before textual analysis.6 He particularly stresses the need to examine the extant form of a text to see what it says, implying that if it makes sense as it stands, textual modifications are unnecessary. Note, for example, the following arguments. 3 Sanders, Sacred Story, 134. Cf. Ferdinand E. Deist, “Is die Massoretiese Teks die Ou Testament?” Skrif en kerk 10 (1989): 9-20, especially 16-19. 4 The inclusion of two forms of Esther, however, is based primarily on the acceptance of the two

forms by various modern, rather than ancient, religious groups. If the use of different forms of a book by various ancient communities were included as a criterion for translating more than one form, variant forms of such books as Jeremiah, Proverbs, and Job would be included on the basis of the differences between the Hebrew and (Old) Greek texts, and two forms of Judges and Daniel (three forms?) would also be translated on the basis of greatly divergent Greek manuscripts, as Sanders has in fact suggested. 5 Others would claim that the presumed “original” text is the only one that carries divine authority, so it is the one that should be sought. This, too, is a theological choice. (The textual critic should avoid the equation “original text” = “correct text”; cf. Peter D. Miscall, 1 Samuel: A Literary Reading, Indiana Studies in Biblical Literature [Bloomington: Indiana University Press, 1986], viii.) However, the search for earlier forms of the text need not be theologically motivated; it can also be a purely text-critical exercise, unfettered by theological presumptions of the superiority of earlier forms of the text. Because textual criticism is not a theological discipline, it cannot tell the contemporary community which ancient witness, extant or reconstructed, is best for modern translation. 6 Gooding, “Approach to the Problems,” 63: “Let us consider first the context of the book as a

whole.” Twenty pages later he says, “The final question is one of textual criticism,” ibid., p. 83. Cf. also the approach of Dominique Barthélemy, “Trois niveaux d’analyse (a propos de David et Goliath),” in David and Goliath, by Barthélemy et al., 47, who begins his detailed discussion of the story by saying, Si nous envisageons ces trois récits [i.e., 1 Sam 16:1-13 (A); 16:14-23 (B); 17:12ff. (C)] comme des essais de légitimation des prétentions de David à la royauté, le récit A est celui qui atteint ce résultat de la manière la plus efficace. Si les auteurs des récits B et C l’avaient connu, ils auraient pu renoncer à leurs essais moins convaincants. Il est donc vraisemblable que A est le dernier-né des trois récits. On a de cela un indice complémentaire dans le fait que les autres récits ignorent cette onction de David enfant.

11 [Critics should not try] to claim to prove irreconcilable discrepancy, and therefore multiple authorship, by selecting a detailed feature from one context, contrasting it with a detailed feature from another context, without first carefully examining the place each feature holds and the function it performs within the thought-flow of its own particular movement. . . . To understand a narrative from a literary point of view, we must first listen to the narrative as it stands, trying to see where and how each part fits into the thought-flow of the whole. In other words we must initially give the narrative the benefit of the doubt. . . . Only after a sustained and sympathetic attempt at making sense of the narrative as it stands should we reluctantly conclude that the narrative is an irreconcilably discrepant hotch-potch.7 Several points may be made in response. First, it must be said that scholars like Gooding who are once again examining the final form of a biblical text to see what its message is as it stands are filling a needed and much neglected void in earlier critical scholarship, which tended to focus on earlier, hypothetical forms of the text and ignore the present form. Having said that, however, the question remains: should this examination of the final form affect text-critical judgments? Just because the final form of a text “makes sense” as it stands, does this imply that it is the original form?8 To say that it does is to agree that any authors or redactors that might have used earlier sources must indeed have used a “scissors and paste” approach in their composition. If, in fact, the author of the final form was truly an author in some sense, it is quite possible that he or she would have smoothed over more obvious difficulties or discrepancies. Thus, an understandable final form does not imply either verbal proximity to the original form or unitary authorship. Second, while it is true that critics in the past have sometimes exaggerated differences of detail between different parts of a biblical text, it remains true that discrepancies and notable differences of presentation do provide both textual and literary critics with clues pointing to deviations from a unitary tradition, whether textual or literary. The phrases “benefit of the doubt,” “sympathetic attempt,” and “reluctantly conclude” seem more appropriately applied to persons standing trial than to critical evaluation of textual data.9 While one must guard against assuming multiple authorship or textual disruption before examining the data, so must one avoid either simplistic or contrived explanations of apparently incongruous data that unfairly prejudice the case in favor of 7 Gooding, “Approach to the Problems,” 58. 8 Cf., e.g., the approach of Gnuse to “retain the Massoretic text when it seems logical”; Robert

Karl Gnuse, The Dream Theophany of Samuel: Its Structure in Relation to Ancient Near Eastern Dreams and Its Theological Significance (Lanham, MD: University Press of America, 1984), 121. See also Stoebe, Erste Buch Samuelis, 25-32; P. A. H. de Boer, I Samuel i-xvi, passim. De Boer also produced the textual notes for Samuel in BHS. 9 Indeed, conservative critics have often used the Anglo-American legal tradition of “innocent until proven guilty” as an analogy to suggest that scholars should respect traditional views, sometimes including acceptance of MT readings, unless overwhelming evidence proves them to be untenable (e.g., Robert Dick Wilson, A Scientific Investigation of the Old Testament [Philadelphia: The Sunday School Times Company, 1926], 19-20). However, the reason for the legal dictum is protection of the innocent, even if injustice is sometimes the result. Since no one is at risk of injury in scholarly discussions of the text, favoritism of traditional views is unwarranted. Of course, Gooding’s views of the development of the text do not correspond to those of these conservative critics, but his use of language at this point is similar.

12

unitary authorship or faithfulness to the textual tradition. Third, when Gooding says, “To understand a narrative from a literary point of view, we must first listen to the narrative as it stands,” he makes a valid point. The exegete should examine the final form of a text. However, though the textual critic does do exegesis, to interpret is not the goal of his work but rather a means to an end, namely, discovering an earlier form of the text. An important difference exists between examining a text “from a literary point of view” and doing so from a textual point of view. The two must not be confused. Stanley D. Walters also believes that literary analysis is primary, and he uses an excerpt from 1 Samuel 1 as an example. His primary contention is that the narratives of Samuel’s birth as recorded in MT and LXX (specifically, ms B) are “discrete narratives, each with its own Tendenz.” 10 He interprets the two versions of the story as they stand, and he compares them with other biblical texts with similar vocabulary or motifs. He notes that “it is a signal advantage of the comparative method that we can see features in each story that could otherwise easily have been missed.”11 The proposed interpretations are interesting, even ingenious at times. However, his understanding of the text-critical task requires scrutiny. He contends that scholars from the time of Thenius have often tried to rewrite the text rather than explain it, and the result has been the creation of various forms of the text that probably never existed. Even the discovery of the Qumran mss, revealing the existence of Hebrew texts closer to LXX than MT, has confused rather than helped the situation. In light of this situation, he says, “we must return to the commentator’s primary task, that of explaining the text.”12 It is at this point that an important distinction must be made. While it may be that a commentator’s primary task is to explain the text as it is (though this assertion in itself is debatable), it is certainly not the textual critic’s task. Again, though the textual critic may in fact create a text that never existed in all its details, it is the nature of the text-critical task to reconstruct hypothetical texts. The amount and nature of the data will determine the chances of success in each case, but the discovery of more material (such as the Qumran mss) enhances rather than hinders the attempt. Literary analysis is important for the text-critical task, as well as for the exegetical task. Examination of the larger literary context may well clear up apparent discrepancies or reveal certain stylistic or theological tendencies in a particular witness or within the tradition as a whole. Nevertheless, it seems clear that for the textual critic, literary analysis must accompany, not supersede or supplant, a thorough examination of all existing textual data.13 10 S. D. Walters, “Hannah and Anna,” 409. 11 Ibid., 410. 12 Ibid. 13 Cf. also Miscall, who says that “text-critical study should be preceded by extended readings,

13 Textual Analysis before Literary Analysis In The Story of David and Goliath, both Johan Lust and Emanuel Tov say that textual analysis is primary. Lust, in his response to the initial papers by the other contributors, states this point clearly. While agreeing that textual and literary analysis are both important, he says, A close reading of the four papers strengthened my conviction that the data and questions of textual criticism should be handled first. To a certain extent, they present facts, providing a sound basis for further more hypothetical theories. The differences between the Hebrew and the Greek texts should be noted carefully, including both pluses and minuses. . . . Once these data [Hebrew and Greek mss, Qumran, patristic evidence] are listed and checked, one should ask whether the difference between the Greek and the Hebrew texts are intentional or unintentional and whether they are due to a scribe, an editor or a translator.14 Tov likewise affirms the necessity of examining the textual data first: For the story of David and Goliath we possess two main sources, MT and the LXX, and these are by implication textual sources, witnessing that abstract entity which we call the biblical text. In our discussion we must attempt to approach these sources as impartially as possible. This means by implication that we should not speak first about the literary problems of the story nor about its textual difficulties, but we should first assess the value of the LXX because the LXX is one of the sources which contain the story. . . . The starting point of the discussion of the Greek version of David and Goliath should be the translation technique of the Greek translator. . . .15 Thus, both Lust and Tov stress the importance of beginning textual criticism with an examination of the text. Tov particularly stresses the importance of determining the translation technique of LXX. It is at this point that a comparison between the text-critical methodology employed in the OT and the NT will be helpful. It has already been noted that many Greek mss the NT exist that were written within the first few centuries of the original compositions.16 In contrast, few if any Hebrew mss date from so close a time to the originals, and the vast majority come from the medieval period, more than one thousand years after composition. Perhaps more importantly, the thousands of mss of the Greek NT contain a rich variety of readings in almost every verse, and at least two, perhaps three or four, major text types can be distinguished among the mss, in addition to numerous other wherever possible, of each version”; Miscall, 1 Samuel, viii. 14 Johan Lust, “Second Thoughts on David and Goliath,” in David and Goliath, by Barthélemy et

al., 87. 15 Emanuel Tov, “Response by E. Tov,” in David and Goliath, by Barthélemy et al., 92-93 (italics original). Cf. also Natalio Fernández Marcos, Introducción a las versiones griegas de la biblia, Textos y estudios “Cardinal Cisneros”, no. 23 (Madrid: Instituto “Arias Montano” Consejo Superior de Investigaciones Científicas, 1979), 25: “El análysis de las técnicas de traducción de cada libro o de cada unidad de traducción ha de preceder a todo estudio de sintaxis.” 16 Kurt and Barbara Aland list forty-five NT mss that date prior to the fourth century; Kurt Aland

and Barbara Aland, The Text of the New Testament, trans. Erroll F. Rhodes (Grand Rapids: William B. Eerdmans, 1987; Leiden: E. J. Brill, 1987), 57.

14

early readings.17 Medieval Hebrew mss, on the other hand, represent one text type and contain almost no variants that might reflect a competing text.18 OT textual criticism, then, is much more dependent on witnesses that are translations than is NT textual criticism. Whereas NT textual critics can almost ignore the versional evidence, since almost every variant within a version with any claim to originality also occurs in at least one Greek ms,19 OT textual critics cannot. In fact, versions of the OT, especially LXX, have long been used to provide textual evidence, though not always in a consistent manner. One desideratum of developing a methodology for OT textual criticism is to determine how to use versional alongside Hebrew evidence. Though NT textual critics have employed various approaches in the past, the majority today use one of two methods: rigorous eclecticism or rational (modified) eclecticism.20 Rigorous eclecticism, as employed, for example, by G. D. Kilpatrick and J. K. Elliott, focuses exclusively on internal evidence in evaluating variant readings. External evidence such as age, geographical setting, quantity of witnesses, and distribution among text types are considered of little or no importance in determining the correct reading. Instead, the critic examines the author’s style and vocabulary to determine what he was more likely to have said. He or she also attempts to discern among the readings which one was most likely to have given rise to the others.21 The scholar who uses rational eclecticism does consider internal evidence, but he or she also takes into 17 Aland and Aland, Text of NT, 56-67; Bruce M. Metzger, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 2d ed. (New York: Oxford University Press, 1968), 211-19. Cf. also Bruce M. Metzger, A Textual Commentary on the Greek New Testament, corrected ed. (London: United Bible Societies, 1975), xv-xx. 18 Goshen-Gottstein, “Hebrew Biblical Manuscripts,” 287. He identifies Kennicott’s manuscripts

70, 89, 174, and 187 as having several readings that differ from the main stream of MT in 1 Samuel. The variations between the Ben Asher and the Ben Naphtali mss are minimal (only eight variations in the consonantal text); cf. Ernst Würthwein, The Text of the Old Testament: An Introduction to the Biblia Hebraica, trans. Erroll F. Rhodes (Grand Rapids: William B. Eerdmans, 1979), 24-25. 19 Cf. the statement of the Alands: “It must be emphasized that the value of the early versions for establishing the original Greek text and for the history of the text has frequently been misconceived, i.e., they have been considerably overrated”; Aland and Aland, Text of NT, 182. They say later on the same page that the versions are useful in “confirming the identity of the regional or provincial text where they were produced,” but this is mere confirmation in most cases of the text found in Greek mss. See also p. 275, “The primary authority for a critical textual decision lies with the Greek manuscript tradition, with the versions and the Fathers serving no more than a supplementary and corroborative function . . . .” (italics mine). Even those scholars who might see somewhat more text-critical value in the NT versions would probably still attribute far less importance to them than most OT textual critics do to LXX or even P. 20 For a brief overview of these methods, see James A. Brooks, “The Text of the New Testament

and Biblical Authority,” Southwestern Journal of Theology 34 (1992): 19-20. 21 “Each reading must be judged on its intrinsic merits and not by weighing manuscripts, whatever we may mean by weighing”; G. D. Kilpatrick, “Literary Fashion and the Transmission of Texts in the Graeco-Roman World,” chap. in The Principles and Practice of New Testament Textual Criticism: Collected Essays of G. D. Kilpatrick, ed. J. K. Elliott (Leuven: University Press, 1990), 63. Cf. also other chapters in the same volume: “Atticism and the Text of the Greek New Testament” and “Eclecticism and Atticism.”

15

consideration the external factors mentioned above. When external and internal evidence conflict, the scholar may choose to follow the lead of either the one or the other, depending on the text and the variants in question (though it is often the external evidence that predominates). This approach is that used by the Alands, Metzger, and many others.22 It is a compromise between the exclusive use of internal evidence in rigorous eclecticism and the exclusive use of external evidence in genealogical methods. Whereas most NT textual critics use a form of rational eclecticism, often stressing external evidence over internal evidence, many OT textual critics tend to put more stress on the internal evidence. Various reasons for this preference for internal criteria may be offered. First, OT witnesses have not been divided into distinct text types that are recognized by all scholars in the field. It would be particularly useful if all the mss of LXX were grouped by text type.23 Second, the age and textual history of some witnesses, particularly P, are disputed, so scholars’ opinions of the value of the witnesses vary. Third, since they have not been reared in the school of Westcott and Hort, OT textual critics have less tradition pulling them toward an extensive use of external criteria. One group of OT textual critics that puts more emphasis on external evidence than others is that heavily influenced by Cross’s local text theory.24 Cross’s theory has influenced many other OT textual critics to a lesser extent, so that they acknowledge some of Cross’s arguments, but are not convinced of all the details.25 22 Rational eclecticism is based largely on the work of Westcott and Hort (Brooke Foss Westcott and Fenton John Anthony Hort, The New Testament in the Original Greek [New York: Harper & Bros., 1882]) and the local text theory of B. H. Streeter (B. H. Streeter, The Four Gospels: A Study of Origins [London: Macmillan & Co., 1924]). A summary of this approach may be found, with some variations, in Metzger, Text of NT, 209-11, and in Aland and Aland, Text of NT, 275-77. Note particularly the Alands’ method of relating external and internal evidence (excerpted from a list of twelve basic rules of textual criticism) (Aland and Aland, Text of NT, 275):

2. Only the reading which best satisfies the requirements of both external and internal criteria can be original. 3. Criticism of the text must always begin from the evidence of the manuscript tradition and only afterward turn to a consideration of internal criteria. 4. Internal criteria . . . can never be the sole basis for a critical decision, especially when it stands in opposition to the external evidence. 23 Cf. the classification of Samuel LXX mss offered by Bo Johnson, Die hexaplarische Rezension des 1. Samuelbuches der Septuaginta, Studia Theologica Lundensia, no. 22 (Lund: CWK Gleerup, 1963), 19. 24 Frank Moore Cross, “Theory of Local Texts.” Cross’s theory is itself an amplification of

previous work, particularly that of Dominique Barthélemy, “Redécouverte d’un chaînon manquant de l’histoire de la Septante,” Revue biblique 60 (1953): 18-29; idem, Les devanciers d’Aquila, Supplements to Vetus Testamentum, no. 10 (Leiden: E. J. Brill, 1963). See also R. W. Klein, Textual Criticism, 69-73; Ulrich, Qumran Text of Samuel, 4-9. 25 See, for example, Talmon, “Textual Study of the Bible,” 323: “[Cross’s] theory, which is yet in the process of being revised and further refined by Cross and his students, appears to have attracted surprisingly little comment either from European scholars, with few exceptions, or for that matter, Israeli students of the Bible text.” Cf. also Emanuel Tov, “A Modern Textual Outlook Based on the Qumran Scrolls,” Hebrew Union College Annual 53 (1982): 11-27, who argues for a scenario more complicated

16

One other area in which the practice of NT textual criticism might shed some light on OT methodology is the statistical analysis of the data. Many NT scholars advocate the use of statistical analysis to delineate text types and smaller groups of related texts within text types. One of the oldest approaches is called the Claremont Profile Method, a method which uses selected passages within a given corpus to test similarity of readings among mss. The goal of this approach is to generate a group profile, that is, a list of readings characteristic to one particular group of mss (shared by at least two-thirds of them). An alternative, though similar, method appears in the work of Larry Hurtado on the so-called Caesarean text. Yet another variation of this theme is found in the dissertation of Mark Dunn.26 The importance for OT textual criticism of these attempts to classify the witnesses should be apparent. Proposed text types, families, and groups of witnesses to the OT text are based more on random observation than on rigorous statistical analysis of the data. Such methods as these could easily be applied to witnesses in a single language (e.g., to Greek witnesses or Syriac witnesses). No method has yet been developed to apply these methods across the language barrier, but such a method is needed. The advantage of statistical studies like these is that OT textual critics would be able to use external evidence with a greater degree of certainty concerning the relationship of witnesses and their relative value in determining earlier stages of the text.27 In conclusion, it is crucial for the textual critic to engage in textual analysis prior to literary or theological analysis. This analysis should ideally include a clear understanding of both the textual history of each witness (external evidence) and the translation technique of each versional witness (internal evidence). After textual analysis has been done, further internal evidence may be gathered from literary and theological analyses of the sources. Genealogical Method Before describing the proposed model for analysis that will be used in this study, a brief discussion of another method is in order. Textual criticism of classical texts arose during the Renaissance. Since relatively few mss of most classical works are extant, it is usually possible to construct a stemma (or tree), graphically showing the relationships of each extant ms to all the others and to the archetype. Reconstruction of the archetype than Cross’s local text theory. 26 Paul Robert McReynolds, “The Claremont Profile Method and the Grouping of Byzantine New

Testament Manuscripts,” Ph.D. diss., Claremont Graduate School, 1968, 3-8; Larry W. Hurtado, TextCritical Methodology and the Pre-Caesarean Text: Codex W in the Gospel of Mark, Studies and Documents, no. 43 (Grand Rapids: William B. Eerdmans, 1981), 10-12; Mark R. Dunn, “An Examination of the Textual Character of Codex Ephraemi Rescriptus (C, 04) in the Four Gospels,” Ph.D. diss., Southwestern Baptist Theological Seminary, 1990. 27 For a more detailed comparison of OT and NT textual criticism, as commonly practiced, see James R. Adair, “Old and New in Textual Criticism: Similarities, Differences, and Prospects for Cooperation,” TC: A Journal of Biblical Textual Criticism 1 (1996).

17

requires three stages: constructing the stemma (recensio), examining the readings presumed for the archetype to see if they may all be considered original (examinatio), and attempting to reconstruct original readings where they no longer exist (divinatio). 28 The stemma is constructed by comparing extant witnesses to one another and isolating shared readings that diverge from readings of the other mss. Trivial divergences that could have been corrected easily by an intelligent scribe (e.g., misspellings or nonsensical errors in word division) are ignored. Once extant mss are grouped into the smallest possible groups, common ancestors, or hyparchetypes, are constructed by comparing the readings of all the independent witnesses to that hyparchetype. If only two mss are descended from a hyparchetype, the reading of the hyparchetype for each unit of variation must be chosen on some grounds (usually internal evidence), or, if neither seems original, a conjecture may be suggested. If three or more mss are descended from a hyparchetype, the reading shared by more than one independent ms, or by one ms and the hyparchetype in the adjacent branch of the tree, is the reading of the hyparchetype (this rule is based on the assumption that, other than trivial or purely mechanical errors, two scribes would not make the same error in any particular place). The process is continued up the tree until the archetype itself is reconstructed.29 It is clear that this method makes several assumptions, not all of which may be valid for the textual criticism of the OT. One assumption that should at least be questioned is that two independent copyists would not make the same non-trivial or non-mechanical error. This assumption does sound intuitively likely, but the possibility of two scribes making the same error cannot absolutely be ruled out. Moreover, the identification of trivial or purely mechanical errors is not as straightforward as it might seem. Is the presence or absence of an article or a conjunction trivial? Is a change in tense or number trivial? Also, some purely mechanical errors may in fact be of a sort that two scribes would not make independently. A more troubling assumption is the idea that a ms was copied from only one other ms. The fact of the matter is that scribes copying OT mss frequently used more than one exemplar when creating a new ms. At other times, scribes would correct a ms, either their own or an older one, from a ms different from the original exemplar. But using more than one exemplar for a ms destroys the simple tree concept and seriously complicates the methodology for determining the relationships among the mss. Clearly the simple genealogical method that works so well with a few classical mss needs revision in order to accommodate the much more complicated textual history of OT mss. Probably the most thorough attempt to account for the possibility of contamination in the simple tree is the work of Vinton A. Dearing. Dearing’s method is rigorous, 28 Maas, Textual Criticism, 1-19. 29 Ibid., 2-9. See especially the diagram on p. 5.

18

allowing little room for subjective selection among variants.30 His method has seven steps: (1) decide on the states31 to analyze; (2) decide on variations to analyze; (3) rewrite any variations that would introduce rings into the tree; (4) find any terminal groups not in the simple variations; (5) connect the terminal groups; (6) locate the archetype; (7) emend the archetype, if necessary.32 This fascinating and thorough approach is quite precise in its methodology, covering every possible textual variation and leaving little room for human variation in its application. Nevertheless, it is open to criticism, both on the philosophical level and on the level of applicability to the text of the OT. M. P. Weitzman begins his critique of Dearing’s book, Principles and Practice of Textual Analysis, with a quotation from A. E. Housman, a textual critic of the early twentieth century: “Textual criticism is not a branch of mathematics, nor indeed an exact science at all.”33 Immediately, then, a philosophical difference emerges: is it even possible to use methods such as Dearing proposes to do textual criticism? In fact, Dearing’s method is not nearly as exact as one might be led to suppose from some of his statements. For example, he lists at least twenty-five rules for recognizing what he calls “directional variants,” variants that indicate which state explains the origin of the others. These rules, however, are of varying degrees of certainty, as he himself admits, with some (“harsher reading,” “reading that smells of a gloss”) quite subjective.34 The fact that the genealogical method works well with works extant in much smaller numbers than OT mss indicates that it could very well be helpful in OT textual criticism as well. Thus, to deny the genealogical approach a hearing on the grounds that textual criticism is an art rather than a science is not justified. A second critique of Weitzman’s, namely, that Dearing’s method is too complex, even though he provides a sample computer program for analyzing the data, may have had some validity in 1977. However, advances in technology greatly reduce the force of this argument. With the help of an algorithm, which Dearing provides, one can easily design a 30 Dearing, Principles and Practice, 83: “Textual analysis, having absolute rules, is not an art.”

Cf. also p. 58, “Textual analysis is a logic engine, like a computer.” He also gives three advantages of the genealogical method over the eclectic method: (1) in the genealogical method, a fact about the text (i.e., the genealogy) is determined; (2) in most cases, the preferable variant is determined without resort to eclectic methods; (3) it is possible to reconstruct other states of the text besides the archetype (ibid., 8). He goes so far as to say that the textual analyst describes not what was (the actual transmission of mss) but what is (the relationship among the extant states of the text), so he cannot be wrong (ibid., 19). 31 For a definition of a state of the text, see above, p. 4, n. 16. 32 Dearing, Principles and Practice, 21, where these steps are listed. He elaborates on each of them

in great detail on pp. 21-83. 33 A. E. Housman, “The Application of Thought to Textual Criticism,” chap. in Selected Prose, ed. J. Carter (Cambridge: Oxford University Press, 1961), 132; quoted in M. P. Weitzman, review of Principles and Practice of Textual Analysis, by Vinton A. Dearing, in Vetus Testamentum 27 (1977), 225. 34 Dearing, Principles and Practice, 44-54. Of course, the seventh step, emending the archetype, is

highly subjective, but it lies outside of textual analysis proper (as he defines it), steps three through six.

19

computer program to do the work of comparing the states and generating a tree (steps three through six). The most time consuming part of the process, once the program itself is written, is data input; but considering that Dearing and Weitzman were considering using punch cards (before the advent of the affordable personal computer), today’s data input is much faster. Still faster methods of data input may be achieved through the use of optical scanners, touch-sensitive screens or mouse technology, and improved data entry algorithms. However, data entry will probably remain the most difficult part of the process for the foreseeable future. A more telling criticism, indeed, the argument that alone threatens the entire validity of Dearing’s method, is the observation that the genealogical method deals with the transmission of the text as though each state were copied from one and only one exemplar, without contamination from others. Weitzman says, If one were dealing with a textual tradition of which each ms (except, of course, the original) had been copied from a single exemplar, without the scribe introducing any reading from a subsidiary source (e.g. by consulting a second ms, or by adopting variants which may have been recorded in the margins of his exemplar), these tasks would be simple enough. . . . But in a textual tradition wherein subsidiary sources were utilised to a significant extent, albeit by no more than a minority of the copyists involved in the whole process of transmission, all neat and exact methods have been found, at least until now, to break down.35 Dearing, of course, is aware that contamination can occur in the process of transmission, and his third step, rewriting the variations that would introduce rings, is designed to deal with this problem. Rings occur when two mss share readings not found in every intervening ms in the tree. Dearing’s solution to this problem is simply to discard some of the conflicting readings by treating them as through they did not conflict. After analyzing the relationships among the variants and the mss, he breaks the ring at the point of its weakest link, where there are the fewest agreements between different states. In fact, he breaks all the links if more than one are equally the weakest. He justifies this action by saying, “Breaking the weakest instead of some other standard connection reflects the principle of parsimony. The fewest possible readings are treated as something different from what they really are.”36 He further differentiates the work of the textual critic from that of the bibliographer. Because the textual critic analyzes extant states of the text 35 Weitzman, review, 226-27 (italics original). Weitzman downplays the amount of contamination in OT mss, but in fact, contamination from other branches of the genealogical tree is quite common, especially among LXX mss. In particular, hexaplaric readings, introduced into Greek mss from Hebrew mss, infect a large number of LXX mss, even those whose scribes used only one (hexaplaric) exemplar. 36 Dearing, Principles and Practice, 88. On pp. 16-19, Dearing attempts to assume the mentality

of a scribe copying a text and using more than one source, claiming that, from the scribe’s point of view, the result is not conflation. He supports this conclusion by distinguishing between constructing a genealogy of the states of a text (the work of a textual analyst) and constructing a genealogy of the records themselves (the work of a bibliographer). He says, “In short, from the point of view of one who deals with states of a message instead of with its records, and the textual analyst is such a one, conflation does not exist at all” (p. 17).

20

(results) and the bibliographer with the presumed transmission of the text (causes), the textual critic defines the weakest agreement between states in terms of number of agreements, whereas the bibliographer defines the weakest agreement as that having the least “weight,” or intrinsic probability.37 His arguments, however, fail to convince that his method yields accurate results, either in archetypical readings or in accurate trees. Weitzman constructs four hypothetical mss descended from a lost archetype, introducing a degree of contamination. After applying all of Dearing’s rules to the hypothetical mss, he arrives at the correct solution in 56% of the cases and at the wrong solution in 44% of the cases.38 These results challenge Dearing’s contention that textual analysis can be done mechanically, with little or no critical judgment. In Dearing’s response to Weitzman, he points out that, given the data, the tree predicted by his method explains the data just as well as the tree presupposed by Weitzman. Less convincing, however, is his contention that the failure of his method to predict the correct archetypical readings in 44% of the cases is no reflection on the usefulness of the method. He says, With real problems, one either knows the facts in some way, in which case no method of reconstructing them is required, or one doesn’t, in which case the best one can do is reason correctly from all the evidence, on the basis of axioms that seem (and one hopes will also seem to one’s fellow-critics) to be satisfactory.39 However, contrary to Dearing’s opinion, the results of a method indicate whether the theory on which it is based is true, not vice versa. Of course, if the method yields results that are better than other methods, even though not perfect, it may still be preferred. The application of traditional methods to the hypothetical data, however, belies Dearing’s claim that his method is superior. Without attempting to construct a genealogical tree, a textual critic using traditional methods of subjective selection based on internal evidence (i.e., eclecticism) would select 25 correct readings (the “x errors”), where the critic could determine on the basis of comparing the readings to one another that one was original. Of the other 25 errors, assuming a 50% chance of picking the right reading from two equally possible readings, the critic would select 12 correct readings, yielding a total of 37 correct readings out of 50, or 74%, a result considerably better than Dearing’s 56%. Consequently, Dearing’s method cannot be expected to be helpful in evaluating the text of the OT. Unless or until an improvement is made in the genealogical method that will deal more effectively with the problem of contamination, it must be set aside. 37 Ibid., 89. 38 Weitzman, review, 231. An alternate interpretation of Dearing’s rules yields the following results: correct 44%, incorrect 36%, indeterminate 20%. The errors are of two types: “x errors” that are clearly not original (and can be determined not to be so on the basis of internal evidence) and “+ errors” that cannot easily be distinguished from the correct readings. 39 Vinton Adams Dearing, “Textual Analysis: A Consideration of Some Questions Raised by

M. P. Weitzman,” Vetus Testamentum 29 (1979): 358-59.

21 Determining the Translation Technique Having examined various proposed methodologies for OT textual criticism, the time has come for developing the methodology to be used in evaluating the text of 1 Samuel 3. To summarize the previous discussion, both textual analysis and literary/theological analysis are important for determining earlier forms of the text, but textual analysis should have priority. Since no genealogical approach has yet been developed that is helpful in the textual criticism of the OT, it can have no place in the present methodology. Also, no particular theory of the relationship among the various extant witnesses (e.g., Cross’s local text theory) will be assumed. Tov has repeatedly insisted that the first step of textual criticism, at least as far as the versions are concerned, is the determination of the translation technique,40 and the methodology to be tested in this study follows Tov in this regard. Scholars often characterize a particular translation with terms such as “literal,” “fairly literal,” “free,” and so forth. While these labels may characterize certain aspects of the translation, they are not sufficiently nuanced to be of much help to the textual critic who wants to know how the translator rendered certain types of constructions in order to use that information to determine whether a particular reading is likely to reflect a different Vorlage. After all, it is entirely possible that a translator could render every Hebrew word with a single word in the target language, yet be relatively indifferent about word order. Would such a translation be literal or not? The answer is that it would be literal in regard to lexical choice but not in regard to word order. The need to push beyond such general terms as “literal” and “free” is apparent.41 In order to quantify these terms and to provide information that is more useful to the textual critic, a methodology consisting of two major phases is proposed. The first major phase of the methodology involves an analysis of the lexical and grammatical characteristics of each witness in order to understand better the translation technique. Prior to analyzing these characteristics, however, each target language must be examined to determine its limitations for rendering Hebrew.42 If these limitations are not 40 Tov, “Response,” 92-93; idem, Text-Critical Use, 50-52; idem, “The Composition of 1 Samuel 16-18 in the Light of the Septuagint Version,” in Empirical Models for Biblical Criticism, ed. Jeffrey H. Tigay (Philadelphia: University of Pennsylvania Press, 1985), 100-102. 41 Cf. Benjamin G. Wright, “The Quantitative Representation of Elements: Evaluating

‘Literalism’ in the LXX,” in VI Congress of the International Organization for Septuagint and Cognate Studies, ed. Claude E. Cox (Atlanta: Scholars Press, 1987), 312-13. 42 Cf. J. W. Wevers, “The Use of Versions for Text Criticism: The Septuagint,” in La Septuaginta en la investigación contemporánea (V congreso de la IOSCS), ed. Natalio Fernández Marcos, Textos y estudios “Cardinal Cisneros” de la Biblia Políglota Matritense, no. 34 (Madrid: Instituto “Arias Montano” Consejo Superior de Investigaciones Científicas, 1985), 15: “Before a version can be used textcritically the nature and limitations of the target language in contrast to those of the source language must be thoroughly understood”; see also p. 20: “Before a translation can be properly used in the text criticism of the text of a source language, one must fully understand just how and from what points of view this translation was done by a particular translator.” Models for such an analysis of the various languages may be found in the various articles found in Fernández Marcos, La Septuaginta, and in Bruce M. Metzger,

22

recognized, some differences in representation may wrongly be thought to be deviations from a literal rendering. Grammatical impossibilities (e.g., a Latin article), grammatical improbabilities (e.g., rendering the Hebrew in Greek), and stylistic characteristics of the target language (e.g., the Syriac anticipatory pronoun) must all be noted. Any rendering in the target language that meets standards for literalness that the translator may have thought to be imposed by the language itself (or at least by proper use of the language) should not be counted against the literalness of the translation, though all such differences should be noted. Tov suggests four criteria for determining the relative literalness of a particular witness’s translation technique: consistency, the representation of the constituents of compound words in the source language by individual equivalents in the target language, word order, and quantitative representation.43 Each of these criteria will be examined in turn, and certain modifications or amplifications will be suggested. The result will be a somewhat more detailed and nuanced approach to determining the translation technique. By consistency, Tov refers to the translators’ tendency to render every occurrence of a given word in the source language by the same word in the target language. This tendency toward consistent representation of words is called stereotyping, and Tov notes that stereotyping was probably the rule rather than the exception from the beginning. However, he also says, “It has yet to be examined which types of words and elements were rendered stereotypically, in which books and under what circumstances.”44 It is one of the goals of this study to undertake just that examination for 1 Samuel 3. One type of consistent representation commonly found in the secondary witnesses in question is Hebraism. Hebraisms are renderings in the target language that are unusual or even unidiomatic, but occur because the translator has attempted to render his Vorlage literally. Because they preserve some structure of the source language not usually found in the target language, Hebraisms can be useful tools for reconstructing the Vorlage. However, it was also possible for the translators to have written Hebraisms that did not reflect a particular text before them, simply because they thought or spoke the target language in terms of structures from the source language.45 Another type of consistent representation that The Early Versions of the New Testament: Their Origin, Transmission and Limitations (Oxford: Clarendon Press, 1977). 43 Tov, Text-Critical Use, 54-60. He adds that a fifth criterion, linguistic adequacy of lexical choices, cannot be described statistically and is highly subjective, so it cannot be used profitably in the analysis of translation units. 44 Ibid., 55. 45 It is common for a people who learn a new language to fuse it with aspects of their own language. See the discussion of pidgin and creole languages in Encyclopædia Britannica, 1985 ed., Macropædia, s.v. “Languages of the World,” 803-5. Cf. also the comment by Conybeare and Stock, that the Greek of the OT (and NT) “is so deeply affected by Semitic influence as often to be hardly Greek at all,

23

sometimes occurs in the versions of the OT is etymologizing, the tendency to render words derived from one Hebrew root (presumably) by words derived from a single root in the target language. All of the examples Tov gives of consistent representation are lexical in nature. However, translators often rendered certain grammatical structures in the source language (e.g., stem, inflection, and person in Hebrew) consistently by particular grammatical structures in the target language (e.g., voice, tense, and person in Greek). Another aspect of consistency that should be measured is whether the target language regularly rendered a certain class of words in one language by a particular class in the other, for example, participles by participles. An examination of the grammatical and syntactical correspondences between MT and the secondary witnesses is another aspect of consistency that should be investigated. In the proposed methodology, the lexical consistency of (1) verbs, nouns, and adjectives; (2) adverbs, prepositions, and particles; and (3) conjunctions will be tallied separately in a variety of ways.46 First, the number of different Hebrew words occurring more than once will be compared with the number of different words in the target language occurring more than once. Second, the total number of Hebrew words occurring more than once will be compared with the sum of the most frequently used words that correspond to each of these Hebrew words. Third, the factor measuring deviation from complete consistency will be figured.47 The final aspect of lexical consistency to be examined will be a measure of the level of etymologizing apparent in the translation. The first element of grammatical consistency to be measured is the percentage of one class of Hebrew words (verbs, nouns and adjectives, pronouns) represented by the same class of words in the target language. Next, various aspects of verbs, nouns and adjectives, and pronouns will be evaluated separately and compared to their corresponding words in the target language. For verbs, the Hebrew inflection, stem, person, and number will be compared with the corresponding grammatical structures in the target language to but rather Hebrew in disguise”; F. C. Conybeare and St. George Stock, Grammar of Septuagint Greek (Boston: Ginn & Co., 1905; repr., Peabody, MA: Hendrickson, 1988), 21. This statement is an exaggeration, but it does show that the Jews often spoke a form of Greek influenced by their Semitic linguistic background. 46 Proper nouns will be excluded from the evaluation of lexical consistency and from the grammatical analyses of nouns, with two exceptions. First, the proper name , since it is equivalent in most contexts to and / , which are common nouns, will be included in the evaluation of lexical consistency. Second, all proper nouns will be included in the analysis of the use (i.e., grammatical function) of Hebrew nouns and adjectives in the sentence. 47 The algorithm for arriving at the deviation factor will be explained more fully in Chapter 3, and

in detail in Appendix 4. The deviation factor itself is a numerical representation of the deviation from absolute consistency, i.e., one-to-one correspondence, with a deviation factor of zero indicating absolute consistency. The deviation factor is a number that is useful primarily in comparison with the deviation factors of other elements in the same version or of similar elements in another version.

24

determine consistency of grammatical representation. For nouns and adjectives, Hebrew gender, number, syntactical use in the sentence, articles, and definiteness will be compared. For pronouns, the Hebrew gender, number, and syntactical use in the sentences will be compared. The level of consistency in each category will be indicated by the deviation factor. Tov’s second criterion for determining the translation technique is the representation of the constituents of words in the source language by individual equivalents in the target language, also called segmentation. Because translators were interested in being faithful to the text before them, they were often meticulous in rendering each element of a compound Hebrew word (including such things as articles, prepositions, and suffixes) with a corresponding element in the target language, even in cases where such a rendering was not idiomatic. Consistency in the segmentation will be measured by the ratio of the number of Hebrew compound words rendered exactly to the total number of compounds.48 The third aspect of translation technique which Tov discusses is word order. Unlike Hebrew, which relies heavily on the position of words within a sentence to determine their function, inflected languages (e.g., Greek and Latin) are capable of great freedom in word placement without altering the basic meaning of a sentence. Whether the translators took advantage of the abilities of their languages or whether they limited themselves to the order present in their Hebrew Vorlagen is an important measure of translation technique. Of course, variations in word order forced by the grammar of the target language (e.g., postpositive conjunctions) do not count against consistency of word order. The total number of Hebrew semantic units that are represented in MT will be compared with the number of unforced deviations from the Hebrew word order in the translation. The fourth criterion for determining translation technique, quantitative representation, refers to the tendency of some translators to render every element of their Vorlage without omitting superfluous or redundant elements and without adding clarifying words. The target language might put certain restrictions on the translator, so that a one-toone correspondence is impossible; however, grammatically imposed strictures (e.g., absence of an article in Latin) will not be counted against consistency. To measure consistency in quantitative representation, it is not sufficient to compare the total number of semantic units in MT and in the version, since one addition and one omission in the version would yield a number identical to that of MT. Instead, the total number of differences 48 Wright notes that the representation of the constituents of a word by elements in the target language “is actually a sub-category of quantitative representation [the fourth criterion, below] because it involves a type of one-to-one representation albeit a very specific type”; Wright, “Quantitative Representation,” 316. Cf. also Barr’s discussion of segmentation and his examples of extremes from the version of Aquila; James Barr, The Typology of Literalism in Ancient Biblical Translations, Mitteilungen des Septuaginta-Unternehmens, no. 15 (Göttingen: Vandenhoeck & Ruprecht, 1979), 26-29.

25

between MT and the version will be calculated and then compared with the total number of Hebrew semantic units to produce a ratio. One other note about the grammatical analysis is necessary at this point. It will become apparent when comparing MT with each version that certain add-oms occur.49 The question of whether to incorporate add-oms in the statistics is important, for a considerable number of add-oms could significantly alter the picture of the translation technique. It is clear that add-oms should play a part in determining the translation technique. On the other hand, it is difficult to compare something to nothing, and calculations of consistency that tried to take add-oms into account might well produce less than meaningful results. Therefore, add-oms will be omitted from the computation of statistics in the first three areas (consistency, segmentation, and word order). However, they will be counted in the measure of quantitative representation. Once the initial statistics for each of the categories in a particular version being investigated have been figured, the process of refining those statistics begins. The first statistics will give only a rough idea of the translation technique, because certain trends in the translation not apparent at the beginning of the investigation will become evident. Each new discovery will lead to a modification of the previously calculated partial translation technique, until finally all the grammatical data will be accounted for. Then, using the partial translation technique as determined to that point, variants in the version that do not reflect a different Vorlage, either because they are nonreproducible in the target language or because they are theoretically reproducible but are not consistently rendered, will be eliminated.50 All other variants that are not purely orthographic will be considered significant, because they have some probability of reflecting a Vorlage different from MT. The second major phase in the proposed model involves determining the literary 49 For a definition of add-om, see above, p. 7. 50 The variants which are eliminated correspond for the most part to the non-variants that Tov discusses in Text-Critical Use, 217-28. Tov’s pseudo-variants, discussed on pp. 228-40, require special comment. He defines pseudo-variants as readings that “presumably were not found in the translator’s Vorlage, but existed only in his mind” (p. 228, italics his). The problem with this definition (as he notes, pp. 228-29; cf. pp. 140-41) is that distinguishing real variants from pseudo-variants is usually subjective. After all, if the translator could mistake one word for another, so could a Hebrew scribe, and there is no way of convincingly demonstrating at what point the mistake occurred, whether in the transmission of the Hebrew or of the versional text. Cf., e.g., P. Kyle McCarter, Jr., Textual Criticism: Recovering the Text of the Hebrew Bible, Guides to Biblical Scholarship, Old Testament Series (Philadelphia: Fortress Press, 1986), 26-57, passim. An exception to this contention is the case of demonstrably inner-version (e.g., inner-Greek) variants, which, of course, could only occur in the transmission of the versional text (cf. Tov, Text-Critical Use, 237-40). It is undoubtedly true that the translators sometimes read their Hebrew Vorlagen incorrectly, thus creating variants that never existed in a Hebrew ms. However, it seems methodologically preferable, except in the case of inner-version corruptions, to treat what might be pseudovariants in exactly the same way as other variants. The decision to eliminate them on the grounds that they probably never occurred in Hebrew is best left to the final stage of the process of textual criticism of the Hebrew text, namely, choosing the best reading from the various Hebrew and retroverted readings (see below, chap. 7).

26

and theological characteristics of the witness. This process will require an examination of the larger context to determine theological tendencies and other changes probably related to literary, cultural, or historical circumstances. If a witness is determined to have a specific tendency (e.g., a tendency to harmonize parallel passages), variants that exhibit this tendency cannot be considered likely indicators of a different Vorlage. An examination of various literary and theological studies of Samuel will be particularly helpful in this analysis. Numerous recent commentaries and articles deal with the larger literary shape of 1 Samuel in a particular tradition (especially MT and LXX), and others, though not dealing specifically with 1 Samuel, use a literary approach that might fruitfully be applied to the text. Literary approaches to be investigated include close reading, rhetorical criticism, structuralism, narrative analysis, deconstruction, and analysis of comparative structures. Literary analyses that borrow from several different approaches will also be consulted. An examination of the theological themes identified in 1 Samuel will be studied for the light they might be able to shed on the state of the text. Finally, any historical or archaeological data that might be relevant to the meaning of the text will be analyzed. The purpose of investigating these different approaches is to determine whether further characteristics of the translators, such as cultural or theological biases, should be indicated in a full description of the translation technique of each of the versions. Once as many characteristics of this nature have been identified as seem probable, the partial translation technique will be adjusted again, and the result will be the final translation technique. On the basis of the new information gathered in the second phase of study, other variants will be eliminated which probably do not reflect a different Vorlage. After the process of eliminating variants has been completed, all the remaining variants may be considered significant. Since Hebrew variants that are not purely orthographic are significant by definition (see above, p. 6), all the variants found in those Masoretic mss that possibly preserve pre-Masoretic readings51 and all the variants found in 51 As identified by Goshen-Gottstein; see above, p. 16, n. 18. Though all such variants will be

listed, Goshen-Gottstein’s warnings concerning the use of variants from medieval mss must be stressed. In his article “Hebrew Biblical Manuscripts,” he makes the following statements: “Among all the MSS and fragments known so far there is not even one the deviations of which can be significantly connected with any non-Massoretic tradition. We possess no medieval manuscript which, on the strength of its readings, may be termed ‘valuable’ or be worthy of our attention more than any other” (p. 277); “For those books of the Bible which have already been checked the results gained for Isaiah are paralleled, and there is no reason to assume that further investigations of MSS and fragments from other books are going to enable us to uphold the notion of a ‘valuable’ medieval MS” (pp. 282-83); “The possibility of infiltration of extraMassoretical variants into the Massoretical ‘central current’ is so negligible that for all practical purposes it may be disregarded” (p. 286). Nevertheless, “the analysis of readings such as in the Book of Kings may at least justify the mentioning of the possibility that different results may be obtained for different books (or parts) of the Bible and that, accordingly, we may have to reckon with different ‘breadths’ of the ‘central current’ and different strengths of the ‘trickle’ from the side” (p. 287). (It is at this point that he lists the four mss that “stand out” from the rest in Samuel.) For example, he finds only seven variants in medieval mss of Isaiah that “might be possibly more than the result of harmonization” or other scribal habits, and only one of these “makes one really prick up one’s ears.” It will be in the light of these cautions that

27 4QSama, the only other primary witness to the text of 1 Samuel 3, may simply be listed. Once all the significant variants have been determined, the process of retroversion and evaluation of variants may begin. Before beginning the analysis of the secondary witnesses, however, the text of the secondary witnesses themselves must be established, and one other preliminary matter requires some discussion, namely, the script of the Vorlagen of the various Samuel translations. The Script of the Vorlagen Although it is often assumed that the Vorlagen of the secondary versions of Samuel were written in the later square script, this assumption needs further discussion and demonstration, particularly in the case of LXX. Rabbinic tradition says that Ezra and the other returnees from Babylonia introduced the square (“Assyrian”) script into Palestine in the sixth and fifth centuries B. C. E. (b. Sanh. 21b-22a; j. Meg. 1.9, etc.), whereupon it replaced the Old Hebrew script then in use. However, evidence from Qumran, coupled with evidence already known (script of the Samaritan Pentateuch, numismatic evidence from the Hasmonean and Herodian periods, revival of the older script during the time of Bar Kokhba), necessitates a new look at the issue. Shemaryahu Talmon concludes on the basis of all the data, especially the existence of several scrolls from Qumran written in the Old Hebrew script, that “the ancient Hebrew script retained a lease on life until after the destruction of the Second Temple and the cessation of Jewish political sovereignty.”52 Qumran data demonstrates that the Old Hebrew script was sometimes used for transmitting biblical material in the late Second Temple period. Most of the mss in Old Hebrew script are from the Pentateuch; however, one Old Hebrew ms of Job does exist (4QpaleoJobc). 53 It is probable that Samuel was composed, and subsequently transmitted, in the Old Hebrew script.54 However, S. R. Driver concludes that the Vorlagen used by the translators of most of the books in LXX (presumably including Samuel) were written in an earlier form of the square script.55 In order to demonstrate that the Vorlage of LXX variants from medieval mss will be listed. 52 Shemaryahu Talmon, “The Ancient Hebrew Alphabet and Biblical Text Criticism,” in Mélanges bibliques et orientaux en l’honneur de M. Mathias Delcor, ed. A. Coquot, S. Légasse, and M. Tardieu, Alter Orient und Altes Testament, no. 215 (Kevelaer: Verlag Butzon & Bercker, 1985; Neukirchen-Vluyn: Neukirchener Verlag, 1985), 391. See his discussion of examples of the alternation of and in the Old Hebrew script, ibid., 393-401, and of the alternation of and , idem, “The Ancient Hebrew Alphabet and Biblical Text Criticism,” in Mélanges Dominique Barthélemy: Études bibliques offertes a l’occasion de son 60e anniversaire, ed. Pierre Cassetti, Othmar Keel, and Adrian Schenker, Orbis Biblicus et Orientalis, no. 38 (Fribourg: Éditions Universitaires, 1981; Göttingen: Vandenhoeck & Ruprecht, 1981), 504-22. 53 Tov also mentions 4QpaleoJoshpara, containing parts of Joshua 21; Emanuel Tov, Textual

Criticism of the Hebrew Bible (Philadelphia: Fortress Press, 1992), 105, n. 79. 54 Cf. ibid., 220. 55 S. R. Driver, Notes on the Books of Samuel, lxiv. Cf. the similar conclusion of Johann

28

Samuel was written in Old Hebrew, it would not be sufficient to point to evidence of confusion that can occur only in Old Hebrew, since this confusion could have occurred during the process of transmission of the Hebrew text that led to the Vorlage. Instead, it would be necessary to show that errors resulting from graphic similarity which only exist in the square script (e.g., and , and , and ) do not appear with any great frequency in the translation. (Furthermore, the confusion of and , that could only have occurred in square script, must be ignored if matres lectionis are involved.) However, such graphic confusion based on the square script does seem to play a major role in Samuel (cf. 3:2, 6, 11), so the Vorlage used by the translators of LXX (as well as those used by the other translators) does appear to have been written in square script. The ingenious, but discredited, transcription theory of Franz Wutz may be mentioned briefly at this point. In several studies produced early in the current century, Wutz propounded the theory that the translators of LXX worked from texts in which the Hebrew had been transcribed in Greek characters, checking the actual Hebrew text only occasionally.56 Though he won several influential early supporters, most notably Rudolf Kittel and Paul Kahle, support quickly faded, and Wutz himself seems to have abandoned the view in later years. Johann Fischer, who supported the transcription theory at least in part, pointed out one of the flaws in Wutz’s theory, namely, that the frequency of such graphic errors as for , which do not resemble one another in Greek, tend to contradict any reliance on mss written with Greek characters.57 One last matter of a methodological nature needs to be mentioned in connection with work on the script of the Vorlage of the translations, a matter that applies equally to attempts to trace the early history of the Hebrew text itself. Many studies, including those of Wutz and Johann Fischer, have based much of their analysis on the evidence provided by the transmission of proper names, on the grounds that since the names are transcribed rather than translated, they should provide a stable interface between Hebrew and the target Fischer for the Pentateuch: “Wie ersichtlich, steht das A[lphabet] der LXX-V[orlage] zwischen neuaramäischer Schrift und Quadratschrift und man wird es als neuaramäisches Alphabet mit starker Neigung zur Quadratschrift bezeichnen dürfen”; Johann Fischer, Das Alphabet der LXX-Vorlage im Pentateuch: Eine textkritische Studie, Alttestamentliche Abhandlungen, vol. 10, no. 2 (Münster: Aschendorffschen Verlagsbuchhandlung, 1924), 93. However, cf. Driver’s comment, “The confusion of and , and and (which cannot be explained from the old character) is in the Pent. so uncommon that it may be due to accidental causes: the books in which it is frequent can only have been translated after the change of character had been effected; the Pent., as tradition states, may have been translated earlier”; S. R. Driver, Notes on the Books of Samuel, lxiv, n. 3. 56 Franz Wutz, Die Transkriptionen von der Septuaginta bis zu Hieronymus, pt. 1, Beiträge zur

Wissenschaft vom Alten Testament, n.s., no. 9, Texte und Untersuchungen zur vormasoretischen Grammatik des Hebräischen, no. 2 (Berlin: W. Kohlhammer, 1925), 5: “Da wir weiterhin beweisen können, dass nur in ganz wenigen Fällen ein hebräischer Konsonantentext eingesehen wurde, so sind selbst die hebräischen Verlesungen nicht direkt mit der Septuaginta in Zusammenhang zu bringen, sondern mit ihrer Transkriptionsvorlage.” 57 Johann Fischer, Alphabet der LXX-Vorlage, 95.

29

language. However, the evidence suggests that the opposite is rather the case. Names, particularly lists of names like those found in Joshua, Chronicles, Ezra, and Nehemiah, are subject to deviation on a grand scale. The problem in using proper names to determine script of the Vorlage is that, unless they refer to well-known individuals, they are meaningless collections of letters, subject to uncontrolled variation, unlike other words, whose variation is controlled to some extent by the context of the sentence. Thus, common nouns, verbs, and adjectives are better suited for studies on script than are proper names.58 Having determined that the square script apparently lay behind all of the secondary versions of 1 Samuel, the archetypical secondary witnesses may now be reconstructed. This is the task of the second chapter of this study. 58 Cf. S. R. Driver, Notes on the Books of Samuel, lxiv, n. 3.

CHAPTER 2 RECONSTRUCTING ARCHETYPICAL SECONDARY WITNESSES Method for Reconstruction Ideally, the steps of the model just described would be applied to each particular versional tradition to determine as nearly as possible the original text. However, that task is beyond the scope of the present study, so traditional methods alone will be employed to determine a standard text for each tradition. This step is particularly important for those versions that do not have critical texts (e.g., the Cambridge edition of LXX, which prints a diplomatic text), but the variants in those traditions with critical texts will also be checked. First, though, a brief textual history of the four secondary witnesses will be given in order to ascertain the relationships of these witnesses to one another and to MT.1 LXX has the most complex textual history of the secondary witnesses. It also exhibits both the greatest number of witnesses and the greatest number of variants within the tradition. The term LXX is often used rather loosely. Though it originally referred to a particular translation of the Pentateuch, it was extended to include the Greek translation of the other books in the OT. Herein lies the problem, for not one, but many different Greek versions existed in the ancient world. Though Paul Kahle proposed the theory that various Greek “targums,” in use among different Jewish communities, lay behind the main LXX tradition,2 the theory that won the day was that of Paul de Lagarde, who proposed a single original translation of the Hebrew, from which the different Greek traditions developed.3 1 Johann Cook, “Die pluraliteit van ou-testamentiese tekste en eksegetiese metodologie,” paper

presented at the annual meeting of Die Ou-Testamentiese Werkgemeenskap van Suid-Afrika, 1988, pp. 1-2, stresses the importance of having a grasp of the history of the text before textual criticism (or exegesis) is undertaken: ['n tekshistoriese perspektiewe] kan enersyds die gevolg wees van bepaalde teoretiese uitgangspunte waarvolgens die tekskritiek, die wetenskaplike dissipline war oa gerig is op die ontstaans- en teksgeskiedenis van bepaade tekste, tot bepaalde voorarbeid beperk word. . . . Andersyds, let 'n oordrewe klem op die finale vorm van die teks (die sg. strukturele metode) en/of op die uitweking van die teks op die ontvanger van die boodskap van die teks (resepsie-kritiek) tot 'n onderwaardering van die tekstuele sy van die teks. 2 Paul E. Kahle, Cairo Geniza, 132-79. His first statement of his theory was in idem,

“Untersuchungen zur Geshichte des Pentateuchtextes,” Theologische Studien und Kritiken 88 (1915): 399439. 3 For a discussion of Lagarde’s views, see Sidney Jellicoe, The Septuagint and Modern Study (Oxford: Oxford University Press, 1968; repr., Winona Lake, IN: Eisenbrauns, 1989), 5-9. It is better to

30

31

Lagarde believed that three recensions of LXX existed by the third century C. E.— associated with the names of Origen, Lucian, and Hesychius—and that all extant mss of LXX present mixed texts, preserving readings from all of these recensions. Many modern scholars question the existence of a Hesychian text, though they accept the Origenic (Hexaplaric) and Lucianic texts. The focus of many modern discussions of the history of LXX is the recovery of the Old Greek (OG) text, the original LXX that lies behind all existing mss. Though no single ms is believed to contain a perfect OG text, those that show little evidence of Hexaplaric or Lucianic readings are believed to be fairly good representatives of it. A problem arises, however, in the text of the four books of Kingdoms in LXX (i.e., Samuel and Kings in MT). Henry St. John Thackeray demonstrated that the text of the major uncial mss was not uniform throughout these books, but represented a mixture of two different Greek versions.4 He divided the books into the following sections: (1 Kingdoms), (2 Kgdms 1:1-9:13), (2 Kgdms 10:1-3 Kgdms 2:11), (3 Kgdms 2:12-21:29), and (3 Kgdms 22:1-4 Kgdms 25:30). Sections , , and he identified as authentic OG sections. Sections and represented another version with different characteristics.5 Barthélemy built on Thackeray’s theory with evidence from a ms found at Nah@al H9ever by suggesting that the sections and do not contain a Greek translation independent of OG but rather a revision of OG, called the kaige recension.6 He later modified his view somewhat under the influence of studies by Robert A. Kraft and Sebastian Brock, suggesting that another layer of revision, which is sometimes called proto-Lucian, probably underlay the kaige recension. The kaige text in turn was the basis for the versions of Aquila ( ´), Symmachus ( ´), and Theodotion ( ´).7 As its name indicates, proto-Lucian is a layer of the text that many scholars find underneath the Lucianic speak of original translations of LXX, since different translators were responsible for the various books. 4 Henry St. John Thackeray, The Septuagint and Jewish Worship, 2d ed. (London: Oxford

University Press, 1923). 5 R. W. Klein, Textual Criticism, 25. The use of the LXX title Kingdoms instead of the Hebrew Samuel and Kings explains the names Thackeray gave to the sections; the Greek letters represent their numerical values, e.g., contains parts of 2 and 3 Kingdoms. Thackeray dates the Old Greek sections to the late second century B. C. E. and the kaige sections to sometime after the turn of the era; Henry St. John Thackeray, A Grammar of the Old Testament in Greek, vol. 1, Introduction, Orthography and Accidence (Cambridge: Cambridge University Press, 1909; reprint, Hildesheim: Georg Olms, 1978), ix. 6 Barthélemy, Devanciers d’Aquila, 126-27. The kaige recension is sometimes called proto-

Theodotion or kaige-Theodotion; ibid., 144-57. 7 Idem, “Les problèmes textuels de 2 Sam 11,2 - 1 Rois 2,11 reconsidérés à la lumière de certaines critiques des ‘Devanciers d’Aquila,’” in 1972 Proceedings, ed. Kraft, 28. The reviews of his book by Kraft and Brock are: Robert A. Kraft, review of Les devanciers d’Aquila, by Dominique Barthélemy, in Gnomon 37 (1965): 474-83; Sebastian P. Brock, “Lucian Redivivus: Some Reflections on Barthélemy’s Les devanciers d’Aquila,” Studia Evangelica 5 (1968): 176-81. This view is further developed in the commentaries by McCarter and Klein, as well as in numerous other studies.

32

readings, itself a revision of OG.8 The importance of these finds to the present study is that they show the importance of the Lucianic text-tradition (mss boc2 e2 in the Cambridge edition) as an independent witness to an early form of the Hebrew text. This summary of the history of the Greek text has been necessarily brief, since what is of primary importance here is the relationship of the various Greek witnesses to their Hebrew Vorlagen.9 Aside from OG itself, (proto-)Lucianic readings that are revisions of OG toward a Hebrew text will be important, especially since it has been shown that this Hebrew text had substantial differences from the later MT.10 Similarly, though Origen (LXXO), ´, ´, and ´ all revised their Greek texts toward a Hebrew text very similar to MT, they will be examined to see if their respective Hebrew exemplars varied at all from it. All the readings from Greek witnesses outside the OG tradition will be considered partial secondary witnesses, since they are not independent translations of their Hebrew exemplars. Unlike the textual history of LXX, on which scholars exhibit general agreement, that of P is hotly debated. The date, provenance, and translators of P are all contested issues. These issues, however, are closely related to one another. Scholars who see a Jewish origin for P believe that it was first used in the province of Adiabene, on the border of the Roman Empire, in the middle of the first Christian century. The rulers of Adiabene had been converted to Judaism about 40 C. E. and needed a translation of the Bible. Other scholars posit a Jewish-Christian origin for P, also in Adiabene, but somewhat later, when missionaries first reached the area with the gospel. Still others place the origin of the version in Edessa, the most important city in Syria (outside of Antioch, which was largely Greek). Some Syriac traditions point to a Christian origin for P in Edessa. The discovery of two Old Syriac gospels has also raised the issue of whether P in the OT also had an Old Syriac predecessor.11 The resolution of these issues also involves identifying the type of text from which 8 Natalio Fernández Marcos, “The Lucianic Text in the Book of Kingdoms,” in De Septuaginta:

Studies in Honor of John William Wevers on His Sixty-Fifth Birthday, ed. A. Pietersma and C. Cox (Mississauga, ON: Benben, 1984), 166-71; Sebastian P. Brock, “The Recensions of the Septuagint Version of 1 Samuel,” D.Phil. diss., Oxford University, 1966. Tov expresses doubts that a separate proto-Lucian can be distinguished from Lucian, believing instead that Lucian itself is a direct revision of OG; Emanuel Tov, “Lucian and Proto-Lucian,” Revue biblique 83 (1976): 51-54. 9 For a fuller discussion of the Greek text, see James Donald Shenkel, Chronology and Recensional Development in the Greek Text of Kings, Harvard Semitic Monographs, no. 1 (Cambridge: Harvard University Press, 1968), 5-21. 10 Ulrich, Qumran Text of Samuel, 257-59. 11 IDB, Supplementary Volume [hereafter IDBS], s.v. “Syriac Versions,” by A. Vööbus, 848-49; Bleddyn J. Roberts, The Old Testament Text and Versions (Cardiff: University of Wales, 1951), 217-23; Svenskt Bibliskt Uppslagsverk, s.v. “Bibeln,” by C.-M. Edsman, col. 256; ibid., s.v. “Gamla Testamentet,” by Ivan Engnell, col. 654; J. B. Segal, Edessa: The Blessed City (Oxford: Clarendon Press, 1970), 42-43, 165.

33

P was translated. Several scholars have noted parallels between P and T, both in language (Western Aramaic elements embedded in the Eastern Aramaic version) and content (certain shared deviations from other witnesses), and have concluded that P was originally transcribed into Syriac from a Western Aramaic Palestinian targum. Subsequently, the text was brought more and more into conformity with the proto-MT tradition. However, not all scholars have accepted this reconstruction of P’s textual history. M. D. Koster, in his study of P in Exodus, comes to the conclusion that P was originally translated from a Hebrew text similar to MT; it was only later that targumic additions were made to the text. Both M. J. Mulder and Johann Cook support Koster’s position, explaining that similarities between P and T might come from a shared Jewish exegetical tradition, rather than direct influence of T on P. Another theory, which stands in the middle of these two, is proposed by Alexander Sperber. He believes that a real textual connection between P and Targum Onkelos does exist, but rather than seeing P as a revision of a Palestinian targum stripped of its paraphrases, he says that both P and Onkelos derive from a common ancestor, and the characteristic targumic paraphrases are later additions to the targum tradition.12 The question must be said to be still unresolved, particularly with regard to 1 Samuel, since recent studies have focused on the Pentateuch. One other note about the text of P must be addressed, namely, its relationship with LXX. The text of P has often been denigrated as being of little value to the textual critic. Since readings shared with LXX have been seen as the result of the influence of LXX on P, readings in P have only been considered valuable in conjunction with LXX. However, Cook points out that though LXX did influence P to some extent, its influence has been greatly overestimated in the past.13 Mulder goes even farther, asserting the essential independence of LXX and P.14 In this study, each reading of P will be examined individually in order to determine its relationship with other versions and, more importantly, the likelihood that it is a witness to an independent Hebrew reading in certain places. 12 A. Vööbus, “Der Einfluss des altpalästinischen Targums in der Textgeschichte der Peschitta des Alten Testament,” Le muséon 68 (1955): 215-18; Paul E. Kahle, Cairo Geniza, 272-73; M. D. Koster, The Peshi a of Exodus: The Development of Its Text in the Course of Fifteen Centuries, Studia Semitica Neerlandica, no. 19 (Assen: Van Gorcum, 1977); idem, “Which Came First: The Chicken or the Egg? The Development of the Text of the Peshit@t a of Genesis and Exodus in the Light of Recent Studies,” in The Peshi a: Its Early Text and History. Papers Read at the Peshi a Symposium Held at Leiden 30-31 August 1985, ed. P. B. Dirksen and M. J. Mulder, 147-68, Monographs of the Peshit@t a Institute, Leiden, no. 4 (Leiden: E. J. Brill, 1988), 99-126; M. J. Mulder, “The Use of the Peshit@t a in Textual Criticism,” in La Septuaginta, ed. Fernández Marcos, 52; Johann Cook, “The Composition of the Peshit@t a Version of the Old Testament (Pentateuch),” in The Peshi a: Its Early Text and History. Papers Read at the Peshi a Symposium Held at Leiden 30-31 August 1985, ed. P. B. Dirksen and M. J. Mulder, 147-68, Monographs of the Peshit@t a Institute, Leiden, no. 4 (Leiden: E. J. Brill, 1988), 153-59; Sperber, Bible in Aramaic, 4b:409-17. 13 Johann Cook, “Composition of the Peshit@t a,” 159. 14 Mulder, “Use of the Peshit@t a,” 53.

34

The various targums have long traditions lying behind their commitment to parchment or papyrus. Oral renderings of the OT in Aramaic probably originated in the synagogue at a time when the general populace no longer understood Hebrew.15 The meturgemanîn (translators) were careful to preserve traditional interpretations of the texts, but, as is clear from a comparison of the targums to the Pentateuch (Onkelos, PseudoJonathan, Neofiti, the Fragmentary targums, and the fragments from the Cairo Geniza), conflicting renderings did develop in different places and at different times. The state of affairs with Targum Jonathan is somewhat different from that of the targums to the Pentateuch. It is the only Aramaic targum preserved in the prophets (except for a number of fragments and glosses), though many scholars believe that Palestinian targums of this material once existed as in the Pentateuch. Jonathan, like Onkelos, attained its final form in Babylonia and served as the official targum of the books it contained. The wording of Jonathan was definitively established sometime in the fifth century after a long and complex history. 16 Though many of the traditions underlying Jonathan antedate the choice of the proto-Masoretic tradition as the official text for all Judaism about 100 C. E., it was revised toward that official tradition, so that few variations from MT remain. Because of this revision, though, any variations that do remain will be important if it can be shown that they probably reflect a Hebrew Vorlage different from MT. The textual history of V is probably the least complicated and controversial of all the secondary witnesses, though it is not as simple as it might appear at first glance. Jerome was commissioned by Pope Damasus I to produce a Latin version of the Bible about 382. His translation proceeded in two steps. First, he revised existing Old Latin texts according to the Greek text of Origen. Dissatisfied with this initial effort, he then began to translate the entire OT directly from the Hebrew, and the end result was V. Since the Hebrew text Jerome used as the basis for this second translation was very similar to the present MT, it might seem as though the textual history of V were straightforward and that V would be of little use as an independent witness to the text. However, the situation is somewhat more complicated. In the first place, the scribes who transmitted the text of V were often less than faithful copyists, frequently mixing various it readings into their texts (not to mention numerous scribal errors). As a result, comprehensive revisions of the text began before V had supplanted it as the preferred Latin version. The most important early revisions are associated with the names of Cassiodorus (sixth century), Alcuin (eighth century), and Theodulf (eighth century).17 15 On the basis of fragments of Palestinian targums from the Cairo Geniza, Kahle states that a targum existed in Palestine by the second century B. C. E. at the latest; Paul E. Kahle, Cairo Geniza, 207. 16 Würthwein, Text of OT, 78; B. J. Roberts, OT Text and Versions, 207-9. 17 IDB, s.v., “Versions, Ancient,” by Bruce Manning Metzger, 752-53; B. J. Roberts, OT Text and Versions, 259-60; Fischer, Biblia Sacra Stuttgartensia, 1: xx-xxi.

35

As Metzger notes, “the more than eight thousand MSS of the Vulg. known today exhibit the greatest degree of bewildering cross-contamination of textual type.”18 Thus, before V can be consulted, a relatively pristine form of the text must be produced, either by Quentin’s method (see below, p. 66**) or by some other. Another problem is the identity of the sources of the various readings Jerome used in V. Though he attained a good knowledge of Hebrew, he himself admitted in letters to friends that he frequently consulted with rabbis in the vicinity of Bethlehem (where he was working on his translation). This contact accounts for some similarities between V and T. He also had access to Origen’s Hexapla and to mss of the secondary Greek versions, and he often relied on them to clarify a difficulty. Finally, the various it versions (no official it version ever existed) undoubtedly played a strong role in his decisions.19 Thus, the textual affiliations of V are quite complex, and agreements between V and other ancient versions must be considered carefully before one can conclude that a Hebrew text different from MT lies behind a reading in V.20 Nevertheless, a reading in V that supports that of another witness at least testifies to the authenticity of that reading in the fifth century, and an independent reading is of great value. Further research would be necessary to see where and to what extent each of the versions mentioned influenced Jerome’s text. Now that the history of the text of each of the secondary witnesses has been reviewed, a method for reconstructing their texts can be stated briefly. For the text of those witnesses that do not have an eclectic critical text (i.e., LXX, P, and T), each variant given in the critical apparatus of the diplomatic text will be evaluated to see if it is preferable to that of the basic text. This evaluation will be done primarily on intrinsic grounds, though extrinsic grounds will be a factor in some cases. Other editions of the version will also be consulted where appropriate, especially in the case of LXX. The text of V, which has two eclectic critical texts, will be easier, since the variants have already been evaluated. Nevertheless, all the variants in the critical apparatuses will be checked again, and the two critical editions of the text will be compared with one another, to see if any differences in evaluation have been made. Finally, the partial secondary witnesses will be presented in fragmentary form. All attested readings from ´, ´, and ´ will be presented. In the case of LXXO, only those readings that deviate from both MT and OG will be evaluated. The entire text of LXXL will be reconstructed, but only those readings that vary from OG will be considered in Chapter 3. The full analysis applied to the complete secondary witnesses will not be done for the partial secondary witnesses. Instead, their variants will be evaluated in a more 18 Metzger, “Versions, Ancient”, 753. 19 Ibid.; Würthwein, Text of OT, 92-93. 20 Friedrich Stummer, Einführung in die lateinische Bibel: Ein Handbuch für Vorlesungen und

Selbstunterrich, (Paderborn, Germany: F. Schoning, 1928), 123; cited in Würthwein, Text of OT, 93.

36

traditional way.21 Location of Textual Data The primary source for the textual data of LXX will be the text and apparatus of the Cambridge edition, but Rahlfs’s smaller edition will be considered as well, especially since Rahlfs makes a preliminary attempt to create a more eclectic text. The editions of LXX edited by Holmes and Parsons and by Swete will also be checked.22 Variant readings from the daughter versions of LXX will be drawn primarily from the critical apparatus in the Cambridge edition, though several of them will be checked to determine the accuracy of citation; however, only secondary text-traditions will be analyzed in detail. The Leiden edition of P provides a diplomatic text of P, along with a critical apparatus. This will be the text of P used in the study. Sperber’s edition of T presents the text of one ms as the basic text, with other readings in the apparatus, and it will serve as the text of T for this study. Daniel J. Harrington and Anthony J. Saldarini, in the introduction to their translation of T, note the variation present in the different mss of T, saying that “the individual manuscripts of what we call Targum Jonathan tend almost to constitute separate works.”23 Many of these variations, though, are simply haggadic or halakic expansions, so are of little value in indicating the Hebrew Vorlage. V is the only secondary witness published as an eclectic critical text, so the basic text has already been evaluated by scholars and determined to be as close as possible to the original. Thus, little reconstruction need be done, except to evaluate the variants presented in the apparatus and to compare the critical edition from the Abbey of St. Jerome with the Stuttgart version, which does have some variations in both the basic text and in the apparatus. For the partial secondary witnesses, the readings of ´, ´, and ´ will be culled from both Field’s edition and the apparatus of the Cambridge LXX. The readings of 21 The importance of the partial secondary witnesses for reconstructing the text of the OT may be seen in Driver’s comment on the Lucianic text: “Whether these renderings were derived by him from MSS. of the LXX of which all other traces have disappeared, or whether they were based directly upon Hebrew MSS. which had preserved the genuine reading intact, . . . is a matter of subordinate moment: the fact remains that Lucian’s recension contains elements resting ultimately upon Hebrew sources, which enable us to correct, with absolute certainty, corrupt passages of the Massoretic text”; S. R. Driver, Notes on the Books of Samuel, xlix. The absolute certainty of the reconstructions may be questionable, but the importance of LXXL and other partial secondary witnesses cannot be denied. 22 Robert Holmes and James Parsons, eds.,Vetus Testamentum Graecum cum variis lectionibus, 5

vols. (Oxford: Clarendon Press, 1798-1827); Henry Barclay Swete, ed., The Old Testament in Greek According to the Septuagint, 3 vols (Cambridge: Cambridge University Press, 1887-1912). 23 Daniel J. Harrington and Anthony J. Saldarini, Targum Jonathan of the Former Prophets, The Aramaic Bible, vol. 10, ed. Kevin Cathcart, Michael Maher, and Martin McNamara (Wilmington: Michael Glazier, 1987), 2.

37 LXXL will be taken from Lagarde’s edition of the Lucianic text24 and the Cambridge apparatus. LXXO will also be taken from the Cambridge apparatus, particularly from mss Acx. Presentation of Reconstructed Witnesses The reconstructed secondary witnesses will not be presented in their entirety. Instead, only those readings that have a reasonable claim to being original (to the version) and that differ from the collating bases (Cambridge LXX, Leiden P, Sperber’s T, and VR) will be listed, along with an explanation of the reason for the choice. For the partial secondary witnesses, all the fragments of ´, ´, and ´ will be given, along with their corresponding LXX reading. Since some mss refer to these three witnesses as a group or rather vaguely, a separate section will be devoted to listing all those readings that might belong to the three but are not clearly indicated. The readings of LXXL that deviate from LXX will be given in the next section. Finally, all the variants of LXXO from LXX will be listed. The variants will be presented in a form compatible with most textual apparatuses, but a few notes about sigla are in order. The reading of the base text will be given first, followed by the witnesses that support that reading, followed in turn by a right bracket ‘]’. If more than one such reading occurs in the verse, the one in question will be identified by a numeral followed by a small superscript o: ‘1°’, ‘2°’, and so forth. Next, the variant reading(s) will each be given, along with the witnesses that support the reading. Each of the variants following the base reading will be separated from the previous one by a vertical bar ‘|’. Witnesses will be listed in the following order: mss in the base language, versional witnesses, patristic sources. Mss in the base language will be separated from other witnesses by a semicolon in order to avoid possible confusion. Witnesses that support the variant in most respects (or the most important respects) but differ in small details will be enclosed in parentheses ‘()’. Witnesses that are inferred from the list of extant witnesses given in the apparatus of the base text but that are not explicitly listed will be enclosed in braces ‘{}’. Transposition of words from the base text or transposition combined with omission will be signified by numerals representing the variant word order (e.g., 132 signifies that the second word in the base text is the third word in the variant). Next, the symbols ‘+’ and ‘>’ represent an addition and an omission with respect to the base text, respectively. The abbreviations ‘pr’ and ‘post’ mean that the variant precedes or follows the word or phrase given as the base text, respectively. Finally, ‘idem’ indicates a reading that is identical with the base text. 24 Paul de Lagarde, Librorum Veteris Testamenti Canonicorum, vol. 1 (Göttingen: Aedibus Dieterich, Arnold Hoyer, 1883).

38 Septuagint As a general introduction to the evaluation of LXX readings, the following additional LXX resources should be mentioned: Hatch and Redpath’s concordance, Swete’s introduction, and Conybeare and Stock’s grammar. Bo Johnson’s study of hexaplaric mss is also helpful, for he identifies certain family groups within LXX mss of 1 Samuel, namely, B (sometimes associated with mss ya2 ), Acx (hexaplaric mss), boc2 e2 (Lucianic mss), dlpqtz (referred to as d+ [l is lacking in the chapter]), and fmsw (referred to as f+). The other mss (MVaeghijnvb2 , and often ya2 ) do not fall into any clear group.25 In addition to Greek mss and daughter versions, the church fathers Theodoret (Thdt) and Chrysostom (Chr) are sometimes cited. 3:1

{Bya2 Acx boc2 e2 d+ f+ MNaeginvb2 }]

h

Despite only one witness to this reading, it is interesting because it preserves Semitic, not Greek, idiom. However, it seems likely that the scribe was imitating the idiom found in both P and T in the same place, an idiom current in Hebrew as well. The only other two occurrences of in Kingdoms with God as the object are 1 Kgdms 2:11, 18. 2:11 reads , and 2:18 reads . In both cases a circumlocution occurs, so the scribe of h could have been simply conforming his text to the idiom of LXX itself. 3:2

B* c] cop eth syhj

{ya2 } A{x} {boc2 e2 d+ f+} B abMN{aeghinvb2 }; arm

The reading is attested only by the original hand of B and by one fifteenth century ms. It could be argued that should be preferred as the reading different from similar statements in the context (i.e., 3:3, 5 bis, 6, 9) and from MT. Furthermore, a closer examination shows that here refers to Eli, whereas all examples of and related forms refer to Samuel. It is possible that a Hebrew scribe (or the translator) could have misread as . However, the extreme paucity of external support makes the preference of questionable. A scribe transmitting the Greek text could just as easily have misread as . 26 If is 25 Edwin Hatch and Henry A. Redpath, A Condordance to the Septuagint and the Other Greek

Versions of the Old Testament (Including the Apocryphal Books), 3 vols. (Oxford: Clarendon Press, 18971906; repr., Grand Rapids: Baker Book House, 1983) [hereafter HR]; Henry Barclay Swete, An Introduction to the Old Testament in Greek, with an Appendix by Henry St. John Thackeray, rev. Richard Rusden Ottley (Cambridge: Cambridge University Press, 1914; repr., Peabody, MA: Hendrickson, 1989); Conybeare and Stock, Grammar; Bo Johnson, Die hexaplarische Rezension des 1. Samuelbuches der Septuaginta, Studia Theologica Lundensia, no. 22 (Lund: CWK Gleerup, 1963). For a summary of the family groups in LXX, see ibid., 19. In the following pages, LXX evidence for particular readings will be grouped according to Johnson's family groups. Mss y and a2 , when supporting reading in the absence of the other ms or B, will be listed with the ungrouped mss. 26 Cf. a similar error in the Ethiopic of 3:3, where

is apparently read for

.

39

indeed original, the scribes responsible for B and c could well have made the graphic error independently of one another or of any text in the direct genetic line between the two. Thus, the majority reading is to be preferred, with Rahlfs. 3:2

{Ba2 Acx bc2 e2 f+ Mghiv}]

o d+ Naenyb2

At issue in this reading is whether the subject of the verb is Eli, thus requiring a singular verb, or Eli’s eyes, requiring a plural verb. Though the main reading of both MT and V is singular, one ms of each has a plural. Both readings are well attested in Greek mss. It is likely that was influenced by the plural verb in the same verse to become . However, a similar change could have happened in Hebrew just as easily. The agreement of the plural with the mss from MT (ms 187, one of the mss Goshen-Gottstein says perhaps contains some non-Masoretic readings) and V suggests the possibility that the plural did exist in an early Hebrew ms; however, the original Greek was probably . 3:3

{Ba2 Acx dpqt MNaeghnb2 }] +

3:7

Ba2 ] +

boc2 e2 f+ vyz; Thdt

A{cx boc2 e2 d+ f+} MN{aeghinvyb2 }

These two readings need to be considered together, inasmuch as the preferred reading in one place is likely to be the same as in the other. The simple is predominantly an Attic reading, whereas the compound is originally Ionic; however, the latter reading predominates in the koine. 27 is well attested in verse 3 (apparently read by Ba2 Acx dpqt MNaeghnb2 ) but is poorly attested in verse 7. If the normal koine form were original in both places, later Atticizing scribes might have omitted the to create a more Atticistic reading. It is unlikely that later scribes would have corrected an original Attic reading in the direction of the koine, especially in the numbers attested in verse 7. As far as the difference in the attestation patterns between the two verses, two mss (c and x) read rather than in verse 3; since even Attic idiom allowed before an infinitive (both of Homer’s uses are before infinitives), it is possible that Atticistic revisers, after changing to in verse 3, would have considered the of verse 7 perfectly acceptable. Though the difference in the verb is attested in only two mss, other such mss may have exercised their influence on the reading of the preposition in some extant mss. Whichever reading is to be preferred, 27 Walter Bauer, A Greek-English Lexicon of the New Testament and Other Early Christian

Literature, 2d ed., trans. and ed. William F. Arndt, F. Wilbur Gingrich, and Frederick W. Danker (Chicago: University of Chicago Press, 1979) [hereafter BAG], s.v. “ ”; ibid., s.v. “h.” Bauer characterizes the compound form as an “Ionism, very rare in Attic wr., but common in the Koine” in the latter article. Smyth notes that though the compound form occasionally occurs in Homer and Herodotus, it is rare and suspect in Attic Greek; Herbert Weir Smyth, Greek Grammar, ed. Gordon M. Messing (Cambridge: Harvard University Press, 1956), 555.

40

both verses probably originally read the same. Rahlfs’s decision to read in verse 3 and in verse 7 seems to have been due to the combined witness of A and B in the former passage and their split witness in the latter. Nevertheless, it seems preferable in light of both internal and external evidence to read in both verses. 3:3

{Bya2 f+ MNaeghinvb2 }]

Ac | +

x boc2 e2 d+

Since MT here reads , it seems likely that the third variant is a correction to a similar Hebrew text. This supposition is strengthened when one realizes that the mss that support this reading are the Lucianic group (boc2 e2 ) and the group that Johnson calls d+, both of which often correct the Greek text in the direction of the Hebrew.28 The hexaplaric ms x also supports the reading. Harder to account for is the second reading, supported only by the hexaplaric witnesses Ac. Though the addition of supports MT in part, renders in only one other instance out of hundreds of occurrences of in the books of Kingdoms (4 Kgdms 20:18). is rendered much more frequently by . These factors suggest that represents a correction of OG to the reading of a Hebrew text different from MT. The first reading, , is almost certainly the OG reading. 3:4

1° {Bya2 Acx d+ fmw aeinb2 }] pr (M)Nhv; (itv )

g | pr

boc2 e2 s

The second and third readings may be taken together, since the omitted in the second is probably the result of haplography. The first reading is clearly the better attested of the two main readings, containing all of Johnson’s groups except the Lucianic mss, as well as a few mss that do not fit in any particular group. The third reading is attested by two uncials, the Lucianic mss, and a few other mss. It is also supported by a similar reading in an it ms: et astitit ei dominus et clamabat. One could argue that the third reading originated from the first as the result of assimilation to verse 10. The first reading may also claim precedence on the grounds that it is the shorter reading. Scribes might also have considered God’s call of Samuel abrupt, leading them to introduce God’s words with mention of his arrival in the sanctuary. On the other hand, the omission of (or even , cf. verse 10) could be due to parablepsis ( … ). The possibility that the shorter reading represents a correction to a Hebrew text similar to MT seems unlikely, for, though some of the witnesses are hexaplaric (Acx) and others sometimes have hexaplaric readings 28 Johnson says of the group d+, “d+ zeigt gelegentlich, besonders bei Eigennamen, eine Textform, die besser mit MT übereinstimmt, als es bei A cx der Fall ist. Dagegen hat d+ kaum eigene Varianten von hexaplarischem Charakter aufzuweisen.” Johnson, Hexaplarische Rezension, 107.

41

(d+, f+), there is little evidence that B is ever corrected toward a Hebrew text.29 If the longer reading is taken to be original, another factor must be considered, namely, that the subject must also be displaced, since a text ( ) , though perfectly acceptable in Greek, does not represent normal Hebrew idiom, since the subject is too far from the initial verb (cf. verse 10: ). It may be questioned why, if is a harmonizing addition, the subject was not moved forward in the sentence to match verse 10. The answer may be that the Greek scribes, unfamiliar with Hebrew idiom, felt no discomfort at the distance of the subject from the initial verb. A Greek form of the works of Ephraem Syrus, preserved in the apparatus of Holmes and Parsons, provides a form of the text that may also have some bearing on the question of the position of the subject. Ephraem reads . Although “Samuel” cannot be the original subject, its presence is an attempt to clarify the subject of , an indication of the difficulty Ephraem (or the scribes) felt with the dangling verb, so the reading presupposes a text identical with the third reading above. Thus, if a longer reading was original in OG, it was probably not that of the third reading. Since internal evidence is ambiguous, it is perhaps best to rely on the reading that has the most, the oldest, and the most diverse witnesses. Though some doubt must exist because of the possibility of parablepsis, the reading of OG was probably that of B and its congeners. 3:5

{Bya2 A dpqt i}] pr msw Mgv; it | + Nf

3:6

Bya2 i; cop eths] MN{aeghnvb2 }

cx aehnb2 ; arm |

244 | +

3:6

{B*y A dpqt i}] pr

3:9

B A qt i] pr

Acx d+; arm | +

boc2 e2 z {boc2 e2 f+}

cx boc2 e2 z f+ BabMNaeghnva2 b2 ; arm

{ya2 cx boc2 e2 dpz f+} MN{aeghnvb2 }; arm eth

All these variants must be considered together from two perspectives, first looking at the presence or absence of between and in verses 5, 6, and 9; and next looking at the presence or absence of (or ) in each of these verses. First, it may be noted that the majority of the witnesses in each verse include , whereas MT does not have a conjunction between the corresponding words. When one looks at the grouping of witnesses, it is surprising to see that the hexaplaric mss cx consistently include the , with A alone omitting it. The group d+ (except z, which often has Lucianic affinities) omits in the first two instances, but two of its members, dp, apparently join the majority in the third instance. The group f+ and the Lucianic group include in all 29 Johnson, Hexaplarische Rezension, 53: “Zusammenfassend wäre zu sagen, dass B nur in sehr

geringen Umfang Stellen enthält, die auf hexaplarische Korrecturen hindeuten könnten. . . . Hexaplarische Korrecturen können in vereinzelten Fällen eingedrungen sein, doch lassen sich die betreffenden Stellen auch auf andere Weise erklären.”

42

three cases. What does this evidence imply? It is clear from an examination of both Greek and Hebrew mss that the addition or omission of a conjunction is common, so it is impossible to tell for certain whether the change occurred before or after the initial translation into Greek. Though Hebrew uses conjunctions with great frequency, asyndeton in certain constructions is fairly common. Such a construction occurs here, where the first imperative is almost an auxiliary to the main idea expressed in the second verb . 30 However, similar constructions do appear with the conjunction. It may have seemed to a scribe or translator that the addition of a between the imperatives provided a better balance to the that was between the indicatives in the next phrase. Since internal evidence is questionable, it seems best to go with the preponderance of external evidence in this case, which favors the inclusion of the in the first instance and strongly 31 favors its inclusion in the last two instances. With regard to the question of the inclusion or exclusion of or in verses 5 and 6, the shortest text, supported by the group Bya2 in both cases, excludes the terms, while MT includes in verse 6. It is significant that in verse 9, all Greek mss include after , but there is no corresponding Hebrew term in MT. The easiest issue to address is the expression in verse 6, an obvious correction to a Hebrew text similar to MT by the hexaplaric and d+ groups. is a common translational equivalent for , rendering it twenty times in Kingdoms (in A, nineteen times in B), though always in sections that are not kaige. It is not nearly as common as , however, which renders some eight hundred times, in both OG and kaige sections. Thus, is an attempt to yield a text closer to the Hebrew. The short reading in verse 6 is the key to determining the OG reading in verses 5 and 6. Though supported by only a small number of witnesses, both Greek and versional, this reading is almost certainly the OG reading. If original, it is easy to understand that the longer text of the third reading could be an assimilation to verse 9 or to some non-Masoretic Hebrew text that added , but it is difficult to explain the origin of the shorter text if the longer were original. If some scribe felt the need to add or in verse 6, or to assimilate the verse to verse 9, the same process could have occurred in verse 5, where the group d+ joins the witnesses to the shorter text. Therefore, it appears that the shorter text without or is the OG reading in both verses 5 and 6, but in verse 5 could very well reflect a Hebrew text different from MT. 3:6

{Ba2 Acx d+}]

y|

MNaghb2 | +

30 E. Kautzsch, ed., Gesenius’ Hebrew Grammar, trans. A. E. Cowley, 2d ed. (Oxford: Clarendon

Press, 1910), § 120g. 31 Jellicoe, along with many others, believes that the Lucianic witnesses, which consistently

include the in all three verses, sometimes preserve primitive readings which have disappeared from most or all other witnesses, including B. See Jellicoe, LXX and Modern Study, 168-71.

43 i|+ 3:6 3:8 3:8

boc2 e2 f+ env {Bya2 boc2 e2 f+ MNaeghinvb2 }] +

Acx d+

1° {Bya2 Acx boc2 e2 d+ f+ MNaeghinv}] + {Bya2 Mehinvb2 }] pr boc2 e2 Ng

a | pr

Acx | pr

b2 ; arm it d+ | pr

f+ | +

The repetition and variation in the story is evident again in verses 6 and 8. At the beginning of verse 6, MT reads ; it lacks an equivalent for the in verse 8. The first item to note is the two forms of the clause that appear in the third and fifth variants in the first unit of variation. Both of these variants conflate the text by adding a clause closer to the Hebrew text that lay before the scribes, namely, the reading of MT. The result is an unwieldy and redundant sentence that certainly is not original.32 Little doubt can exist that is an addition not present in OG. The next bit of variation involves the position of in verses 6 and 8. In verse 6, does it belong before , after , or after ? Or should it be omitted altogether, with Ba2 ? The hexaplaric and d+ groups place in a position equivalent to that of in MT. Most other witnesses put it only in the first clause, either before or after . The variety of position suggests that is a later addition to the text and that the OG reading is that of Ba2 . 33 In verse 8, each of Johnson’s groups supports a different reading: Bya2 variant one, hexaplaric variant three, d+ variant four, f+ variant five, and Lucianic variant six. (Only those readings in v. 8 that deal with will be discussed here; see below for the other variants in this verse). The base text is equivalent to MT, so there is no question of other groups altering the text toward one with the same reading as MT. The third reading is clearly an inner-Greek error for , so variants three, four, and five may be considered together. As in verse 6, the hexaplaric and d+ groups support the inclusion of after a form of , and they are joined by the f+ group. However, the two possible positions for in the verse, and the likelihood of assimilation to the readings of verse 6, make the reading questionable. The failure of the Lucianic group to support the reading also tends to make one suspect that it is not original. Probably, then, is original in neither verse 6 nor verse 8. It may be noted quickly that the omission of in verse 6 (variant 2) is 32 Ms g omits the second phrase

, but this omission is probably due to

parablepsis rather than a conscious effort to improve the text. 33 There are three main occasions for transposition. The first is when a scribe simply inverts the

order of words or phrases that are adjacent or nearly so. The second is when a unit of material is accidentally omitted by a scribe, only to be added back by another scribe in the wrong place. Cf. Albert Curtis Clark, The Descent of Manuscripts (Oxford: Clarendon Press, 1918; reprint, Norwich: Fletcher and Son, 1969), 256. The third occasion for transposition occurs when an addition is made to the text, often as the result of a scribal gloss in the margin. See Shemaryahu Talmon, “Aspects of the Textual Transmission of the Bible in the Light of Qumran Manuscripts,” Textus 4 (1964): 100-103, who discusses the marginal and interlinear notes in Qumran biblical mss and their role in the production of conflate readings.

44

probably the result of haplography (perhaps omitting the first word of 3:6

3° {Bya2 boe2 f+ MNaeghinvb2 }] pr Acx c2 d+; arm eths

).

44 | pr

The third variant is equivalent to the reading of MT, and it appears in the hexaplaric and d+ groups, plus c 2 and two versions. The real questions are (1) whether is an addition added as an assimilation to verse 8 or a clause omitted by parablepsis ( ... ) and (2) whether the assimilation or parablepsis occurred in Hebrew or in Greek. Verse 6 MT at this place reads , but verse 8 reads (also reflected in ms 44 in v. 6). It is clear that if the longer text of verse 6 is the result of assimilation, it is not perfect assimilation, since is missing. Rather than imperfect assimilation, it seems more likely that (or its Hebrew equivalent) was accidentally omitted. The descriptions of Samuel’s actions the first three times God calls him are all different in detail, but in MT Samuel always does two things. In verses 4 and 5, Samuel says, “Here I am!” and then runs to Eli. In verses 6 and 8, Samuel gets up and then goes to Eli. This balance is destroyed if verse 6 omits his getting up, but if both actions are included, the variety of presentation is preserved by the presence of Samuel’s name a second time in verse 6 ( ). It is difficult to decide where in the process of transmission the parablepsis occurred, in a Greek or a Hebrew ms. However, if the longer reading was present in OG, then disappeared, it apparently left no trace in the ms tradition, since the hexaplaric and d+ groups probably derived the reading from Origen’s correction to the Hebrew. It is possible that a trace might remain in the Lucianic witnesses c2 arm, but it is more likely that these, too, reflect the influence of the Hexapla. It seems probable, then, that the parablepsis was already present in the Hebrew Vorlage of OG, so the first variant represents the OG text. 3:6

{Bya2 Acx d+ i}]

boc2 e2 f+ MNaeghnvb2

The first reading is closer to normal Greek idiom (though the article is usually omitted in classical Greek) than the second.34 occurs twice more in Kingdoms—2 Kgdms 14:29 and 3 Kgdms 19:7—both times for ; both of these occurrences are also adverbial. 2 Kgdms 16:19 is also adverbial and renders in MT as , but the adverbial idea is different: “moreover” rather than “a second time.” 3 Kgdms 9:2 renders , “a second time,” as . A broader look at the word in LXX reveals that the Pentateuch uses ( ) exclusively to render 34 Smyth, Grammar, 288. Cf. also p. 317, where he notes that

numerals as partitive genitives; the use of be idiomatic, at least in classical Greek.

is sometimes used with with an ordinal to represent an adverbial idea does not seem to

45

when it means “a second time,” a total of nine times. On the other hand, the other books almost always render , “a second time,” by . 35 It is interesting that B omits two of these occurrences, Josh 5:2 and 1 Chr 29:22. The only occurrence of to render outside the Pentateuch and Kingdoms is B’s text of Jer 40:1 (Heb 33:1), but A has . It appears, then, that the tendency of LXX outside the Pentateuch is to use , but B differs three out of eight times. Since B sometimes displays a tendency towards Atticism (see above, pp. 39-40, on vs. in verses 3 and 7), and since it elsewhere avoids , it seems more likely that the scribe of B, or some previous scribe, would have changed to the more idiomatic than vice versa. 3:7

{Bya2 MNaeghinvb2 }] pr

Acx boc2 e2 d+ f+ r

occurs frequently in Kingdoms both with and without the article. When the examination is restricted to the accusative , the vast majority of anarthrous cases fall into one of two types. The first common use of without the article is in the expression ( ) x, where x is a pronoun or noun in the genitive case (e.g., [1 Kgdms 7:8], [3 Kgdms 15:30]). On one occasion, 1 Kgdms 5:7, the similar construction occurs. The second common use of the accusative of without the article is in plural references to other gods (e.g., 1 Kgdms 28:13; 3 Kgdms 14:9). Once these cases are eliminated, only thirteen instances of the anarthrous remain, ten of them in 4 Kingdoms (a kaige section). Of eleven occurrences of the accusative of with the article, two are plural references to other gods (1 Kgdms 7:3; 2 Kgdms 5:21) and one is in the expression (3 Kgdms 16:13), though AN and many other witnesses do not have before . Only two cases of with the article occur in kaige sections (2 Kgdms 12:16; 22:7). It appears, then, that, apart from the expression x (genitive), the singular accusative references to in Kingdoms generally use the article in OG sections (six of nine times with the article) and generally do not in kaige sections (ten of twelve without the article, all in 4 Kingdoms). Nevertheless, without the article is still fairly common in OG, occurring in 33% of the cases, 40% if this case is considered anarthrous. Furthermore, one of the other instances of anarthrous is in the same chapter, 1 Kgdms 3:13. It seems best, then, to omit the article here, taking its inclusion as an attempt to conform to the more usual pattern. 35 Josh 5:2; 1 Chr 29:22; Jon 3:1; Hag 2:2; Zech 4:12; Jer 1:13. Dan 2:7 LXX renders the

Aramaic

with

, while Theodotion uses

.

46 3:7

{Bya2 Acx p*qtz}] pr pb? | pr MNaeghnvb2 | pr i | antequam adaperiret ; it

boc2 e2 d f+

The second, fourth, and fifth variants in this unit of variation should be seen as witnesses to the reading of the third variant, , the second and fourth variants being inner-Greek corruptions of the reading. The real textual question, then, is whether the preposition appears once or twice. MT reads before both verbs, so the second could be seen as an attempt to conform the Greek text to one similar to MT. Whereas Hebrew prepositions usually govern only one word,36 Greek prepositions frequently govern more than one object.37 Another reason for the second , then, might be conformity to Hebrew idiom. It is particularly significant that the hexaplaric mss Acx agree with B against MT. Thus, it is probable that the reading of the base text is the original reading of LXX. 3:8

{Bya2 Mehinvb2 }] pr boc2 e2 Ng

a | pr

Acx | pr

d+ | pr

f+ | +

The variants that concern have been dealt with above, but it remains to consider the article and the second . The second is probably an assimilation to the previous two accounts of God’s attempt to communicate with Samuel, verses 4 and 6. In both of those cases, and in the Lucianic witnesses here, is clearly considered to be a vocative. The insertion of the accusative article in variants two and five indicates an understanding that here is the direct object of the infinitive. The article was probably inserted to clarify this point. Therefore, the reading of the base text is to be preferred. 3:9

{Bya2 f+ Maghivb2 }] + Nen; ethvid | + cop | + Acx boc2 e2 pqtz

d | + Heli ei ;

Readings two through five present four different attempts to specify the participants in the action. The longest reading, supported by the hexaplaric, d+ (-d), and Lucianic groups, is the same as MT. Since no reason seems to exist for shortening the text, all of the longer readings can be seen as attempts at specification or at conformity to a Hebrew text similar to MT. The variety of readings also suggests that none of the longer texts is original. 36 But see Kautzsch, ed., Gesenius’ Hebrew Grammar, § 119hh. O’Connor calls this phenomenon prepositional override, citing 1 Sam 15:22 as an example. See Bruce K. Waltke and M. O’Connor, An Introduction to Biblical Hebrew Syntax (Winona Lake, IN: Eisenbrauns, 1990), 222-23; M. O’Connor, Hebrew Verse Structure (Winona Lake, IN: Eisenbrauns, 1980), 310-11. 37 Smyth, Grammar, 369-70.

47 3:9

{Bya2 A dpqt ein}] +

cx | +

boc2 e2 z f+ MNaghvb2 ; Thdt

Two hexaplaric mss again add , almost certainly secondarily. More significant is the longer reading that includes and is supported by many witnesses, including the f+ and Lucianic groups. Although no equivalent for appears in MT, it is conceivable that the reading could have arisen from a Hebrew text whose verb could be read as a participle rather than an imperfect.38 The reading seems to be an attempt to specify the subject of the verb, though in a rather unusual way. A participle used in proximity to a verbal form of the same word is probably a Hebraism, but Greek scribes were certainly capable of creating their own Hebraisms. Since the Lucianic text is known for expansions of this type, it is probable that the variant is secondary and within the Greek tradition, so the first reading is preferable. boc2 e2 d+ f+ AMNaa?cegnvxa; arm etha39

3:9

{Bya2 x a*hi*b2 }] +

3:10

{Bya2 A o dpqt fm*sw MNaeghinb2 }] +

cx bc2 e2 z mbv; arm

These two sets of variants exist because scribes were troubled at the lack of agreement between what Samuel was supposed to say and what he actually said. The fact that many witnesses that support the longer reading in the first unit of variation support the shorter reading in the second unit suggests that scribes attempted to correct the problem either by deleting in the first instance or adding it in the second. On the other hand, might have been added in verse 9 or omitted in verse 10 in order to conform the text to a Hebrew text similar to MT. However, other units of variation indicate that most scribes were more concerned with consistency than conformity to a Hebrew text, especially since few scribes could read Hebrew.40 The hexaplaric witnesses are usually an exception to this rule, but here they are split, indicating the importance of internal consistency to at least two of the scribes (the scribes of cx or their predecessors, though x* does read alone in verse 9). These considerations, plus the stronger external support for the longer reading in verse 9 and the shorter in verse 10, indicate that these readings should be preferred, with Rahlfs. 41 38 The possibility that LXX was translated from Old Hebrew script rather than square script has

been raised by various scholars. If so, the present case may be explained as a confusion of the of the imperfect with a , taken as an article, since the two letters are similar in the older script ( and , respectively). Cf. Shemaryahu Talmon, “Ancient Hebrew Alphabet,” in Mélanges Delcor, 387-402. 39 Thdt reads

; Natalio Fernández Marcos and José Ramón Busto Saiz, Theodoreti Cyrensis quaestiones in Reges et paralipomena, Textos y estudios “Cardinal Cisneros”, no. 32 (Madrid: Instituto “Arias Montano” Consejo Superior de Investigaciones Científicas, 1984), 12. 40 Cf. Wellhausen, Bücher Samuelis, 4-5. 41 Several commentators refer to ms B in v. 9 as though it were the LXX reading. The fact that B

here may not be the original LXX reading illustrates the importance of analyzing the variants within LXX before using one of the mss to reconstruct the original Hebrew text. Cf. P. Paul Dhorme, Les livres de Samuel, Études bibliques (Paris: Librairie Victor Lecoffre, 1910), 43; R. W. Klein, 1 Samuel, 30.

48 3:10

{Bya2 f+ MNaeghinvb2 }] Mmg; it | boc2 e2 | +

d|

cx | A pqtz; arm

MT here reads , which corresponds to the sixth variant, supported by A, d+ (-d), arm. The variants which substitute for the second clearly preserve an attempt to produce a better Greek text. It is interesting that in the only verse where MT reads , most Greek mss do not (cf. vv. 4, 6). Since the addition of can be explained as an attempt to conform to the common Hebrew text, the shorter text (variant one) is preferable to the others. 3:11

Bya2 ] pr

A(cx) MN {boc2 e2 d+ f+ aeghinvb2 }; arm eth

Though the vast majority of mss have , it seems more likely that it was added to make the text closer to a Hebrew text similar to MT (which reads ) than that it was deleted by Bya2 . A similar expression appears in two other passages in LXX: 2 Kings 21:12 and Jer 19:3. Though is found in both these passages, other elements of the formula are different, indicating the lack of a fixed form for this stock phrase. Of course, it is possible that was omitted by haplography, but B elsewhere lacks a tendency to conform to the Hebrew text where others do (cf. v. 6). In addition, the omission of the conjunction is supported by T. Thus, the shorter reading is to be preferred. 3:13 |

Bya2 Nhi(v)b2 ] boc2 e2 z | A| M {f+ aegn}| praedixi ; it

dpqt |

cx

The it reading is not particularly helpful in determining which Greek reading it reflects, so it must be set aside. The reading of A seems to be an error, perhaps an aural confusion of the first variant. The other variants reflect two sets of differences: aorist or perfect tense, and a prefix of - or -. The perfect tense, being used less frequently than the aorist, is more probable because scribes would be more likely to change perfect to aorist than vice versa. External evidence also supports the originality of the perfect. The choice between the prefixes is not so easy. Both prefixes are well represented, though appears in a variety of forms. A survey of and in Kingdoms reveals that is more common then in the OG sections (fiftyseven and twelve times in A, sixty-seven and seven times in B, respectively), but is more common than in the kaige sections (thirty-one and seventeen times in A, thirty-one and twenty-one times in B, respectively). Furthermore, is somewhat more frequent in LXX as a whole. These figures suggest that, in the OG sections especially, scribes would have a greater tendency to replace with than vice versa. B in particular strongly favors in the OG sections. The fact that B keeps here, perhaps along

49

with a predecessor of A, indicates that the first variant, the reading of the base text, is the best. 3:13

{Bya2 cx Maehinb2 }] pr

A boc2 e2 d+ f+ Ngv

MT here reads , with no preposition such as corresponding to . In fact, there seem to be no cases in MT of the verb with a preposition before the personal object (excluding , used of deciding between parties). However, frequently takes after it (in about 16 of 36 cases), particularly when rendering (whether with or without in MT, in about 14 of 25 cases). It seems, then, that was added during the process of transmission of the Greek text and that the shorter reading is preferable. 3:13

2° {Bya2 Acx Mag}] quas ipsi scivit ; itb | + Chrvid | + bc2 e2 z | + e|+

b´ dpqt f+ Nhinvb2 ; o | quas ego scio ; itv

All of the variants that differ from the base text are modeled on the phrase found in MT, . Variants five and seven change the subject of the verb to God (“which I knew”), probably by inner-Greek and/or inner-Latin corruption. The main Lucianic reading, , differs from the majority reading, , because the case of the relative pronoun has been attracted to that of its antecedent, . 42 The additional found in variant six is apparently a corruption of the main Lucianic reading. The second variant is basically equivalent to the majority reading. Thus, all the other variants from the base text are derived from the third variant. It is more likely that the phrase was added to agree with a Hebrew text similar to MT than that it was deleted, since no problem with the reading is evident. Thus, the base text should be accepted as original. 3:13 3:15 3:15

B Ac c2 e2 N] pr

{ya2 x bo d+ f+} M{aeghinvb2 }; Chr

{Bya2 A dp MNaeghinvb2 }] pr Bya2 ] pr

cx boc2 e2 qtz f+

A{cx boc2 e2 d+ f+} MN{aeghinvb2 }

The ease with which scribes added and omitted articles has been mentioned above, so it is difficult to have any certainty with regard to them. However, since articles were present more often than not, lacking evidence of haplography, it seems more likely that a scribe would have added the article than omitted it. Construct relationships like the first case occur 17 times with the article and 5 times without in the chapter, excluding this one, so it is likely that a scribe would have changed the reading of his exemplar in favor of the more common reading. In the second case, the phrase occurs three other times in Samuel, none of which has an article. The addition of the article in this case was a 42 Smyth, Grammar, 567.

50

move in the direction of more articles, which reflected better Greek idiom. Thus, the article should be omitted in the first two cases. The third unit of variation is somewhat different, both in the scantiness of external support for the base text and in the fact that the use of the article with the dative of the direct and indirect objects is a special case. Since most Hebrew names are indeclinable as transliterated into Greek, those forms of the dative that cannot take prepositions (i.e., the pure dative) have no way of indicating case without the article. Since the article is not used in Hebrew, its use in LXX is clearly a semantic indicator inserted by the Greek translators. Of about twelve other cases of proper names used as datives of direct or indirect objects in the first twelve chapters of 1 Samuel, all of them have the article. Thus, it seems probable that the article should be included in the present case as well. 3:17

Ba2 ; cop it] + arm (eth)

A(x) {c boc2 e2 d+ f+} MN{aeghinvyb2 };

The expression ... occurs eleven other times in LXX: ten times in Kingdoms and once in Ruth. In each of these cases the full expression is found; is never omitted. Though it is conceivable that the second part of the phrase could be omitted as being unidiomatic in Greek, none of the mss ever do so in any of the other cases. On the other hand, could have been added to conform the passage to the usual expression and to match the Hebrew, but if the omission is original in LXX, the reason for this uncommon readings would still be unexplained. It seems more likely, then, that the original was omitted by B or one of its predecessors, perhaps because of some graphic similarity between and . 43 3:17

Ba2 A] > cx e2 b2 ; ethvid | pr

{boc 2 d+ f+} MN{abeghinovyc2 }

In this unit of variation, the omission in the second variant is the result of parablepsis. Since more nouns have articles than do not, one’s initial reaction, in spite of the extremely thin external support, might be to exclude the article as a later addition. However, two factors argue against this evaluation. First, MT here reads , so the third variant cannot be an attempt to conform to MT (at least not completely), since is singular and is plural. Second and more importantly, the meaning of is not really appropriate to the context, for the phrase 43 A possible two step sequence of events that could lead to the omission of the phrase is as

follows. First, the scribe might have accidentally written a second time for . The resulting phrase would be . Since this new phrase would make no sense to a second scribe reading the exemplar, the easiest correction he could make would be to omit .

51

means, “from all conceivable words that were spoken,”44 but the context clearly indicates that specific words (i.e., God’s words) are being requested. The article probably dropped out as a result of parablepsis. 3:17

2° {Ba2 Acx dpqt i}]

y | > boc2 e2 z f+ MNaeghnvb2 ; arm cop

MT here reads ; it has nothing that corresponds to . Since (or ) functions as the equivalent of , the longer Greek text found in the first variant is redundant. It is possible that is a simple equivalent of the circumlocution and that the longer Greek text is a conflation of these two equivalents.45 Even so, the question remains whether the conflation occurred subsequent to the original translation or whether it was taken from the Hebrew Vorlage. If the longer text were in the Vorlage of LXX, no good reason exists for the omission of . However, if the shorter text represented OG, later scribes might well have added to bring the Greek closer to a text like MT. It is important to observe that among the mss that support this apparent addition are the hexaplaric mss and group d+, which often has hexaplaric readings. The fact that Ba2 (y) also support the longer reading demonstrates the occasional secondary readings of the group in general, and of B in particular. is probably an inner-Greek variant, but it could also be an independent correction to the same Hebrew text. It seems, then, that the shorter text represented by the third variant has a greater claim to originality. 3:18 d(p)qtz; arm | +

{Bya2 v}] f+ MNaeghinb2

Acx boc2 e2

The second variant, though supported by many witnesses, should be considered an assimilation to a text similar to MT. It is supported by the hexaplaric, d+, and Lucianic groups, all of which either occasionally or frequently correct their Greek texts toward the reading of Hebrew texts. The third variant represents an alternative attempt to specify to whom Samuel was speaking, and so is secondary. This variant presupposes the first, rather than the second, variant as the text which needed correction. Thus, the first variant is to be preferred. 3:21

{Bya2 Acx dpqtz* aein}] 21 MNgb2 ; eth itv | boc2 e2 f(h)msvwza?; arm | et factum est profetam; itb Nothing in 1 Sam 3:21 MT corresponds to this phrase. However, it is possible 44 Smyth, Grammar, 296. 45 For a possible parallel, see 1 Sam 1:23. MT reads, in part,

LXX reads is a simplification of 400, 410-11.

, and . Many commentators contend that , but cf. S. D. Walters, “Hannah and Anna,”

52

that the phrase (or whichever is original) is itself a variant of in 3:20, probably a variant based on a different Hebrew text (or different pointing of the same consonantal text). If so, the in the third variant should be seen as assimilation to the previous verse. The genitive article is a matter of Greek idiom rather than Hebrew Vorlage. 46 Once these matters are settled, all that remains is to determine the original word order. Though the reading of the base text has much stronger external support than the second variant, the third and fourth variants also put the verb before the noun. However, the word order in the third variant may be explained as better Greek style, so the first variant may be preferred on the basis of its external support. 3:21 Magb2 |

Bya2 Nm] boc2 e2 h | domini ; eth itb syhj | A {d+ fsw einv}

x|>c|

A look at the rest of the books of Kingdoms reveals that the phrase ( one of these times) occurs seven times outside this chapter, and the similar phrase occurs four times. No other occurrences of ( ) appear, so it is evident that the genitive construction is much more common than the dative. The question is whether an original ( ) was corrected toward the more common or whether ( ) was corrected toward the of the previous verse. The similarity of the phrases in verses 20 and 21 might suggest that the correction was made toward the of verse 20. However, if the theory is true that the phrase in verse 21 is really a variant of that in verse 20, the two phrases should actually be very similar to one another, and later scribes, without an extant Hebrew text to prevent modifications of OG, would have been more inclined to substitute the more common construction. Thus, the reading of the base text is probably the best. 3:21

3° B boc2 e2 ] > {ya2 } A{cx d+ f+} MN {aeghinvb2 }; arm cop eth it

It is better Greek style to write ... than ... , which is a Hebraism. It is apparently better style in the languages of the various daughter versions of LXX as well. That is omitted for stylistic reasons in these versions is evident from a comparison of the same pattern in verse 20, where the same versions omit the in , but with Greek support only from ea2 . The multitude of Greek witnesses that join the versions in the present case can again be explained by the apparent lack of an extant Hebrew text for the purposes of comparison, since scribes that would have a tendency to correct towards the Hebrew would not be able to. Thus, though the omission of the has many more witnesses supporting it, the reading of B and the Lucianic witnesses should be preferred. 46 Smyth, Grammar, 451.

53 Summary of Deviations from the Base Text 3:2 3:3 3:5 3:6 3:6 3:7 3:9 3:9 3:15 3:17 3:17 3:17

] ]+ ] pr ] ] pr ]+ ] pr ]+ ] pr ]+ ] pr 2°] >

Peshitta The family of mss dependent upon 9a1 displays frequent deviations from the text of 7a1 that are best explained as improvements in the direction of more authentic Syriac style. In addition, several deviations seem to be in the direction of a Hebrew text similar to MT. Because of these observations, deviations in 9a1fam will not be considered apart from other supporting witnesses. A second identifiable group of Syriac mss consists of those mss based on 12a1, referred to as 12a1fam. This group shows a strong affinity for agreement with a third group, the witnesses to the standard Nestorian version (9c1, 10c1, 11c1). The corrector of 8a1 is also a witness to the standard Nestorian version, and restored portions of the ms may also sometimes be Nestorian in orientation.47 3:1

{6h1 6h4 7a1 7h12 7k3 8a1 9c1 9l1 10c1 11c1 12a1}] > 9a1fam

The base text of 7a1 agrees with LXX, whereas the omission of by 9a1fam is in agreement with MT. Since 9a1fam often corrects an earlier Syriac text on the basis of mss similar to MT, the reading of the base text is to be preferred here. 3:1

{6h1 6h4 7a1 7h12 7k3 8a1 9c1 9l1 10c1 11c1}]

9a1fam 12a1fam

Both , an active participle, and , a passive participle or adjective, have the same meaning. It is doubtful that variant one can be explained as assimilation to the Hebrew as in MT; it is more likely that one form developed from the other within Syriac itself. The form is more common in Syriac than , so it is probable that 47 OT in Syriac, VI-VII.

54

later scribes changed the active participle to the more commonly employed passive participle. This conclusion is supported by the rather free nature of many renderings in 12a1fam and especially 9a1fam. 3:3

{6h1 6h4 7a1 7h12 7k3 8a1 9c1 9l1 10c1 11c1 12a1}] 9a1fam

The base text agrees with MT and LXX in reading “the ark of God,” but 9a1fam and other mss read “the ark of the Lord God.” An examination of the thirty-four references to the ark in 1 Samuel 4-748 reveals that P agrees with both MT and LXX fourteen times, P agrees with MT against LXX nine times, P agrees with LXX against MT once, P agrees with neither MT nor LXX when the latter two agree seven times, and all three disagree three times. S. R. Driver, in his commentary on Samuel, says that P often agrees with the Lucianic witnesses when both differ from MT, but no significant agreement with LXXL occurs in these cases.49 An examination of the shorter expressions “ark of God” and “ark of the Lord” sheds some light on the solution to the textual problem. When MT and LXX agree in one or the other of these short expressions (nine times), P agrees with them in every case but one (4:11), and this one case is the only time MT reads instead of . When MT and LXX differ in these short phrases (seven times), P agrees with MT four times, agrees with LXX once, and goes its own way twice. Since MT and LXX are equivalent in the present case, it seems likely that P agreed with them. This deduction is confirmed by the fact that, though the expression “ark of the Lord God of Israel” is frequent in various mss of P (as well as in 6:3 LXX; 6:2 LXXcx), no other occurrence of “ark of the Lord God” without “of Israel” exists in any of the mss. Thus, the longer expression of 9a1fam should be seen as an inner-Syriac variant, and the reading of the base text should be accepted as original. 48 The second MT and LXX occurrence in 5:10 is missing from P by parablepsis, so is excluded

from consideration here. 49 P agrees with both MT and LXX L 14 times, P agrees with MT against LXXL 10 times, P agrees with LXXL against MT 2 times, P agrees with neither MT nor LXXL when the latter two agree 2

times, and all three disagree 6 times. It should also be noted that a shift in vocabulary in P occurs in 6:13. Before 6:13, P consistently uses , derived from LXX’s , to render . Frequent divergences from MT (and LXX) occur in these cases, as already noted. From 6:13 on, however, P renders with , clearly derived from Hebrew. In the 8 cases found from 6:13 on, P agrees with MT in every one of them. Before 6:13, P never agrees with MT more than five times in a row.

55

3:5

3:6

{7a1 8a1 9c1 10c1 11c1 12a1}] 7k3 9a1fam 9l1 ]

6h4 9a1fam | +

6h1 7h12

9a1fam

Though the addition of the pronominal suffix “you” in both verses might be considered a correction toward LXX, 9a1fam shows no particular tendency to correct toward LXX. Rather, it frequently demonstrates an independence in modifying the Syriac text that lay before it, apparently without the support of another exemplar. Thus, the additional may also be seen as an independent rendering. The additional in verse 5, derived from verse 6, is not found in either LXX or most mss of MT (though many LXX mss add after ; see above, p. 41). Kennicott lists two mss, one of them considered important by Goshen-Gottstein (the margin of ms 70), as adding the phrase, and two it mss do as well, according to the apparatus of BHS. Since scribes who corrected their Syriac mss with a Hebrew exemplar seem to have used one very similar to MT, the readings of the divergent Hebrew mss and the it mss perhaps suggest a different Hebrew tradition at this point in the text, one that has almost disappeared from the extant witnesses. On the other hand, the phrase could have arisen in the various witnesses independently by assimilation to the following verse. Since internal evidence is ambiguous, a look at external evidence will prove helpful. The first and third variants contain about the same number of pre-fifteenth century witnesses, but the third contains the earliest ms, 6h1, a sixth (or fifth) century ms. However, 6h4, another sixth century ms, also omits , though it adds . It is significant that half of the witnesses that support the first reading are representatives of the standard Nestorian version (9c1, 10c1, 11c1). It is also relevant that, when 8a1 lost two folios, they were replaced with text in a Nestorian hand. Since elsewhere in the first twelve chapters of 1 Samuel the standard Nestorian witnesses show no tendency to correct towards a Hebrew ms similar to MT, their agreement with 7a1 here suggests that the reading is original. Thus, the third reading, despite the support of 6h1, should be seen as an assimilation to the following verse that occurred early within the course of transmission of the book in Syriac. 3:6

1° {6h1 7a1 7k3 8a1 9a1 9c1 9l1 10c1 11c1 12a1}] +

3:6

2° {6h1 7a1 8a1 9c1 10c1 11c1 12a1}] +

6h4 7h12

6h4 7h12 7k3 9a1fam 9l1

Neither MT nor LXX support the longer readings in either case. The similarity of the ms grouping in the second unit of variation to that of the second unit in the previous set should be noted. The addition of 6h1 to 7a1, 8a1, and the witnesses to the standard Nestorian version makes the ms support of the base text almost overwhelming. The early witness 6h4 does support the second variant in each unit, but the ms elsewhere shows a

56

tendency to amplify the text for greater specificity (cf. the previous set of variants; 2:28 + ; 10:14 + ). On internal grounds, the shorter reading is preferable, since the longer reading is more specific. Thus, both internal and external probability support the readings of the base text in both cases. Summary of Deviations from the Base Text No deviations from the text of 7a1 are considered original in the chapter. Targum Sperber groups the witnesses to T that he uses in five groups. The first group, witnesses with Babylonian vocalization, includes the biblical mss p (the base text) and mwxy. Also included among the witnesses with Babylonian pointing are the haphtaroth mss, jk. The second group, witnesses with Tiberian (or no) vocalization, includes the biblical mss acf. The printed editions dbo make up the third category. Sperber’s fourth category, fragments of various targum mss, he refers to as Fr. Finally, the fifth group of witnesses, rabbinic citations, includes Aruk of R. Nathan (Ar), Yonah ibn Ganah (Gan), Rashi (Ra), and Kimh@i (K). Sperber uses subscripted numerals to refer to various marginal notes, often additional toseftoth, that occur in all the mss (e.g., w 1 ). Sperber divides his critical apparatus into two sections, the first dealing with differences in vocalization and the second with differences in the consonantal text. In the present study, differences in vocalization will be ignored unless they indicate significantly different understandings of the Hebrew text. In particular, when one set of consonants allows more than one set of vowel points (e.g., to indicate perfect or participle), the reading of the base text will be accepted, since it is the consonantal text that is being reconstructed. Of course, differences in punctuation are important indicators of how the text is to be understood and of how the translators understood the underlying Hebrew text. However, unless the difference in vowel points in T could reflect a difference in the underlying Hebrew consonantal text, pointing will be ignored. The history of the development of T raises the issue, discussed above for the Hebrew text, of what text to attempt to construct. Though T was an official targum, it was not a creation ex nihilo but relied on centuries of earlier oral and written renderings, particularly from Palestine. In light of this fact, the focus here will be to construct not the earliest form of any given verse but rather the “official” form. This approach will result in the exclusion from consideration of earlier, even pre-Christian, material (e.g., the marginal readings in Codex Reuchlinianus), but in the light of T’s textual history, this method seems best.

57 3:2

{mp f b}]

wy a do

Both the inclusion and the exclusion of the conjunction are supported by Babylonian and Tiberian witnesses. Though the Masoretic ms L does not have the conjunction, Kennicott lists nine mss that include it, one of which (187) is in GoshenGottstein’s list of important mss. LXX and P include the conjunction as well. It is always difficult to evaluate the inclusion and exclusion of conjunctions, since it was easy for scribes to add or delete them inadvertently. However, because T presupposes a text similar to MT in most cases, and because the conjunction is supported by LXX, P, and some Masoretic mss, it seems likely that it was also original in T and was only later deleted, either accidentally or in an attempt to conform to a Hebrew text like MT. 3:6

{mpwy af do}]

b

3:8

{mpwy a do}]

fb

The lack of a before the infinitive corresponds to the lack of a in the Hebrew Vorlage (also lacking in MT). The weakness of the textual support for the second reading in each instance is evident, especially since b is a printed edition rather than a ms. Also, the tendency of the scribes would normally be to add a , since the form without it was uncommon. Thus, the secondary nature of the forms with is apparent on both external and internal grounds. However, these readings have significance for the next set of variants. 3:7

{p af db}]

mwy o

The first variant is clearly to be taken as an infinitive, as the prefix indicates. The second variant could be an infinitive (as pointed by w) or a noun (as pointed by myo). The Aramaic construction, ( ) , is different from that of both MT and LXX, and is certainly the result of the meturgemanîn, so there is no question of a different Hebrew Vorlage here. The external evidence leans somewhat toward the first variant, since both Babylonian and Tiberian mss support it. As for internal evidence, infinitives with are far more frequent than those without. The is even retained after prepositions (e.g., 3:15; cf. also 10:13; 12:23 bis). In fact, the two occurrences of in 3:6 and 3:8 appear to be the only infinitives without in the first twelve chapters of the book. Lest one suppose that the root has some unique tendency toward the lack of a , an examination of all the occurrences of the infinitive of in T in the former prophets where there is a corresponding infinitive in MT reveals that the only other two occurrences (1 Sam 22:11; 1 Kings 22:13) do have . In two other passages, Judges 18:1 and 2 Kings 5:7, the infinitive of in MT has no corresponding infinitive in T (i.e., some other form of appears). The reason for the omission of the in 3:6, 8 is that no

58

was present in the Hebrew Vorlage. Since the construction found in T varies from that in MT, it is almost certain that the presence or absence of was not based on the Vorlage. Since the occurrence of infinitives with is much more common than those without , one would expect the meturgeman to use the more common form. It might be argued that since is the more difficult form, later scribes might have added the to make the text conform to the normal usage. However, the presence of with infinitives is so overwhelming, even in passages that have no corresponding infinitive in Hebrew (e.g., 1:21; 2:1 ter, 3 bis, 5 bis, 6 bis, etc.), that it seems more likely that the original was dropped by a scribe influenced by the two infinitives without in the preceding and following verses (i.e., ). The difficulty that scribes had in accepting as an infinitive may be reflected in the pointing of myo, which take the word as a noun. 3:9

{mpwy f dbo}]

a

in T generally renders Hebrew , the word found in most mss of MT. However, two mss listed by Kennicott read , the reading reflected by in a. Both and (taken as a temporal conjunction) fit the context equally well. However, is a more common conjunction than , and the scribe of a might have unconsciously replaced the latter with the former. Alternatively, because of graphic similarity, the scribe might have misread as , especially if the in his exemplar were smudged or faded at that point. These considerations, coupled with the preponderance of the external evidence, clearly indicate that the first variant is to be preferred. Nevertheless, it cannot be ruled out that the second variant reflects a correction to a varying Hebrew text. 3:10

{p a do}]

mwy f b

The phrase is a common circumlocution for the Hebrew in T.50 It occurs particularly frequently when God is portrayed in a manner that might be considered irreverent.51 The surprising reading is the simple , for one might have expected the circumlocution. It is easy to suppose that a scribe confronted with the shorter reading would have inserted , probably intentionally. It seems highly improbable that the longer reading, were it original, would have been shortened, regardless of the fact that a modern individual might consider the shorter reading closer to MT. The external evidence also supports the shorter reading somewhat. Though both variants are supported by Babylonian and Tiberian mss and by printed editions, the second variant has stronger 50 Etan Levine, The Aramaic Version of the Bible: Contents and Context, Beihefte zur Zeitschrift für die alttestamentliche Wissenschaft, no. 174 (Berlin: Walter de Gruyter, 1988), 58-59. 51 Scholars frequently explain such substitutions as attempts to avoid anthropomorphism, but, as

Levine notes, many much more blatant anthropomorphisms remain in the targumic texts; Levine, Aramaic Version, 55.

59

Babylonian support. However, it is the internal evidence that is most convincing, and it favors the shorter reading. 3:11 3:11

{p af d}]

mwy bo |

f

{mpwy a dbo}]

f

The external evidence in the first unit of variation favors the second variant, from which the third differs only orthographically. However, it is possible that the before in f in the first unit of variation is simply misplaced from the beginning of in the second unit. If so, f cannot be considered to support a deviation from the base text. Moreover, internal factors do not favor the alternatives to the base text. First, the reading of MT, , is equivalent to variants 2 and 3. Scribes would have been more likely to change the text to conform with the standard Hebrew text than to delete the ( ) . Second, the inclusion of ( ) creates a smoother text, connecting and subordinating the following clause to the preceding one. Third, the absence of in LXX mss Bya2 suggests that a Hebrew text lacking was current at one time (see above, p. 48). Thus, despite the stronger external attestation for the second and third variants, the base text should be preferred on the basis of the internal evidence. 3:16

{mp a bo}]

wy f d

These two variants could be considered to be simply inner-Aramaic variations, since as well as can function as the marker of the definite direct object. However, it is more common for T to render with , so it is possible that reflects a different Hebrew word. In fact, though Codex Leningradensis reads , many Masoretic mss (Kennicott lists twenty-three) read . LXX also supports a Hebrew by its reading, . It seems likely, then, that some Aramaic scribe, feeling that was not quite as literal as might be desired, corrected the text to agree more closely with the dominant Hebrew ms tradition, reflected in most Masoretic mss. The external evidence does not contradict this conclusion, so the first variant should be retained. 3:18

{mp a do}]

wy f b

These two expressions are basically equivalent, but the second is the more exact rendering of the Hebrew found in many Masoretic mss.52 The standard rendering of expressions based on and referring to people in T is the similar Aramaic phrase . This rendering is found consistently in 1 Samuel in all twenty-six cases. When God is the object of the expression (either , , or a pronominal suffix referring to 52 Though L reads the singular , many other mss either read the plural in the consonantal text or have the plural as a kethib-qere variant in the margin.

60

God), the standard rendering is (i.e., , , etc.), occurring all sixty times, excluding this one, in Samuel and Kings. In four of these instances, variants containing appear in one or two mss (2 Sam 11:27 f; 15:26 a; 1 Kings 11:38 yb; 2 Kings 18:3 a). However, these readings are surely secondary. The solidarity of the ms tradition leads to the conclusion that renderings including are either corrections to Hebrew idiom or, perhaps more likely, echoes of a lost Palestinian targum. Nevertheless, the official targum represented by T almost certainly read in all these cases, including the present one. 3:19

{pwy af do}]

b|

w1 | +

m

A look at the whole clause is helpful here: . MT here has . All the other versions support MT here, except L LXX , which adds after . It seems clear that the second and third variants are corrections to MT, and m is a conflation of these two renderings, similar to, but probably not genetically related to, that found in LXXL. All that remains is the reading of the base text, and it is by far the most strongly attested reading. Thus, it should be retained. One other item in the chapter that needs some comment is the form in 3:13. The vowels imply that the form is a peal, meaning “to be dull,” but the context demands a pael, whose meaning corresponds with the Hebrew piel of , meaning “to rebuke, punish.” Levy compounds the difficulty by citing the form as , which is anomalous.53 Because the context requires it, this verb is taken to be a pael in the statistics. Summary of Deviations from the Base Text 3:2 3:10

] ]

Vulgate V has the distinction among all the versions being considered, including the Hebrew, of being the only one to have a critical eclectic text. In fact, as noted above, two critical texts of V exist, the smaller Stuttgart version and the larger Roman version. The production of VR was entrusted to the Benedictine order, and in particular to the supervision of Dom H. Quentin, at different times president of the order and abbot of the 53 He designates it a pael, however; J. Levy, Chaldäisches Wörterbuch über die Targumim und einen grossen Theil des rabbinischen Schriftthums, 3d ed. (Leipzig: G. Engel, 1866; reprint, Cologne: Joseph Melzer Verlag, 1959), s.v. “ , ”. He mistakenly gives the reference to the present verse as 1 Sam 2:13.

61

monastery of St. Jerome. Quentin’s methodology for determining the original text of Jerome is to divide the mss into families. The three great families he discerns in Samuel and Kings are the Hispanic family (C LHX FG TOMB), the Alcuinian family ( RZGVP), and the Theodulfian family ( HSAMG). Three less important groups of mss are the Italian ( BDF), Parisian ( SJM ), and the Ambrosian ( AB) groups. Other mss have mixed texts, but D shares many readings with the Alcuinian family, and E shares readings with the Theodulfian family. The two oldest complete mss, R and A, belong to none of these groups,54 having a text that is often reflected in one or more of them but that is more primitive. When these two mss agree, Quentin says, they manifest Jerome’s text. When they disagree, an agreement with another old ms, C of the Hispanic group, provides Jerome’s text.55 Roberts criticizes this approach as too mechanical, but he notes that Quentin himself and the Benedictines who continued his work do not use this method without critical acumen.56 Because of this criticism, it will be important to compare the text of the Roman edition with that of the Stuttgart edition. When citing mss from a family or subfamily in which each ms has a common Greek letter and different superscripts, the agreement of all the mss with a certain reading will be indicated by giving merely the Greek letter (e.g., = AB). None of the fragmentary mss cited in VR has any readings in 1 Samuel 3. Printed editions are cited as lower case italic characters (i.e., agrelvwsc). Finally, patristic sources are occasionally noted, as follows: Gothic Breviary (Brev.goth.), Gregory the Great (Greg.M.), Isidore (Isid.), Bede (Beda). 3:3

lucerna RA C L* F K* wc] lucernam { Lmg HX G B D E Brev.goth. Greg.M. Isid. Beda

PHK mgI};

Most mss and many patristic sources prefer the accusative lucernam to the nominative lucerna, thereby restructuring verses 2 and 3. The deviation is clearly an innerLatin one, perhaps caused by confusion over whether lucerna had a line, representing m, over the a ( ). Another possible explanation for the variation is that scribes might have considered the construction with the nominative to be difficult and so substituted the accusative. This latter reason supports the nominative as the more difficult reading, as it seems unlikely that a scribe who had the accusative in his exemplar would have changed it to a nominative. Thus, though the preponderance of mss have the accusative, the nominative read by the early mss RAC and a few others is to be preferred. 54 Ferdinand Deist suggests that A may be associated with the work of Cassidorus, mentioned

above, p. 34; Deist, Text of OT, 212. 55 Biblia Sacra Romana, 5:xiv-xv. 56 B. J. Roberts, OT Text and Versions, 261-62.

62 3:4

Samuhel {RA CLXPSB DF JHSAG C W GA PHKI}] Samu[h]elem EJM GB

3:6

Samuhel 1° {RA CLXPSB DF JHSAG C W GA PHKI}] Samu[h]elem EJM GB relvsc

3:8

Samuhel {RA CLXPSB DF JHSAG C W GA PHKI}] Samu[h]elem EJM GB relvsc

3:9

ad Samuhel {RA CLXPSB DFRZGP JHSAG C W PHKI}] Samu[h]elem FV EJM G K relvsc

3:11

ad Samuhel {RA CLXP DFRZGV*P JHSAG C WSJM PHKI}] Samu[h]elem SB FV2 EJM WM G relvsc

Several mss and most editions treat Samuhel as a declinable proper noun, whereas the other witnesses consider it to be indeclinable (or at least they do not distinguish between nominative and accusative).57 Though the declinability of the word is an inner-Latin matter, the syntactic use of the indeclinable form is questionable in the first three units of variation. Whereas Samuhel in verses 9 and 11 are clearly accusative objects of the preposition ad (and thus the phrase ad Samuhel functions as the equivalent of a dative of indirect object), the same word in verses 4, 6, and 8 could be either accusative or vocative. Verse 10, where V reads Samuhel Samuhel as vocatives, must also be considered. MT in verse 4 reads . In verses 6 and 8 is ambiguous, since one would expect the direct object to be preceded by or a preposition;58 it is possible that is vocative.59 in verse 10 are certainly both vocative. LXX clearly assumes a vocative in verses 4 and 6 and a direct object in verse 8 (most mss do not include in verse 10; see above, p. 48). T agrees with MT in all four cases. Only P consistently and clearly renders the word as a direct object in the first three cases, but in verse 10 it reads a double vocative. When the context of V is examined, it seems probable that verse 8 should be taken as a direct object, but verses 4 and 6 could be vocatives. It is interesting that the editions relvsc do not have a variant Samuelem in verse 4 (or, of course, in verse 10). In conclusion, then, it is certain that the indeclinable form of the name is to be preferred in all the units of variation, but the function that Samuhel plays within verses 4 and 6 is debatable. 3:5

et dixit {RA C LH FG TB D RZGVP E H*SAMG* B

JM AB P mg}]

+ ei

57 Most mss are inconsistent in their use of case endings with Samuhel. Cf. v. 16, where almost

all mss (except D) read Samuhelem, and v. 21, where Samuheli and Samuhelis are both found. See Stummer, “Einige Beobachtungen über die Arbeitsweise des Hieronymus bei der Übersetzung des Alten Testaments aus der Hebraica Veritas,” Biblica 10 (1929): 1-30. 58 in Samuel and Kings is usually followed by , , , or , but cf. 1 Kings 22:13; 2 Sam 5:20; 1 Sam 9:24 (though this last passage may be corrupt). 59 Kautzsch, ed., Gesenius' Hebrew Grammar, § 115b, says, “The object after the infinitive

construct must also always be regarded as in the accusative, even when it is not expressly introduced . . . by the nota accusativi - .” This is indeed the case in 1 Kings 22:13. However, a verb like introducing a vocative or direct discourse does not seem to have been considered (cf. 2 Kings 11:14 for an example of direct discourse after infinitive construct).

63 X OM 3:6

H1 G2

dixit {A C X

P*HKI

DF

BD E

S

PHKI}] + ei R

It was a common scribal practice to provide more details when the context might be ambiguous (and often when it was not). Several mss in the first case, and one the of most important mss in the second case, specify to whom Samuel was speaking. However, these additions are clearly inner-Latin and secondary. 3:5

revertere {RA C LH* T* D RZGV H*SAMG BDF* Jmg M P*K}] + et H2 X T2 OMB P E H2 P 2HI F 2 SJ* agrelvsc; Greg.M. Brev.goth.

3:5

et 3° {R C LHX T* D RZGVP E Hmg AMG BDF T2 OMB H*S

SJM AB PHKI}]

>A

The addition and omission of conjunctions in V, as in other versions, was a common scribal practice, and external considerations must of necessity play a larger role in determining the original reading. Nevertheless, internal considerations do have a role to play. In the first unit of variation, a number of mss have a conjunction between the two imperatives. The addition of a conjunction at this point of the text is also found in several LXX mss, and, significantly, the it. However, the absence of the conjunction in any of the most ancient witnesses to V suggests that the presence of the conjunction is secondary, perhaps introduced on the basis of it texts; alternatively, it could have been added for stylistic reasons. The absence of the conjunction in the second unit of variation is not supported by as many witnesses as in the first case, but the important ms A supports the variant, as do important Hispanic and Theodulfian mss. However, no witnesses in any other version support the second variant. It is possible that the omission of the conjunction indicates that, in the mind of at least one scribe, abiit et dormivit properly belonged to the following phrase et adiecit dominus vocare rursum Samuhel. Since the preponderance of external evidence points to the first variant, and since no evidence is compelling enough to suggest preferring the second to the first, the first variant should be accepted. 3:9

ait 1° {RA C B E

PHKI*}] dixit X

D I2

; Greg.M.

These two variants are equivalent, and both aio and dico commonly render . In the first two accounts of God’s calling Samuel (verses 4 and 5), after Samuel runs to Eli, dixit is used. It is possible that later scribes, influenced by these similar passages, changed ait to dixit in order to achieve a greater homogeneity. Otherwise, the change might be considered an unconscious lexical substitution, influenced, no doubt, by the earlier passages. 3:10

loquere R*A C

G2

rw] + domine R2

G* {

X

B D E HSAM

PHKI

64 agelvsc}; Greg.M. The second variant agrees with LXXL, arm, and P. More importantly, it reflects Samuel actually saying what Eli instructed him to say in verse 9. The fact that Samuel did not say exactly what he was told to say in many witnesses (including MT, LXX, it, and T) caused the scribes some consternation, and some of them remedied the situation by adding “the Lord.” The question here is whether or not domine was original with Jerome. The age and importance of mss RAC supporting the first variant outweigh the numerical superiority of the second variant. Furthermore, the intrinsic probability of adding domine to an originally shorter text is high, while the probability of dropping an original domine is low. Though extant it mss agree with MT, their propensity for agreeing with LXXL suggests the possibility that other it mss no longer extant did add domine. If so, the second variant might be a correction to this text. Otherwise, it is an independent “improvement” of the text. 3:15

timebat indicare visionem {A C X S | 213 TO

D E HAMG

PHKI}] 132 R MB

The variable in these variants is the position of indicare. Should it be second as in variant one, third as in variant two, or first as in variant three? The other secondary witnesses agree with the first variant and with MT that indicare should be second in the phrase. The origin of the other variants may be a ms of V that inadvertently omitted indicare when the text was first written. The same scribe or a later one, noticing the mistake, added the word in the margin next to its proper place. Later copyists, however, unable to discern the exact location (although the infinitive in Hebrew generally follows the verb, Latin has no such constraint), inserted it in a variety of places. Notwithstanding the testimony of R, the first variant should be accepted as the preferred reading on the basis of superior external attestation. 3:17

ad te RA C D RZGVP2 M2 G2 P*; Goth.Brev.] dominus ad te X DF aelvsc; Greg.M. | ad te dominus P* E2 HSAM*G* B

B P2HKI

The problem with the first variant is that the subject is unspecified, either in the verse or in the immediately surrounding verses. The implied subject in variant one, in fact, is found in verse 11, six verses earlier. Variants two and three have no such problem, since the subject (dominus) is explicit. MT here also omits the subject, as does LXX, but with an important exception. Whereas the pointing in MT indicates an active verb with an implied subject, LXX has a passive verb, . The passive verb, of course, takes no subject, and so is no problem. LXXL, apparently still troubled by the lack of specificity, reads . P addresses the problem in a manner similar to variant three, . It is

65

unlikely that any direct connection exists between the rendering in P and that in mss of V, but the similar solution does indicate the manner in which scribes in various traditions addressed what they saw as textual difficulties. The variant position of dominus in variants two and three suggests that the word might be secondary to the text, as does the observation that the first variant is the most difficult reading. The fact that mss RAC all agree in the omission, as does the it ms Belsheim, sets the external evidence on the side of the first variant as well. Thus, the first variant should be accepted as original. Summary of Deviations from the Base Text No deviations from the text found in VR (which agrees completely with VS in the chapter) are considered original in chapter three. Aquila When presenting the readings of Aquila and the other partial secondary witnesses, the reading of LXX (or another complete secondary witness, in the case of other possible Hebrew readings) will be given as the collating base for identification purposes. Contested Readings 3:3

´3 witnesses |

] ´1 witness

The reading is reflected in three witnesses, while is found in only one. This latter variant, since it combines two essentially equivalent phrases, seems to be a conflation of ´ and another Greek reading related to ´. The first of these readings, then, should be preferred as the authentic reading of ´. Accepted Readings 1 2 3 4 5 6

3:1

]

´ ]

3:3 3:10 3:13 3:21

] pr

´

´ ]

´ ]

]

´ ´

Symmachus Contested Readings There are no contested readings in the chapter, aside from minor orthographical differences among mss.

66 Accepted Readings 1 2 3 4 5 6 7 8

3:1

]

´ ]

3:3 3:4 3:7

´ ]

]

´

´ ]

´ ]

3:8 3:21

] ]

´ ´ ´

Theodotion Contested Readings There are no contested readings in the chapter. Accepted Readings 1 2 3

3:8 3:13 3:21

Other Readings Attributed to

]

´ ]

]

´ ´

´

No evaluation of the readings will be attempted, since it is not clear that different readings in fact represent the same tradition. Purely orthographical variants will be omitted. To avoid confusion, the abbreviation ´ will be shortened to ´ in the collation. 1 3:1 ] ´ 2 3:2 ] ´ 3 ] ´ 4 3:3 ] ´ 5 ] ´ 6 ] ´ 7 3:4 ] ´ 8 ] ´ 9 3:6 ] ´ 10 ] ´ 11 ] ´ 12 ]+ ´ 13 ]+ ´ 14 3:10 ] ´

67

15 16 17 18 19 20 21 22 23

]

´

3:13

] ] ]

´ ´ ´

] ] 3:14 3:18 3:21

´ ´

] ]

´ ´ ...

] ´

The Lucianic Recension Lagarde’s edition of the Lucianic text has been criticized for various reasons, not least because he failed to include a critical apparatus with his text.60 Furthermore, he sometimes accepted the reading of one Lucianic witness against the combined testimony of the other three. For these reasons, Lagarde’s edition will be used, but all of his readings will not be accepted as being the pure Lucianic text. However, all of Lagarde’s readings that are rejected will be listed in a separate section. Studies on the Lucianic text have shown that the purest witnesses to that form of the text are boc2 e2 . The witness referred to as b is in fact the consensus of b´ and b. Of course, the Lucianic mss frequently agree with other groups of mss, but some individual mss occasionally agree with the main Lucianic mss on occasions in which all others differ. The ms which agrees the most often when others do not is z. In the collation below, the reading of LXX, as determined above, is given first, followed by the reading accepted as Lucianic.61 The Lucianic mss that support the reading will be listed, as will any mss or versions that agree uniquely with LXXL or are joined by only one more witness. If more than five other witnesses agree with the Lucianic reading against LXX, this fact will be indicated by the word “many” in parentheses. The agreement of from three to five mss with LXXL against LXX will be indicated by “few” in parentheses. It should be noted that any Lucianic reading that has no non-Lucianic witnesses listed in the collation as agreeing with it is a uniquely Lucianic reading.62 60 Jellicoe, LXX and Modern Study, 7-9; Alfred Rahlfs, Septuaginta-Studien, 2d ed., vols. 1-3 (Göttingen: Vandenhoeck & Ruprecht, 1965), 3:29. 61 The Lucianic reading will be determined primarily on external grounds, by counting the number

of Lucianic mss that support the reading and taking the majority reading. If the witnesses are split two to two, the reading that differs from the main LXX will be accepted. 62 That is, no other LXX witnesses or daughter versions of LXX support the reading. A unique Lucianic reading may be supported by a witness in another tradition (e.g., P) or by one or more of the minor Greek witnesses.

68 Readings Accepted as Lucianic 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

3:1 3:3

3:4 3:5 3:6 3:7

3:8 3:9 3:10 3:11 3:12 3:13

3:14

3:15

3:16 3:17

]

] boc2 e2 ; Thdt ]+ boc2 e2 (many) 2°] pr boc2 e2 z; arm 1°] pr boc2 e2 (many) 3°] boe2 z ]+ boc2 e2 (many) ]+ boc2 e2 (many) ]+ boc2 e2 (many) ] pr boc2 e2 (many) ] pr boc2 e2 (many) ] boc2 e2 ]+ boc2 e2 Ng ]+ boc2 e2 ]+ boc2 e2 (many) ]+ boc2 e2 (many) ] boc2 e2 ]+ bc2 e2 ] pr boc2 e2 (many) ] boc2 e2 ; eth ] boc2 e2 ] boc2 e2 z ] pr boc2 e2 (many) 2°] + bc2 e2 z ] boc2 e2 (many) ] boc2 e2 ] boc2 e2 ] pr boc2 e2 ] pr boc2 e2 (many) 2°] post boc2 e2 ] boc2 e2 (many) 2°] > boc2 e2 (many) ]+ boc2 e2 z ]+ boc2 e2 z ] post

36

bc2

3:18

1° boc2 e2 ]+ boc2 e2 (many)

69

37 38 39 40 41 42 43 44 45 46 47 48 49

1°] + boc2 e2 z ] boc2 e2 (few) ] pr bc2 e2 3°] > boc2 e2 ]+ boc2 e2 z ] boc2 e2 A ] boc2 e2 Mmg ] boc2 e2 (many) ] boc2 e2 za? 2°] > boc2 e2 ] ] boc2 e2 h (many read 1°] + boc2 e2 gzmg

3:19

3:20 3:21

boc2 e2 (many) )

Readings from Lagarde’s Edition Not Considered to Be True Lucianic Readings 3:2 3:2 3:3 3:7 3:8 3:13 3:13 3:20

b zmg o (many)

] ] ]

(no LXX witnesses)

]+ ]

b; it b Mmg ] b (many) ] pr bo (many) ] idem b´ (many)63

The Hexaplaric Recension The hexaplaric text-tradition arose historically out of Origen’s monumental work in creating the Hexapla. In the fifth column of this work, he included the text of LXX common in his day, but he made some changes. When he found a section in LXX that was not reflected in his Hebrew text in column one, he marked the section with an obelus (÷). When he found a section in the Hebrew not reflected in his text of LXX, he added it, usually from one of the minor Greek versions, and marked the section with an asterisk (Ë). The hexaplaric text arose when scribes copied the fifth column of Origen’s work, either ignoring or not paying careful attention to the asterisks and obeli. Even if the original 63 The reconstructed Lucianic text used in this study may be found in Appendix 1, below, p. ***. Except for minor matters such as orthography, punctuation, versification, and the use of square brackets, it agrees with the Lucianic text reconstructed by Bernard Taylor, The Lucianic Manuscripts of 1 Reigns, vol 1: Majority Text, Harvard Semitic Monographs, no. 50 (Atlanta: Scholars Press, 1992), 9-12, in all respects except one: the text reconstructed in the present study reads in v. 13, whereas Taylor’s text reads [ ] .

70

copyists were faithful to preserve the text-critical marks, later copyists were not. The result was the creation of a hybrid text, based on LXX, but supplemented from other Greek versions closer to the proto-Masoretic text. The importance of hexaplaric readings for textual criticism is two-fold. First, its renderings that vary from OG (especially the pluses) often reflect readings of one or more of the minor Greek versions, which are only preserved in fragments and glosses scattered over several mss. Second, and more important for this study, any renderings that differ from both OG and MT might reflect a correction to a Hebrew text somewhat different than MT. Thus, it is only this latter category of readings that will be considered here. Determining the hexaplaric text is problematic, since most mss and versions contain at least some hexaplaric readings. However, the studies of Bo Johnson help to isolate those witnesses that are particularly likely to preserve hexaplaric readings. In particular, Johnson concludes that the Greek mss Acx and, to a lesser extent, family d+ are the best representatives of the hexaplaric text in 1 Samuel.64 It is important to note that, since the hexaplaric readings are being compared with MT as well as LXX, a reading which reflects only variation from LXX will not be listed, even if it might be the authentic hexaplaric Greek text. This limitation is legitimate in light of the fact that the hexaplaric text-tradition is a partial secondary witness rather than a complete secondary witness. Two other important witnesses to the hexaplaric text in 1 Samuel are arm and syh. 65 Those hexaplaric readings that will be considered in this study are listed below, collated in the same manner as the Lucianic text was above. Accepted Hexaplaric Readings Differing from LXX and MT 1 2 3 4 5 6

3:3 3:8 3:10 3:13

]

Ac ] pr d+ f+; (arm)66 ]+ cx b+; arm (few) ] A; arm (cx b; cop read ] pr A d+ (many) ] cx; armvid itb

)

64 Johnson, Hexaplarische Rezension, 88. 65 Ibid., 88-89. Syh is a translation of the fifth column of the Hexapla, including the diacritical

marks. The textual history of the Armenian version is somewhat complex, having gone through three major stages: translation from Syriac, translation from Greek, and translation from another Greek texttradition. The last stage, when it was translated from a hexaplaric Greek text, is the most evident in the present Armenian text, but remnants of the earlier history of the version remain. These can be observed in agreements between arm and P and between arm and non-hexaplaric Greek texts. See Bo Johnson, Die armenische Bibelübersetzung als hexaplarischer Zeuge im 1. Samuelbuch, Coniectanea Biblica, Old Testament Series, no. 2 (Lund: CWK Gleerup, 1968), 13-17, 158-60. 66 Acx reads instead of the of d+ f+, almost certainly an inner-Greek corruption rather than a rendering of a Hebrew or . Arm, along with b2 and it v , places the equivalent of before the infinitive; cf. Johnson, Armenische Bibelübersetzung, 86.

71

7 8 9 10 11

3:14 3:15 3:21

2°] > cx d+; it ] cx 2°] > cx d ] A d+; syhj eth itb (c omits phrase, x reads ) 3°] > Acx; arm (many)

Rejected Readings from Hexaplaric Witnesses The following list contains those readings that are shared by at least two Greek hexaplaric mss but are not considered the true hexaplaric text. Other instances of witnesses that are often hexaplaric but that have unique or almost unique readings (especially readings found in arm and no other hexaplaric witness) are numerous but are not listed. 3:8 ] cx 3:9 ] cx (many) ]+ cx 3:11 ] pr cx 3:14 ] cx ] cx Ne2 * 3:17 1°] cx b´ 3:21 ] cx fm All these readings are unique to cx, among Hexaplaric witnesses. In each case, the reading of cx is an inner-Greek development from the main LXX text and so is textually inferior. These readings will not be considered further. 3:17 ] > cx b2 e2 ; ethvid The omission of here is probably the result of parablepsis, so this reading should not be considered an authentic hexaplaric reading. Other Possible Hebrew Readings In this final section, other readings from witnesses that might possibly reflect a Hebrew Vorlage different from the reading of MT are given. Only those variants are listed that are not reflected in any other secondary or partial secondary witness. 1 3:2 ] Nadenopqtyzb2 2 3:5 ]+ 6h1 7h12 7k3 9a1fam 9l1 3 3:9 ] a

CHAPTER 3 GRAMMATICAL/STYLISTIC CHARACTERISTICS OF THE SECONDARY WITNESSES Having determined the texts to be used in this study, the stage is now set to analyze the secondary witnesses and compare them with MT. The procedure to be followed is described above in Chapter 1, but it may be summarized briefly at this point. Simply put, the method starts with the mass of readings in the secondary witness that might possibly be considered variants and proceeds to weed out those that have no value for textual criticism, a sieve of Eratosthenes approach. The first readings to be eliminated are those which were likely forced by the nature of the differences between the source language (Hebrew) and the target language in question. Since the limitations of the target languages for rendering Hebrew are discussed at the beginning of the analysis of each version, these readings are never even listed. Those readings that are considered are listed under the category of variation that they represent. Next, a summary of the comparison of the secondary witness with MT is given, with specific statistics for each of Tov’s four categories and many sub-categories within these four. Armed with these findings, a preliminary partial translation technique is described, and significant and nonsignificant variants will be distinguished. Once the variants that are probably reflections of the translator’s rendering of the Hebrew have been determined, the partial translation technique will be recalculated for each category and subcategory, eliminating the significant variants from consideration in order to refine the translation technique. The rationale behind eliminating significant rather than nonsignificant variants is as follows (described in terms of LXX). The preliminary partial translation technique, that is, the one based on MT, is a comparison of the deviations in LXX from MT, which, as described above, is used initially as though it were the Vorlage that lay before the translators. Since it is not identical to their Vorlage, the deviation of LXX from MT will be greater than that between LXX and its actual Vorlage. The variants that are determined to be significant are those that probably reflect a difference in the Vorlage, and, although they do not agree with MT, they presumably agree with their Vorlage, and so should be not counted in the statistics as deviations from a literal translation technique. Whereas the partial translation technique based solely on MT will imply a disproportionately high degree of variation in the translation, the revised partial translation technique will imply a

72

73

disproportionately low degree of variation, since some of the variants considered significant at this stage will be eliminated later in Chapter 4. However, most of the nonsignificant variants will be identified in the present chapter, so the partial translation technique for each secondary version at the end of Chapter 3 should be reasonably close to the final translation technique as determined in Chapter 4. It may occasionally be necessary to repeat the process of recognizing significant variants and recalculating the translation technique for a particular category if the preliminary partial translation technique was skewed a large amount because of significant variants. Septuagint Limitations of Greek for Rendering Hebrew Since Greek is a member of that group of languages known as the Indo-European family of languages, whereas Hebrew is a Semitic language from the larger Afro-Asiatic family,1 it is not surprising to find that syntactic structure in the two is different. Nevertheless, Greek translators of Hebrew texts were capable of rendering almost every detail of their Hebrew Vorlagen, if they so chose. Different translators varied in their degree of conformity to Hebrew style, with Aquila being the most slavishly literal, even rendering by . The nominal system in Greek is a highly inflected one, employing five cases (nominative, genitive,2 dative,3 accusative, and vocative), three genders (masculine, feminine, and neuter), and three numbers (singular, plural, and dual). By contrast, biblical Hebrew relies primarily on word order and the use of prepositions to express the nuances present in the Greek cases,4 and it has only two genders (masculine and feminine), but it does have three numbers. Since the association of gender with inanimate objects, abstract ideas, and so on, is largely arbitrary (as far as can be determined now),5 any significant correlation between the Hebrew gender of a word and the gender of its Greek counterpart 1 Also known as the Hamito-Semitic family; Encyclopædia Britannica, “Languages of the World,”

740. 2 Includes ablative uses. 3 Includes locative and instrumental uses. 4 The morphological changes that occur in nouns in the construct state reflect phonetic and

rhythmical phenomena rather than remnants of case endings. Occasional nominative, genitive, and accusative endings on nouns support a picture of an earlier form of the Hebrew language that had three distinct case endings, like its Proto-Semitic ancestor. See Kautzsch, ed., Gesenius’ Hebrew Grammar, §§ 89-90; Waltke and O’Connor, Introduction, 127-28; Gotthelf Bersträsser, Einführung in die semitischen Sprachen, with an appendix, “Zur Syntax der Sprache von Ugarit,” by Carl Brockelmann (Munich: Max Hueber, 1928), 14-15. 5 For a comparison between the use of gender in Hebrew and in other languages, see Waltke and O’Connor, Introduction, 95-110.

74

seems unlikely, and the data from 1 Samuel supports this supposition. On the other hand, it would have been natural for the translators to have rendered the number of the Hebrew nouns by their Greek equivalents. Not surprisingly, the translators did do so for the most part, with the exception that, since the dual was no longer used in koine Greek, they expressed Hebrew duals with Greek plurals.6 The options open to the Greek translators regarding the use of Greek cases to express various Hebrew constructions were much broader than those concerning the rendering of gender or number. Greek translators, even those concerned with being literal, had the option of rendering prepositional phrases with nouns in the appropriate case rather than rendering both the preposition and the noun.7 Both methods of translation could be considered literal, though the translators themselves might have thought that rendering Hebrew prepositions with Greek ones was a more faithful rendering. However, the multiplicity of available options in Greek for rendering these Hebrew constructions requires that the translation technique be examined from several different angles to determine the translators’ own approach to translating them; labeling the translation as “literal” is not sufficient. Another aspect of the nominal system in Hebrew is the (definite) article. Greek also has an article, so one-to-one correspondence was possible in translation. However, the use of the article in idiomatic Greek often varied from what would be required of a strict one-toone rendering. For example, Greek articles could stand for relative pronouns or with infinitives, something that was not possible in Hebrew. Moreover, two barriers to a strict representation of the Hebrew article by the Greek article existed. The first was the failure of the Hebrew to use an article to identify definite nouns in the construct case; the second was the assimilation of the Hebrew article after an inseparable preposition. Whether the Greek translators would render the articles that they could see in the text or whether they would insert Greek articles for definite Hebrew nouns in which the article did not appear— or whether they would simply be inconsistent—is a matter for investigation. Many differences also exist between the verbal systems of the Hebrew and Greek languages. The Hebrew verb can be classified by stem, inflection, person, gender, and number. Greeks verbs have tense, voice, mood, person, and number. A correlation clearly exists between person and number,8 and it is equally clear that Greek will have to render both masculine and feminine Hebrew forms by one common form, since gender is not indicated in Greek verbs. More difficult is the relationship between stem and inflection in Hebrew and tense, voice, and mood in Greek. 6 Conybeare and Stock, Grammar, 25. 7 For example, with indirect object could be expressed in Greek simply by the dative case, and with a noun could be rendered by a Greek noun in the dative (locative) case. 8 A dual number for verbs does not exist in either Hebrew or Greek.

75

The combination of Greek tense and mood corresponds fairly closely in translation to Hebrew inflection. The debate over the exact meanings of the classical Hebrew perfect and imperfect have raged for years, and no universal consensus has yet been reached. However, the Greek translators often rendered the simple Hebrew perfect with a past tense (aorist, perfect, imperfect,9 or pluperfect) and the simple Hebrew imperfect with a present or future tense (present, future, or future perfect), and least when rendering the verb with the Greek indicative mood. The addition of the waw consecutive to the beginning of the Hebrew verb generally resulted in a reversal of the characterization just given in order to make the Greek rendering match the Hebrew intention. This reversal was not required by the Greek language but rather reflects the translators’ understanding of the Hebrew that lay before them. When rendering conditions, possibilities, and wishes, the Greek translators had available to them moods other than the indicative, namely, the subjunctive and optative moods, and they used these moods frequently. However, when the Greek translators used subjunctive or optative moods, the correlation between Greek tense and Hebrew inflection often disappeared: since the time element of the tense no longer mattered in these moods, the translator would generally choose the tense on the basis of its Aktionsart, whether punctiliar (aorist), durative (present), or perfected (perfect).10 One Greek mood which does have an almost exact parallel in Hebrew is the imperative, and Hebrew imperatives were regularly rendered by Greek imperatives, though the Greek tense could vary. Hebrew stems may be divided into three groups: simple stems (qal, niphal), intensive stems (piel, pual, hithpael), and causative stems (hiphil, hophal).11 These stems may also be classified according to type of action (similar to voice in Greek): active stems (qal, piel, hiphil), passive stems (niphal, pual, hophal), and reflexive stems (niphal, hithpael).12 There is no Greek equivalent to the intensive and causative stems, though the 9 A terminological difficulty exists in comparing Greek or Latin to Hebrew, since the Hebrew

imperfect, often reflecting present or future time, functions quite differently from the Greek and Latin imperfects, which reflect past time. Though other terms are available for the Hebrew inflections—notably the suffix and prefix conjugations, referring to the perfect and the imperfect, respectively—perfect and imperfect are still the most commonly used. Furthermore, the term “conjugation” itself presents terminological difficulties, since Greek and Latin conjugations are merely morphological categories, whereas the two Hebrew conjugations reflect semantic differences. It may be best to refer to the inflections simply as the qtl and yqtl inflections, as is sometimes done. See Waltke and O’Connor, Introduction, 45558. 10 A. T. Robertson, A Grammar of the Greek New Testament in the Light of Historical Research, 4th ed. (Nashville: Broadman, 1934), 824. 11 Hebrew also has various other stems (poel, pilpal, etc.) which are related to these primary

stems. For a discussion of the relation of the Hebrew stems to the proto-Semitic language, see Hans Bauer, Pontus Leander, and Paul Kahle, Historische Grammatik des hebräischen Sprache des Alten Testaments, vol. 1 (Tübingen: Max Niemeyer, 1922; reprint, Hildesheim: Georg Olms, 1962), 279-88. The names of the three groups is traditional and not always particularly helpful, especially in the case of the name “intensive.” For a discussion of the different uses of the various stems, see ibid., 289-94. 12 The niphal was originally strictly a reflexive stem, but it came to be used also as the passive of the qal, replacing an earlier qal passive, of which only remnants remain in the Hebrew Bible (esp. the qal

76

LXX translators did occasionally render Greek verbs as though they were causative.13 A greater correlation exists between the type of action reflected in the Hebrew stem and voice of the verb. Hebrew active and passive stems are generally rendered by the Greek active and passive voices.14 The correlation between Hebrew reflexive stems and the Greek middle voice is much smaller, if not nonexistent. The reasons for this lack of correlation are many. First, the niphal is often used with a purely passive meaning. Second, the meaning of the Greek verb used to render the Hebrew in the reflexive stem may not correspond in the middle voice to the meaning required by the Hebrew. Next, many Greek verbs, whether deponent or not, have what appears to be an active meaning expressed by the middle voice.15 When these verbs are used in the middle voice to translate Hebrew verbs, it is unlikely that a non-active Hebrew stem can be assumed. Finally, in later Greek a blending of the middle and passive voices occurred, so that middle verbs were sometimes used with passive meaning, and vice versa.16 The last verbal forms to be considered, infinitives and participles, exist in both languages, and a precise rendering in Greek of these two forms was possible. The fairly common use of the participle as equivalent to a full verb in Hebrew was rare or perhaps unknown in classical Greek. However, the use of the independent Greek participle for the verb was a permissible, if still uncommon, construction in koine, as demonstrated by the papyri and the New Testament.17 Undoubtedly, many of the instances of independent Greek participles used as verbs in LXX were based on a similar construction in the Hebrew Vorlage. Nevertheless, some of the translators seem to have preferred rendering independent Hebrew participles used as verbs by Greek verbs rather than participles, particularly when such a rendering could be supported by the consonantal text. Thus, the rendering of a Hebrew participle by a Greek verb in certain instances cannot be considered indicative of a differing Hebrew Vorlage, though, of course, it does not rule it out. Greek has only one infinitive whereby to render the Hebrew infinitive absolute and infinitive construct. Moreover, the common addition of the preposition to the Hebrew infinitive construct is not rendered by a Greek preposition, which would be unidiomatic. It is possible that some translators may have rendered the by the genitive of the article with passive participle); cf. C. L. Seow, A Grammar for Biblical Hebrew (Nashville: Abingdon Press, 1987), 250; for a more complete discussion see Bauer, Leander, and Kahle, Grammatik, 285-88. 13 Conybeare and Stock, Grammar, 76-77. 14 Except when the Greek verb used did not have the required form, so substituted another. See

Smyth, Grammar, 218-19. 15 A. T. Robertson, Grammar, 811-13. 16 Conybeare and Stock, Grammar, 75-76; A. T. Robertson, Grammar, 333-34. For other

examples of the use of one voice with the meaning of another in Greek, see Smyth, Grammar, 219-24. 17 A. T. Robertson, Grammar, 944-46. Cf. also Conybeare and Stock, Grammar, 74.

77

an infinitive of purpose. This rendering, however, is not consistent.18 This brief discussion of the similarities and differences between Hebrew and Greek is not complete, and other matters concerning the rendering of the Hebrew text by the translators of LXX are discussed in longer works and articles.19 Partial Translation Technique Based on an analysis of the Greek language in comparison with Hebrew, the following are potentially significant variants in LXX that need further scrutiny. Add-Oms Since an analysis of the first three categories of variants will not include add-oms, they must first be identified. Add-oms, as mentioned above, are variants in which the reading of the base text (MT) is either longer or shorter by at least one semantic unit20 than the corresponding section in the translation (in this case, LXX). In general, the list of addoms will be almost the same as the list of quantitative variants (category four), but in each language exceptions may be made, and certain quantitative variants may not be classified as add-oms; in other words, these exceptional cases would be analyzed in the first three categories of variants. It is sometimes difficult to determine whether a difference between the Hebrew and the Greek should be classified as an add-om or not. In particular, a decision must be made concerning Hebrew particles ( ) and prepositions (especially and ) that are not rendered in LXX in a particular passage. Such particles and prepositions will be included in the add-om list if the Hebrew word in question is usually rendered by some equivalent Greek word. The same consideration will apply as well to the other versions analyzed. In LXX, the omission of , , or will not be considered add-oms, so they will be taken into account when analyzing both representation of Hebrew lexemes by Greek lexemes and quantitative representation. A full list of the add-oms of LXX is given in Appendix 2; it may be compared with the list of quantitative variants given later in this section. 18 Conybeare and Stock, Grammar, 58-59. Cf. the tables in Ilmari Soisalon-Soininen, Die

Infinitive in der Septuaginta, Annales academiæ scientiarum fennicæ, no. 132, 1 (Helsinki: Suomalainen Tiedeakatemia, 1965), 180-85. 19 See especially Conybeare and Stock, Grammar, 25-97; Thackeray, Grammar; Jellicoe, LXX and Modern Study, 314-37; and works listed in the Bibliography by Anneli Aejmelaeus, Francis I. Andersen, Albert Debrunner, Kenneth James Dover, H. S. Gehman, Martin Johannessohn, Max Leopold Margolis, E. Nestle, Alfred Rahlfs, Ilmari Soisalon-Soininen, Raija Sollamo, and Emanuel Tov. 20 A semantic unit is a phrase, word, or part of a word that represents a single concept. Of course,

the most common semantic units are words, but the Hebrew pronominal suffixes and inseparable prepositions are also semantic units, as is the Greek compound .

78 Consistency The first criterion that Tov lays out for evaluating the translation technique of a version is consistency. Here, those variants in LXX that involve consistent rendering in comparison with MT are listed. The columns entitled “LXX” and “MT” describe the nature of difference between the LXX reading and MT. For example, if LXX has a full verb where MT has a participle, the respective columns will read “verb” and “participle.” If the LXX rendering reflects a lexical choice other than the main rendering (if one exists), the LXX column will read “lexeme,” and the MT column will be empty. Since MT does have a tradition of interpretation represented by the vowel points, the complete Masoretic understanding of the consonants will be the initial basis for comparison. However, when variants come to be eliminated, those that reflect a possible alternate understanding of the consonants of MT will also be eliminated. Three common forms that would often be ambiguous without vowel points are (1) words that could be considered participles, infinitives, or various forms of the full verb; (2) nouns with inseparable prepositions, which may or may not include an article that has been assimilated; and (3) verbs with a waw prefix that could be either waw consecutive or waw conjunctive. Also, it should be noted that the kethib form of MT is used as the basis for evaluation, though qere forms will be discussed later. Each variant is numbered individually for reference in later discussions. When more than one variation is associated with a particular word, each one has its own reference number. Ref Variation Septuagint Masoretic Text 1 3:1 periphrasis participle 2 periphrasis participle 3 active niphal 4 3:2 verb participle 5 plural singular 6 lexeme 7 infinitive adjective 8 imperfect imperfect 9 3:3 lexeme 10 infinitive verb 11 verb participle 12 3:5 lexeme 13 3:6 verb infinitive 14 3:7 infinitive verb 15 lexeme 16 infinitive verb

79

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

3:8 3:9

verb participle lexeme verb participle lexeme 3:10 verb participle 3:11 verb participle article not definite plural singular plural singular 3:12 lexeme lexeme verb infinitive verb infinitive 3:13 perfect perfect w/c verb participle plural singular lexeme word division 3:14 active niphal plural singular 3:15 lexeme present imperfect w/c adverb noun 3:16 lexeme 3:17 participle verb passive piel plural singular participle verb passive piel plural singular 3:18 lexeme 3:19 active hiphil 3:20 plural singular adjective participle Now that the potential variants have been listed, it is time to examine the level of consistency reflected by various aspects of the LXX rendering. First, lexical consistency (tables 1-3) will be measured by counting the different Hebrew words used more than once

80

and comparing that number with the number of Greek words used more than once.21 Next, the number of Hebrew words (occurring more than once) rendered by the primary Greek rendering will be calculated. This statistic will measure the translator’s loyalty to a particular vocabulary item when an equivalent lexical choice is available. Next, the lexical deviation factor (described in the next paragraph) will be computed. Finally, a comparison of Hebrew and Greek roots, rather than words, will be made, in an effort to see if the translators have a tendency toward etymologizing.22 Table 4 is an analysis of how the translators rendered words in a particular Hebrew word class, namely, verbs and verbals, nouns and adjectives, and pronouns. Tables 5-16 analyze the translators’ tendencies in rendering various grammatical categories and syntactic relationships (i.e., for verbs: inflection, stem, person, and number; for nouns and adjectives: gender, number, use in the sentence, articles, and definiteness; for pronouns: gender, number, and use in the sentence).23 The deviation factor is a measure of the deviation from absolute consistency, whether lexical or grammatical, by a translator. Absolute lexical consistency, for example, is defined as the use of one and only one Greek word for each distinct Hebrew word. The phenomenon of using the same Greek word for more than Hebrew word is not considered in the calculations. The algorithm for computing the deviation factor is given in detail in Appendix 4, but its rationale is given here, described in terms of lexical consistency. If a single Hebrew word is always rendered by the same Greek word, the deviation factor for that Hebrew word is zero. If more than one Greek word is used to render a single Hebrew word, the deviation factor is a positive number. Given ten occurrences of a given Hebrew word, it is considered a greater deviation from absolute consistency for three Greek words to be employed in rendering that word than for two to be so employed (all other things being equal). Furthermore, if two Greek words are used to render a single Hebrew word, the deviation factor will be greater if each Greek word appears five times than if one appears nine times and the other only once. It is considered a greater deviation when one case out of five differs from the standard than when one out of ten does. The deviation factor is a modification of a chi-square test, adapted to account for the number of discrete renderings in the target language as well as the total number of deviations from the most 21 The number of Greek words will always be greater than or equal to the number of Hebrew

words, for if a Greek word is used to render more than one Hebrew word, it is counted more than once. 22 Tov identifies the technique of etymologizing, the rendering of all words based on a single Hebrew root (real or imagined) by words based on a single Greek root, as an aspect of consistency calling for special attention; Tov, Text-Critical Use, 57. In table 1, the last column identifies different words that share the same root by assigning a common reference number, given in italics. Roots represented by only one word in the chapter are not explicitly indicated, though they are of course figured in the data. 23 A shortened form of the tables, containing the results of the various computations, is given

below. For an expanded form of the tables, containing the data on which the computed figures are based, see Appendix 2 (the category number here corresponds to the table number in Appendix 2).

81

commonly used rendering. Larger deviation factors indicate more variation from absolute consistency, but it must be stressed again that the numbers themselves are not helpful until they are compared with either the corresponding deviation factor in another secondary witness or other deviation factors within the same witness. Of course, the larger the statistical base that is being analyzed, the more meaningful will be the results. Table 1.—Verbs, Nouns, and Adjectives: Lexemes Heb words (>1x): 31 Gk words: 42 1.35 Gk words/Heb word deviation factor: 0.38 total Heb (>1x): 132 primary Gk: 117 88.6% of Heb words by primary rendering Heb roots (>1x): 29 Gk roots: 42 1.45 Gk roots/Heb root Table 2.—Adverbs, Prepositions, and Particles: Lexemes Heb advs (>1x): 13 Gk advs: 26 2.00 Gk advs/Heb adv total Heb (>1x): 85 primary Gk: 67 78.8% of Heb advs by primary rendering deviation factor: 2.78 Table 3.—Conjunctions: Lexemes Heb conjs (>1x): 2 Gk conjs: 2 1.00 Gk conjs/Heb conj total Heb (>1x): 66 primary Gk: 66 100% of Heb conjs by primary rendering deviation factor: 0.00 Table 4.—Word Classes percentage of verbs/verbals represented by verbs/verbals: 94/95 = 98.9% percentage of verbs represented by verbs: 71/76 = 93.4% percentage of participles represented by participles: 4/12 = 33.3% percentage of infinitives represented by infinitives: 4/7 = 57.1% percentage of nouns/adjs represented by nouns/adjs: 52/55 = 94.5% percentage of pronouns represented by pronouns: 37/38 = 97.4% Table 5.—Verbs: Hebrew Inflection, Greek Tense and Mood deviation factor (discrete tense/mood combinations): 6.68 deviation factor (grouped): 0.97 deviation factor (grouped, without verbals): 0.85 Table 6.—Verbs: Hebrew Stem, Greek Voice deviation factor: 3.57 Table 7.—Verbs: Person deviation factor: 0.00 Table 8.—Verbs: Number deviation factor: 0.12 Table 9.—Nouns and Adjectives: Gender deviation factor: 21.54

82

Table 10.—Nouns and Adjectives: Number deviation factor: 2.29 Table 11.—Nouns and Adjectives: Use vs. Case deviation factor (discrete use/case combinations): 6.45 deviation factor (grouped): 0.37 Table 12.—Nouns and Adjectives: Articles deviation factor: 11.37 deviation factor (without implied articles): 12.23 Table 13.—Nouns and Adjectives: Hebrew Definiteness, Greek Articles deviation factor: 8.04 Table 14.—Pronouns: Gender deviation factor: 1.44 Table 15.—Pronouns: Number deviation factor: 0.06 Table 16.—Pronouns: Use vs. Case deviation factor (discrete use/case combinations): 3.76 deviation factor (grouped): 0.06 Representation of Hebrew Lexemes by Greek Lexemes (Segmentation) Translators often had a tendency to represent compound words in the source language with compound words in the target language. Thus, if a Hebrew word were composed of a preposition and a noun, for example, the Greek translator might have preferred to render the expression with a Greek preposition and noun, even though an adverb would have carried the same meaning. Here are listed all the Hebrew compounds which the Greek translators did not render all of the constituent parts. Only the consonantal text is considered, so articles indicated by pointing alone that are not represented are not listed. Ref Hebrew Compound Greek Rendering 51 3:1 52 3:2 53 3:5 54 3:6 55 3:7 56 3:8 57 58 3:9 59 3:13 60

83

61 3:14 62 3:15 63 3:17 64 3:18 65 3:20 66 3:21 Heb compounds: 43 Gk equivalents: 27

62.8% of Heb compounds rendered exactly

Word Order As noted above, since Greek is an inflected language, its writers and speakers had much greater freedom to vary word order and still maintain the thought of the sentence. Whether the translators of LXX in fact chose to take advantage of this facet of their language, or whether instead they attempted to follow closely the word order of their exemplar, is another aspect of the literalness of the translation. The list below contains the deviations from the Hebrew word order found in LXX. As already stated, Greek postpositive conjunctions that appear as near as the language permits to the equivalent Hebrew conjunction will not be considered variations from the Hebrew word order. The number of variations is calculated by ascertaining the minimum number of shifts of position a Hebrew word (including a compound word) would have to make in order to reflect the word order found in LXX. Ref Number of Variations Greek Variant 67 3:16 3 68 3:19 1 Heb semantic units: 373 variations: 4 Heb word order followed 98.9% of the time Quantitative Representation The tendency of literal translators was to render every element of their Vorlage, without adding or subtracting anything unless it was required by the target language. The following list analyzes the literalness of LXX in terms of quantitative representation, in order to determine the translator’s commitment to render all the elements of the Vorlage, and only those elements. The number of Hebrew semantic units represented by the variation is given, prefixed with a plus if the excess is in Greek and with a minus if the excess is in Hebrew. In the last column, each element of MT not found in LXX is listed in Hebrew, and each element of LXX not found in MT is listed in Greek. As noted above, add-oms will be included in these calculations, but quantitative differences arising from the representation of compound words have already been discussed and will not be included here. Furthermore, it should be noted that in verse 1 is rendered by , and the he locale in verse 19 is rendered by ; these renderings are not considered differences in

84

quantitative representation, since Greek has no exact equivalents for the Hebrew expressions. , the sign of the definite direct object, has no real equivalent in Greek and so cannot be rendered idiomatically (Aquila’s notwithstanding). Finally, articles are considered only when they appear or are implied (e.g., in 3:1 requires that an article be assumed in ) in the consonantal text (or when they would appear in the presumed Hebrew underlying an addition in LXX). In order to avoid begging the question as to which witness (if either) contains an addition to or omission from the original, the excess text is given in whichever language it appears. Ref Number of Variations Variant 69 3:1 +1 1° 70 +2 71 3:2 +1 4° 72 3:3 -1 73 -1 74 3:4 -1 75 +1 76 3:5 +1 77 +1 4° 78 3:6 +1 2° 79 -1 80 +1 2° 81 -3 82 +1 83 -2 84 +1 85 +1 6° 86 3:7 -1 2° 87 3:9 -3 88 +1 2° 89 +1 90 3:10 +1 91 -2 92 3:11 +1 93 -1 94 3:13 -2 95 +2 96 3:15 +4 97 -1

85

98 99 100 101 102 103 104 105

3:16 3:17 3:18 3:21

-2 -1 -2 +3 -2 +1 -5 +31

Heb semantic units: 438 Gk deviations: 88

1° 2°

79.9% agreement

Elimination of Variants The LXX translation of 1 Samuel is often characterized as literal or fairly literal.24 A glance at the preceding tables, however, indicates that this characterization, though perhaps true in general, does not describe certain aspects of the translation technique with precision. Whereas conjunctions are rendered with absolute consistency (i.e., exactly one Greek term per Hebrew term), verbs, nouns, and adjectives, and especially adverbs, prepositions, and particles, are not. The translators were reasonably consistent in rendering a particular class of Hebrew words (verbs/verbals, nouns/adjectives, pronouns) with the equivalent Greek class. However, the percentage of Hebrew verbals rendered by the equivalent class of verbals in Greek is low. The level of grammatical consistency is 24 Thenius, Bücher Samuels, XVII-XXII; S. R. Driver, Notes on the Books of Samuel, lix-lxii; Thackeray, Grammar, 13; Swete, Introduction, 323; Fernández Marcos, Introducción, 25. Cf. also Soisalon-Soininen, Infinitive, 171-72; Raija Sollamo, Renderings of Hebrew Semi-Prepositions in the Septuagint, Annales academiæ scientiarum fennicæ, Dissertationes humanarum litterarum, no. 19 (Helsinki: Suomalainen Tiedeakatemia, 1979), 286. De Boer disputes this claim: “We have already observed that the translators did not have literal rendering as motive for translation, but that the thread of the story was the cause for omissions, additions and differences,” de Boer, I Samuel i-xvi, 51. However, he seems to approach his analysis with a foregone conclusion. Note the following incredible conclusion: “Even though certain passages remain without a clear explanation of why G differs from M, and although some of the proposed explanations are nothing more than probabilities, the independence of the translated story, the agreement with Tg and S and elucidation as the main tendency leave us in no doubt that we in G have to do with the same Hebrew text as the one offered by M” (italics mine).

86

more complicated. The translators are absolutely consistent (deviation factor of 0.00) in rendering the person of verbs. They are very consistent (deviation factor less than 1.00) in rendering the number of verbs, the number of pronouns, the inflection of verbs, the use of nouns and adjectives, and the use of pronouns (when the Greek data is grouped appropriately). They are fairly consistent (deviation factor between 1.00 and 3.00) in rendering the number of nouns and adjectives and the gender of pronouns. They are less consistent (deviation factor between 3.00 and 6.00) in rendering verb stems. They are fairly inconsistent (deviation factor between 6.00 and 12.00) in rendering articles found with nouns and adjectives and the definiteness of nouns and adjectives. Finally, they are very inconsistent (deviation factor greater than 12.00), as expected, in rendering the gender of nouns and adjectives. Apart from the category of consistency, the following general results may be mentioned. The translators did not feel compelled to render each component of Hebrew compounds with a Greek equivalent, choosing in 40% of the cases to translate the compound as a whole. By contrast, Hebrew word order is followed very closely, with LXX deviating only 1.1% of the time, and that in only two units. Finally, an initial look at quantitative representation suggests that the Greek translators frequently varied from the Hebrew text, either adding (fifty-seven times) or omitting (thirty-one times) elements. It must be emphasized that these results are preliminary, based on a rigid application of the principles for determining translation technique discussed in Chapter 1, and assuming that MT is the same as the Vorlage of LXX. As individual variants and groups of variants are weighed, a more nuanced picture of the translation technique will emerge, and many of the above results will have to be modified. Variants Related to Consistency The first aspect of consistency to be examined is lexical consistency, of which twelve variants from MT have been identified in LXX. The identification of a lexical variant that might indicate a varying Hebrew Vorlage is based on two factors: similarity in meaning as reflected in the lexicons and Greek equivalents for the same word elsewhere in the chapter. The first lexical variant is variant 6, where LXX reads for MT . The lexicons list two roots , the first meaning “to be dim, faint, blind,” and the second meaning “to rebuke.” The second root is found only in 3:13, where it is a hapax; the present instance comes from the first root. Of the fifteen probable renderings in LXX of either the verb or the adjective , the only time LXX uses a root related to is here. (There is no single Greek root used to render I that dominates, but the words related to are used six times). and related words are often used to render words related to the Hebrew : twenty-eight of thirty-five total

87

uses of the Greek word, twenty-five of thirty uses, one of one, one of one. The close association between and suggests that some form of appeared in the Vorlage of LXX. On the other hand, the use of (rendering ) of blinded eyes in Gen 48:10 suggests that the translators might have chosen as appropriate in the present case (cf. Isa 33:15; 59:1; Zech 7:11). However, there is enough graphic similarity between and to allow the possibility of textual confusion in the transmission of the Hebrew text, or maybe at the point of translation. The reading of P supports the possibility that the confusion occurred during the transmission of the Hebrew text. Thus, it is likely that the Vorlage of LXX at this point was different from MT, so the variant is significant. In variant 9, (“to be made ready”) renders (“had gone out”). Though the context might stretch the semantic field of enough to make it roughly equivalent to , the former renders the latter only here out of seven occurrences in the OT. renders four other words, three of which mean “to restore” and are used in the context of restoring the temple. Ex 30:8 uses to render the hiphil of , referring to Aaron setting up the lamps in the evening. occurs thirteen times in the qal and ten times in the piel that are rendered in LXX. Twentyone of these occurrences are rendered by , and one by . It is likely that scribes confused an original or with 25 , and since no Hebrew verb with a like meaning appears to be similar enough orthographically to , 26 variant 9 should probably be considered an inner-Greek corruption and not significant. in variants 12 and 18 is the normal translation for in the passage (cf. 3:5, 6) and elsewhere in LXX, including the books of Kingdoms. In fact, apart from one other occurrence in several mss (3 Kgdms 6:12 [Rahlfs’s versification] mss AMN etc.), never translates , while rendering some forty times. It seems that either the translator or the Vorlage changed the verbs in these two instances to match the other occurrences in the immediate context. Since no tendency of the translators to harmonize the LXX of this chapter has been demonstrated yet, and since is an unusual translation of in the book, it is quite possible that the Vorlage itself contained rather than in these verses, so these variants are significant. Variant 15 is a variation between the Hebrew (usually rendered by ) 25 Klostermann suggests that the LXX reading is a corruption of

(Erich Klostermann, Die Bücher Samuelis und der Könige, Kurzgefaßter Kommentar zu den heiligen Schriften Alten und Neuen Testamentes sowie zu den Apokryphen, ed. Hermann Strack and Otto Böckler, vol. 3 [Nördlingen: C. H. Beck, 1887], 11). Cf. the readings of ´ ( ), ´ ( ), ´ ( ). 26 Wellhausen suggests , which he says is implied by Josephus, Ant. 3.8.3 (Bücher Samuelis, 52), but this section is not a reference to Samuel but to the Mosaic regulations concerning the lamps in the tabernacle. Cf. de Boer, I Samuel i-xvi, 62.

88

and the Greek . Though the words are of course equivalent in the sense that they refer to the God of the Israelites, the translators of 1 Samuel are fairly consistent in their renderings of both and . 27 All of the other fifteen occurrences of in the chapter are rendered by . Thus, this variant should be considered significant. Variants 20 and 37 both use a form of the verb to render the Hebrew verb , even though forms of are the more common rendering in the chapter (six of eight occurrences of are rendered by ). However, a check of Hatch and Redpath reveals that both Greek verbs are used frequently to render , and they render few other words (in the books of Kingdoms, renders forty-nine times out of fifty occurrences of ; the numbers for are thirteen of fifteen). In fact, as these figures show, is the more common of the two main renderings outside this chapter. Thus, these variants are not significant. Variants 26, 27, and 40 each reflect Greek prepositions that are different from what one might have expected based on the main renderings of the corresponding Hebrew prepositions. Of the thirteen occurrences of in the chapter, only in verse 12 is it rendered by or (one time each; the usual renderings are [six times] or nothing [five times]). Unlike the other renderings of , carries a somewhat different meaning, and one suspects that it might reflect a Hebrew rather than . 28 The semantic field of , on the other hand, overlaps that of to a large extent. However, a perusal of several passages in 1 Kingdoms (Hatch and Redpath do not give the equivalents for each occurrence) indicates that rarely renders , though it is frequently used for - , - , and - (he locale). The readings of the other versions might suggest a reading in their Vorlagen, but this preposition does not lie behind the LXX reading. It is possible that the preposition or lay before the LXX translators in this passage. could have been lost through simple haplography ( became ) and inserted by a scribe who sensed that a preposition was needed. On the other hand, the inseparable preposition might have been replaced with its near equivalent , perhaps as a result of similarity in pronunciation. Either of these two possibilities could account for the origin of as well. in verse 16 is the only instance in the chapter (of nine possibilities) where an apparent equivalent appears for . However, the whole Greek clause is different from the Hebrew: for . It is probable that rather than lies behind the in LXX. Therefore, variants 26, 27, and 40 are all significant. Variant 33 in verse 13 reflects an ancient scribal correction, one of the tiqqune 27

occurrences of

in LXX reflects in MT about 12 out of 100 times in 1 Samuel. The ratio of in LXX corresponding to in MT is even smaller.

28 So Thenius, Bücher Samuels, 17; McCarter, I Samuel, 96.

89

sopherim. In order to avoid reading that Eli’s sons cursed God ( ), lest the reader join in the blasphemy by juxtaposing these two words, scribes altered the text to read , “they cursed themselves.” LXX preserves the original reading, since obviously reflects and not . 29 The final LXX variant that might reflect a different Hebrew lexeme is number 47, which reads for . Though ( ) ( ) might be considered more literal, the phrase that appears here carries the same meaning as the Hebrew of MT. In fact, the translators of 1 Samuel often did use the expression for (especially in kaige sections), but they did not do so consistently, often substituting the more idiomatic (e.g., 1 Kgdms 11:10; 12:17; 14:36). 30 Therefore, this variant cannot be considered significant. Variant 34, (end of 3:13 LXX) for (beginning of 3:14 MT) seems to reflect a difference in word division and a slight change in spelling: instead of . 31 This variant must have arisen in Hebrew rather than Greek, so it is significant. All the other variants that reflect on the consistency of the LXX translators deal with grammatical differences. Variants 1, 2, 4, 11, 17, 19, 21, 22, and 31 all render Hebrew participles by Greek verbs (or periphrastic constructions in variants 1 and 2). In fact, of the eleven participles rendered by Greek verbs or verbals, only two (3:11, 13) are rendered by participles. When it is remembered that the LXX translators worked from a purely consonantal text, it becomes clear that all of MT’s participles that could be understood, with different pointing, as full verbs (i.e., the qal active participles that function as the main verbs in their respective clauses in MT) were in fact rendered by verbs. The piel participle in 3:13 is rendered with a Greek participle. The piel and niphal participles in 3:1 are rendered as periphrastic constructions, consisting of a form of and a participle. The only qal participle actually rendered by a participle is the one in 3:11, but the construction of the clause (the participle is part of a construct chain) mandates that the word be understood as a participle. Although the translators may have had a different reading tradition than that of the Masoretes concerning many of the participles, the consonantal text lying behind their renderings seems identical to that found in MT, so none of these variants is significant. In variants 41, 42, 44, and 45 in 3:17, the reverse situation occurs: verbs in MT are 29 So BHK and most commentators. R. Althann, “Northwest Semitic Notes on Some Texts in

1 Samuel,” Journal of Northwest Semitic Languages 12 (1984): 28-29, proposes on the basis of Ugaritic evidence that “lhm can be understood as a by-form of l m, parsed as the stative participle of the root ym, ‘to inspire awe,’ preceded by the emphatic lamedh.” However, his analysis is not convincing. 30 See the table in Sollamo, Semiprepositions, 138, and, for more discussion, ibid., 123-55. 31 Wellhausen, Bücher Samuelis, 53. S. R. Driver, Notes on the Books of Samuel, 44, gives

several other examples of the same phenomenon.

90

read as participles in LXX. Furthermore, the participles are passive, whereas the verbs in MT are both piel. Since no subject is explicitly stated (the nearest reference to God speaking is three verses earlier), the LXX translators, if they had the same consonantal text as MT, may have considered the forms pual in both instances, in spite of the infrequency of in this stem. The use of the Greek article as the equivalent of in both cases is an idiomatic Greek rendering of the construction, a construction which has no exact parallel in Hebrew. Thus, some degree of flexibility has already been shown in each phrase. Though it is conceivable that the Vorlage of the translators had participles (with the prefixed ), it seems more likely that the translators in these two cases rendered idiomatic Hebrew expressions with idiomatic Greek ones. Even if, as appears likely, the translators read the verbs as puals rather than piels, there is no indication of any variation from the consonantal MT, so the variants are not significant. On one occasion, variant 50, the translators used an adjective, , to render a Hebrew participle. Though this is the only time in the chapter in which a Hebrew participle is rendered by something other than a verb, the participle functions in the Hebrew clause as a subjective complement (predicate adjective). The niphal participle does function elsewhere as a predicate adjective.32 The Greek rendering is surely the meaning intended by the use of the niphal participle, and it may be that the translators felt that the adjective carried the meaning more faithfully than would a passive participle such as . Thus, the variant should not be seen as significant. Three times in chapter 3 the translators use infinitives to render what are full verbs in MT, namely, variants 10, 14, and 16. In each case, the Hebrew verb follows ; two of the verbs are imperfect (3:3, 7), and one is perfect (3:7).33 The Greek equivalent for is (see above, pp. 39-40), and though LXX has only two occurrences of this preposition, all three Greek infinitives are governed by . Since the use of the infinitive rather than some form of the full verb is required when means “before,”34 these variants cannot be taken to be significant. Variants 13, 28, and 29 have Greek verbs for Hebrew infinitives. In variant 13, LXX reads for . Both the infinitive and the imperfect (usually with waw consecutive) can follow in Hebrew when the meaning is “to do again,” though the infinitive is more common. Idiomatic Greek does not 32 Waltke and O’Connor, Introduction, 619-20. 33 Kautzsch, ed., Gesenius’ Hebrew Grammar, § 107c, says that the perfect should be emended to

an imperfect, since the imperfect regularly follows and since the perfect is coordinated with the imperfect ; so also BHK, most of the commentaries. Of course, the imperfect requires no change in the consonantal text. 34 Smyth, Grammar, 549, 553-55. Soisalon-Soininen, Infinitive, 131, notes that the expression ( ) + full verb is usually rendered by either + infinitive or + infinitive throughout LXX.

91

have a similar construction, and the reading of LXX here is definitely unidiomatic. This consideration alone implies that the translators are following a Hebrew text rather than changing the Hebrew construction, especially since it is the less common Hebrew construction that is reflected in LXX. Thus, variant 13 is significant. Variants 28 and 29 also have Greek verbs for Hebrew infinitives. The two infinitives form an expression that, though idiomatic and fairly common in Hebrew, has no exact parallel in Greek.35 The translators managed to create a reading in that represented both infinitives idiomatically, deriving the person of these verbs from . In view of the nature of the Hebrew construction, the LXX reading seems to effectively capture the flavor of the Hebrew, despite avoiding the normal grammatical parallels, so these two variants are not significant. Variant 7 reads the infinitive for the adjective . Greek idiom certainly requires the use of an infinitive after , but Hebrew idiom would seem to require it, too. It is probable that the letters were meant to stand for the infinitive . 36 However, since Greek idiom requires an infinitive, and since such a change in Hebrew would not change the consonantal text, the variant is not significant. In the case of variant 39, a Greek adverb, , is used to render the Hebrew noun and article . Although none of the other fifty-four Hebrew nouns in the chapter is rendered by an adverb, here functions as an adverbial of time, and as such, is an appropriate translation. In fact, is the regular equivalent for in LXX, so this variant is not significant. The next three variants deal with the rendering of the inflection of Hebrew verbs in a manner contrary to the most common representation. First, variant 8 uses the Greek imperfect (a past tense) to render the Hebrew imperfect (usually rendered by the present or the future tense). The context clearly shows that an event in the past in being described, so one would usually expect a past tense. However, the Hebrew imperfect does not correspond completely with the Greek present or future tenses. In the case at hand, the 35 Though is classified as an infinitive construct by BDB and Holladay, it seems more likely that it should be seen with